Don’t tell God I am knocking on Devil’s door:
The risks of intertwining the delivery of aid with the military
Author:
Drs. Thijs van Vulpen
Supervisors:
Dhr. A.J.M. Strik (ret BrigGen-NLA) Dr. X. Guillaume
Dr. M. Madej
Study programme:
MA International Humanitarian Action (NOHA)
Universities: University of Groningen University of Warsaw Date: 8 July 2019 Abstract
The hearts and minds campaigns in Afghanistan and Iraq have reheated the debate about incorporating aid in counterinsurgency operations. Humanitarian organisations point toward the risks for their employees and the implications for the delivery of aid in general. However, quantitative research on the effects for civilians seems limited. This research aims to examine the relationship between aid disbursements and insurgent violence towards civilians by looking through the lens of insurgents and adopting a case study of Afghanistan and Iraq. The statistical model shows that aid does indeed lead to increased attacks on civilians, but appears to be violence-reducing during the counterinsurgency years in Afghanistan and Iraq. A possible explanation could be that aid can complement military activities if civilians feel that they are credibly protected.
Key words: Militarization of aid, hearts and minds campaigns, insurgents, violence against
THE SUPREME ART OF WAR
IS TO SUBDUE THE ENEMY WITHOUT FIGHTING
1. Introduction
The militarization of aid is not a new phenomenon, but is in fact, it is a strategy that appears in different forms throughout history. Examples are stability operations, proxy wars, and the embedment of aid agencies into the military. Neither is it a strategy that is solely applied by the military as the word ‘militarization’ suggests. Aid organisations and agencies have also contributed to the militarization of aid by openly picking sides and cooperating with public and private militaries for self-defence. In particular self-defence was catalysed by the rise of ‘new wars’ which increased the complexity of conflicts and included more non-state actors with a predatory nature (Pérouse de Montclos, 2014). The debate about the militarization of aid caught new fire when light was shed on the failures of the so-called ‘hearts and minds campaigns’ in Afghanistan and Iraq (see e.g. Berman, Shapiro, & Felter, 2008; Fast, 2010; Ford, 2005; Sexton, 2016; Williamson, 2011). Hearts and minds found its origin in British counterinsurgency and is a strategy with a population-centric approach that applies soft measures including the distribution of aid. The United States reshaped and modernised the concept for its counterinsurgency operations in Afghanistan and Iraq based on lessons learned from the Vietnam War (Valeyre, 2012). The strategy is centralised around the idea of winning the support of the local population by inquiring on the population’s needs and persuasively providing aid and security. The objective is to separate insurgents, referring to those who fight against a government or invading forces, from their pre-conflict established social networks by penetrating and manipulating these networks, building trust, and gathering intelligence. In a conflict where insurgents hide among the population in an area unfamiliar to the counterinsurgent, civilians can provide crucial information on the insurgents’ whereabouts (see e.g. Ford, 2005). For the counterinsurgent this would enhance its troops’ safety and security and a step closer to winning the conflict.
But over the years this strategy has received fierce resistance. Aid organisations have released statements and have publicized shocking stories relates to the consequences of such campaigns (see e.g. Krahenbuhl, 2011). Intertwining the delivery of aid with counterinsurgency, which is defined as the military or political actions applied by national or international actors to oppose or supress insurgencies, can lead to a confusing situation for both civilians and insurgents. Traditionally the military is focused on combat and not impartial, independent or neutral to a conflict, while these are the fundamental pillars of the humanitarian sector (see e.g. Mosely; Perito, 2005). Due to the confusing created insurgents view those who deliver aid as part of the hostile campaign that aims to
undermine their established local authority and those who receive aid as traitors (see e.g. Williamson, 2011; Fast, 2010). The insurgency’s answer is to deliberately attack aid workers and indiscriminately attack villages in an attempt to remain in control and to punish civilians (Hirose, Imai, & Lyall, 2017). Such danger can make it impossible to deliver aid and forces aid organisations to seek safety and security with public and private militaries. Hence, the self-defence mechanism.
While the debate focuses on the recent failures of hearts and minds campaigns in Afghanistan and Iraq and the accompanied implications for the aid sector, it seems considerable less attention is being paid to the consequences for civilians. This is surprising since civilians are clearly considered the key to winning the conflict by both insurgents and counterinsurgents. While the counterinsurgents try to persuade civilians to their side, the insurgents often aim to prevent civilians switching sides by applying violent repercussions (Hirose et al., 2017). Civilians, due to the valuable information they can potentially provide, seem to be trapped between two fires.
Even if quantitative research is being conducted on the consequences for civilians, much of the work uses financial data from the Commander Emergency Response Program (CERP) that provides funds for development projects in Afghanistan and Iraq (see e.g. Berman et al., 2008; Sexton, 2016). This, however, does not seem to capture the perspective of insurgents. CERP provides funds for projects implemented by the United States-led coalition (i.e. the counterinsurgent), but the discussion above entails that aid organisations are deliberately targeted due to the insurgents’ association with the embedment of aid into counterinsurgency campaigns. If insurgents consider the provision of aid to be part of a hostile campaign, it is more likely that they will oppose any aid distribution independent of the organisation, agency or institution providing it.
This research aims to complement the discussion by providing insight in the risks for national civilians living in conflict area where aid is used as a counterinsurgency strategy to persuade the local population. It hypothesises that incorporating aid into a counterinsurgency strategy leads to more insurgent attacks. The underlying reasoning is that insurgents, afraid of losing their authority, attack civilians to punish them for making use of aid provided by the insurgents’ perceived enemies. As a result insurgent attacks are expected to increase in an attempt to remain in control when more aid is provided. To provide structure, this paper will first address the channels of aid distribution, the risks and dangers for aid organisations, the different forms of militarization, the history and current debate on hearts and minds campaigns, and the risks for local civilians. Answering
these sub-questions should provide a well-layered theoretical base before continuing with the quantitative part. Based on the work of Eck and Hultman (2007), a model will be presented that focuses on a case study of the counterinsurgency operations in Afghanistan and Iraq. Both cases are of interest, because they describe how hearts and minds conducts have been reshaped for modern counterinsurgency operations. The model aims to show a positive statistical relation between aid disbursements and the number of incidents during which national civilians were targeted by insurgents. Additionally it tests whether aid disbursed during the counterinsurgency years in Iraq and Afghanistan had a positive effect on insurgent violence. Next, the model is then re-tested for robustness by reducing the sample size to countries that experienced at least a ‘minor conflict’ or ‘war’ as defined by the standard of the Stockholm International Research Institute (SIPRI).
Contrary to other research, the problem will be explained through the lens of the insurgent instead of the counterinsurgent. For the statistical model this means using alternative data sources and providing a glimpse into data on insurgent attacks. The main objective is to provide a complete story on whether local civilians are indeed punished by insurgents (i.e. ‘God’) when knocking on the door of the insurgents’ perceived enemies (i.e. ‘Devil’).
In short, the results confirm that aid is positively related to insurgent violence, but, against expectations and most theory, aid delivered during the counterinsurgency years in Afghanistan and Iraq appears to be violence-reducing. Although purely speculative, a possible explanation could be the counterinsurgent’s troop presence in the area because of which civilians feel credibly protected and are less reluctant to cooperate (based on Sexton, 2016).
Before continuing with the theoretical part certain remarks and notifications are in place that should ease reading this paper. First of all, aid workers who conduct activities of a primarily humanitarian nature should be respected and protected by law (see e.g. article 20 in Geneva Convention IV or article 8.2(b)(iii) in the Rome statute of the International Crime Court). In such a situation an aid worker is considered to have a ‘civilian status’ which means (s)he enjoys the same legal protection as civilians. However, for the purpose of this paper aid workers are considered to be ‘civilian workers’ rather than ‘civilians’ to allow for making a distinction between both groups.
Second, humanitarian aid often refers to short-term aid delivered in emergency situations with the purpose to alleviate suffering and to provide civilians in their basic needs. Development aid often implies long-term aid to enhance a country’s social,
economic, political or environmental development. Although this difference is duly noted, the text in this paper refers to aid in general terms which can either have an emergency or development nature, unless otherwise specified.
Third, channelling aid to beneficiaries is done by many organisations, agencies, institutions, or entities which are structured, organized, and financed differently and have a distinct operationalization. Although it is useful to understand the channels through which aid can be provided, and therefore the text will address this matter later on, it should be emphasized that no specific distinction is made between the terms ‘aid agencies’ and ‘aid organisations’ despite their differences, in particular with regard to government cooperation. Again, the text refers to aid in general terms which include all actors supplying aid to beneficiaries.
Fourth, although there is a distinction between ‘insurgents’ and ‘terrorists’, the terms are sometimes used interchangeably in the literature. As defined before, insurgents are combatants fighting against a government or invading power while terrorists use mass violence to inflict harm on civilians and to spread fear. The existence of this difference does not mean that insurgents cannot commit terrorist attacks to achieve their goal, but does neither label all violent activity against governments as an act of terrorism. This paper prefers to use the broader term ‘insurgents’ when speaking about those who oppose a government or invading force(s).
Finally, the terms ‘local’ and ‘national’ civilians both refer to a civilian person living in the country of his or her nationality. The terms are used interchangeably since the data are aggregated on country rather than regional level.
The structure of this paper is as follows. Section two presents the theoretical part that addresses the aid delivery channels, the dangers for aid organisations, the militarization of aid, hearts and minds campaigns and the risks for the local population. Section three describes the research methodology, presents the data sources, and shows the preliminary results. Section four shows the regression results and its implications, and re-tests the model for robustness. Section five concludes.
2 Channels of aid delivery
Supplying aid to beneficiaries is not solely limited to aid organisations, but can be undertaken by many different actors to a conflict. Aid distribution is split into five broad channels. It is useful to consider and distinguish these channels since each channel has its own characteristics. Governments are not mentioned as a separate channel because governments are assumed to outsource their aid distribution to any of channels mentioned below via donations (e.g. financial or material donations or providing expertise). The first channel is through aid organisations including international organisations, nongovernmental organisations (NGOs), charities, voluntary organisations, and the International Committee of the Red Cross (ICRC).1 Examples are Médecins Sans Frontières, Doctors with Africa, Oxfam Novib, Care International, and Save the Children. Although during emergencies some of these organisations might cooperate in the cluster approach for coordination purposes, they usually operate on a stand-alone basis. Day-to-day management is focused on delivering aid to beneficiaries by sticking to the core principles of the humanitarian sector, namely humanity, independence, impartiality, and neutrality2, while avoiding interference of governments. But the dodging for political agendas does not imply that these organisations do not receive funding from governments. Some of these organisations mainly rest on the generosity of private and civilians donors like Médecins Sans Frontières (Médecins Sans Frontières, 2019), while others receive substantial donations from governments like Oxfam Novib (Oxfam Novib, 2019). In particular organisations receiving considerable amounts of government donations risk being vulnerable for the enforcement of political agendas by governments, especially when those organisations receive protection from national militaries.3
The second channel is through aid agencies which is a term often used interchangeably with aid organisations while there is a clear distinction. Aid agencies refer to organisations operating under the umbrella of a government or multiple governments (e.g. bilateral or multilateral organisations) or those organisations of which
1 The ICRC is mentioned separately because it is a private association registered in Switzerland owning an unique legal status to carry out its mandate domestically and abroad. For more information visit the website:
https://www.icrc.org/en/international-review/article/tools-do-job-icrcs-legal-status-privileges-and-immunities
2 The principle of neutrality is not always met by aid organisations since some organisations tend to speak out against specific war crimes or attacks against themselves, other agencies or civilians. A good example is Médecins Sans Frontières.
3 This point will be elaborated upon in subsection 2.3 “Humanitarian aid protected by public or private militaries”
the operations are directly influenced by the political agenda of a government. Examples being the agencies of the United Nations, USAID, or Australian Aid. Hence, the contrast with aid organisations that clinch to their independence and impartiality. Aid agencies align their activities with the interests of governments which affects the agency’s independence, impartiality, and neutrality. Operating in favour of governments means that the barrier to seek safety and security with national militaries is expected to be lower for agencies compared to aid organisations that try to avoid political agendas.4
A third channel is through public militaries which overlaps with the previous two channels. Public militaries can contribute to aid delivery in two ways. First, such militaries can provide security services to aid actors to facilitate the delivery of aid. For instance the protection of food convoys (Fast, 2010). In addition security services can assist aid organisations in creating safe and controlled working conditions. An example is peacekeeping operations which focus on maintaining peace, but at the same time providing aid organisations with the opportunity to ‘safely’ implement projects. Second, public militaries can directly supply aid to beneficiaries for security or political reasons (Williamson, 2011). In hard-to-reach or dangerous areas these militaries often possess superior skills and intelligence to provide aid compared to aid organisations. But public militaries, in particular national militaries, can also provide aid to civilians in an attempt to gain local support (e.g. hearts and minds campaigns).
A fourth channel is through private militaries, such as private security companies and private military companies, which can provide similar security services to aid organisations like public militaries and can deliver aid to beneficiaries living in insecure areas (Leander, 2005a; Leander, 2005b; Spearin 2007; Spearin, 2008; Von Boemcken, 2007). The main advantages of private militaries versus national militaries will be discussed, but for now it should be emphasized that a clear difference with public militaries is that private militaries are profit-seeking entities and not necessarily influenced by a political ideology (Spearin, 2007). However, private militaries are increasingly hired by states to replace specific, if not all, operations of the national military, because states consider this more economical than maintaining an army (Isenberg, 2009). Therefore it is possible that these entities are intertwined with military campaigns and pursue a client’s political objective. An example is the hiring of private militaries to facilitate logistics (and security and offensive services) during the war in
4 Again, the term ‘aid agency’ and ‘aid organisation’ will be used interchangeably throughout the text despite the difference made here.
Afghanistan and Iraq (Isenberg, 2009).
A fifth channel is through the private sector with the exclusion of private militaries. Development aid in the form of favourable loans or grants is channelled to (inter)national businesses to boost the local economy or trade. An example is the official development assistance flows to private sector instruments to enhance welfare (OECD, 2019). But aid can also be distributed by the private sector in many different forms. For instance, multinational or large corporations invest in corporate social responsibility by funding social, economic or environmental projects (Jamali & Keshishian, 2009). Another example is that some profit-seeking organisations are directly linked to a self-established foundation that focuses on social goals (e.g. Waka Waka5).
2.1 The risks for humanitarian workers
Humanitarian aid workers are defined as those working for a national or international emergency or development aid agency who, for the sake of this research, are considered to be distinguished from the legal umbrella ‘civilians’ but are rather classified as ‘civilian workers’. They often work in challenging environments and it is therefore not surprising that they face increased risk when conducting their employment. In 2018, a total of 360 aid workers got killed, wounded, or kidnapped which illustrates the dangerous working conditions.6
Attacking aid workers seems illogic at first since they comply to help all people in need and attempt to relief them of pain and suffering. However, it remains unclear how insurgents perceive this impartial and neutral behaviour of humanitarian agencies and whether they even consider it (Fast, 2010). Insurgents might regard that working under these two fundamental humanitarian principles also means helping the enemy which could be a reason to attack those who help the enemy (Fast, 2007).
Yet security issues can also arise from individual behaviour of organisations and their employees or the subsequent failure to act accordingly to the security matter (Fast, 2010). A related factor is the (in)experience of aid workers since most incidents happen at the beginning of the deployment (Sheik et al., 2000). Not for nothing security manuals have been published to establish awareness about the link between personal conduct and security incidents (see e.g. Van Brabant, 2000).
5 For more information visit the website: https://waka-waka.com/en/mission/
6 This figure is retrieved from the Aid Worker Security Database which will be further discussed in the next subsection.
Another factor is the financial gain from looting, hijacking, stealing, and robbing of aid organisations. Those organisations import vehicles, equipment, food and medicines which are often scarce resources, are useable or can be sold to generate funds to finance the war (LeRiche, 2004; Pérouse de Montclos, 2014). Part of this thievery and criminality is attributable to organised crime that is primarily after the financial gains, but this does not exclude the involvement of insurgencies in this organised crime. In particular organised terrorist groups that follow a long-term ideological campaign tend to cooperate with organised crime to generate a continuous income stream to finance their operations (Hutchinson & O’Malley, 2007). Banditry and insurgency can cooperate and thus overlap and it would even be possible for banditry to turn into an insurgency if the bandits’ interests are to sustain or reactivate a conflict due to the financial potential. The reasoning is simple: an ongoing conflict means an ongoing investment of aid organisations in the country which could mean a potential ongoing stream of income (LeRich, 2004). It is important to notify that insurgents or bandits are not the only criminals or violent actors in a conflict. Some governments have been manipulating the aid distribution away from insurgency-supporting population groups and national forces have been reported to loot aid organisations as well (LeRiche, 2004). For example, during the early 1980s the Thai army entered refugee camps near theThai-Kampuchean border at night to take whatever they considered to be theirs. Also security guards in Somalia have been reported to steal from the aid organisations they were meant to protect (LeRiche, 2004; Bakonyi, 2010) and even beneficiaries are guilty of aid manipulation and thievery. For instance, UNCHR reported that refugees living in camps often possess more ration cards than allowed (Ryle, 1992).
It should be emphasized that criminality is not solely motivated by financial prospects. Other motivations can be desperation or survival due to hunger or poverty, to balance social injustices, power and domination, or war strategy. Revenge could be a motive to engage in criminal activities as well (Bakonyi, 2010). Even if not directly aimed at aid organisations, these motivating factors can lead to increased risk and deteriorating working conditions.
There are many more reasons why aid workers are being attacked. For instance because of the type of aid or the size of the project. Humanitarian emergency projects tend to have no significant effect on violence while protective measure projects do have a violence-increasing impact and small projects could be considered easier targets compared to large projects (Sexton, 2016). Surely, there are many more reasons to think
of why aid workers are being targeted.
It is striking that the vast majority of the victims are national rather than international aid workers (Fast, 2010; Egeland, Harmer, & Stoddard, 2011; Sheik et al., 2010). Out of the 360 aid workers who were attacked in 2018, 340 (94%) were locally hired. This is partly due to national staff being more deployed in the field, but also attributable to different security perspectives of national and international humanitarian organisations. It has been reported that international agencies provide national staff with less security trainings or preparations for reasons based on incorrect assumptions such as that national staff have their families and communities and therefore would require less moral or psychological support or that national staff does not stand out as much as foreign staff while they actually might stand out more when working in other communities or environments (Egeland et al., 2011; Stoddard, Harmer, & Haver, 2006). In addition research suggests that national staff tends to view domestic situations less risky compared to international staff (Egeland et al., 2011). National organisations would, on their turn, consider security measures of lesser importance and receive a smaller part of the funding cake which means they have less funds to devote to security (Hasenstab, 2018; Redvers, 2017).
To consolidate the discussion above data on the number of victims are presented in the next subsection along with a brief analysis to create a better visual overview.
2.2 A graphical representation
The Aid Worker Security Database (AWSD) provides data on violent incidents involving humanitarian aid workers starting from 1997 up to the present. Data are selected from 2000 up to 2018 to conduct a brief analysis of the number of victims and to discover a trend during the twenty-first century. Figure 1 and 2 graphically depict this data. Clearly 2013 has been the most violent year for humanitarian aid workers with a total of 474 victims of which 82 percent were national personnel. Although proportionally most international aid workers were attacked in 2001, their most violent year is 2009. For national aid workers proportionally the most violent year was 2018, while in 2013 most victims were claimed.
In the beginning of the twenty-first century most attacks resulted in the death of aid workers rather than kidnapping or injury. This appears to balance more evenly over the years with killings, injuries or abductions to happen roughly equally frequent. In general, there seems to be an increasing overall trend over the years with 91 victims in
2000 while there were 309 victims in 2011 and 360 victims in 2018.
The upward sloping trend is shocking, though not completely surprising. The humanitarian sector has grown exponentially over the past decades due to the increase in financial contributions, thereby allowing for more projects and the hiring of more aid workers which, frankly, means more employees at risk. Moreover, aid workers are most needed in intense conflicts, but that is exactly where they are in the most danger
57 27 38 88 56 53 87 88 128 108 72 86 70 156 122 110 107 139 120 23 20 23 49 46 96 87 87 91 94 84 126 115 178 88 109 98 102 116 11 43 24 7 23 23 66 45 59 93 98 97 92 140 123 69 89 72 124 0 50 100 150 200 250 300 350 400 450 500 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Nu m b er o f v ictim s Year
Killed aid workers Wounded aid workers Kidnapped aid workers
Figure 1: Total victims among humanitarian employees from 2000 to 2018
The figure shows the total number of national and international humanitarian employees affected by attacks from 2000 to 2018. Bars are coloured for killed, wounded or kidnapped aid workers with the number of victims presented in absolute numbers in the graph. The x-axis shows the number of victims and the y-axis shows the year. Data is retrieved from the Aid Worker Security Database.
77% 69% 80% 81% 81% 91% 89%85% 82% 75%82%91%82%88% 90% 90% 85% 91% 94% 23% 31% 20% 19% 19% 9% 11%15% 18% 25%18% 9% 18%12% 10% 10% 15% 9% 6% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 P er ce n tag es Year
National victims International victims
Figure 2: Percentual presentation of national versus international victims from 2000 to 2018
The figure shows the proportion of national versus international victims among humanitarian employees affected by attacks which include employees being killed, wounded, or kidnapped from 2000 to 2018. The rounded percentages are presented in the graph. The x-axis shows the total percentages and the y-axis shows the year. Data is retrieved from the Aid Worker Security Database.
(Hoelscher, Miklian, & Nygård, 2017). The logic behind an upward sloping trend in this industry seems somewhat understandable, though still unacceptable, and is known as the denominator effect (Fast, 2010; Sheik et al. 2000).
2.3 Humanitarian aid protected by public or private militaries
It is no wonder that aid agencies try to protect their employees and property from these attacks by seeking their salvation with public or private militaries. The definition of public militaries is rather broad since the term does not only include the militaries of governments. There are other forces that could qualify as public military, like NATO or United Nations’ peacekeeping forces, but those militaries pursue different objectives and play a distinct role in a conflict. This does not fit the argument made in the text below per se and for that reason a distinction is made in the terminology. The term ‘national militaries’ refers to national militaries of states which are directly controlled by governments while the term ‘public militaries’ refers to all qualifying public militaries possibly influenced by a political agenda, but not necessarily controlled by a single government of a state. The discussion in this subsection will mainly focus on national militaries.
Cooperating with national militaries has several advantages of which the most obvious is general security. Moreover, the military can provide access to restricted and insecure areas for aid convoys or for personnel to safely travel into these areas (Fast, 2010). Additionally, if the area is too ‘hot’, meaning too insecure or dangerous, the military can supplement aid agencies by delivering goods and services to beneficiaries in the area (Williamson, 2011). Furthermore, the military disposes of better intelligence than most aid agencies which could help agencies to plan and coordinate their missions. Finally, the military can consult agencies on security matters and train aid workers on working in dangerous contexts before or during a deployment (e.g. Hazardous Environment Awareness Training). Note that those security services are general and can be provided by other public militaries.
However, seeking safety and security with such national militaries has a great downside. Governments are known to be the biggest donors of aid organisations and might expect to enforce their political agenda through those organisations in return for their donations. This makes it harder to refuse for recipient organisations seeking safety and security through national militaries and can affect their neutrality and independence. It could even affect the impartiality of the organisation if the government limits the
provision of aid to specific population groups (Spearin 2007; Spearin 2008). Of course the key here is whether the agency is actually operating in the field and not distributing received donations to local aid organisations. If that is the case the distributing agency does not require immediate security from national militaries. However, the agency could still be vulnerable to cuts in government donations which can affect their respective activities.
Another downside is that an agency physically protected by a national military could become an actual target of belligerents opposing this military (Spearin, 2007). This is probably more likely if the military is a direct party to the conflict and is deploying soldiers in uniforms to protect certain agencies. It could even occur when agencies are cooperating with militaries to provide aid to beneficiaries or when militaries incorporate aid provision in their campaigns. This blurring of lines between agencies and militaries leads to a confusing situation for belligerents and civilians (Williamson, 2011).
Also other frictions can occur. For example, soldiers of a national military to a conflict are not neutral and could be idealistic and not impartial (Leander, 2005a). This could hamper aid operations or prohibit agencies to deliver aid to certain population groups.
As a solution many organisations have turned to private militaries over the past years by collaborating with Private Security Companies (PSCs) (Stoddard, Harmer, DiDomenico, 2008). Private Security Companies are non-governmental profit-seeking organisations offering a range of defensive security services to protect individuals or assets. Although sometimes mentioned interchangeably with Private Military Companies (PMCs) in the literature, PSCs should be distinguished from PMCs since the latter also offers offensive services (O’Brien, 2000). PCSs often employ former military personnel and provide essentially the same services as public militaries such as physical security, trainings and consultancy, intelligence, complementary aid (e.g. logistical, engineering, or technical services), and supplementary aid (e.g. by delivering services and personnel trained to replace aid workers when it becomes too dangerous for them to operate in the area). But contrary to national or public militaries PSCs are not necessarily involved in political agendas and specifically train personnel that can be deployed in a versatile way and refrains of any ideological views on the respective conflict. (Leander, 2005a; Leander, 2005b; Spearin 2007; Spearin, 2008; Von Boemcken, 2007).
However, there are caveats concerned to utilising PSCs for managing the security of aid agencies, mainly related to dependence, reliance, and the structure of PSCs. First
of all, the choice of personnel, quality of the service and implementation of activities are harder to influence if aid agencies depend too heavily on the services of PSCs (Leander 2005b). This dependents on whether legal contracts are adequately constructed. A worst case scenario is when their guardian could force organisations to increase payments to ensure the continuation of security services or suddenly terminate the contract and leave the aid organisation unprotected. This criticism is countered by emphasizing that PSCs are profit-seeking entities which makes them vulnerable for reputational damage and therefore seek to maintain the confidence of the employer (Spearin, 2007).
Second, PSCs are often large organisations with ties to governments to enlarge their market or to resolve conflicts in some cases.7 This would mean that governments could still indirectly pressure aid agencies by influencing their security companies (Leander, 2005b; Van Boemcken 2007). Furthermore, these large security companies possess an enormous clientele base possibly including notorious clients for whom they undertake sketchy activities (e.g. the protection of blood diamonds mines). This could indirectly inflict serious reputational harm on an aid agency and consequently affect its position in the local community when associated with these activities or clients (Von Boemcken, 2007).
Finally, the information flow to aid agencies could be distorted if they use PSCs as their most important source since those organisations have a commercial self-interest in sustaining insecurity (Leander 2005b).
An agency seeking safety and security for its staff and assets seems to undertake a nearly impossible task. While PSCs provide similar services as public militaries, these organisations contain serious disadvantages as well. On top of that, agencies are often hesitant to deploy armed guards near their compounds or warehouses because they are considered to evoke aggression (Von Boemcken, 2007). In most cases agencies use third parties just for soft measures such as trainings, consultancy, or unarmed guards (Stoddard et al. 2008). Frankly, cooperating with a third party specialised in fighting for security matters is a first step towards the militarization of aid (Von Boemcken, 2007).
2.4 Different forms of the militarization of aid
Seeking safety and security with national, public or private militaries is not always
7 Governments hiring private militaries to resolve a conflict or to replace their public militaries would likely hire PMCs rather than PSCs. The main reason being the emphasized difference in definition: PMCs offer offensive services which PSCs do not. Therefore the latter part of this argument should be slightly nuanced.
entirely up to aid organisations. Donors are important for aid organisations and can force them to ‘militarize’ their activities. For example, donors can approve aid programmes in terms of national or international security or decide to give priority to specific countries or sectors based on security analyses (Brown & Grävingholt, 2016). Although this argument is related to securitization it is inherent to the militarization of aid in this case. Stability operations such as peace-making, peace enforcement, peace-building, and peacekeeping have also contributed to the militarization of aid. Peace operators have been involved in aid distribution and provide protection to civilians and civilian workers (Gheciu, 2011). Creating, maintaining, and sustaining peace involves investments in development which requires creating a relatively safe working environment for aid institutions.
Although it is useful to understand that these operations have different definitions and objectives, boundaries between them have blurred since creating and maintaining sustainable peace is rarely limited to only one type of peace operation (United Nations, n.d.). Peace-making usually involves diplomatic action in an attempt to build a negotiated agreement between hostile parties during active conflict. These operations are not necessarily conducted by the United Nations, but can likewise be undertaken by governments, regional organisations, non-governmental groups, or even a prominent individual. Peace enforcement refers to applying coercive actions, including military force, in order to restore peace and security, when peace is threatened or breached, or in case of an act of aggression. Such action requires authorization of the Security Council. Peace-building aims at creating stable and sustainable peace and development in the long-term by addressing the core issues of the respective society and state and investing in capacity-building so that the state can execute its core functions (e.g. the training of security forces). Peacekeeping aims at maintaining peace, such as maintaining a ceasefire or peace agreement, after active conflict has halted but not been solved yet. This requires the deployment of peace forces assisting in the disarmament and reintegration of combatants, promoting human rights, and supporting elections, for instance. Hence, peacekeeping operations can overlap with peace-making and peace-building activities (United Nations, n.d.).
But conducting stability operations and deploying peace forces on the ground can lead to friction with humanitarian actors. The North Atlantic Treaty Organisation (NATO) has been criticized by aid organisations for their peace operations in Kosovo in 1999 (i.e. Kosovo Force) and in Afghanistan in 2001 (i.e. International Security
Assistance Force). According to relief workers, NATO made it impossible to sustain local trust in the impartiality of international aid organisations due to its involvement in aid distribution, cannot provide impartial reports on human right violations given the political stakes, failed in providing protection to civilians, suggestively created adverse implications for development and reconstruction, and unlike aid organisations would not possess the unique cultural and technical expertise that is particular valuable for peace operations (Gheciu, 2011). The existence of such a struggling relationship between NATO and the aid community is not entirely surprising. NATO is a military organisation that is structured and hierarchical whereas aid organisations might adopt a less focused or efficient approach and tend to attach less importance to structure and hierarchy (Gheciu, 2011).
There are many other forms in which the militarization of aid occurs. Historically this is not a new phenomenon and has not always been unwanted by aid organisations. This is best described by Pérouse de Montclos (2014) who divides the militarization of aid into four categories.
The first category is related to the discussion in the previous sub-section. Self-defence is a form of militarization of aid which includes aid organisations seeking safety and security with national, public, or private militaries. As discussed, the rise of so-called ‘new wars’, which is a conflict characterized by fighting non-state actors, has contributed to this development. In the past this self-defence has triggered even more extreme forms. Aid charities have resorted to force and have adopted their own armies in order to protect themselves (Pérouse de Montclos, 2014). A textbook example is the Order of Malta whose knights have fought many battles on both land and sea.
The second category is called ‘international brigades’ which refers to the political leaning of relief organisations which choose to help only one side to a conflict and mix military support with relief operations. In this perspective aid organisations are not neutral or impartial to a conflict, but instead show solidarity to a party. This enables aid organisations to gain access to restricted (guerrilla) zones and assist communities that could otherwise not be reached. On the other side such organisations legitimize violence and support forces which are not necessarily representative of the people. It could even shift their activities from aid to military orientated (Pérouse de Montclos, 2014). It has been reported that some of these organisations have supplied arms and other war materials. A well-fitting example is the Norwegian People’s Aid (NPA) which openly
supported specific parties to certain conflicts, such as the Kurds in Iraq, and supplied war material to rebels in South Sudan (European-Sudanese Public Affairs Council, 1999). A third category is related to war by proxy. Relief organisations have been used by governments to support allies. Those allies receive material aid, financial support, weaponry and military training from government-supported aid agencies. A recent example is the Syrian Civil War which is said to show the continuation of the post-Cold War proxy war between the United States and Russia (e.g. Pengelly, 2015). The United States openly supports the Free Syrian Army and provides them with weapons, equipment, and financial aid. The financial and logistical support is provided by the Syrian Support Group which is a Washington-based NGO (Rozen, 2012).
A final category is the embedment of aid agencies in the military. The role of these organisations is to save the lives of their nationals on the battlefield or to win the ‘hearts and minds’ of civilians living in occupied territories (Pérouse de Montclos, 2014). The most famous example is the Red Cross whose national associations were embedded in national militaries to aid the wounded, in particular during World War One (Hutchinson, 1996).
Whereas the first two categories are based on the perspective of aid organisations, the latter two categories focus on the perspective of governments. While the described forms of militarization all blur the lines between aid delivery and military activities for different reasons or objectives, it is the last category that encompasses an important distinction. It refers to the mixture of aid with a military, or rather counterinsurgency, strategy that is focused on the civilian population (i.e. ‘hearts and minds’) while the other categories describe the relation between aid organisations and public militaries or the political commitments of aid organisations to parties to a conflict. The aim is to address insurgent violence against civilians and therefore the text will focus on the dangers related to using aid as a counterinsurgency strategy.
2.5 Hearts and minds
The discussion encompasses an important point: aid organisations seek safety and security while trying to avoid the political agendas of governments. But it is not only aid organisations that aspire cooperation between aid and the military in certain contexts. National militaries have used so-called ‘hearts and minds’ campaigns as a counterinsurgent strategy. By adopting a population-centric approach that includes soft (e.g. securing basic needs, providing security, and propaganda that particularly focuses
on the cruelty of insurgents) rather than hard (e.g. harsh regulations or sanctions) measures, the military aims to penetrate local social networks in an attempt to isolate the insurgency. The local population is promised to receive aid in exchange for information and intelligence on the insurgents’ whereabouts in order to guarantee safety and security for its troops. The idea is to defeat the insurgency by building on the trust and confidence of the population (Berman et al., 2008).
The modern concept of hearts and minds found its origin in the British counterinsurgency war in Malaya between 1948 and 1960. After encountering fierce resistance of Malayan insurgents and some brutal conducts of British forces and their allies, the security situation worsened and the Malayan population increasingly sided with the insurgents. As a response the British developed and implemented the ‘Briggs Plan’, based on a memorandum of Lieutenant General Sir Harold Briggs, in 1952 that focused on population control and security instead of adopting an enemy-centric approach. This plan changed the British counterinsurgency doctrine and, due to its success, has laid the foundation for many other counterinsurgency operations of the British (e.g. Kenya 1952-1959, Cyprus 1955-1952-1959, Aden 1963-1967, Oman 1962-1975) and other foreign armies (Valeyre, 2012). The strategy did not remain a Western concept either. Russia, for instance, has applied similar military campaigns during the Afghanistan War in the 1980s by supplying aid to the Afghan population (Robinson, 2010).
The United States, convinced by the successes of the British military, adopted the hearts and minds doctrine during the Vietnam War. In 1967 the United States launched the Civil Operations and Revolutionary Development Support (CORDS) that would oversee all Civil-Military Cooperation (CIMIC) projects and was jointly managed by the South Vietnamese government. CORDS aimed at providing security to the rural population in order to isolate them from insurgencies (i.e. Viet Cong and North Vietnamese People’s Army) and implementing socio-economic projects to gain support for the South Vietnamese government (Herring, 2014; Valeyre, 2012). Wherever the program was implemented it appeared to be successful as shown by a decline in the insurgency’s influence. However, the program was overshadowed by the extreme levels of violence used in a war that had grown costly in terms of casualties. The war was considered to be truly military of nature and CORDS was quickly abandoned after the United States troop withdrawal in 1973 (Valeyre, 2012).
Having learned from the lessons of the Vietnam War, the United States adapted the hearts and minds doctrine to counterinsurgency operations in Afghanistan (e.g. the
McChrystal Plan) and Iraq (e.g. the Petraeus Doctrine) in the 21st century. The aim was
to tap into social networks and isolate the insurgency which rely on their pre-conflict established network. The objectives were to secure the environment by ending the insurgents’ presence, regain political control, restore the order of law, rebuild national institutions, and gain support of the local population. Soldiers were instructed to network with community members, identify their basic needs, and establish a band of trust. The population-centric approach was based on a continuous presence, partnerships with the community, and investing in security, reconstruction, and development (Valeyre, 2012). The hearts and minds doctrines adopted in Afghanistan and Iraq emphasized the need to train Afghan military and police forces to fight corruption, intimidation, and violence. International security forces, such as the International Security Assistance Force (ISAF), were meant to shift away from the ‘search and destroy’ doctrine to a much more population-centric and self-sustaining approach (Valeyre, 2012). The United States army even deployed the Human Terrain System (HTS)8 that introduced teams with specialists working in social sciences such as anthropology, ethnography, geography, and languages. The objective of these teams was to create a better understanding of the local customs and habits which could prove to be useful information for building trust among tribes or gaining control over areas (CEAUSSIC, 2009).
Although a historical strategy throughout (Pérouse de Montclos, 2014), most literature emphasizes the dangers of carrying out campaigns that mix, or rather ‘confuse’, the counterinsurgency operations with aid delivery. Some of the contra arguments relate back to the discussion in the previous subsection. Hearts and minds campaigns blend aid with military or political objectives thereby potentially affecting the local perception of civilians and insurgents on the impartiality and neutrality of humanitarian actors (Fast, 2010; Krahenbul, 2011; Moseley, 2009; Williamson, 2011). In the worst case scenario aid agencies are perceived as working under the umbrella of the invading forces and become actual targets, especially when these agencies obtain funds from local or foreign governments or when governments exercise some form of control over how and where aid is distributed. Such a perception of loss in neutrality is difficult to undo and can harm the aid community as a whole (Williamson, 2011).
These increased security concerns can affect the regular flow of aid to populations
8 The program was heavily criticised by the American Anthropology Association for mixing anthropology with counterinsurgency intelligence, collecting socio-cultural sensitive data, putting academic researchers in danger, among others (CEAUSSIC, 2009).
in dire need of assistance if access to these areas is restricted (Krahenbul, 2011). It can also affect long-term projects if they are put on hold due to the increased danger. Although the military can complement these organisations by delivering aid in high risk areas, it could become a vicious circle if these high risk areas are due to their own hearts and minds conducts. Suppose the military uses aid to win local support in order to create stability and security, but this leads to more insurgent violence towards aid workers then it requires the military to deliver more aid to the region because of reduced humanitarian access. Even though the military could provide expertise on delivering aid, traditionally their priority is combat or constructing temporary infrastructure. While often it is assumed that aid workers possess soft skills with historical and cultural knowledge combined with technical expertise required for effective long-term development (Moseley, 2009). A positive relationship between the delivery of aid services and violence does not necessarily imply that the service provision itself worsens the situation. In fact, the optimal distribution of aid may dictate it to be delivered where it is expected to have the most violence-reducing effect (Berman et al., 2008). Using aid to consolidate military activities can lead to reduced insurgent violence in areas controlled by pro-government forces beforehand (Sexton, 2016). It remains important for governments and aid agencies to acquire better intelligence on locations where investments in service provision will yield the highest returns in terms of social order, program effectiveness, and reduced violence (Berman et al., 2008). Of course, acquiring such intelligence is time costly and depends on the time frame of the supplier. Aid organisations and government are unlikely to conduct such inquiries if the need for aid is urgent.
Even if hearts and minds might actually proof that security and benign actions are complementary, it becomes irrelevant if actors such as militaries, aid agencies or governments do not have the capacity or competence to act (Berman et al., 2008). For example if there are no resources to access a restricted area, providing (emergency) aid or setting-up projects becomes nearly impossible, which applies as well to improving security. A lack of security in combination with institutional disorganization enhances the ineffectiveness of the system which could fuel criminal or terroristic activities. Insurgencies will use this failure and the occupation of the coalition as propaganda. Fundamental to defeating an insurgency is to ensure the population that the government alone has authority over the people (Ford, 2005). This confidence is fostered by executing traditional state functions in an honest, just, and competent way while it is undermined
by the weaknesses of the government.
2.6 The risks for the local population
Unfortunately the story is not over yet. The militarization of aid has negative implications for both aid organisations and civilians. Especially hearts and minds campaigns, which embed the delivery of aid into a counterinsurgency strategy, puts civilians between two fires. Both insurgents and counterinsurgents understand that the local population plays a determinant role in defeating the insurgency (Ford, 2005). The counterinsurgent aims to persuade civilians to their side in order to receive useful information on the insurgents’ whereabouts by offering soft measures such as aid. It could harm civilians if insurgents alter their view on civilians and use violence to punish the beneficiaries of this aid. Insurgents might label all civilians living in a certain village, city or region as ‘traitors’ if they feel that they are losing authority (Williamson, 2011; Wood & Sullivan, 2015). A response is to use indiscriminate violence to any of those civilians in an attempt to remain in control after spreading fear (Hirose et al., 2017). Such attacks could be amplified if insurgents consider not only aid delivered by the military to challenge their local authority, but rather all aid delivered independent of the supplier. Like aid organisations the local population could become a target of insurgents and as a consequence more aid delivery could mean more insurgent violence.
Civilians might also not be used to receive aid, especially those living in hard-to-reach areas. For them receiving aid would not necessarily mean cooperating with a (foreign) military but rather reaping the benefits, in particular if the cost for cooperation becomes too high (e.g. would result in kidnapping, extortion, or death). This appears to be true for reconstruction projects in the more restless areas with frequent attacks of insurgents. Civilians prefer to remain passive to avoid diminishing their personal safety while gaining the benefits of the projects (Ford, 2005).
The delivery of aid could even distort the community. As discussed before, insurgencies sometimes cooperate with organised crime to generate a continuous income stream to finance their operations. Looting and aid predation is a reason for agencies to increase their security, while civilians remain unprotected and thus are easier targets. Delivering aid to civilians could therefore lead to insurgents targeting and abusing the local population for financial gain (Wood & Sullivan, 2015). Insurgents engaging in criminal activities might encourage aid delivery to the local population to maintain a certain income (Hutchinson & O’Malley, 2007, LeRiche, 2004).
It is clear that the tactics of insurgencies are meant to have an adverse effect on the counterinsurgency strategy. The tactics applied are meant to improve or maintain the insurgency’s position relative to the counterinsurgent, but affects civilians since they play a crucial role in securing this position. Tactics of insurgents include (i) punishing villages that support the counterinsurgent, (ii) forestalling the loss of neutral or ‘swinging villages’ by using indiscriminate violence (e.g. against counterinsurgent-supporting villages to state an example), and (iii) creating the perception that the counterinsurgent is not able to help or defend the local population (Hirose et al., 2017). The problem is that for insurgents neutrality among civilians is enough to hide themselves, while the counterinsurgent needs positive support of the population to acquire cooperation (Ford, 2005). This means that the counterinsurgent is already one step behind from the start and it is not surprising that they raise the benefits for civilians by providing employment through cash-for-work programs or increasing their troop strength in an attempt to grow confidence (Hirose et al., 2017). A neutral civilian will not provide information on the insurgents’ whereabouts if the cost of sharing information outweigh the benefits of cooperating. In particular the ‘costs’ relate to how protected civilians feel by the counterinsurgent. This suggests that aid programs should not be intertwined with military campaigns if the military cannot credibly protect civilians.
The discussion above is best illustrated by the counterinsurgency operations in Afghanistan and Iraq because much of the debate in the literature is based on evidence from the Global War on Terrorism in those countries. The United States reapplied the strategy after reshaping it based on lessons learned from the Vietnam War. A ‘modern’ counterinsurgency strategy was created for both countries, but this strategy was heavily criticised for similar problems discussed before. Among the biggest problem was aid delivered by the United Stated-led coalition to convince civilians to cooperate. Therefore a case study of those countries seems appropriate to inquire whether the relation between aid delivery and insurgent incidents is indeed positive.
2.7 Afghanistan and Iraq
The following subsections provide the reader with a short recap on the time line of the conflicts in Afghanistan and Iraq and the counterinsurgency operations. It is not meant to provide the reader with an extensive analysis, but to complement the methodology in the next section.
2.7.1 Counterinsurgency in Afghanistan (2001 – present)9
Following the 9/11-attacks in 2001, a United States-led coalition invaded Afghanistan under code name Operation Enduring Freedom. The initial objective was to dismantle Al-Qaeda and to remove the Taliban from power which provided a safe haven for Al-Al-Qaeda, the suspected attacker of the 9/11-attacks. Its objectives were rapidly accomplished at the end of 2001, after which most of the fighting shifts to the mountains. Through the Bonn Agreement an interim-regime was installed and the International Security Assistance Force (ISAF) was established to maintain security. Over the following years ISAF gradually assumed command over the international forces, mainly provided by NATO, to address security matters in Afghanistan.
In the meanwhile the United States’ military achieves to establish a framework for redevelopment and reconstruction in coordination with the United Nations (U.N.) and nongovernmental organisations. Simultaneously the United States and NATO commit to contribute to the training of Afghan police and military forces. These programmes are named Provincial Reconstruction Teams (PRTs) and its command was eventually handed to NATO. However, PRTs are heavily criticised for reasons similar to those mentioned before. PRTs would (i) blur lines between combatants and aid workers, (ii) lack coordination and civilian representation, and (iii) over-represent military organisations which prefer focusing on security or combat matters (see e.g. Perito, 2005).
Multiple other frictions occur during this long-lasting counterinsurgency mission. For example, insurgency attacks dramatically increase in 2006 for which ISAF is partly blamed for lack of assistance in security matters and the Afghan government for struggling with setting up a stable police force. Additionally, there are multiple reports of the United States’ military unintentionally killing civilians during combat, but this ‘collateral damage’ is used as propaganda by the Taliban.
At the end of 2014, the Obama-administration withdraws all United States’ troops from Afghanistan (United States Department of Defence, 2014) and ISAF combat operations are terminated. Security is left in the hands of national forces, but starting January 2015 counterinsurgency continues through missions like Operation Freedom’s Sentinel which operates under the umbrella of the NATO-led Resolute Support Mission
9 This recap of the Afghanistan War is based on a publication of the Council on Foreign Affairs. Visit their website for a more detailed chronological description of the Afghanistan War (Council on Foreign Affairs (n.d.A). U.S. war in Afghanistan. Retrieved from https://web.archive.org/web/20150302014738 /http://www.cfr.org/afghanistan/us-war-afghanistan/p20018).
that prioritizes on training, assisting and advising Afghan forces (North Atlantic Treaty Organization, 2018a). At the time of writing this mission is still operational.
2.7.2 Counterinsurgency in Iraq (2003 – 2011 and 2014 – present)
In March 2003, a United States-led coalition invaded Iraq to overthrow the regime of Saddam Hussein and to search and destroy Weapons of Mass Destruction (WMDs). The Iraqi army was quickly overrun and Saddam Hussein was later captured and sentenced to death by hanging, but WMDs were never found. Over the years violence in the country originated from former Hussein-supporters, who turned against coalition and Iraqi security forces, and Al-Qaeda which deliberately targeted those groups cooperating with United States-led coalition forces. Attempts to fight the insurgents were successful, but occasionally resulted in many civilian deaths (Council on Foreign Affairs, n.d.B). During the invasion and subsequent occupation the U.S.-led coalition forces use humanitarian emergency and development aid under the umbrella of the so-called Commander Emergency Response Projects (CERP) which forms a “non-lethal weapon” to assist the local population and provides funds for the PRTs in Afghanistan (Department of the Army, 2017). However, during the occupation many Iraqis complain about the lack of security and provision in basic services such as electricity, water and sewage. On top of that, many Iraqis are opposing the coalition forces due to the high civilian death toll during counterinsurgency operations and because they simply do not want to live in occupation (see e.g. Shehata, 2005; O’Hanlon, 2006).
In 2011, the United States announces the end of the Iraq War and withdraws most of its troops from the country. A remainder stayed to continue the coalition’s main goal to train and equip Iraqi security forces projected to take over security in their own country. Many feared that these forces would be unable to establish credible security and stability in a country that had become a breed nest for insurgents. This became reality when insurgency and sectarian violence increased after the departure of the coalition forces (see e.g. “Iraq profile – timeline”, 2018).
In 2014, a United States-led coalition returned to Iraq under Operation Inherent Resolve to respond to the rapid up rise of Islamic State of Iraq and the Levant (ISIL) that had taken control over a large territory in Iraq and Syria. Fighting was intense, especially in cities like Mosul where insurgents could easily hide, but in December 2017 the coalition, together with Kurdish and Iraqi forces, successfully recaptured all territory from ISIL. However, the insurgents went underground thereby forcing counterinsurgency
operations to continue up to the time of writing (see e.g. Najjar, 2019; Chulov, 2019). Since the defeat of ISIL at the end of 2017 reconstruction of Iraq has begun. Unfortunately the process advances slowly due to many human corpses, explosives and booby-traps left behind by the war(s) (see e.g. Whiteside, 2019).
3.1 Methodology
As explained in the literature in the previous section, the delivery of aid can contribute to an increase in insurgent attacks directed at civilians for simple motives like financial gain or more comprehensive reasons like the fear to lose local authority. The aim of this research is to examine whether this positive relation exists and is statistical relevant by constructing and testing a model. The model will test two hypothesis. First, the number of insurgent attacks is expected to be positively correlated to the amount of aid disbursements. ‘Insurgent attacks’ refer to events during which insurgents attempt killing, wounding or kidnapping national civilians. By using the word ‘attempt’, the definition highlights that such attacks do not necessarily have to be successful for insurgents. Failed attacks are included to capture the complete picture of violence against civilians. Second, this research will test whether aid delivered during the counterinsurgency years in Afghanistan and Iraq is positively related to violent incidents of insurgents. While the first hypothesis is rather general and applied to the whole dataset, this hypothesis is specifically focused on the case study in the previous section.
The established methodology is partly based on the work of Eck and Hultman (2007) who perform an event time series regression using a logit regression model for the incidence of one-sided attacks and a negative binomial regression model for the number of one-sided attacks on civilians. They included all actors involved in armed conflict between 1989 and 2004 and focus on conflicts with at least 25 battle-related deaths which follows the standard of the Stockholm International Peace Research Institute (SIPRI). Variables for their regression are derived from the work of Harff (2003) who searches for explanatory factors related to genocide and finds some of these factors to be statistically relevant for explaining conflict. Eck and Hultman (2007) code a dummy variable for previous conflicts in the country, a dyadic dummy for civil war to reflect the intensity level of the conflict between parties, a scale polity dummy variable between -10 and +10 describing the level of autocracy, a scale dummy variable between -10 and 7 reflecting the level of democracy, the countries’ openness to trade, whether the aggressor is a government, and finally include a lagged dependent variable to correct for non-independence among the time-varying observations.
In this research the methodology is similar, but slightly different. The approach is distinctive from other research because it attempts to explain the relationship through the eyes of insurgents rather than counterinsurgents. This has several implications for the model and data selection procedures.
The model of Eck and Hultman (2007) is modified for four reasons. First, a dated panel regression is applied to examine the effects of counterinsurgency operations in Iraq and Afghanistan only while Eck and Hultman (2007) use a panel regression for all actors involved in armed conflict between 1989 and 2004. The cases of Afghanistan and Iraq are good examples of the current hearts and minds campaigns of the United States, but have been heavily criticised. Unfortunately quantitative research on these cases through the lens of insurgents seems to be absent and therefore this research aims to verify whether the findings in the literature are confirmed. However, focusing on ‘just’ Afghanistan and Iraq leads to the problem that the number of observations is too small in order for this research to be statistical relevant. As a solution the panel dataset includes most countries of sub regions South and Central Asia and the transcontinental Middle East which are the regions Afghanistan and Iraq belong to, respectively. Next, following the description of the case studies, a counterinsurgency dummy variable is coded ‘1’ for Afghanistan and Iraq for the years during which the coalition forces conducted counterinsurgency operations, and ‘0’ otherwise. The dummy variable also includes ‘gap years 2012 and 2013 for Iraq to correct for the rise of many insurgencies and the continuation of aid distribution. After all, Iraqi troops were still trained by coalition forces during those years so they could take over their own national security.
Second, Eck and Hultman (2007) also consider governments to be a potential aggressor while the model in this research is purely aimed at insurgent violence. The only relevant aggressors who commit violence against civilians in this dataset are insurgents which implies that a variable for whether the aggressor is a government is unnecessary and is omitted from the model. On the side this seems supported by the results of Eck and Hultman (2007) which show the variable to be insignificant.
Third, following Eck and Hultman (2007), variables for polity and trade openness are included in the model, but a scale dummy to reflect the level of democracy is omitted to avoid overfitting and because it is already captured in the polity dummy variable. The variables for polity and trade openness are not necessarily important, but are merely added to increase the predictability of the model. Furthermore, the model includes the dyadic conflict intensity dummy variable, but excludes the dummy variable for previous conflict since the former proves to be more relevant and is meant to replace the latter. The dyadic conflict intensity dummy variable is based on battle-related deaths and is ‘1’ for a minor
