The deliberate targeting of aid workers has increased in recent years, encouraging analysts to question the cause of these events. The 2014 Aid Worker Security Report reveals that 251 incidents of major violence against aid workers occurred globally in 2013, resulting in 460 workers being killed, kidnapped, or seriously wounded.1 Afghanistan alone was home to 81 attacks, the highest number of attacks in 2013. It was followed by Syria (43), South Sudan (35), Pakistan (17), and Sudan (16) (Stoddard et al. 2014).

As intentional violence is now the leading cause of death among aid workers worldwide, analysts have attempted to decipher why relief groups—which largely claim to adhere to the core humanitarian principles of independence, impartiality, and neutrality—have become the deliberate target of combatants. Although some observers attribute these acts to part of a broader military strategy employed by recalcitrant insurgencies (Hammond 2008), others within the aid community believe that a 'blurred line' between military engagement and humanitarian work has contributed to the violence. For example, certain nongovernmental organizations (NGOs) have specifically attributed decreased aid worker security to Provincial Reconstruction Teams (PRTs) operating in Afghanistan. PRTs were small military units (100–250 troops) with a civilian component that served three broad functions: reinforcement of the authority and legitimacy of the Afghan government at the provincial and district levels; development of national, provincial, district, and local governance and infrastructure; and establishment of economic and social stability and security for the people (SIGAR 2010: 75).2

Those who attribute PRT presence with decreased aid worker security believe that military involvement in humanitarian activities ‘blurred the line’ between soldiers and civilians operating in the same realm. Therefore, insurgent actors no longer distinguish between combatants and civilians. As Fast (2010: 381) has noted, although the reasoning that militarization of aid increases incidents is compelling, ‘the evidentiary basis is not, at present, equally compelling even though anecdotal evidence does exist.’

In accordance with the Enduring Strategic Partnership Agreement, PRTs began to close in the summer of 2012 and were phased out by the end of 2014. Now that one of the most significant 21st century experimental undertakings by military forces is at an end, it is important to analyze the successes and difficulties encountered along the way so that future endeavors are best equipped for success. With that in mind, this study hopes to offer a small contribution to the discussions regarding humanitarian insecurity generally and PRT-NGO relations specifically.

This study uses panel-corrected standard errors of all 34 provinces in Afghanistan in 2010 and 2011 to test the hypothesis that military involvement in humanitarian activity has contributed to an increase in violence against aid workers. The article is divided into four sections. Part I provides a brief historical background of the PRT initiative in Afghanistan and a review of the existing literature. Part II outlines the research design, followed by empirical analysis in Part III. Part IV provides a few concluding remarks and policy implications.

A Brief Background of PRTs in Afghanistan

Prior to the development of PRTs, Coalition Humanitarian Liaison Cells (CHLCs) and Afghanistan Civil Affairs Teams (CAT-As) supported humanitarian assistance, relief, and reconstruction (Dziedzic and Seidl 2005: 3). CHLCs were staffed by a dozen Civil Affairs (CA) soldiers in small outposts referred to as ‘Chiclets.’ These units were tasked with small-scale reconstruction projects and the assessment of humanitarian needs. They also established relations with the UN Assistance Mission in Afghanistan (UNAMA) and NGOs already in the field (Perito 2005: 2).

In November 2002 this concept was further developed in Washington by the Department of Defense (DoD), Department of State (DoS), and the US Agency for International Development (USAID). New Joint Regional Teams (JRTs) were proposed to conduct village assessments, identify and coordinate reconstruction projects, and liaise with regional commanders (CHC 2003). The interim president at that time of the Afghan Transitional Authority, Hamid Karzai, enthusiastically supported the new concept, but requested the name be changed (Dziedzic and Seidl 2005: 3). The name was initially altered to Provisional Reconstruction Teams and later, following the pilot phase, to Provincial Reconstruction Teams.

PRTs combined diplomatic, military, and development components in an effort to ‘improve stability by building up the capacity of the host nation to govern; enhance economic viability; and deliver essential public services such as security, law and order, justice, health care, and education’ (CALL 2011: 2). Teams did not engage in combat operations, but the integration of military and humanitarian work enabled civilian representatives to conduct activity with force-protection capabilities in hostile regions. All PRTs eventually fell under the operational command of the International Security Assistance Force (ISAF), while individual nations led the teams. The official PRT mission statement as incorporated into the ISAF Operational Plan states that PRTs ‘will assist The Islamic Republic of Afghanistan to extend its authority, in order to facilitate the development of a stable and secure environment in the identified area of operations, and enable Security Sector Reform (SSR) and reconstruction efforts’ (ISAF 2009: 3).

The first PRT was established in the city of Gardez in Paktia province in November 2002 and became fully operational in February 2003. By the end of 2003, seven more teams were established in Kunduz, Bamyan, Balkh, Parwan, Herat, Nangarhar, and Kandahar. Initially, US-led PRTs operated in the volatile southern and eastern regions while ISAF coalition-led teams remained in the relatively stable northern and western regions. As the PRT network continued to expand, eleven additional teams were added in 2004. There was an increased number of US-led teams in the south while ISAF deployed from the north to the west in a counterclockwise motion, eventually extending to the south and east. By October 2006, all teams came under ISAF command (Jalali 2007: 36). ISAF commanded 26 PRTs with 14 separate lead nations; the United States led 12 of these teams.

US PRTs were structured so that the DoD was assigned responsibility for improving security while also providing logistical support and force protection for PRT members. USAID was tasked with leading reconstruction while the DoS was responsible for political oversight, coordination, and reporting. Examples of PRT activity included political development (e.g. election support), governance (e.g. development of effective legislative oversight capabilities), education (e.g. refurbishment of schools and provision of materials), health (e.g. medical training), human rights, economic development (e.g. infrastructure projects, banking, and enhancement of agricultural capacity), and rule of law (e.g. police training, judicial reform, and development of law school curricula and standards) (CALL 2011: 41, 48–50).

Guidance for PRT activity came from two sources: ISAF and the member nations leading the teams. Although ISAF was the primary source of guidance, individual nations provided the funding for PRTs. The political apparatus of each country often influenced how each team operated, ‘resulting in different types of operations being conducted from PRT to PRT.’ Additionally, staff was subject to a high turnover rate; civilians often served for twelve months while military members served for nine. As a result of frequent personnel changes, PRTs would regularly shift their direction, objectives, and programs based on the expertise of individuals in charge (CALL 2011: 15, 28).

Following President Obama’s June 2011 announcement that the United States would remove 10,000 troops from the country by the end of 2011 and 23,000 additional troops by summer 2012, ISAF and the Afghan Transition Coordination Commission established a plan to guide the transition. As part of this process, the Afghan government asked the international community to begin phasing out PRTs to ensure Afghan institutions would develop the capacity to manage development projects. In May 2012, President Obama and President Karzai signed the Enduring Strategic Partnership, outlining the US-Afghanistan relationship during transition and beyond. As part of the agreement, the United States committed to eliminate PRTs throughout the country by the end of 2014. Although all 26 PRTs remained at the start of 2012, the phase-out began that summer with the closure of the first US-operated team. At the end of the transition, the functions of all PRTs would be handed over to the Afghan government, development agencies, NGOs, or the private sector. In January 2013, 22 PRTs were still operational, but as the phase-out continued only 13 remained by that summer (SIGAR 2011: 30; SIGAR 2012a: 4–5; SIGAR 2012b: 8; SIGAR 2013: 126). The majority of PRTs were eliminated by mid-2014.

The civilian humanitarian community has a different approach to aid facilitation that unsurprisingly conflicted at times with the PRT concept. Although PRTs engaged in activities similar to those of their NGO counterparts, their mission was tied into the overall counter insurgency strategy of the coalition. NGOs are independent, non-profit civilian organizations that conduct a wide array of humanitarian missions. There are both international and national NGOs with different areas of expertise, resources, structures, governing bodies, approaches, and areas of operations; they can be secular or ‘faith-based.’ Though each is unique, many of their goals and objectives are similar. However, there is a divide in the field between organizations that strictly adhere to a politically impartial distribution and those that embrace political advocacy and structural reform. Whereas groups in the former camp focus their efforts strictly on providing relief, those in the latter engage in such activities as post-conflict reconstruction, human rights advocacy, democracy promotion, conflict resolution, and peacebuilding.3 All official organizations are registered under the Ministry of Economy and the Government of Afghanistan’s Law on NGOs (ISAF 2009: 229).

In 2005, four main NGO coordinating bodies (ACBAR, ANCM, AWN, and SWABAC) developed a code of conduct for NGOs engaged in humanitarian action, reconstruction, and development in Afghanistan. The mission statement outlined in the code was defined as follows:

Our general mission as NGOs operating in Afghanistan is to address humanitarian, reconstruction and sustainable development needs in Afghanistan, with a special focus on the rights of those who are disadvantaged and vulnerable. We work in partnership with each other, the government, donors, and communities. (ACBAR 2005: 3)

The purpose of the code was to promote improved understanding of NGOs; transparency, accountability, and good management practices; and improved quality of services. The code also highlighted the independence, impartiality, and neutrality of organizations. Considering their adherence to this creed, it is unsurprising that many NGOs were reluctant to interact with PRTs whose primary objective through relief work was the suppression of insurgency. According to a 2004 report by Save the Children, ‘the divergence of views relates not so much to the type of activity considered to be humanitarian assistance, but more to the process of delivery and motivations behind it’ (McHugh and Gostelow 2004: 1). A 2009 report compiled by eleven NGOs operating in Afghanistan at the time stated, ‘In the same way that NGOs are not expected to take the lead in the security sector, predominantly military institutions should not be expected, or presume, to take a leading role in local development or governance’ (Waldman 2009: 13).

Many within the aid community have suggested that the decline in security for NGOs is the result of a blurring of military operations and humanitarianism. The 2009 NGO report claims that the global increase in violence against aid workers can be attributed, in part, to military engagement in aid activity, which has ‘blurred the line between military and humanitarian actors’ and thus ‘adversely affected NGO security [and] endangered the lives of NGO workers.’ The report goes on to declare, ‘[W]e do not believe that PRT engagement in development activities is effective or sustainable, and strongly oppose the expansion of such activities or the establishment of new PRTs’ (Waldman 2009: 16).

The eleven signatory NGOs to the report are not alone; additional organizations have echoed similar sentiments. A 2003 policy brief by CARE International argues that extensive coalition engagement in reconstruction may put aid workers at risk and should therefore be left to the Afghan government, UN, and other civilian aid agencies. The brief recommended that coalition forces ‘should take all necessary steps to ensure that communities, policy makers and the general public do not confuse military- and civilian- implemented assistance’ (CARE 2003: 6). Additionally, InterAction—an alliance organization of over 180 NGOs—has accused the US military of ‘increasing risks to aid workers and undermining sustainable development efforts’ through the provision of humanitarian assistance generally and PRT activity specifically (InterAction 2013).

However, the ‘blurred lines’ explanation for decreasing aid worker security is not unanimously accepted. For example, Hammond (2008) believes that the attacks are deliberate and ‘intended to demonstrate the might of the attacker, the weakness of the victim, and the inability of the opposing force to prevent such attacks.’ Thus, the deliberate targeting of aid workers is part of a broader military strategy and ‘should be seen not as the result of extreme confusion but as deeply conscious acts’ (Hammond 2008: 175, 177).

Other observers believe that humanitarian activity is inherently political in nature (Barnett 2011). This view holds that, regardless of whether aid workers publicly claim to adhere to the core principles of independence, impartiality, and neutrality, their work seeks to alter existing structures which can influence local perceptions. In recent decades, many humanitarian groups have shifted toward human rights advocacy, peacebuilding, and post-conflict reconstruction to include gender equality, religious rights, legal reform, and election monitoring among a host of other politically-motivated, democracy-promoting endeavors. Organizations no longer solely offer succor to those in need; they also seek to restructure environments to address the root causes of suffering and instability. As insurgents are not likely to embrace the alteration of existing norms in favor of Western democratic values, NGOs could face increased vulnerability due to the very nature of their activity.

Nonetheless, the dispute between the two communities was highlighted in June 2004 when Doctors without Borders withdrew from Afghanistan, claiming that PRT operations in Badghis province contributed to the murder of five of its staff. A spokesman for the Taliban took responsibility for the attacks, stating, ‘We killed them because they worked for the Americans against us using the cover of aid work. We will kill more foreign aid workers.’ McHugh and Gostelow of Save the Children believe this quote ‘highlights the readiness of insurgents and disaffected groups within the population to associate humanitarian aid workers with the military’ (McHugh and Gostelow 2004: 34–35).

One of the most notable instances of controversy between the military and NGOs concerned uniforms and clothing during the preliminary phase of the PRT initiative. To minimize the negative psychological impact of his military unit’s interaction with the civilian population, one commander allowed his troops to ‘wear baseball caps, grow beards, and mix with local people in the market’ (Perito 2005: 8). NGOs expressed concern that their personal safety was at risk, as the populace were unable to differentiate between civilian aid workers and military personnel.

Although events like this strained relations with NGOs, the military showed initiative to alter approaches in an effort to build trust with the civilian aid community. In 2004, UNAMA established a civil-military working group to address ongoing concerns. Following the ‘civilian clothing’ backlash, military personnel ceased the practice after addressing the grievances expressed by the NGO community. PRTs also altered their approach toward NGOs by no longer visiting unannounced in full ‘battle-rattle.’ The PRT handbook states that ‘[i]nitial contact with NGOs is best done through a non-military agency (NGO coordinating body, UNAMA or relevant IRA department). It is requested that armed PRT personnel, including civilians with a military escort, do not visit an NGO office without first obtaining permission from the NGO’ (ISAF 2009: 240–241). Additional guidelines in the handbook outline how to properly interact with NGOs, security protocols, and weapons policies.

It should be noted that NGOs may have also helped contribute to ‘blurred lines’ early in the conflict, as they often drove the same vehicles and used the same equipment as the military. Moreover, their bases of operations were frequently located near the military. According to sources with deployed experience who were interviewed by Watts (2004), these factors made it difficult for locals to distinguish between actors (Watts 2004: 11).

PRTs also often operated in violent areas that NGOs tended to avoid. For example, when the Taliban killed two German civilians in Paktia in 2003, most NGOs fled the province for over a year. Similar instances occurred in Kandahar and Helmand in 2005; however, PRTs remained to carry out reconstruction in the provinces (Malkasian and Meyerle 2009: 30).4

Prior Studies

A rather limited number of studies have been conducted regarding NGO security. This is due to multiple factors, one of which is that the movement to target civilians is relatively recent. Additionally, it has historically been difficult to collect data on the structure, finances, and activities of humanitarian organizations. Prior to the 1980s, few organizations kept track of their financial activities, missions, and relationships with partners. Data collected during this period often proves unreliable because there was no industry-wide standardization (Barnett and Weiss 2008: 30). Although reporting improved during the 1990s, several organizations either refused to release information or actively misrepresented figures due to safety concerns. As a result, empirical analyses are often a daunting task, as it is nearly impossible to determine the number of organizations, their structure and employee makeup, relationships with donors (individuals, states, and corporations), and field activity.

However, recent developments have made it easier for analysts to collect reliable information for studies on NGO security. Organizations such as the Afghanistan NGO Safety Office (ANSO) coordinate, track, and report on security incidents with the consent and help of NGOs in the field. Furthermore, greater numbers of humanitarian groups are now providing general information on projects and expenditures. Nonetheless, the seemingly simple task of analyzing the data is quite difficult. Fast (2014) persuasively argues that the intangible internal dynamics of organizations are of great importance when attempting to comprehend attacks. Although contemporary data are by no means perfect, it is now possible to engage in quantitative empirical analysis in areas that have previously posed great challenges.

Two studies were conducted regarding NGO security and its possible relationship to military operations. Following a marked increase in the death of civilian aid workers in Afghanistan in 2004, McHugh and Gostelow (2004) of Save the Children conducted an anecdotal analysis of the possible impact of PRTs on humanitarian security. The limited analysis sought to identify the linkages between PRT activities and possible changes in humanitarian security. Four areas of PRT activity were assessed: security, reconstruction, strengthening government authority, and relief operations. The results of the analysis suggested a positive effect of PRT activities on humanitarian security when the teams were involved in security, reconstruction, and expanding central authority. However, there were negative effects on humanitarian security when teams were involved in relief activities. The authors of the report conceded that their analysis was not sufficient to draw general conclusions regarding the impact of PRTs on humanitarian security. However, their findings demonstrated it was plausible that PRTs operating in similar areas as NGOs represented ‘the most significant potential threat to humanitarian security’ (McHugh and Gostelow 2004: 30–35).

A study by Watts (2004) of the Combating Terrorism Center (CTC) at West Point analyzed NGO security incidents from January 2003 to February 2004 at the provincial level in Afghanistan. The results of his empirical study indicated that provinces which shared a border with Pakistan and those with a higher percentage of home radios displayed a positive relationship with violence against NGOs. Conversely, provinces with higher levels of poppy cultivation witnessed decreased levels of violence against NGOs. All other variables—including US military presence, distance from each provincial capital to Kabul, warlord control, shared border between a province and a neighboring country, percentage of population in need of food aid, number of beneficiary population receiving food aid, cereal allocation, number of learning spaces, and number of formal/informal schools—were found to be statistically insignificant determinants of aid worker security. As a result of his analysis, Watts suggested a reevaluation of US and Afghan policy directed at eradicating poppy cultivation (Watts 2004).

Studies by Save the Children and CTC did not identify aid worker proximity to PRTs and military presence respectively as significant determinants of NGO security. However, the landscape in Afghanistan has changed significantly over the years, and an updated analysis would be helpful to validate prior findings. This study uses multiple regression to test the hypothesis that the presence of PRTs within a province is likely to lead to attacks against NGOs.

Research Design

This study relies on a panel-corrected standard error regression model of all 34 Afghan provinces for the years 2010 and 2011. The research builds upon Watts’ 2004 study, with the inclusion of additional control variables. Watts broadly measured the relationship between US military presence in a province and NGO security. However, it should be noted that only three PRTs were operational by mid-2003, so his inclusion of the variable was likely not feasible. As mentioned previously, the PRT initiative expanded significantly throughout the country shortly thereafter.

Operational Definitions of Variables

Dependent Variable

The dependent variable is the number of NGO security incidents by province. Consistent with Watts (2004), data on security incidents were obtained from ANSO. ANSO is sponsored by the International Rescue Committee (IRC) and funded by the European Commission’s Humanitarian Aid and Civil Protection Department (ECHO) and—among other tasks—collects information on the security environment around the country. ANSO uses their security incident reports to disseminate information and advice for NGOs operating in Afghanistan. ANSO classifies ‘incidents’ as fatalities, injuries, abductions, and criminal activity toward NGO workers. Provincial level ANSO data were readily available on ReliefWeb for 2010, 2011, and 2012. Data from 2012 were not included in this study because the PRT phase-out began the summer following the Enduring Strategic Partnership Agreement.

Independent Variable

PRT presence within a province is the independent variable. Information on PRT locations and team leads was obtained from the Institute for the Study of War (ISW). US-led teams are defined as those operated by the US, while coalition-led teams are defined as those operated by ISAF nations other than the US. The PRT presence variable is a dummy and is given the value ‘1’ if present within a province or ‘0’ if not present.

Control Variables

The following variables were included in Watts’ study and have been updated with data from 2010–2011: US military presence, poppy cultivation, border with a neighboring country, border with Pakistan, distance of the provincial capital from Kabul, population, and home radio ownership. Information on US military presence was obtained from ISW (Morgan 2012). The placement of combat forces is provided down to the battalion level and includes the following categories: maneuver (i.e., infantry, armor, and cavalry) units, artillery units, military police units, most types of engineer and explosive ordnance disposal units, and ‘white’ special operations forces. Information on ‘black’ special operations units is not provided. This variable is a dummy and is given the value ‘1’ if present within a province and ‘0’ if not present.

The number of ‘formal’ and ‘informal’ schools has been replaced by the number of students enrolled in government-run general education and private general education schools. As information on food aid—a variable included by Watts to test economic prosperity and NGO activity—was not readily available, it has been replaced by the number of NGO projects conducted and total NGO expenditures by province. NGOs often move from province to province, and monitoring individual agency activity over an extended period of time can be difficult. Therefore, the number of projects and the amount of money spent by all organizations within a province are the best determinants of overall NGO activity.

Watts included the percentage of home radio ownership in his study as a means of testing economic prosperity. However, he was later informed by a Pakistani military officer that the US military provided radios to locals in dangerous provinces as part of reconstruction efforts and this measurement may not be an adequate indicator of modernization and economic advancement. Therefore, this study includes the percentage of those who own a mobile device, in addition to those who have access to the Internet. Coalition fatalities, number of people incarcerated, and number of security incidents against journalists have also been included as a means of measuring the overall security of each province.

Data on population figures, enrollment in government-run general education programs, enrollment in private general education schools, number of people incarcerated, NGO projects, and NGO expenditures are from the Afghanistan Central Statistics Organization (ACSO). Information on poppy cultivation was gathered from the UN Office on Drugs and Crime’s (UNODC) 2011 Afghanistan Opium Survey. Provinces considered ‘poppy-free’ by UNODC are those estimated to have less than 100 hectares of cultivation, and are coded as ‘0’ in the UN survey.5 Poppy cultivation and total NGO expenditures were logged to provide a normal distribution.

The distance of each province’s capital from Kabul was calculated using, which provides a distance calculator using Google Maps’ application program interface (API). Data on coalition fatalities are from, while information on journalist security incidents is from Nai. Nai, an Afghan NGO, supports open media with the assistance of Internews and has been tracking threats against media workers throughout Afghanistan for the past ten years. Percentages of radio and mobile device ownership and access to the Internet were obtained from Internews, which made available data from Altai Consulting’s comprehensive Afghan Media in 2010 report that was commissioned by USAID. Finally, provincial border information was derived from maps published by ISAF. As provinces which share a border with Pakistan tended to experience greater overall levels of violence, this variable was included to test if the security situation was also more volatile for aid workers. Both border variables are dummies.

Empirical Analysis

Descriptive Statistics

Table 1 provides descriptive statistics of the variables included in the model by province. There was an average of four NGO security incidents per province per year in 2010 and 2011, with a high of 18 incidents in 2010 in Herat (see Figure 1). As Figure 1 indicates, no security incidents occurred in 2010–2011 in Badghis, Daykundi, Nimroz, Paktika, Panjshir, and Zabul. Some provinces had incidents in one year only, including Bamyan (2010), Helmand (2010), Samangan (2011), and Uruzgan (2011). Maps 1 and 2 provide a comparative visual overview of the security situation for both years.

Table 1

Descriptive Statistics (by Province).

Variable Observations Mean SD Minimum Maximum

NGO Security Incidents 68 4.49 4.88 0.00 18.00
PRT Presence 68 0.76 0.43 0.00 1.00
US PRT Presence 68 0.41 0.50 0.00 1.00
Coalition PRT Presence 68 0.44 0.50 0.00 1.00
US Military Presence 68 0.74 0.44 0.00 1.00
NGO Projects 68 173.76 183.05 15.00 1,250.00
Total NGO Expenditures (in millions $USD) 68 21.29 34.52 0.64 188.84
Coalition Fatalities 68 17.22 41.37 0.00 290.00
Journalist Security Incidents 68 1.44 4.03 0.00 32.00
People Incarcerated 88 651.35 1,261.25 1.00 7,591.00
Hectares of Poppy Cultivation 68 3,72915 11,860.82 0.00 65,045.00
Afghan Border 68 0.56 0.50 0.00 1.00
Border with Pakistan 68 0.32 0.47 0.00 1.00
Distance of Province Capital from Kabul (in miles) 68 173.88 122.81 0.00 492.37
% of Population Own a Radio 68 70.62 21.74 22.00 99.00
% of Population Own a Mobile Device 68 61.26 25.09 8.00 97.00
% of Population with Internet Access 58 7.32 8.78 0.00 33.00
People Enrolled in Govt General Education (in thousands) 68 215.98 207.69 21.11 1,086.82
People Enrolled in Private General Education 68 2,209.60 7,713.19 0.00 49,371.00
Population (in thousands) 68 727.55 645.03 136.30 3,818.70
Figure 1 

NGO Security Incidents 2010 vs 2011.

Map 1 

2010 NGO Security Incidents.

Map 2 

2011 NGO Security Incidents.

The average number of NGO projects per province was 174, with a minimum of 15 in 2010 in Zabul and a maximum of 1,250 in 2011 in Kabul. The number of NGO projects increased exponentially across all provinces, from 1,165 in 2010 to 5,140 in 2011, an increase of 341 per cent.6 Average annual NGO expenditures were $21.29 million, with a minimum of $640,000 in 2010 in Panjshir and a maximum of $188.835 million in 2011 in Kabul.

There was an average of 17 coalition fatalities per province per year, with a high of 290 fatalities in 2010 in Helmand. The average number of journalist security incidents was one, although Kabul in 2011 experienced a high of 32. The average number of people incarcerated was 651, with a maximum of 7,591 in 2011 in Kabul. The average amount of poppy cultivation was 3,729 hectares, with a high of 65,045 in 2010 in Helmand.

Data Analysis

Table 2 provides statistical results using panel-corrected standard errors of all 34 provinces in 2010 and 2011.7 Models 1 and 2 fit the data well, as indicated by their highly statistically significant chi-squares. Model 1 reveals that all variables except NGO projects, coalition fatalities, and the number of people incarcerated were statistically significant at ten per cent or better.8 Consistent with NGO complaints, the model suggests that if PRTs were present within a province, NGOs were likely to encounter more security incidents. However, the findings also reveal that US military presence was likely to be associated with fewer incidents. The results suggest that PRT presence was likely to increase NGO security incidents by three; however, US military presence decreased incidents by one, holding other variables constant. A second model was run in which general PRT presence was replaced with both US- and coalition-led PRT presence. This model was developed to assess if there was a difference between teams with different leads, as NGOs were less likely to encounter incidents in areas where US troops were present.

Table 2

Panel-Corrected Standard Error Models of NGO Security Incidents by Province.

Model 1 Model 2

Variable Coefficients SE Coefficients SE
PRT Presence 3.3328*** 0.6421 -- --
US PRT Presence -- -- 1.6548 1.8711
Coalition PRT Presence -- -- 2.8056*** 0.2588
US Military Presence -1.2755** 0.7266 -0.7159 1.5382
NGO Projects -0.0022 0.0064 -0.0011 0.0057
LOG Total NGO Expenditures (in millions $USD) 1.3680* 0.9258 1.2718* 0.9517
Coalition Fatalities -0.0039 0.0032 -0.0094*** 0.0028
Journalist Security Incidents 0.2062*** 0.0537 0.1150* 0.0732
People Incarcerated -0.0006 0.0008 -0.0005 0.0007
LOG Hectares of Poppy Cultivation -0.4652*** 0.1187 -0.4136*** 0.1217
Afghan Border -3.1086*** 0.5739 -2.4404*** 0.5491
Border with Pakistan 2.7003* 1.7939 3.0242** 1.4465
Distance of Province Capital from Kabul (in miles) 0.0264*** 0.0054 0.0243*** 0.0044
% of Population Own a Radio 0.0987*** 0.0068 0.1006*** 0.0181
% of Population Own a Mobile Device -0.0224* 0.0168 -0.0266*** 0.0115
% of Population with Internet Access -0.0584*** 0.0173 -0.0599*** 0.0270
People Enrolled in Govt General Education (in thousands) 0.0298*** 0.0033 0.0285*** 0.0033
People Enrolled in Private General Education -0.0003** 0.0002 -0.0003* 0.0002
Constant -12.8101*** 1.3384 -12.3337*** 1.2981

R2 0.58 R2 0.58
Wald Chi2 76.99*** Wald Chi2 57.78***
N = 68 N = 68

***p < 0.01; **p < 0.05; *p < 0.10, one-tailed test.

The results of Model 2 are revealing. ‘US-led teams’ was not statistically significant; however, ‘coalition-led teams’ was significant at the one per cent level. The model suggests that if a coalition-led team was present within a province, NGOs were likely to experience three more incidents, holding all other variables constant. This accounts for all incidents indicated by the general ‘PRT Presence’ variable in Model 1. Model 2 therefore suggests that PRT presence within a province tended to result in decreased security for NGOs, but only in provinces with coalition-led teams; US PRTs were insignificant indicators of NGO security.9 US and coalition teams were structured differently and engaged the populace in unique ways. Although it is possible that specific methods utilized by non-US teams may have resulted in decreased NGO security, this needs to be analyzed in greater detail.

Similar to Watts’ findings, provinces which share a border with Pakistan and those with a higher percentage of home radios were likely to experience violence against NGOs. The AfPak border was largely perceived as a generally volatile region for military actors, and both studies suggest these dangers extended to NGOs. As previously mentioned, although Watts initially used radio ownership to control for economic prosperity, he was later informed that the US military provided radios to locals in dangerous provinces as part of reconstruction efforts during the early stages of the conflict. This suggests higher levels of radio ownership may equate to greater volatility (Watts 2004: 16). The findings of this study appear to support this hypothesis. Additionally, results call into question the effectiveness of expanded radio access as a means of quelling violence.

Although radio ownership holds a positive relationship, mobile device ownership and access to the Internet were found to be negative. These variables were incorporated to test for economic advancement within a province. Given that provinces with greater access to these technologies experienced decreased attacks on aid workers, perhaps more economically developed areas were less likely to either produce or host actors who targeted NGOs.

Also consistent with Watts, provinces with higher levels of poppy cultivation tended to be safer areas for NGOs. Thus, cultivation may actually have had a stabilizing effect on the region. One speculation is that cultivation only occurs in regions with existing stability, as farmers may not want to risk growing crops in conflict-ridden zones. Another possibility is that many of the projects NGOs engage in are beneficial to cultivators. These organizations not only provide the necessary materials and equipment to farmers, but also engage in infrastructure projects such as roads and bridges which facilitate crop transportation within rural areas of the province.

Moreover, the findings reveal that, although the number of NGO projects conducted was statistically insignificant, total expenditures held a statistically significant relationship with increased incidents. One speculation is that NGOs may have been more likely to encounter attacks while engaged in a ‘large scale’ activity that utilized a greater number of resources. When NGOs spend significant amounts of time and money in one area, insurgents may be able to monitor daily activity closely, thus rendering aid workers more vulnerable to attacks.

Although the number of coalition fatalities and the number of people incarcerated were statistically insignificant, journalist security incidents held a statistically significant relationship with NGO security incidents. This suggests that provinces where NGOs encountered attacks may have been hostile areas for other civilian actors as well. Information on Afghan civilian casualties should be included in a future analysis to further test this hypothesis. After accounting for the difference in team leads in Model 2, coalition fatalities became statistically significant at the one per cent level.

Finally, educational enrollment was also shown to be a significant determinant of incidents. Interestingly, provinces with a high number of citizens enrolled in government-run education programs held a positive relationship with NGO attacks, while provinces with a high number of people enrolled in private education programs held a negative relationship. Similar to mobile phone ownership and Internet access, perhaps a greater number of people enrolled in private education indicates higher levels of economic prosperity in a region and therefore results in less incentive to turn to violence.


In an effort to curtail the insurgency in Afghanistan, the US and coalition forces have blended military and humanitarian operations, much to the dismay of many within the NGO community. One of the major debates surrounding this effort concerns the PRT initiative, which many NGOs have faulted for causing ‘blurred lines’ between military and aid activity. Although certain NGOs have accused PRTs of decreasing the overall security situation for aid workers within a province, there was a lack of empirical evidence to support this claim. A report that often surfaces in this debate is Clinton Watts’ 2004 study, which analyzed NGO security incidents in Afghanistan that occurred from 1 January 2003 to 15 February 2004. Although Watts’ findings revealed that US military presence was not a significant variable concerning NGO security, PRTs were not included in his analysis.

To build upon Watts’ findings, this study ran a panel-corrected standard error regression model with the inclusion of PRT presence as an independent variable. Results indicated that if PRTs were present within a province, NGOs were likely to encounter a greater number of security incidents; however, if the US military was present, NGOs were likely to experience fewer incidents.

Given that the findings revealed a positive relationship between increased NGO incidents and PRT presence, yet a negative relationship with US military presence, a second model was run substituting PRTs with US- and coalition-led PRTs more specifically. Model 2 results indicated that US PRTs were insignificant indicators of NGO security; however, coalition PRTs were found to be highly significant. Given these results, it is difficult to conclude that PRTs in general are culpable for lax NGO security.

As the ‘blurred line’ explanation is not supported in this context, policymakers seeking remedies for aid worker insecurity should examine other possible causes of the increased attacks. The results of this study concur with Watts that a reevaluation of anti-drug policies directed towards eradicating poppy cultivation is needed. Additionally, provinces with higher levels of private school enrollment, mobile phone ownership, and access to the Internet all held a negative relationship with NGO incidents, suggesting that greater economic advancement may result in less incentive to turn to violence. However, the paradox faced by aid workers is that they tend to operate in less developed regions of the country by design.

Additionally, this study’s findings reveal a positive relationship between higher levels of NGO expenditures and security incidents. This suggests that aid workers may have experienced greater attacks in regions where they were most active (although total NGOs projects was statistically insignificant). Nonetheless, without access to the number of aid workers operating in each province, it is difficult to test for incident proportionality. If data from the projects that NGOs were engaged in when they were attacked—including length of time in the field, manpower, and public visibility—is made available in the future, this should be analyzed in greater detail.

Given the results of this study, future research should seek to explain the different approaches of lead nations in greater detail. Although it is widely understood that coalition forces approached their PRT mandate in unique ways, an in-depth comparative study would be beneficial for researchers. As the scholarly literature on PRTs is limited, a project of this sort may best be served by extensive interviews with those who have served on teams in the field.

If information becomes available, future studies on NGO insecurity should include the nature and affiliation of the assailant, whether the NGO was directly targeted, if the NGO was local or international, what type of activity or project the NGO was engaged in (including length of time in the field, manpower, and public visibility), and where the incident occurred in proximity to PRT operations. In smaller provinces, PRTs may have been operational in most areas; however, in much larger provinces, an NGO may come under attack in an area with no history of PRT activity.

In addition to external variables, it is equally important to analyze the internal dynamics of NGOs operating in the field. This is a critical factor which is often omitted from studies concerning aid worker security. The inclusion of internal vulnerabilities in analyses has recently been suggested by Fast (2014). Aid workers—both as individuals and within organizations—approach their work in diverse ways, and not all are uniformly perceived as favorable or unfavorable by the communities in which they operate. NGOs can contribute to these perceptions through their individual interactions and organizational practices, and internal dynamics must be analyzed for a more complete understanding of the cause of these attacks.

Replication Data

The data set and STATA output for this study are available at