Domestic versus transnational terrorism: Data, decomposition, and dynamics

This article devises a method to separate the Global Terrorism Database (GTD) into transnational and domestic terrorist incidents. This decomposition is essential for the understanding of some terrorism phenomena when the two types of terrorism are hypothesized to have different impacts. For example, transnational terrorism may have a greater adverse effect than domestic terrorism on economic growth. Moreover, the causes of the two types of terrorism may differ. Once the data are separated, we apply a calibration method to address some issues with GTD data - namely, the missing data for 1993 and different coding procedures used before 1998. In particular, we calibrate the GTD transnational terrorist incidents to ITERATE transnational terrorist incidents to address GTD's undercounting of incidents in much of the 1970s and its overcounting of incidents in much of the 1990s. Given our assumption that analogous errors characterize domestic terrorist events in GTD, we apply the same calibrations to adjust GTD domestic incidents. The second part of the article investigates the dynamic aspects of GTD domestic and transnational terrorist incidents, based on the calibrated data. Contemporaneous and lagged cross-correlations for the two types of terrorist incidents are computed for component time series involving casualties, deaths, assassinations, bombings, and armed attacks. We find a large cross-correlation between domestic and transnational terrorist incidents that persists over a number of periods. A key finding is that shocks to domestic terrorism result in persistent effects on transnational terrorism; however, the reverse is not true. This finding suggests that domestic terrorism can spill over to transnational terrorism, so that prime-target countries cannot ignore domestic terrorism abroad and may need to assist in curbing this homegrown terrorism.