Regime Age and Terrorism: Are New Democracies Prone to Terrorism?

Abstract

One of the earlier empirical studies of the relationship between regime type and terrorism published in International Interactions determined that while established democracies were significantly less likely to experience terrorist attacks than were nondemocratic countries, newly established democracies were highly vulnerable to terrorism. Subsequent empirical studies have routinely controlled for both regime type and age, but scholarly understanding of the effect of regime longevity on terrorism remains underdeveloped. This study revisits the relationship between terrorism and regime type and regime age using updated data, analytical techniques, and time-series and finds that while young democracies experience more terrorism than older democracies, dictatorships of any age experience less terrorism than any other type of regime.

Keywords:

• democracy
• dictatorship
• regime age
• regime type
• terrorism

Acknowledgments

The author would like to thank Erica Chenoweth, Todd Sandler, Burcu Savun, James Forest, and the three anonymous reviewers for their comments and suggestions about this piece.

Notes

¹For a good summary of the relative strengths and weaknesses of ITERATE and GTD, see Enders, Sandler, and Gaibulloev (2011).

²Source: RDWTI. Data http://www.rand.org/nsrd/projects/terrorism-incidents.html. However, average annual total counts of terrorism—combined domestic and transnational—did increase between the 1970 and 1986 and 1987 and 2010 period from 1,863.6 to 2,969.3 incidents according to GTD data. Data http://www.start.umd.edu/gtd/.

³Results available from author.

⁴In these exploratory models, I coded “Anocracies” or “Hybrid Dictatorships”—those that possess and engage in both democratic and nondemocratic political institutions and practices—as those observations where the Polity IV was between 5 and −5, and “Dictatorships” as those observations where Polity IV was −5 and lower. This is a standard coding practice (see Fearon and Latin 2003; Marshall and Gurr 2003). Results available from the author.

⁵Results available from author.

⁶One-year lag of Polity durable score is a negative predictor of change of regime, the zero score for durable (coeff. = −.092; p > |z| = .000).

⁷One-year lag of Polity durable score is a negative predictor of combined score for MEPV—Major Episodes of Political Violence (Marshall 2010)—civil violence and civil war intensity (coeff. = −.070; p > |z| = .000).

⁸One-year lag of Polity durable score is a negative predictor of Aggregate State Failure score—Political Instability Task Force (Marshall et al. 2009) (coeff. = −.025; p > |z| = .000).

⁹The group of countries with low average Polity durable scores during the period also includes the countries of the former Soviet Union, which became independent in 1991. Concerned that these countries might affect the results of the study—several of them are particularly terrorism-prone, such as Armenia, Azerbaijan, Russia, and Tajikistan—I ran two sets of robustness models, including dummies coded 1 for all former Soviet republics and coded 1 for post-1991 years. Inclusion of these controls do not change the core results.

¹⁰This design produces more and more evenly grouped regime-age cohorts to analyze than those in the main analysis, providing a further test of the relationships.

¹¹Results available from author.

¹²Clearly, all of the regime qualities tested in Table 6 are also predictors of one another (they are significant correlates and predict one another at the p > |z| = .000 level), and it is also possible that regime age has interaction with the other factors that affect a country's susceptibility to terrorism in important ways. These tests merely determine that regime age itself is a crucial and substantive predictor of terrorism in democracies.