State Instability and Terrorism

Abstract

I explore the relationship between political instability and terrorism in this dissertation, using the Global Terrorism Database (GTD), which contains both domestic and transnational terrorism. I use the Political Instability Task Force data to measure political instability. Breakdown theory suggests that the occurrence of political instability should increase levels of terrorism within a state, because when a rapid social change, such as political instability, occurs, there is a severing of social bonds that tie individuals to society. The effects of the disruption in controls should be to increase levels of non-routine collective action, of which terrorism is a form (Durkheim, 1930 [1951]; Useem, 1998). In addition, different types of instability ought to invite different levels of terrorism based on the degree of disruption to the societal controls. There are four types of political instability: ethnic war, revolutionary war, genocide and adverse regime change. Further, I extrapolate two theoretical extensions from the breakdown model. The first extension is that more instability episodes should produce more terrorism within a state. The second extension is that when two or more instability episodes occur within a year, this increased temporal density should produce more terrorism than when one instability episode occurs within a year.

I test the theoretical framework using the negative binomial regression model with country and time fixed effects. The first model contains control variables that measure country demographics, governance and contiguity to an unstable nation with 147 states from 1970-2005. The second model examines the effects of control variables that measure the population age structure and social and economic development in a smaller sample of 116 states and years from 1981-2005. The third model adds ethnic minority group characteristics from the Minorities at Risk (MAR) dataset and contains 82 countries from 1990-2005. This three-sample strategy allows me to speak to the omitted variable and sample selection biases that may impact the results.

Empirical results demonstrate that political instability is an important predictor of terrorism incidents. The breakdown model itself is supported, but the extensions are not supported. Further research is needed to delineate the boundaries of the instability-terrorism relationship.