| dc.description.abstract | Overseas kidnapping of citizens by nonstate actors is a significant problem for Western countries, a major source of resources for terrorist and criminal organizations, and a traumatic or life-ending event for victims and their families. In this work, I examine what is known about these modern-era kidnappings and their historical parallels, the motivations of their perpetrators, and the utility of hostage rescue operations as a state response, generating causal hypotheses based upon theories of deterrence and social learning. I then apply difference-in-difference and regression discontinuity experimental designs to a dataset of nearly 1,500 hostages constituting more than 600 kidnapping incidents between 2001 and 2015 in order to test these hypotheses.
I find that on average, a rescue operation has a deterrent effect lasting up to 210 days within the same country and subregion. I estimate this effect at .51 potential kidnappings by the same captor group, .52 kidnappings of victims of the same nationalities, and .73 kidnappings in total prevented within 90 days in the same country, equivalent to a 29% decline in kidnappings by the captor group and a 39% reduction in kidnappings of the same nationalities. The deterrent effect is stronger on average if the captor group has material motivations, if the operation has destructive effects or is surprising, or if the group has more alternatives to kidnapping, but weaker if the rescue operation is interpreted as a continuation of ongoing war in country. This effect remains about the same whether an operation is successful or unsuccessful. I also find that on average, a successful ransom payment has an incentive effect on the recipient group, generating an estimated 3.00 additional kidnappings in the subregion by the same captor group over the following 730 days, an increase of 85% over its prior kidnapping rate. This incentive effect appears to be distributed across all nationalities and not highly concentrated on the nationalities ransomed.
I further examine predictors of captor group, hostage, incident, and outcome attributes. I find that given an inconclusive government response, highly significant correlates of hostage death include fewer hostages kidnapped, a hostage who is transferred between captor groups or is a national of a country at war locally, and a captor group which has political motivations or is known to claim responsibility by internet or video. Private ransom payments are associated with a greater number of hostages abducted together, a non-politically-motivated group, a country of kidnapping with poor control of corruption, and a long period of captivity. A longer duration of captivity is associated with hostages who have for-profit occupations, a more potent captor group, and a country of kidnapping that is mountainous or has low unemployment. Additionally, there is limited evidence that survival rates for hostages in separate incidents within the same country improve in the immediate aftermath of rescue operations and deteriorate after ransoms and hostage murders.
I also build a model of government response which incorporates empirical results and find that a government’s assessment of the harm caused by a ransom relative to hostage death is the best indicator of its optimal response. Additionally, when deterrent and incentive effects and the costs of time are considered, the conditions under which it prefers an inconclusive response over conducting a rescue operation or facilitating a ransom become extremely limited. The empirically-tuned model may itself be a useful policy tool. I conclude by developing a “combustion” theory of kidnapping waves, potentially transferable to terrorist or insurgent violence more broadly. In aggregate, my findings are highly supportive of a no-concessions policy toward kidnappings, similar to current US and UK postures, and greater investment in rescue capability. | |
