The likelihood of local allies free-riding: Testing economic theories of alliances in US counterinsurgency interventions

Abstract

In counterinsurgency interventions, free-riding by small, local allies is persistent. Yet, the literature on free-riding by small allies is largely limited to conventional multilateral partnerships, such as the North Atlantic Treaty Organization, neglecting other types of asymmetric alliances. Using new data containing 144 US requests to local allies in Vietnam, Iraq and Afghanistan, this article tests the logic of economic theories of alliances in counterinsurgency interventions. I find even when small allies are explicitly asked to contribute to alliance-wide security goods, they are likely to free-ride almost half the time (45%), and the likelihood of free-riding is dependent on whether local allies can be excluded by larger allies. This conclusion upholds the logic of economic models, since shared defense goods that exclude local allies fail to meet the criteria of public goods.

Keywords

Afghanistan alliances burden-sharing counterinsurgency free-riding Iraq military intervention Vietnam

It is largely assumed that small allies will fail to carry their share of collective burdens, instead opting to free-ride off efforts of larger security partners (Oneal, 1990b; Russett and Sullivan, 1971; Sandler et al., 1980; Sandler and Hartley, 2001). Olson and Zeckhauser instigated a prolific debate by modeling how large states, specifically the USA, would ‘bear a disproportionately large share of the common burden,’ while smaller members free-ride (Olson and Zeckhauser, 1966: 269; Sandler, 1993; Sandler and Hartley, 2001). Although economic theories of alliances traditionally model asymmetric, multilateral security alliances like the North Atlantic Treaty Organization (NATO), commentary on security alliances in counterinsurgency interventions often describe similar free-riding behaviors in their asymmetric alliances between foreign intervening forces and local partners. Despite potential similarities, there has not yet been a systematic analysis comparing the logic of economic theories of alliances and burden-sharing in counterinsurgency coalitions. This article fills this gap, examining the US counterinsurgency wars in Iraq, Afghanistan and Vietnam, specifying the applicability of economic models and analyzing factors that make local free-riding more, or less, likely.

Analyzing free-riding is imperative to understanding long-term outcomes in counterinsurgency interventions. The failure of local partners to carry their share of the burden is a strategic liability. Free-riding hinders local military and governance institutions, potentially affecting the duration, cost or even outcome of the war (Biddle, 2008; Cohen et al., 2006; Cooper, 2009; Elias, 2013; Gompert and Gordon IV, 2008: xlv–lvii, 344; Jaffe and Morris, 2015; Riedel, 2012: 92; Watts et al., 2014). Drawing on existing theories of asymmetric burden-sharing not only enhances understandings of free-riding in these wars, but further expands larger understandings regarding collective goods. As Oppenheimer (1979: 390) argued, testing the range of application for Olson’s model is important for identifying assumptions within the theory.

To test the applicability of economic theories of alliances in counterinsurgencies, I rely on new data from the US wars in Vietnam, Iraq and Afghanistan. Specifically, I examine 144 policy requests from Washington to local allies documenting attempts by the USA to create alliance-wide cooperation against the insurgency. This set of requests does not capture the universe of opportunities afforded to in-country allies to free-ride. However, limiting analysis to US requests for cooperation (a) eliminates instances of independent US action apart from collective joint security efforts and (b) adopts a tough-test case selection where allies were specifically asked to participate, and not to free-ride. This approach identifies the likelihood of free-riding even when a larger ally requested local assistance. As discussed in the methods section, not all instances of non-compliance by small allies with US requests are cases of free-riding, since non-compliance could be motivated by multiple factors, including disagreement with the policy being proposed. Free-riding, on the other hand, is defined as small allies benefiting from the provision of an alliance-wide good aiming to harm the insurgent enemy, or bolster collective counterinsurgency capabilities, without contributing to the production of that security good.

Supporting the logic of economic theories, I hypothesize that free-riding by smaller counterinsurgency allies is more likely the more closely requests from larger allies resemble alliance-wide public goods. Conversely, proposed policies that do not resemble public goods are less likely to result in free-riding. A public good is defined as a commodity or service that is both non-rival (one actor’s consumption does not reduce availability) and non-excludable (no actor can exclude others). This study tests non-excludability, as a first step in examining burden-sharing in counterinsurgency contexts.

These hypotheses build on Sandler’s (1993) ‘joint-products model,’ specifying that Olson’s theory will be likely confirmed, so long as the alliance-wide product in question meets the criteria for being a public good (Murdoch and Sandler, 1984; Oneal and Diehl, 1994; Sandler and Hartley, 2001). Yet, importantly, not all inter-alliance defense products produced in alliances will qualify as alliance-wide public goods and, in those cases, economic models of alliances are less helpful for explaining the behavior of local allies. Appreciating these dynamics can potentially enable policymakers to predict what requests are more likely to result in free-riding, and which are more likely to inspire local participation.

I offer three primary findings. Firstly, even when local allies were explicitly asked by American officials to contribute, they were likely to free-ride at a rate of 45.1%. Secondly, there is significant heterogeneity in alliance-wide security goods, as some requested polices closely resemble public goods, while others resemble private goods. This finding suggests that economic models are helpful for describing the behavior of small allies under certain circumstances, but are insufficient in explaining other significant factors affecting the behavior of small security partners in counterinsurgencies. Thirdly, different policy subjects are associated with different rates of free-riding. Development projects in Iraq and Vietnam were likely to result in free-riding, while requests for political reforms designed to strengthen alliance-wide counterinsurgency efforts were less prone to free-riding.

Free-riding and burden-sharing in alliances

Olson and Zeckhauser (1966) offered an initial model explaining why the costs of defense goods shared between allies would not be distributed proportionally based on the security benefits afforded each member. Smaller allies often free-ride (Olson and Zeckhauser, 1966). Among others, Sandler and Hartley (2001) tested this economic theory of burden-sharing, finding smaller states take advantage of insufficient resources, or an enhanced threat to convince larger states to, ‘shoulder more, if not all, of the burden of defense’ (p. 875).

In addition, Plümper and Neumayer (2014) refute the assumption underpinning economic models that the smaller the ally (measured by gross domestic product (GDP)), the more likely it is to free-ride. This has implications for counterinsurgency alliances. Consider, for example that Afghanistan’s 2014 GDP of US$20.7 billion was 0.1% of US GDP. This disparity might lead to assumptions of unrelenting Afghan free-riding, but as the data offered here demonstrates, there is variation in the prevalence of free-riding, supporting Plümper and Neumayer’s findings.

Furthermore, the logic of burden-sharing rests on the contested assumption that defense among allies is a public good. However, defense goods in alliances have been found to be neither consistently non-rival, nor non-excludable (Goldstein, 2007; Oneal, 1990b; Oneal and Diehl, 1994; Sandler, 1977, 1993; Sandler and Forbes, 1980; Sandler and Hartley, 2001). Instead, there are multiple categories of defense goods in alliances, such as pure public, impure public, private, mixed, multi-use or the use of a joint-product model (Oneal and Diehl, 1994; Plümper and Neumayer, 2014; Sandler and Forbes, 1980; Sandler and Hartley, 2001).

In this study, the public good shared between allies is broadly defined as activities that weaken the insurgent enemy. For example, US military forces combating communist forces in Vietnam were providing a shared public good for the US–South Vietnamese counterinsurgency effort. Saigon could potentially benefit from US military activities aiming to weaken communist insurgents, without directly participating. Saigon could not be excluded from benefiting from US efforts to weaken insurgents, nor would Washington’s consumption diminish Saigon’s consumption of the same good.

Counterinsurgency alliances are notorious for leaving larger partners like the USA with little leverage over seemingly weaker partners. Robert Keohane (1971) described this surprising lack of coercive opportunities for large states as ‘the big influence of small allies.’ This phenomenon was repeatedly lamented in reflections on the war in Vietnam (Cooper et al., 1972; Gravel, 1971; Komer, 1972; Shafer, 2014), detailed in US counterinsurgency alliances in El Salvador (Schwartz, 1991) and noted in the US wars in Afghanistan and Iraq (Berman, 2010; Byman, 2006; Chaudhuri and Farrell, 2011; Elias, 2013; Felbab-Brown, 2013; Maley, 2011; Schaffer, 2002). Policymakers and scholars alike have noted that securing cooperation from local counterinsurgency partners is exceedingly difficult, yet is often curiously glossed over (Biddle, 2008).

The applicability of collective goods theory to counterinsurgency interventions

Although instructive as a foundational approach, there are critical caveats for applying economic models to explain free-riding in counterinsurgencies (Murdoch and Sandler, 1982; Noetzel and Schreer, 2009; Olson and Zeckhauser, 1966; Oneal, 1990a; Oneal and Diehl, 1994; Plümper and Neumayer, 2014; Sandler and Forbes, 1980; Thies, 2002). In this section, I describe four particular reasons why. Many of these issues are familiar to critiques of economic models, but merit specific exploration in counterinsurgencies.

Counterinsurgency alliances provide defense, not deterrence

Economic models tend to focus on alliances that rely on deterrence (using capabilities to threaten an enemy) rather than defense (using capabilities to combat an enemy). This makes a difference, as deterrence remains ‘a relatively pure or inclusive public good,’ whereas with physical defense ‘allies are partially rival’ (Oneal, 1990a: 428). This finding provides caution. Some, but not all, inter-alliance counterinsurgency activities aiming to weaken the enemy are providing public good readily shared between allies.

For example, the USA would be unable to exclude Saigon from benefiting from military operations against Hanoi. However, Washington might be able to exclude the South Vietnamese from benefiting from building an anti-communist labor party designed to weaken the allure of insurgency communist parties (Cooper, 1965). Even as this action aims to weaken the insurgency, the new organization could create political rivals to challenge, not bolster, Saigon’s control, effectively excluding Saigon.

In addition, defense may fail conceptually as a public good because allies can be excluded entirely if abandoned. As Goldstein (2007: 25) argued, ‘the fear of abandonment reflects the fact that alliance security is a benefit that actually fails the key public goods test of non-excludability.’ This fear may be particularly acute in non-colonial counterinsurgency interventions where intervening forces are avowedly temporary missions (Elias, 2013).

Multi-player versus two-player games

Economic models evaluate large multilateral security organizations (Murdoch and Sandler, 1982; Noetzel and Schreer, 2009; Olson and Zeckhauser, 1966; Oneal, 1990a; Oneal and Diehl, 1994; Plümper and Neumayer, 2014; Sandler and Forbes, 1980; Thies, 2002). Interestingly, in counterinsurgency interventions there are similarly multiple allies, including the USA, Germany and the UK in Afghanistan. Also present in the counterinsurgency coalition are multiple local groups, such as the Shi’a government in Baghdad and Kurds in northern Iraq. However, in order to thoroughly investigate one critical, under-examined part of these layered alliances, this study focuses on bilateral negotiations between the USA and local regimes.

The literature specifying potential bias introduced by modeling multi-player events, such as alliance formation, as two-player games, finds that with sampling techniques dynamics between multi-player and two-player games are comparable (Poast, 2010; Signorino, 1999). Indeed, the practice of modeling multilateral alliances bilaterally is a longstanding practice in international relations (Croco and Teo, 2005; Fordham and Poast, 2014). Since I focus on an under-examined component of these multilateral alliances, namely the USA and the primary local ally, the potential bias introduced is minimal. Lastly, multilateral cooperation is typically assumed to be more difficult than bilateral cooperation, and free-riding more likely in multi-player games (Axelrod and Keohane, 1985; Kahler, 1992; Oye, 1985). Yet, the findings here indicate that free-riding is quite common even when modeled bilaterally.

Expanding concepts of allied contributions beyond financial contributions

In addition, economic models of alliances unsurprisingly tend to focus on economic indicators, namely financial contributions, but the literature often also considers non-monetary contributions, since the logic on burden-sharing holds regardless of the form of contribution. Consider, for example, decisions regarding contributing troops for peacekeeping missions (Bove and Elia, 2011).

The logic of providing goods that benefit alliance-wide security is consistent whether goods are financial or political. Political goods, such as legitimacy, may be more difficult to measure; however, limiting burden-sharing analysis to only financial contributions is false since, as Morrow (1991) noted, the very purpose of forming asymmetric alliances is often to trade economic support for political favor (Keohane, 1971; Snyder, 2007; Walt, 1987). Note that valuing political goods may have particular relevance in counterinsurgencies, where local politics is critical to war outcome and is often elusive to intervening militaries (Galula, 2006: viii).

Emergencies, defense as a ‘superior good’ and economic models

Olson and Zeckhauser (1966: 270) noted that ‘during periods of all-out war or exceptional insecurity, it is likely that defense is (or is nearly) a superior good, and in such circumstances alliances will not have any tendency toward disproportionate burden sharing.’ Yet, this concept of defense as a ‘superior good’ in wartime decreasing incentives for free-riding does not accurately describe counterinsurgency conflicts where, despite sustained combat, free-riding can be observed. In long, low-intensity wars, moments of emergency are the exception, creating space for free-riding despite ongoing conflict.

With these limitations in mind, the logic underpinning economic models of alliances is useful in a counterinsurgency context for explaining how decisions regarding defense levels made by allies will cause alliances to function below Pareto-efficient levels of collective allocation, ‘as the marginal benefits that an ally’s defense provision confers on the other allies is ignored’ (Sandler and Hartley, 2001: 873). States allied against an insurgency will look to alliances to optimize their own utility, as opposed to making efforts to optimize the utility of the alliance itself. Thus, decisions will aim to optimize what each state achieves from the alliance, not what the alliance achieves in the war. This simple observation speaks to the inefficiencies and the frustrations expressed by US policymakers as well as highlighting how the structure of inter-alliance politics can undermine collective security.

Theory – Free-riding in counterinsurgency interventions

I test two hypotheses drawn from the logic of economic models of alliances, focusing on the non-excludability requirement of public goods. The hypotheses follow the logic that policies more closely resembling public goods (non-rival and non-excludable) are more likely to produce free-riding by small allies compared to policies that fail to meet these criteria.

I do not test the question of rivalry as criteria for public goods for four reasons. Firstly, in an effort to build the capacity of small allies, Washington is likely to ask for rival goods. Secondly, rivalry is less of a concern when one ally, such as the USA, controls significant resources and is fighting a relatively ‘small’ war. Of course even for a hegemon, warzones are plagued by scarcity, but nevertheless moments of rivalry between allies at the USA and local allies for resources are uncommon. Thirdly, I focus on excludability in order to compensate for a gap in the literature, as existing critiques tend to focus on rivalry (Goldstein, 1995; Sandler, 1993). Lastly, excludability has been a critical factor identified in the counterinsurgency literature regarding alliance politics (Shafer, 2014: 118–120). As one RAND analysis noted regarding Vietnam, the ‘United States can exert significant influence on ally governments only if it can make credible to an ally regime that it has alternatives to collaboration’ (Menges, 1968, cited in Shafer, 2014: 119).

Hypotheses

(1a) Small allies are more likely to free-ride on counterinsurgency policies proposed by large allies that are non-excludable.

(1b) Development-oriented projects will be more likely associated with non-excludability, and prone to free-riding by local counterinsurgency allies. In contrast, US requests regarding political reform will be more likely associated with excludability, and therefore less prone to free-riding.

Hypothesis 1b is counterintuitive in light of work on foreign aid claiming development assistance is effective precisely because recipients can be rewarded for cooperative behavior (Alesina and Dollar, 2000; Berthélemy and Tichit, 2004; Maizels and Nissanke, 1984; Meernik et al., 1998). Yet, hypothesis 1b contends that excludability is less likely for requests pertaining to development. Why? Firstly, bolstering local allies is a long-term goal of US counterinsurgency efforts. As Walt (1987: 43–44) noted, ‘the more important the recipient is to the donor, the more aid it is likely to receive but the less leverage such aid will produce… Foreign aid is likely to be useful in manipulating allies that don’t matter very much.’ The importance of these local counterinsurgency allies to US security undermines the logic that development aid was a useful exclusionary tool. Development often stands at the center of efforts to affect hearts and minds, ideally boosting popular support for counterinsurgents. Did US policymakers feel at liberty to withdraw development support from local allies in Afghanistan, for example, if Kabul did not cooperate? As Former Ambassador Ronald E Neumann quipped, ‘The argument that we could pull out of Afghanistan if Karzai doesn’t do what we say is stupid… we couldn’t accomplish what we’ve set out to do. And Karzai knows that’ (Cooper, 2009).

Secondly, the data only analyzes US policies requesting local allied participation. Other instances where the USA implemented independent development projects likely enabled Washington to reward cooperative local contractors with development aid, as anticipated by the development literature. However, I contend that when development projects explicitly seek local participation, the USA loses this leverage, thus creating opportunities to free-ride.

Data, methods and research design

The study analyzes the behavior of local counterinsurgency allies, and the likelihood that these actors will free-ride in response to requests from foreign intervening allies. I focus on Vietnam, Afghanistan and Iraq for several reasons. Firstly, there is available and comparable data for these conflicts. Secondly, as an initial investigation into the dynamics of free-riding in counterinsurgency, I hold the intervening ally constant in order to investigate dynamics related to US counterinsurgency alliances. This is in keeping with a tradition in economic theories of alliances, usually seeking to explain how the USA in particular would carry a disproportionate burden. In future work the hypotheses offered can be tested in non-US cases.

To gather data, I constructed a list of US requests for cooperation, while tracking the outcomes of each request in order to note free-riding. To identify US requests, over 3000 US primary source documents (8000+ pages) were analyzed, yielding 56 unique demands from the US government to Vietnamese allies (1964–1973), 35 to Iraqis (2003–2010) and 53 to Afghans (2001–2010). Extending the analysis beyond 2010 was not possible because those materials are not public. These 144 requests represent only a portion of the total US policy requests identified. Overall, 359 requests were identified. However, in only 144 of these requests did the USA have an independent capacity to fulfill the request, if necessary. This American unilateral ability to fulfill these requests created the opportunity for allies to free-ride.

Table 1 provides the subject of each US request.¹ Each policy request from the USA is an observation. While not an exhaustive compilation of free-riding opportunities, these are largely representative of US Department of State requests to local allies, documenting where the USA sought local cooperation. Data are drawn from three sources.

Table 1.

US requests to local counterinsurgency allies.

Year	War – Vietnam	US request
1964		US involvement – Government of the Republic of Vietnam (GVN) Military
1964		Strategic Hamlet Program
1964		Operation Hop Tac
1964		Guerilla Force
1964		Port of Saigon
1964		Increased Compensation – Military
1964		Increased Compensation – Political
1964		Saigon Sanitation
1965		Propaganda and Diplomacy
1965		Youth Programs
1965		Montagnard Grievances
1965		Agree – Bombing Halt
1965		GVN Strike North Vietnam
1965		Social/Political Reconstruction
1965		Water Transportation
1965		Daily Press Briefings
1965		GVN Survey Capacity
1965		Increase Pay – Rural Teachers
1965		Train Broadcasters
1965		System of Rewards
1965		Inter-religious Council
1965		Anti-communist Labor Parties
1966		Credit to Farmers
1966		Security Stockpiles
1966		Elementary Schools
1966		Health Personnel
1966		Vocational Training
1966		Limit Spending
1966		Programs – Refugees
1966		Employment – Refugees
1966		Schools – Refugees
1966		Vocational Training – Refugees
1966		Seeds to Farmers
1966		Rural Electrification
1967		Project Take-off
1967		Veterans Benefits
1967		Public Sacrifice Campaign
1967		Transfer Land Authority
1967		Reorient Pacification
1967		Build Prisons
1968		Paris Negotiations
1968		Lien Minh
1969		Accept National Liberation Front (NLF) in Talks
1969		Accept US Withdrawal
1969		Compromise on Table – Talks
1969		NLF Talks
1970		Army of the Republic of Vietnam (ARVN) - Cambodia
1970		Drop Mutual Withdrawal
1971		ARVN – Laos
1971		Drug Problem
1972		Drop Final Paragraph – Talks
1972		Reunification Post-talks
1972		Concurrence Proposal – 1972
1972		Tripartite Commission
1973		International Conference
1973		Concurrence Proposal – 1973
2002	War – Afghanistan	Finding Taliban
2002		Provide Security
2002		Military Factions Under Government
2002		Preparation/Write New Constitution
2002		Women in Government
2002		Fund Ministry of Justice (MOJ) and Ministry of Interior Affairs (MOI)
2002		Poppy Elimination
2003		Anti-corruption Narcotics-related Programs
2003		Drug Intelligence and Interdiction
2003		Drug Treatment, Rehabilitation
2003		Poppy Ban
2003		Post-eradication Areas
2003		Ministry of Rural Rehabilitation and Development - Alternative Livelihood Projects
2003		Low-level Enemy Combatants
2003		Vocational Training
2003		Training in Counternarcotics
2006		Performance-based Reviews
2006		Assistance – Refugees
2006		Assistance – Disabled
2006		Border Management Initiative
2006		Census
2006		Disaster Response
2006		Costs – Elections
2006		Human Resource Study
2006		Monitoring - Human Rights
2006		Ministry of Counternarcotics (MCN) to Poppy Elimination Program (PEP) Teams
2006		Operate the Counter Narcotics Justice Center (CNJC)
2006		National Development Strategy
2006		Benchmark Progress
2006		Reduce Area – Mines
2006		Single National Document
2006		Cross-border Jirgas
2007		Counternarcotics Trust Fund
2007		Electricity Development
2007		Poppy Elimination – Helmand
2007		MCN Pay PEP Salaries
2007		Office Space/Personnel
2007		Information – Poppy/Narcotics
2007		Year-round Poppy Elimination
2009		Anti-corruption Commission
2009		International Community Compact
2009		Fund and Expand the Community Defense Initiative (CDI)
2009		Af-Pak Crossings
2009		Discourage Violence – Election
2009		Reject Dostum
2009		Operations in Marja
2009		Expand CDI
2009		Ministerial Conference
2009		Statement – Taliban Abuses
2009		Status of Forces Agreement
2010		Accept Services Package Approach and Funding Mechanism for District Delivery Program (DDP)
2010		Lessons Learned – DDP
2010		Oversight – NGOs/Charities
2005	War - Iraq	Reconstruction Fallujah
2005		Sunnis – Draft Constitution
2005		Increased Funds – Fallujah
2006		MOI – Human Rights Abuses
2006		Fund Militia Reintegration
2006		Assets from FREs
2006		Baghdad Security Plan
2006		Three Brigades -Baghdad
2006		Reduce Sectarian Violence
2006		US$10 Billion – Reconstruction
2007		Prison Guards
2007		Removal of Yellowcake – Tuwaitha
2007		Non-security, Post-surge Services
2007		Command/Control – Samarra
2007		Rule of Law Complex
2007		Kuwait – Fuel Tankers
2007		Process Detainees
2007		Prosecution of Officials
2007		Negotiated Ceasefires
2007		Fuel Meters
2007		Fund Agribusiness
2007		Vocational Training
2007		Microfinance
2007		Facilities – Fuel Imports
2007		Anti-Corruption Institute
2007		Sheekly – Protection
2007		Kurdistan Regional Government (KRG) - Refuge to Palestinians
2007		Funding Concerned Local Citizens (CLC)
2007		Incorporating CLCs
2008		National Energy Strategy
2008		Protect Central Bank
2008		Sanction Designees
2008		Lobby Russians – UN Resolution
2008		Defend Iraqi Christians
2009		Man-portable air-defense systems (MANPADS)
2009		Hydrocarbon Audit

The first, and the largest source of data for Vietnam, was declassified documents in the Foreign Relations of the United States (FRUS), published by the US Department of State. The second source was declassified documents released to The National Security Archive. The third, and largest source of data for Iraq and Afghanistan, was primary source US Department of State documents published by Wikileaks. A total of 6677 cables from US officials in Baghdad and 2961 from Kabul are contained in the released Wikileaks cable database, providing an extraordinary documentation of alliance politics. The Wikileaks cables contain a notable portion of US requests sent through the US Department of State. The US embassies in Baghdad and Kabul were the center of high-level diplomatic engagement with local allies. While not all interactions with allies are recorded in US State Department cables, and only a portion of State Department records were obtained by Wikileaks, the Wikileaks materials nevertheless contain detailed instructions from Washington to embassy staff on approaching allies, as well as reports from embassy personnel back to Washington, detailing inter-alliance bargaining.² In addition, as a precaution, US requests to allies identified in cables made available by Wikileaks were often matched with declassified documents and public statements to better ensure validity.³

Once identified, all 144 requests were coded for free-riding. Furthermore, since warzones can be demanding environments, small ally government capacity was included as a control variable, defined as a structural limitation within the governments in Saigon, Kabul or Baghdad related to the particular policy requested. Capacity was coded as a dummy variable relying on US assessments of institutional shortcomings in these regimes potentially affecting the implementation of the policy requested. If US documents, including materials from the Department of State, Department of Defense, the Central Intelligence Agency or Congress, noted shortcomings in an ally’s ability (as opposed willingness) to implement a US-requested policy, the request was coded as containing potential capacity shortcomings. Primary source US documents were more reliable than other potential sources of data, such as media accounts. Not all instances of allied non-compliance were coded as free-riding, since non-compliance could be motivated by multiple factors, including disagreement that the policy would help the counterinsurgency effort.

A control variable measuring US efforts to combat free-riding was not created for multiple reasons. Firstly, measuring efforts by large allies to prevent free-riding is not traditionally included in economic models (Sandler, 1993). Secondly, US efforts to minimize free-riding are already accounted for in the variable measuring excludability. US documents repeatedly note an inability to coerce local allies if in-country partners could not be excluded from the intended benefits, a factor accounted for by measuring excludability. In addition, issue linkages such as the US offering side payments, or sanctioning local allies for uncooperative behavior, did not regularly appear in the historical record provided by the US Department of State.⁴ Sanctions were unpopular due to the possibility that they could weaken the very regime the USA was struggling to support. In addition, the provision of side payments by US officials would not invalidate the primary arguments offered here, but note that the policy process of influencing local allies often involves multiple players, organizations, institutions, bureaucratic interests, interactions and possibly cash payments that create complex inter-alliance bargaining processes.

Similar to the process of identifying US policy requests to local allies, coding for free-riding as the dependent variable drew on primary source documents from the three sources previously listed, corroborated with other materials, such as documents from multiple US agencies, media reports, government press briefings and reports from international organizations. Free-riding is defined as receiving benefits of a policy aiming to harm the insurgent enemy, without contributing to the production of that policy. Observations were coded as instances of free-riding when they met three criteria: (a) the USA had unilateral capacity to fulfill the request, if necessary, enabling the possibility of local allies free-riding⁵; (b) evidence indicates that the USA, not the local ally, did the majority of the work to fulfill the policy; and (c) demonstrable benefit for the ally, or the counterinsurgency effort aiming to harm the insurgency.

Inspired by the literature on burden-sharing, the independent variable, excludability, measures if the in-country ally can be excluded from benefiting from the policy if implemented by the USA. Excludability was coded as positive or negative by analyzing if the USA could potentially prevent the local ally from benefiting from the policy outcome. For example, in 2003 the USA requested that Kabul establish anti-corruption, narcotics-related programs. As illustrated in the documents, Washington could not set Afghan government policy within Afghan institutions, yet in 2003 the USA had the unilateral ability to fund and operate an anti-corruption training program targeting Afghan civil servants. Therefore, since Kabul could be prevented from benefiting from the policy, the observation was coded as excludable. This policy produces something short of a shared alliance-wide public good.

An example where exclusion was not possible can be observed in US requests for Iraqis contribution to reconstruct Fallujah (U.S. Department of State, 2005).⁶ Under US pressure, Baghdad pledged a ‘meager,’ few hundred million dollars, which it failed to deliver in full (Schwartz, 2008: 115–117; U.S. Department of State, 2006). By 2008, Americans had taken primary responsibility for reconstructing Fallujah (Garrels, 2008). Baghdad could not be excluded from benefiting from US efforts, temporarily pacifying an insurgent hotbed. The request was coded as non-excludable, a policy producing something closer to an alliance-wide public good. An independent variable was created to test if free-riding and excludability were correlated with different policy issues. Six categories were created classifying US requests by subject. Figure 1 provides a chart illustrating the frequency of each. Development – activities intended to support economic growth and provide social services. Includes land reform, school construction and assistance to refugees (54/144, 37.5% of US requests). Economic Reform – actions intended to change economic policies. Includes limiting spending to prevent inflation and seizing assets from former regime elements (5/144, 3.5% of US requests). Political Reform – actions intended to change government policies and institutions. Includes policies toward opposition parties, counternarcotics and protocol for funding (53/144, 36.8% of US requests). Political-Military Counterinsurgency Strategy – effort intended to implement joint counterinsurgency strategy. Includes specific pacification projects (12/144, 8.3% of US requests). Military Reform – actions intended to change military policies and institutions. Includes treatment of captured insurgents, clarifying command and control in contested areas, and compensation (10/144, 6.9% of US requests). Military Strategy – actions intended to guide military forces in the execution of the war effort. In Vietnam these included putting forces into Laos, invading Cambodia and striking North Vietnam. In Iraq these requests focused on reducing sectarian violence (10/144, 6.9% of US requests).

Figure 1.

Subject of US requests to in-country counterinsurgency allies.

Results

Local allies exhibited free-riding behaviors at a rate of 45.1%. This is likely a conservative estimate of the frequency of free-riding incidences undertaken by small counterinsurgency allies, and is a remarkable for two reasons. Firstly, it indicates that even when collaboration is specifically requested from large allies, small allies are likely to free-ride anyway, almost half the time. Secondly, although the chance of free-riding is substantial, allies are slightly more likely not to free-ride (54.9%) when asked to cooperate, indicating there are influential factors preventing small counterinsurgency allies from free-riding, even when offered the opportunity.

The statistical results aggregating the three wars support both hypotheses. Hypothesis 1a postulated that policies regarding non-excludable goods would more closely resemble public defense goods, resulting in a greater likelihood of local free-riding. This hypothesis was supported by measurements showing that excludability was significant and negatively correlated with free-riding (Table 2). Note that the findings from Afghanistan failed to strongly support either hypothesis. This is an interesting result explored after the aggregate findings.

Table 2.

Local free-riding in Vietnam, Afghanistan and Iraq.

Variable	Model 1: combined wars	Model 2: Vietnam	Model 3: Afghanistan	Model 4: Iraq
Small ally capacity (control)	0.782** (0.230)	0.278 (0.402)	1.148** (0.373)	0.829 (0.573)
Excludability	–0.795** (0.232)	–1.525** (0.549)	−0.283 (0.371)	–1.808** (0.546)
	n = 144	n = 56	n = 53	n = 35

Indicates p < 0.05; ** indicates p < 0.01.

As illustrated in Figures 2 –4, hypothesis 1b is also supported by the aggregate data. Development-related requests were associated with a higher likelihood of non-excludability, increasing the likelihood of free-riding. Requests regarding political reforms were more likely to provide the US opportunity to exclude small allies, and associated with decreased incidence of free-riding.

Figure 2.

In-country ally freeriding and subject of US requests.

Figure 3.

Potential US ability to exclude in-country allies on development projects vs. other US requests.

Figure 4.

Potential US ability to exclude in-country allies on political reforms vs. other US requests.

Examples may be helpful. Consider Washington’s 2009 requests that Kabul discourage violence, affirm support and engage with but refrain from interfering with the Independent Election Commission (U.S. Department of State, Embassy Kabul, 2009a). Fearing exclusion, the Karzai regime opted against free-riding, instead participating in order to better influence outcomes. Karzai feared the USA was funding alternative candidates (U.S. Department of State, Embassy Kabul, 2009a, 2009b). In this instance, the USA was chiefly interested in producing legitimacy, while local allies were focused on maintaining control. Because requests for political reform are prone to allow for the exclusion of small allies (failing to meet the criteria for public goods) small allies are less likely to free-ride, and instead see participation as a way to better ensure they are not undermined (Figure 2).

Results for Afghanistan and validity of findings

Interestingly, although the data on free-riding from Afghanistan supports the proposed hypotheses, the findings are not significant (Table 2). Similarly, despite free-riding in Afghanistan being roughly as prevalent as in Iraq and Vietnam (Table 3), it is not evidently correlated with the subject of the request, thus failing to substantively support hypothesis 1b. Kabul engaged in free-riding on requests for political reform at a rate of 40.7%, as opposed to development projects at a rate of 50.0%. This evidence opposes the strong findings in support of hypothesis 1b in the data from Vietnam and Iraq.⁷

Table 3.

Frequency of local free-riding in Vietnam, Afghanistan and Iraq.

	Combined Wars	Vietnam	Afghanistan	Iraq
Prevalence of free-riding	45.1%	35.7%	47.2%	57.1%
	n = 144	n = 56	n = 53	n = 35

Free-riding is roughly just as common in Afghanistan as other conflicts examined, yet, it less closely adheres to the patterns of free-riding predicted in economic models. This indicates other processes are affecting free-riding decision in Afghanistan. Reports from the US Embassy in Kabul indicate part of this variation is likely rooted in Afghanistan’s notorious issues with corruption (United Nations, 2010). While Baghdad and Saigon engaged in extraordinary corruption, Kabul appears to uniquely emphasize maximizing profits above maximizing security. Sorting through the outcome of US requests in Afghanistan reveals a pattern where Kabul opted to participate (thus failing to free-ride) even when they could not be excluded from the security benefit of the policy, in order to access cash. The logic of Kabul participating in order to capitalize on funding opportunities instead of free-riding may have been particularly acute regarding development, considering the US invested US$89.2 billion in Afghan reconstruction efforts between 2002 and 2012, a significant sum that potentially explains the lack of evidence supporting hypothesis 1b in the Afghan case.

This is not meant to argue that corruption was insignificant in Vietnam or Iraq, but Afghanistan appears to differ in terms of severity and scale. Since 2001, Afghanistan has repeatedly been ranked as one of the worst offenders regarding poor adherence to the rule of law (World Justice Project, n.d.) as well as one of the top five in corruption (Transparency International). Afghanistan is ranked consistently beyond Iraq, which is also notably corrupt. Boosted by the drug trade, corruption in Afghanistan has been described as endemic. As one US counternarcotics officer noted, ‘the big problem in this country is criminality and corruption. It’s huge. It’s just rampant. It’s rife. It’s beyond anything we’ve seen in Colombia or Mexico or any place else’ (Dealey, 2006). This association between drugs, corruption and opting not to free-ride in order to access funding was evident in the data. Consider the 2003 request regarding banning poppy cultivation (Department of State, 2003a). Kabul opted not to free-ride because participating allowed elites to choose which opium operations could be shut down, and which thrived (Mankin, 2011). Select enforcement was documented in Nangarhar Province from 2005 to 2007 (Afghanistan’s Opium Poppies, 2008; Mankin, 2011; Nicoletti, 2011: 38). Afghan officials in Nangarhar shut down competitors, thus decreasing opium production, which boosted profits for the remaining producers, while putting Nangarhar on the US list of ‘good performers’ due to an overall decline in production. Inclusion on this list meant Nangarhar received more aid, which was itself reportedly, ‘all too easily siphoned off’ (Mankin, 2011).

These findings regarding corruption in Afghanistan do not refute the proposed hypotheses, but they demonstrate the complexity of the conflict environment, and suggest that other factors in addition to those specified by economic models influence free-riding, including for example, the motivation to profit financially suggested by the findings from Afghanistan. The Afghan case also suggests that small allies may forgo opportunities to free-ride on collective security goods provided by large allies if there are opportunities to access private goods, such as cash. Such questions do not typically apply to traditional studies of economic theories of alliances analyzing the financial contributions of allies to multilateral institutions such as NATO, but they are apparent in counterinsurgencies where smaller allies can face trade-offs between accessing funding though participating, or free-riding on unilateral efforts of larger partners. The former emphasizes private goods and profit maximization; the latter prioritizes taking advantage of public goods and focusing on security. Depending on the security situation and ally, different patterns of free-riding may be observed in different counterinsurgencies, as reflected in the findings summarized in Tables 2 and 3.

Conclusion

This article applies a familiar literature on unequal burden-sharing between asymmetric allies to security alliances in counterinsurgency interventions. I offer three primary findings. Firstly, even when small allies are explicitly asked to contribute, they are nevertheless likely to free-ride almost half the time (45.1%). Secondly, supporting joint-product models, certain collective security goods more closely resemble public goods, while others resemble private goods (Sandler, 1993). Security requests proposed by large counterinsurgency allies that could exclude small allies failed to meet the non-excludability criteria to qualify as public goods, and thus were unsurprisingly found to be less likely to inspire free-riding, especially in the US interventions in Iraq and Vietnam. This finding was less significant in Afghanistan, potentially due to the severity of corruption in Kabul emphasizing private, not public goods. Thirdly, different policy issues are associated with different rates of free-riding by small allies. In Iraq and Vietnam in particular, development projects were likely to result in free-riding, while requests for political reform designed to strengthen counterinsurgency efforts were less prone to free-riding. Small allies took advantage of US unilateral abilities to fund and manage development efforts, but feared that free-riding on political issues would make them vulnerable to US unilateral action, perhaps exposing them to painful reforms jeopardizing their control.

Overall, the study finds that free-riding behaviors in asymmetric counterinsurgency partnerships largely support the longstanding literature on economic theories of alliances. When defense goods shared between counterinsurgency allies resembled public goods, free-riding by small allies was likely. Conversely, when shared defense goods failed to meet the criteria for public goods, free-riding by small counterinsurgency allies was less likely.

These findings provide a fascinating perspective on the costs and consequences of free-riding in counterinsurgencies. The findings suggest that the moments where small counterinsurgency allies tend not to free-ride (political reforms), as well as the moments where small allies are likely to free-ride (development projects), may both be correlated with perpetuating foreign intervention. Local counterinsurgency allies free-riding on development efforts limits local institutional growth by perpetuating dependencies on intervening forces, while local allies consistently opting to influence proposed political reforms is likely to perpetuate corrupt institutions controlled by local elites, thus limiting the likelihood of necessary political reforms (Bardhan, 1997; Corruption Hampers Development in Afghan Districts, 2014; Le Billon, 2005; Natsios, 2005; Vedantam, 2007).

Footnotes

Funding

The author received no financial support for the research, authorship and/or publication of this article.

Notes

Author biography

Barbara Elias is Assistant Professor of Government at Bowdoin College. She is the former director of the Afghanistan/Pakistan/Taliban Documentation Project at the National Security Archive in Washington D.C.

References

Alesina

Dollar

(2000) Who gives foreign aid to whom and why? Journal of Economic Growth 5(1): 33–63.

Axelrod

Keohane

(1985) Achieving cooperation under anarchy: Strategies and institutions. World Politics 38(1): 226–254.

Bardhan

(1997) Corruption and development: A review of issues. Journal of Economic Literature 35(3): 1320–1346.

BBC (2010) Where the leaked cables came from, 29 November, sec. US & Canada. Available at: http://www.bbc.co.uk/news/world-us-canada-11863618

Beehner

Bruno

(2008) What are Iraq’s benchmarks? Council on Foreign Relations, 11 March. Available at: http://www.cfr.org/iraq/iraqs-benchmarks/p13333 (accessed 20 April 2012).

Berman

(2010) From the Sun King to Karzai: Lessons for state building in Afghanistan. Foreign Affairs 89(2): 2–9.

Berthélemy

Tichit

(2004) Bilateral donors’ aid allocation decisions—A three-dimensional panel analysis. International Review of Economics & Finance, Aid Allocations and Development Financing 13(3): 253–274.

Biddle

(2008) Review of the New U.S. Army/Marine corps counterinsurgency field manual. Perspectives on Politics 6(2): 347–350.

Bove

Elia

(2011) Supplying peace: Participation in and troop contribution to peacekeeping missions. Journal of Peace Research 48(6): 699–714.

10.

Byman

(2006) Friends like these: Counterinsurgency and the war on terrorism. International Security 31(2): 79–115.

11.

Chaudhuri

Farrell

(2011) Campaign disconnect: Operational progress and strategic obstacles in Afghanistan, 2009–2011. International Affairs (Royal Institute of International Affairs 1944) 87(2): 271–296.

12.

Cohen

Crane

Horvath

et al . (2006) Principles, imperatives, and paradoxes of counterinsurgency. Military Review. Available at: http://www.au.af.mil/au/awc/awcgate/milreview/cohen.pdf

13.

Cooper

(1965) Memorandum for Mr. Bundy, subject: Status report on various actions in Vietnam. Lyndon Baines Johnson Library. NLJ 84-130, Declassified August 14, 1984.

14.

Cooper

Corson

Legere

(1972) The American Experience with Pacification in Vietnam – Volume I – An Overview of Pacification (Unclassified). Institute for Defense Analyses (IDA). Available at: www.vietnam.ttu.edu/star/images/1201/1201065051.pdf

15.

Cooper

(2009) In leaning on Karzai, U.S. has limited leverage. The New York Times, 12 November, sec. International/Asia Pacific. Available at: http://www.nytimes.com/2009/11/12/world/asia/12karzai.html

16.

Corruption Hampers Development in Afghan Districts (2014) Institute for War and Peace Reporting. Available at: https://iwpr.net/global-voices/corruption-hampers-development-afghan-districts (accessed 8 February 2015).

17.

Croco

Teo

(2005) Assessing the dyadic approach to interstate conflict processes: A.k.a. ‘Dangerous’ Dyad-years. Conflict Management and Peace Science 22(1): 5–18.

18.

Dealey

(2006) At war in the fields of the drug lords. Available at: http://www.gq.com/story/sam-dealey-drugs-opium-war-afghanistan (accessed 25 July 2016).

19.

Elias

(2013) America’s missing leverage in Afghanistan and Pakistan: A structural analysis. Third World Quarterly 34(8): 1392–1408.

20.

Felbab-Brown

(2013) Afghanistan in 2012: Limited progress and threatening future. Asian Survey 53(1): 22ey.

21.

Fordham

Poast

(2014) All alliances are multilateral: Rethinking alliance formation. Journal of Conflict Resolution 60(5): 840–865.

22.

Galula

(2006) Counterinsurgency Warfare: Theory and Practice. Westport, CT: Praeger.

23.

Garrels

(2008) Long-awaited Fallujah rebuilding shows promise: NPR. NPR.org, 23 January. Available at: http://www.npr.org/templates/story/story.php?storyId=18319948 (accessed 10 May 2012).

24.

Goldstein

(1995) Discounting the free-ride: Alliances and security in the postwar world. International Organization 49(1): 39–71.

25.

Goldstein

(2007) Deterrence and Security in the 21st Century: China, Britain, France, and the Enduring Legacy of the Nuclear Revolution (1st ed). Stanford, CA: Stanford University Press.

26.

Gompert

Gordon IV

(2008) War by Other Means: Building Complete and Balanced Capabilities for Counterinsurgency. RAND Corporation. Available at: http://www.rand.org/pubs/monographs/MG595.2/ (accessed 12 December 2009).

27.

Gravel

(1971) The Pentagon Papers: The Defense Department History of United States Decisionmaking on Vietnam (The Senator Gravel Edition). Beacon Press.

28.

Jaffe

Morris

(2015) Defense secretary carter: Iraqis lack ‘will to fight’ to defeat Islamic State. The Washington Post, 24 May. Available at: http://www.washingtonpost.com/politics/defensesecretary-carter-iraqis-need-will-to-fight-to-defeat-islamicstate/2015/05/24/1f189454-022e-11e5-bc72-f3e16bf50bb6_story.html (accessed 5 June 2015).

29.

Kahler

(1992) Multilateralism with small and large numbers. International Organization 46(3): 681–708.

30.

Keohane

(1971) The big influence of small allies. Foreign Policy (2): 161–182.

31.

Komer

(1972) Bureaucracy does its thing: Institutional constraints on U.S.-GVN performance in Vietnam. Available at: http://www.rand.org/pubs/reports/R967/

32.

Le Billon

(2005) Corruption, reconstruction and oil governance in Iraq. Third World Quarterly 26(4–5): 685–703.

33.

Maizels

Nissanke

(1984) Motivations for aid to developing countries. World Development 12(9): 879–900.

34.

Maley

(2011) Afghanistan in 2010. Asian Survey 51(1): 85–96.

35.

Mankin

(2011) Rotten to the Core. Foreign Policy. Available at: http://www.foreignpolicy.com/articles/2011/05/10/rotten_to_the_core

36.

Meernik

Krueger

Poe

(1998) Testing models of U.S. foreign policy: Foreign aid during and after the Cold War. The Journal of Politics 60(01): 63–85.

37.

Menges

(1968) Democratic Revolutionary Insurgency as an Alternative Strategy. Rand Corporation. March, 3817.

38.

Morrow

(1991) Alliances and asymmetry: An alternative to the capability aggregation model of alliances. American Journal of Political Science 35(4): 904–933.

39.

Murdoch

Sandler

(1982) A theoretical and empirical analysis of NATO. Journal of Conflict Resolution 26(2): 237–263.

40.

Murdoch

Sandler

(1984) Complementarity, free-riding, and the military expenditures of NATO allies. Journal of Public Economics 25(1–2): 83–101.

41.

Natsios

(2005) The nine principles of reconstruction and development. Parameters 35(3): 4.

42.

Nicoletti

(2011) Opium production and distribution: Poppies, profits and power in Afghanistan (Theses and dissertations). DePaul University. Available at: http://via.library.depaul.edu/etd/74

43.

Noetzel

Schreer

(2009) Does a multi-tier NATO matter? The Atlantic Alliance and the Process of Strategic Change. International Affairs (Royal Institute of International Affairs 1944). 85(2): 211–226.

44.

Office of the Special Inspector General for Iraq Reconstruction (SIGIR) (2008) Commander’s emergency response program in Iraq funds many large-scale projects. SIGIR-08-006, SIGIR.

45.

Olson

Zeckhauser

(1966) An economic theory of alliances. The Review of Economics and Statistics 48(3): 266–279.

46.

Oneal

(1990a) Testing the theory of collective action: NATO defense burdens, 1950-1984. The Journal of Conflict Resolution 34(3): 426–448.

47.

Oneal

(1990b) The theory of collective action and burden sharing in NATO. International Organization 44(3): 379–402.

48.

Oneal

Diehl

(1994) The theory of collective action and NATO defense burdens: New empirical tests. Political Research Quarterly 47(2): 373–396.

49.

Oppenheimer

(1979) A reassessment. Journal of Conflict Resolution 23(3): 387–407.

50.

Oye

(1985) Explaining cooperation under anarchy: Hypotheses and strategies. World Politics 38(1): 1–24.

51.

Plümper

Neumayer

(2014) Free-riding in alliances: Testing an old theory with a new method. Conflict Management and Peace Science 32(3): 247–268.

52.

Poast

(2010) (Mis)using dyadic data to analyze multilateral events. Political Analysis 18(4): 403–425.

53.

Riedel

(2012) Deadly Embrace: Pakistan, America, and the Future of the Global Jihad. Washington, DC: Brookings Institution Press.

54.

Rosenberg

(2013) C.I.A. delivers cash to Afghan leader global Jih. The New York Times, 28 April. Available at: http://www.nytimes.com/2013/04/29/world/asia/cia-delivers-cash-to-afghan-leaders-office.html (accessed 24 July 2016).

55.

Russett

Sullivan

(1971) Collective goods and international organization. International Organization 25(4): 845–865.

56.

Sandler

(1977) Impurity of defense: An application to the economics of alliances. Kyklos 30(3): 443–460.

57.

Sandler

(1993) The economic theory of alliances A survey. Journal of Conflict Resolution 37(3): 446–483.

58.

Sandler

Forbes

(1980) Burden sharing, strategy, and the design of NATO. Economic Inquiry 18(3): 425–444.

59.

Sandler

Hartley

(2001) Economics of alliances: The lessons for collective action. Journal of Economic Literature 39(3): 869–896.

60.

Sandler

Cauley

Forbes

(1980) In defense of a collective goods theory of alliances. Journal of Conflict Resolution 24(3): 537–547.

61.

Schaffer

(2002) U.S. influence on Pakistan: Can partners have divergent priorities? The Washington Quarterly 26(1): 169.

62.

Schwartz

(1991) American Counterinsurgency Doctrine and El Salvador. Rand Corporation. Available at: http://www.rand.org/pubs/reports/R4042.html (accessed 24 July 2016).

63.

Schwartz

(2008) War without End: The Iraq War in Context. Chicago: Haymarket Books.

64.

Shafer

(2014) Deadly Paradigms: The Failure of U.S. Counterinsurgency Policy. Princeton, NJ: Princeton University Press.

65.

Signorino

(1999) Strategic interaction and the statistical analysis of international conflict. The American Political Science Review 93(2): 279–297.

66.

Snyder

(2007) Alliance Politics. Ithaca, NY: Cornell University Press.

67.

The Economist (2008) Afghanistan’s opium poppies: No quick fixes, 19 June. Available at: http://www.economist.com/node/11591396

68.

The Guardian (2010) Siprnet: Where America stores its secret cables, 28 November. Available at: http://www.guardian.co.uk/world/2010/nov/28/siprnet-america-stores-secret-cables (accessed 20 April 2012).

69.

The New York Times (2006) Rumsfeld’s memo of options for Iraq war, 3 December. Available at: http://www.nytimes.com/2006/12/03/world/middleeast/03mtext.html (accessed 10 May 2012).

70.

Thies

(2002) Friendly Rivals: Bargaining and Burden-Shifting in NATO. M E Sharpe Inc. Transparency International Corruption Perceptions Index. Available at: http://www.transparency.org/research/cpi/overview (accessed 24 July 2016).

71.

United Nations Office on Drugs and Crime (UNDOC) (2010) Corruption in Afghanistan: Bribery as Reported by Victims. Available at: http://www.unodc.org/unodc/en/frontpage/2010/January/corruption-widespread-in-afghanistan-unodc-survey-says.html

72.

U.S. Department of State (2003) Objectives for certification to the government of afghanistan, June 6.

73.

U.S. Department of State (2005) Fallujans mobilized for election amid increased tension in city, 05BAGHDAD4971, Wikileaks.

74.

U.S. Department of State (2006) The 2006 Iraq budget, 06BAGHDAD955, Wikileaks.

75.

U.S. Department of State, Embassy Kabul (2009a) Karzai on elections and the future: September 1 meeting at the palace, 09KABUL2681, Wikileaks.

76.

U.S. Department of State, Embassy Kabul (2009b) Post-election compact: Karzai team consultations, 09KABUL2422, Wikileaks.

77.

Vedantam

(2007) One thing we can’t build alone in Iraq. The Washington Post, 29 October. Available at: http://www.washingtonpost.com/wp41dyn/content/article/2007/10/28/AR2007102801477.html (accessed 8 February 2015).

78.

Walt

(1987) The Origins of Alliances. Ithaca, NY: Cornell University Press.

79.

Warrick

(2008) Little blue pills among the ways CIA wins friends in Afghanistan. The Washington Post, 26 December. Available at: www.washingtonpost.com/wpdyn/content/article/2008/12/25/AR2008122500931.html (accessed 24 July 2016).

80.

Watts

Campbell

Johnston

et al . (2014) Countering Others’ Insurgencies: Understanding U.S. Small-Footprint Interventions in Local Context. Santa Monica, CA: RAND Corporation.

81.

World Justice Project (n.d.) Rule of Law Index. Available at: www.worldjusticeproject.org/ruleof-law-index (accessed 24 July 2016).