Kidnappings by terrorist groups are on the rise globally. The chances of surviving this ordeal vary greatly according to the nationality of the victim. This is particularly true for aid workers, one of the groups worst hit by this plague. Two journalists - an Italian and an Afghan - were abducted in Helmand, Afghanistan ten years ago: the kidnapping ended with the former alive and the latter beheaded. But it is not as simple as in rich vs poor citizenship, as the fate of American and British victims of kidnapping shows. Let's look into numbers.
Ten years ago these days the Italian journalist and war correspondent Daniele Mastrogiacomo, 53, together with the Afghan journalist Ajmal Naqshbandi, 26, working as his fixer, and the driver Seyed Agha, 24, were kidnapped by the Taliban loyal to Mullah Dadullah near Lashkargah, the main city of Helmand, a province in southern Afghanistan. Seyed Agha was brutally murdered by his captors a few days later: his beheading was captured on video while Mastrogiacomo and Naqshbandi were forced to kneel nearby blindfolded. The Afghan government led by president Hamid Karzai yielded to pressures from Rome and accepted to free five Taleban prisoners in exchange for the release of Mastrogiacomo and Naqshbandi. The prisoners were freed and put on two different cars. A few days later Mastrogiacomo was cheered by authorities, friends and media as he returned safely to Rome. Naqshbandi’s family could not do the same: the Taliban never released him, they brought him back to the place in the desert on the border with Baluchistan where he had been detained together with Mastrogiacomo and threatened to kill him if the Afghan government had not released another two Taliban prisoners. The Karzai administration refused. The Taliban murdered Naqshbandi. Many Afghan journalists and sectors of the public were outraged at the differential treatment and outcome of the kidnapping, blaming Karzai of caring more for a foreigner than an Afghan citizen.
This difference in the outcome of a kidnapping based on the nationality of the victims is far from exceptional: on the contrary, it is the rule. And it does not necessarily follow the same pattern where the prisoner from a rich Western country is released while his or her companion from a poor country is executed.
While kidnapping for ransom is not a new phenomenon, traditionally it is in many contexts a criminal activity with a prominent domestic dimension. The kidnapping of foreigners for political as well as financial reasons has increased significantly in the past two decades. A recent study published by The New America foundation reckons that since 2001 at least 1.185 “westerners” from 32 countries have been taken hostage overseas by “terrorist, militant, and pirate groups.” 90 of them were murdered while in captivity. This translates into a 7.6% crude casualty rate, i.e. without attempting a statistical estimation of what is likely to be so-called fat-tailed phenomenon (whose frequency distribution is represented by a power law rather than Gaussian exponential) and hence where non observed events in the tail have a high impact on the estimated mean. Even with these caveats, the rough figure is still quite high.
Another study on kidnapping by Loertscher and Milton (2015) defines “westerner” as a citizen of any OECD country (as opposed to the more idiosyncratic and restrictive definition adopted by The New America Foundation study that for example does not include Turkey) and is concerned with kidnapping by “non-state actors” in general, including “jihadists, pirates, militant, tribal” groups, hence there is a good overlap in the categories of perpetrators. With a wider definition of who is a “westerner”, Loertscher and Milton recorded a total of 1.483 kidnapping victims, of which 110 were eventually executed by their captors, 11 died while in captivity and 12 died during a rescue operation for a total crude death rate of 8.96%.
Kidnapping outcomes by nationality
Looking at the breakdown by country, more than 60% of the victims are citizens of just six countries (Turkey, United States, Italy, United Kingdom, France, and Germany). Combining data from both studies there are significant differences in the outcome of a kidnapping depending on the nationality of the victim.
|Kidnapped||Executed or died in captivity or during rescue operation||Death rate [%]|
* The New America Foundation database (2016)
§ Loertscher and Milton (2015)
The calculation of death rates does not include "unknown" outcomes, the cases where information is missing or insufficient.
Considering the much lower chances of and American or British or French captive of surviving a kidnapping, one is tempted to link this difference of outcome to the official policy of “not dealing with terrorists” stated and reiterated at least by several US and UK governments (the French government seems to have shifted recently to a “softer” approach). Where the hardline policy towards terrorist kidnappers seem not to work is in the total numbers of kidnappings, but there are many factors to be considered and not just a simplistic cost / benefit analysis from the terrorists side.
Loertscher and Milton comment about the different execution rates (their figures refer to executions only, not including other causes of death during captivity):
With an execution rate of 15%, U.S. citizens are five times more likely to be executed by non-state actor group types other than jihadists than the citizens of all other countries, which have a combined execution rate of 3%. Germany and the United Kingdom have the next highest rates of execution (7% and 6% respectively). France (3%), Italy (3%), and Turkey (2%) all have comparably low execution rates when kidnapped by all other non-state actors. [p. 39]
As noticeable as the relationship between nationality and execution rates is, the correlation between release rates and nationality is perhaps equally remarkable. Turkey and Italy have rates of release that dramatically exceed the average release rate of jihadist groups (85% and 81%, respectively, compared to 63%) while their execution rates (13% and 7%) are well below the average execution rate of 15% for jihadist groups. French release rates (69%) also track above the jihadist average, although they have an execution rate just above the jihadist average. Even though the German release rate (50%) does not exceed the jihadist average, they have had no citizens executed by jihadist groups in our dataset. The U.S. and UK release rates (18% and 24%) are drastically below the jihadist average; in fact, U.S. citizens are almost four times less likely to be released when the combined release rate of all other nations is considered (68%). [p 40]
Kidnapping of aid workers
Looking at a breakdown of victims from the occupational point of view, both sources agree that aid workers are, together with journalists and sailors (due to the spike in piracy in the Indian ocean), among the hardest hit groups. According to Loertscher and Milton:
After the removal of outliers, most (53%) jihadist kidnappings are confined to either NGOs or journalists (NGOs, 34.5%; journalists, 18%).
Journalists are more likely to be kidnapped in open conflict zones (AFPAK, Iraq, Syria). NGOs are abducted almost equally between conflict and non-conflict zones.
That kidnapping in general is increasingly the tactic of choice in attacks against aid workers is one of the grim conclusions of the recent reports published by The Aid Worker Security project, an initiative funded by USAID that also relies on publicly available information to compile its Aid Worker Security Database (AWSD). When looking at kidnappings, this source offers an interesting comparison because it includes not only attacks against “westerners” but all aid workers, with data breakdown between national staff and expatriate staff.
All three databases rely on openly accessible public sources. Although they follow different methodologies in the categorisation of their data, there is some room for comparison. The major difference perhaps is that the perpetrators do not overlap completely. It is possible that part of the kidnappings of aid workers have been carried out by criminal groups that do not necessarily fit in the “terrorist, militant and pirate” categories.
There is a minor inconsistency in the database where some security incidents categorised under the “kidnapped but not killed” rubric as regards the “means of attack” do actually report also some killed staff. We will therefore take both the “kidnapped but not killed” (K) and “kidnapped – killed” (KK) “means of attack” in the following breakdown and reporting the figures of staff killed from both.
|Kidnapped NGO workers||Eventually killed||Released or rescued||of which, wounded||Death rate [%]|
We can compare the total of 265–309 kidnapped expat staff (of all nationalities; the range is due to one coding ambiguity in the database) with the 247 “western” NGO workers kidnapped by non-state actors tallied by Loertscher and Milton in the same period and the two figures seem largely compatible, considering that while not all NGO expats hail from OECD countries, a large majority of them probably does. But when we focus on the fate of those kidnapped the AWSD departs significantly from the L&M database whereby of the 247 “western” aid workers kidnapped, four died during a rescue operation, 5 died in captivity and 18 were executed. Overall, over one in ten (10.9%) did not survive the kidnapping, while the AWSD records 15 killed as a consequence of the kidnapping. But not all expat aid workers are “westerners” and not all kidnappers are “terrorist, militant and pirate”, and the numbers involved are such that a reporting error or difference in interpreting media sources can sway the totals in a significant way and thereby we should refrain from trying to read too much in these figures.
Again with the caveats that we are calculating crude death rates rather than statistical estimators of risk, we can still retain a main observation that is internal to the AWSD dataset, and hence not conditioned by the imperfect overlap with the New America Foundation or Loertscher and Milton’s datasets. When the national staff of NGOs are kidnapped their chances of being killed are almost double those of the expatriate staff.
This does not hold for American or British expatriate aid workers though, whose death rates after a kidnapping are extremely high (28 and 31% respectively). The general finding is averaged across all organisations, so we cannot conclude that this grim difference in outcome holds for the national and international staff of the same aid organisation. But it is still a troubling one in general, and in particular for the NGO and international aid sector.
Around two thirds of all the kidnappings of aid workers (701 of 1168) took place in only three countries: Afghanistan, Somalia, Sudan and South Sudan. Somalia is more dangerous for the kidnapped national staff (death rate of 11.5%) than Sudan (7.8%) or Afghanistan (9.5%).
In order to understand the higher death rate for kidnapped national staff compared with expat aid workers let’s have a look at different kind of aid organisations. The death rate for kidnapped national staff working for local NGOs or for the National Red Crescent Societies is the highest (13.6%), but hardly any expat aid worker among their employees was kidnapped, in fact they probably employ very few of them. The International NGOs are the prime target of kidnapping attacks, with 559 victims, of which 378 national staff and 181 expatriates. The death rate for kidnapped national staff working for INGOs is again higher (10.85%) than the expat staff (7.7%).
The United Nations organisations have had their fair share of kidnappings with a total of 204 (of which 164 national staff, 40 expatriates). However, the outcomes are less deadly for the national staff (5.5%), with no reported deaths among the expatriate UN staff that suffered a kidnapping.
So it seems that it is the national staff working for local NGOs or NRCs that in the past had the highest chance of being killed after being kidnapped. International NGOs are clearly the most targeted by kidnappers. This could be because of the sheer numbers of the staff they employ or because of their foreignness and resources, or both. The national staff employed by INGOs had slightly better chances to survive a kidnapping than their colleagues working for local NGOs, but only marginally so, and significantly worse than the expatriates working for INGOs. This difference in the survival rate of NGO national vs international staff is problematic. But before pointing fingers at how INGOs should care for and protect their national staff, the tragic fate of American and British aid workers (more than a quarter of them did not survive their kidnappings) should give sufficient reasons for pause and reflection.
Cover photo: still image of Ajmal Naqshbandi in the propaganda video taken by his Taliban captors, from the excellent documentary Fixer: the taking of Ajmal Naqshbandi
This post has been updated since its initial publication in order to fix some typos and add kidnapping data for Spain.
Christopher Mellon, Peter Bergen and David Sterman. To Pay Ransom or Not to Pay Ransom? An Examination of Western Hostage Policies. January 8, 2017. Reference link ↩︎
Loertscher, Seth, and Daniel Milton. 2015. Held Hostage: Analyses of Kidnapping Across Time and Among Jihadist Organizations. The Combating Terrorism Center at West Point, The United States Military Academy. Reference link (accessed: February 8, 2017). See in particular Section 2, “Kidnapping of Westerners by Non-State Actors (2001–2015)”, pp.14–32. Database ↩︎
Humanitarian Outcomes. The Aid Worker Security Database (AWSD) https://aidworkersecurity.org ↩︎