Foreign Aid and Policy Concessions: Counterterrorism Assistance as Insurance

Abstract

A puzzling feature of conditional military aid is that its amount is correlated negatively with the level of its recipients’ security cooperation. I propose a new explanation for the apparent failure of security assistance, demonstrating that using foreign aid as a source of income stabilisation for the recipient rationalises the inverse relationship observed between the provision of foreign aid and policy concessions. Aid conditionality is modelled as an implicit interstate agreement that specifies how a recipient allocates domestic and foreign financial resources across different security-related policy areas in an infinitely-repeated agency game. It is shown that efficient security cooperation between countries with conflicting interests involves a certain diversion of conditional assistance. I illustrate the model’s logic by reference to the purported epitomes of the ineffectiveness of foreign aid as a counterterrorism tool, which are the post-9/11 security relationships between the United States and Pakistan as well as Yemen.

Introduction

Although the US sent billions of dollars’ worth of military aid to Islamabad, analysts contend that this vast amount of money failed to generate the intended outcomes. It appears that much of this assistance was not put towards counterterrorism capabilities as Washington desired, but rather towards attaining a more favourable balance of military power with archenemy India (Cohen and Chollet 2007; Fair et al. 2010). Meanwhile, decision-makers in Washington feared that discontinuing foreign aid would bring about the overthrow of the Pakistani government and allow terrorists to seize the country’s nuclear arsenal (e.g. Krasner 2012).

Indeed, extant contributions show that the effectiveness of aid conditionality is mediated by the aid recipient’s ‘state capacity’, which is typically operationalised as his ¹ fiscal capacity, including his ability to conduct counterterrorism operations without security assistance (Bapat, 2011; Bapat et al., 2016; Carter, 2015; Stone Randall, 2010). The proffering of security-motivated bribes as a means of exerting influence can only elicit policy concessions from ‘high-capacity’ recipients, from which foreign aid can be credibly withheld. However, international terrorism predominantly emerges in regions with limited state capacity to dismantle terrorist groups. When issued against these ‘low-capacity’ recipients, the aid donor’s contingent threat to withdraw security support is rendered incredible, which severely compromises the effectiveness of counterterrorism assistance.

In fact, empirical studies (e.g. Sullivan et al. 2011) find a counterintuitive inverse relationship between the volume of aid flowing into a recipient country and the extent of policy concessions made in return. Therefore, some authors conclude that “for a recipient, non-cooperation pays, while cooperation is punished” (Boutton 2014, 742). In view of these theoretical reservations and the seemingly poor empirical record of aid conditionality, why do donors continue to rely on security assistance as an instrument of counterterrorism vis-à-vis ‘low-capacity’ recipients, whose incentives to comply are retarded?

I argue that such a conditional aid relationship may prove effective if evaluated over a longer time horizon. To formalise this claim, I model conditional aid in an infinitely-repeated principal-agent game. Given the recipient’s fluctuating domestic revenue stream, an implicit agreement specifies how the recipient allocates domestic and foreign resources across policy areas that are prioritised by either him or the donor. The recipient is induced to trade off policy concessions (in the form of channelling monies towards policy fields favoured by the donor) in exchange for the donor’s provision of a quasi-insurance against downside risks to the level of spending in the policy areas preferred by the recipient.

The model’s results demonstrate that intertemporal exchange can be sustained in equilibrium notwithstanding the agents’ inability to commit. Crucially, an empirical implication of the findings is that the granting of minor policy concessions is associated with the disbursement of large sums of aid money, and vice versa. Consequently, a discrepancy between a donor’s goals and a recipient’s activities may simply reflect a recurring feature in an actually long-term, mutually beneficial relationship. Efficient collaboration here necessarily involves the diversion of foreign assistance by the recipient.

This paper proceeds as follows: The next section reviews the literature on the efficacy of conditional (counterterrorism) aid. I then set up the model, whose key results are summarised in the subsequent section. This section also contains a plausibility probe of the model's arguments and propositions with reference to the post-9/11 security partnerships between the US and Pakistan as well as Yemen. The final section concludes the paper and suggests possible avenues for future research.

Theory

In principle, a donor can subsidise foreign counterterrorism efforts to eliminate a threat to her security imperatives from a recipient’s soil (Azam and Delacroix 2006; Bandyopadhyay et al. 2011; Young and Findley 2011). Although doling out security assistance is expected to increase a donor’s influence over the counterterrorism policies of a recipient with diverging interests, many studies question the effectiveness of conditional aid on both theoretical and empirical grounds. For instance, Sullivan et al. (2011) find a negative correlation between the amount of US military aid procured by a recipient and his level of cooperation with American objectives. A donor usually faces the problem that a recipient can use her aid payments at his discretion for purposes unrelated or even contrary to the donor’s intentions. For example, after enlisting Islamabad as an ally in the global ‘War on Terror’ in the aftermath of September 11, Washington funnelled counterterrorism assistance through the Pakistani government, ceding any authority or oversight over US aid transfers as they became Pakistani sovereign funds (Fair et al. 2010, 160f.). ²

A donor may try to compel a recipient to honour the terms of their aid agreement by conditioning current aid flows on past expenditures. The donor can reject further military assistance if a recipient fails to meet threshold levels of counterterrorism spending. However, without the donor’s ability to commit, abandoning a ‘fragile’ recipient may not be a time-consistent behaviour for the donor. While the donor prefers to extract short-term policy concessions, she can be unwilling to withdraw foreign assistance from a recipient, especially if he is ‘weak’, for fear of forfeiting the prospective long-term benefits accruing to her from her aid relationship with him. This dilemma between the donor’s motives to provide aid as a ‘reward’ and as ‘insurance’ may cause aid conditionality to fail. Put differently, the lack of donors’ commitment powers results in enforcement problems vis-à-vis impoverished recipients.^{3, 4}

The recipient’s disregard for the conditions placed on even relatively extensive assistance packages is explained by the donor’s inability to credibly threaten to withhold aid. After all, “threats can elicit recipient concessions when their donor can commit to carrying out unsuccessful threats as well” (Krustev 2010, 149). “Thus the paradox of foreign aid is that the greater the potential for influence, the more likely the home state may be inhibited from exercising that influence. Looking at the identity of exposure, one would expect recipients of huge amounts of aid to be highly vulnerable to aid sanctions. If, however, the home state is pouring large sums of money into a country, obviously it considers the recipient to be very important, and will be hesitant to abandon it. Additionally, the recipient state is aware of this – states tend to direct aid to states which they consider important: this also undermines the credibility of threatened aid sanctions” (Kirshner 1997, 47; see also Bearce and Tirone 2010; Dunning 2004; Girod 2012).

The donor’s strategic motives that prompted her to provide foreign assistance in the first place may militate against (threatening) aid sanctions to secure tactical cooperation on other issues. Given that the disbursement of “foreign aid is an investment in a valued regime, representing a direct monetary measure of the importance of a particular recipient to a particular donor” (Stone 2004, 579), leveraging present aid flows for immediate gains may compromise the achievement of the donor’s strategic objectives in the foreseeable future. If the donor then were to “punish […] by withholding aid, […she] risks defeating the purpose for which the aid was originally offered” (Montgomery 1967, 64). For example, Sullivan et al. (2011, 290) suggest that American “military aid is delivered to states that the United States depends on for security reasons. Realizing their leverage over Washington, states that receive high amounts of aid are actually more able to engage in uncooperative behavior than are states that the United States does not depend so heavily upon.” Likewise, Nooruddin and Payton (2010) conclude that, in “the context of the ‘War on Terror’, the USA was loath to endanger relationships with close allies, who were also more likely to […] receive [larger amounts of] US military aid” (p. 719).

Moreover, a compellence dilemma arises if the donor strives to manipulate the behaviour of a ‘low-capacity’ recipient. Because terrorists purposely base themselves in regime-stressed states with limited counterterrorism capacity (Bandyopadhyay et al. 2011; Lai 2007; Piazza 2008), a donor may be concerned about the deleterious effects that financial penalties might have on such ‘fragile’ recipients’ economic prosperity and military resilience. Applying greater pressure on a low-income recipient could impair his ability to advance the donor’s security interests in the future, which may cause the donor to reconsider potential aid cuts in the present. For example, Carter (2015, 474) postulates that “harsh compellent punishments [for Islamabad’s failure to root out terrorists] are simply not credible as it is not at all clear that Pakistan can absorb them without losing ground to militants and their supporters”. The security threat inherent in the collapse of a ‘weak’ recipient can thus be exploited by him to deflect donor pressure to adhere to aid conditionality without jeopardising the steady influx of foreign assistance. In line with such stability interests of Western donors, Licht (2010) shows these donors’ propensity to provide aid to recipient states whose incumbent heads of state are at risk of being deposed. Similarly, Boutton and Carter (2014) report that Washington uses its military assistance to support recipient countries that are perceived to be in danger of becoming safe havens for terrorist organisations. In turn, McGillivray and Stam (2004) as well as Marinov (2005) find that the imposition of sanctions can produce considerable economic and political instability within target countries, raising the chance of leadership turnover.

However, aid payments designed exclusively to preserve the recipient government’s hold on power confer no bargaining power upon the donor to persuade the recipient to increase his counterterrorism efforts. Generally, such stability-oriented subsidies (see, for instance, Steinwand 2015) cannot easily serve anti-terror purposes. Rather, Wagner (1988, 480) demonstrates that “if the donor has already negotiated the best agreement it can concerning the terms of the aid transfer, it will be unable to extract further concessions by threatening to interrupt aid.”

Therefore, the credibility of aid sanction threats is mediated by both the recipient’s strategic value for the donor and his ‘state capacity’ (Bapat, 2011; Bapat et al., 2016; Carter, 2015; Stone Randall, 2010). I define state capacity as the recipient’s disposable fiscal revenue at a certain time (see, for instance, Besley and Persson 2010). ⁵ I now emphasise that these two intervening factors – a sufficiently high valuation of the future and fluctuations in the recipient’s state capacity – not only engender enforcement problems in conditional aid relationships but may also attenuate them at the same time. First, counterterrorism campaigns are typically protracted endeavours. The associated “security externalities provide incentives for donors [and recipients] to take a long-term view of aid relationships” (Stone Randall, 2010, 27). Conversely, “[t]he prospect of future exchanges can make actions enforceable, because of the ability of states to punish opportunistic actors with noncooperation in future rounds” (Drezner 1999, 197; Fearon 1998, 270).

Second, the formal literature on aid conditionality commonly considers short time frames, assuming that the recipient’s state capacity remains constant(ly low) (e.g. Bapat et al. 2016; Carter 2015; Svensson 2000). In reality, however, a recipient’s fiscal revenue actually varies over time. By lengthening the time horizon of the analysis, a momentary lack of credibility of the donor’s aid sanction threat vis-à-vis a low-capacity recipient does not necessarily indicate a failure of her influence attempt, but may instead prove to be merely a temporary inconvenience for the donor in an otherwise beneficial long-term aid relationship. The donor may well foresee that her security investments in a low-income recipient will ‘pay off’ once the attendant enforcement problems are alleviated, i.e. as soon as the recipient’s financial situation improves again.

The donor may hence exploit a recipient’s risk-aversion with regard to spending on issue areas that are important to him. A poor recipient’s domestic income stream is more erratic than that of a wealthy donor, especially if the low-capacity recipient has no access to alternative sources of foreign funding, such as international capital markets (Cordella et al. 2003) or foreign direct investments (Girod and Tobin 2016). For example, “Pakistan’s very limited tax base increases its vulnerability to sharp shifts in revenues, from the domestic tax base or from donors or lenders abroad, or in expenditures” (Fair et al. 2010, 64). The prospect of obtaining foreign aid in the event of drastic budgetary shortfalls may encourage the recipient to abide by the conditions linked to the assistance. The resulting smoothing of expenses is particularly important in the realm of security, ⁶ where maintaining a persistent level of military pressure keeps one’s adversary in check (see Knights 2010). Indeed, Braumoeller (2008, 87) finds that “onset of disputes takes place for the most part when one state has let its guard down and the other tries to take advantage of the opportunity.”

Therefore, the insurance function of security support, which encompasses the promise to cushion a recipient’s expenses against future budget deficits, may induce him to devote more financial resources to combatting terrorist groups. To show this, I model a conditional aid relationship in an infinitely repeated interaction between a donor and a recipient, where the latter’s income stream is volatile. The model demonstrates that, in equilibrium, aid transfers are relatively large and policy concessions are relatively small in low-income phases, and vice versa in high-income periods. While relatively large inflows of foreign assistance facilitate expenditures on recipient-preferred policy issues in low-income phases, the recipient ‘repays’ the donor in high-income periods by raising his spending of domestic and foreign resources on donor-favoured policy issues.

Interestingly, such a consumption-smoothing motive for foreign aid may accommodate some of the recent evidence on the efficacy of conditional assistance. The model’s results show that – for actors with the requisite patience – although the conditional correlation between the volume of aid transfers and the level of security cooperation would be negative in any given period, it would be positive over the long run. Rather than ascribing this empirical pattern to the ineffectiveness of aid conditionality, I offer an alternative interpretation in which conditionality survives the parties’ interest divergence and inability to commit. Conditional aid may simply appear ineffective due to its buffering effect on the recipient’s expenditures.

While some studies recognise foreign aid as an insurance mechanism (Pallage et al. 2006; Savun and Tirone 2012), other contributions on aid volatility highlight its potentially destabilising effect on the viability of the recipient’s regime (Nielsen et al. 2011; Steinwand 2015). However, aid flows necessarily exhibit some degree of variation if they have a dampening impact on a recipient’s income fluctuations. Previous works investigating the temporal dynamics of conditional aid relationships focused on the connection between foreign assistance and debt relief (Cordella et al. 2003), capital flight (Kletzer 2005), and capital accumulation (Scholl 2009).

Model

In this section, I develop the formal model and derive the optimal intertemporal allocation of foreign assistance and policy concessions under various commitment power structures. By moving from the commitment to the non-commitment case, both the self-sustaining nature of the agents’ collaboration and its limits will be illustrated.

A donor country can furnish a recipient state with financial support to gain its cooperation in fighting a resident transnational terrorist group. The differences between the objectives of the donor and the recipient concern the distribution of the recipient government’s spending across two different policy areas. For simplicity, the donor is assumed to care only about the recipient’s expenditures towards the suppression of the terrorist group, whereas the recipient is assumed to be solely interested in financing conventional warfare against a rival country (or a domestic insurgency). Note that objectives do not need to contradict each other strictly but must be at least partly different to provide the underlying conflict of interest here. ⁷ For short-hand, I refer to the recipient’s counterterrorism expenditures as policy concessions. For example, consumed with its long-standing regional rivalry with India, Islamabad was loath to wage the US-led War on Terror (Zaidi 2011, 8). Pakistan’s president “Musharraf saw his government’s [counterterrorism] effort as a concession for which he would pay a domestic price and therefore needed a demonstration of U.S. support in return” (Cohen and Chollet 2007, 10f.; Fair et al. 2010, 160).

A principal-agent problem arises as the donor cannot directly fund her prioritised policy area. At each date, aid intended for counterterrorism purposes is initially forwarded to the recipient, who only thereafter decides how it will be redistributed within the country. The recipient state’s sovereignty implies that it cannot be forced to adhere to a particular spending profile. The recipient can pocket the incoming aid money at any time and refuse to concede policy concessions in return or effectively terminate the aid relationship. The donor thus wants to implement an aid scheme that persuades the recipient to allot as much of his budget as possible to counterterrorism purposes in exchange for using some of the aid to satisfy the recipient’s primary security concerns vis-à-vis his (intra- or) interstate opponents. For example, although being earmarked for the acquisition of advanced military materiel designed for counterterrorism operations against al-Qaeda, Pakistan misappropriated a sizeable portion of American aid monies to bankroll the purchase of conventional weaponry aimed at India (Boutton 2014; Cohen and Chollet 2007).

In an infinitely repeated game, aid flows can be conditional on the recipient’s past use of aid. The donor can punish the recipient for diverting more of her assistance than he should under the implicit ⁸ aid agreement. The sanctions at the donor’s disposal are the suspension of future aid transfers. Since the donor state is also a sovereign, it can choose to retain foreign assistance at any time. Neither party can commit its future actions. I assume that aid cannot be stored. ⁹ At each date, a stage of the repeated game is played.

The Stage Game

The recipient obtains a stochastic income stream. The resulting gains from intertemporal exchange generate a consumption-smoothing motive for foreign aid flows. The recipient’s monetary endowment in state of nature n ∈ {1, 2, …, N} is denoted by e _n and independently drawn each period from an identical distribution over a finite support given by 0 < e₁ < e₂ < … < e _N . The recipient’s endowment represents his state capacity.

The timing of the stage game is as follows: first, the recipient’s endowment e _n realises with probability π _n ∈ (0, 1), where ${\sum }_{n=1}^N{\pi }_n=1$ . Second, the donor chooses her aid transfer τ _n in state n. The non-negativity constraint on aid transfers

${\tau }_n\ge 0$

(1)

implies that aid cannot flow from the recipient to the donor. Finally, the recipient decides how to allocate his endowment and the donor’s aid. He can spend this budget (in part or in full) on combatting a rival country (or quelling domestic unrest) c _n ≥ 0 or defeating a resident terrorist group d _n ≥ 0. The budget constraint in each state n is thus

$c_n+d_n\le e_n+{\tau }_n.$

(2)

Whereas the donor values the counterterrorism effort undertaken by the recipient, less the opportunity costs of her monetary transfers, the recipient derives utility from conventional military expenditures against an (intra- or) interstate rival. The single-period payoffs in state n for the donor and the recipient are given by $\left[u(d_n)-{\tau }_n\right]$ and $\left[v(c_n)\right]$ , respectively. Both utility functions u(d _n ) and v(c _n ) are normalised to u(0) = 0 and v(0) = 0, strictly increasing and concave; both display unbounded marginal utility as expenditures go to zero. The donor is risk-neutral regarding monetary transfers τ _n . The donor and the recipient are risk-averse with respect to spending on counterterrorism d _n and conventional combat c _n , respectively.

In the single-period game, autarky constitutes the unique Nash equilibrium in which the donor and the recipient refuse to grant aid payments and policy concessions, respectively. Under autarky, we have

${\tau }_n=0,d_n=0\text{and}{c}_n=e_n\forall n.$

(3)

Therefore, taking a short-term perspective, aid conditionality is bound to fail.

The Infinitely Repeated Game

For the infinitely repeated game, intertemporal exchange is possible. A conditional aid agreement can capitalise on the recipient’s aspiration to stabilise volatile security expenditures that he deems to be particularly pressing to induce him to adopt counterterrorism measures requested by the donor. In each state of nature n, the stationary aid agreement specifies the recipient’s spending on destroying an indigenous terrorist group d _n and countering a rival country (or domestic adversary) c _n , and the donor’s aid transfer τ _n . In any period t, an agent’s payoff for the remainder of the repeated game is the expected utility from the sequence of state-contingent single-period payoffs from date t onwards. Thus, in each period t, before the realisation of state n, the donor’s and the recipient’s ex ante objective functions read

$\begin{array}{lll}\hfill U& \equiv \sum _{t=0}^{\infty }{\delta }^t\sum _{n=1}^N{\pi }_n\left[u(d_n)-{\tau }_n\right]\hfill & \hfill =\frac{1}{1-\delta }\sum _{n=1}^N{\pi }_n\left[u(d_n)-{\tau }_n\right]\text{and}\end{array}$

(4)

$\begin{array}{lll}\hfill V& \equiv \sum _{t=0}^{\infty }{\delta }^t\sum _{n=1}^N{\pi }_n\left[v(c_n)\right]\hfill & \hfill =\frac{1}{1-\delta }\sum _{n=1}^N{\pi }_n\left[v(c_n)\right],\text{respectively.}\end{array}$

(5)

For simplicity, the discount factor δ ∈ (0, 1) is the same for both actors. ¹⁰ The participation constraint ensures that the recipient’s adherence to the aid agreement renders him at least as well off as under autarky:

$V\ge \frac{1}{1-\delta }\sum _{n=1}^N{\pi }_n\left[v(e_n)\right].$

(6)

Benchmark: Optimal Aid Contract with Commitment

To benchmark the implicit aid agreement without commitment, I first describe the optimal aid contract for the donor if both parties could commit to adhere to it. The efficient aid scheme maximises the donor’s expected per-period utility (1 − δ)U over expenditure choices d _n and c _n as well as aid transfers τ _n , subject to the non-negativity constraint on aid transfers (1), the budget constraint (2), and the recipient’s participation constraint (6) for all states n ∈ {1, …, N}. The equilibrium is characterised by the following proposition, the proof of which can be found in the Online Appendix:

Proposition 1

The optimal aid contract includes a critical state n* ∈ {1, …, N}, which is the highest income state featuring aid payments. For all states n < n*, we have d _n = d_n+1, c _n = c_n+1 as well as τ _n > τ_n+1 > 0 and for all states n > n*, we have d _n > d_n−1, c _n > c_n−1 as well as τ _n = 0. The optimal aid contract stipulates that recipient-preferred expenditures are at least partially smoothed with respect to the recipient’s endowment, i.e. e₁ < c₁ ≤ c _N < e _N .

First, if the critical state n* = N, expenditures are completely smoothed across all income states through aid payments that are strictly decreasing in the state n. In other words, we have d _n = d_n+1, c _n = c_n+1 and τ _n > τ_n+1 > 0 for all states n ∈ {1, …, N − 1}.

Second, if the critical state n* = 1, expenditures are completely smoothed across no income states. In this case, assistance is transferred only in the lowest state n = 1. More precisely, we have d _n < d_n+1 and c _n < c_n+1 for all states n ∈ {1, …, N − 1} as well as τ₁ > τ _n = 0 for all states n ∈ {2, …, N}.

Third, if the critical state n* ∈ (1, N), expenditures are completely smoothed only across some (lowest-)income states. For all states n ≤ n*, expenditures are constant while aid payments are strictly decreasing in the state n. For all states n > n*, expenditures are strictly increasing in the state n while no aid is paid out:

$0<d_1=d_2=\dots =d_{n^{\ast }}<d_{n^{\ast }+1}<\dots <d_N,$

(7)

$0<c_1=c_2=\dots =c_{n^{\ast }}<c_{n^{\ast }+1}<\dots <c_N\text{and}$

(8)

${\tau }_1>{\tau }_2>\dots >{\tau }_{n^{\ast }}>{\tau }_{n^{\ast }+1}=\dots ={\tau }_N=0.$

(9)

In sum, if agents can commit their future behaviour, foreign assistance is constrained-efficient at eliciting policy concessions, however modest they may be. The optimal aid contract exhausts all consumption-smoothing potential between the parties, even though the recipient spends some of the aid on policy issues that are not conducive to the donor’s objective.

Optimal Aid Agreement Without Commitment

The Online Appendix derives the optimal implicit aid agreement without commitment. I characterise the Pareto-efficient set of all perfect equilibria of the game using the reversion to permanent autarky as punishment ¹¹ for any deviation from the equilibrium path. The following proposition highlights a key distinction from the optimal aid contract with commitment:

Proposition 2

If the discount factor δ is sufficiently low under an aid agreement without commitment, Proposition 1 may not hold: there may be no critical state n* and aid payments may be strictly increasing across some states n and n + 1, i.e. τ _n < τ_n+1.

Intuitively, a shrinking temporal outlook diminishes the attainable scope of consumption-smoothing and is progressively incompatible with the insurance rationale for providing foreign assistance. The distortion of optimal risk-sharing under an aid agreement without commitment can result in a strict increase of aid transfers across some income states. Not only can higher discounting of the future lead to a reduction in the monetary support that the recipient may reckon with in times of worst financial hardship, i.e. in phases in which his state capacity is at its lowest, to prevent him from ‘taking the money and running’. A lower value of the discount factor δ can simultaneously entail greater aid inflows in times of replenishing government coffers as compensation, i.e. in some period(s) of higher state capacity, to maintain the recipient’s willingness to cooperate. Empirically, this implies a positive correlation between the volume of aid transfers and the amount of policy concessions across these income states, e.g. d₁ < d₂ and τ₁ < τ₂ for states n ∈ {1, 2}.

Findings and Case Illustrations

Propositions 1 and 2 yield two main findings. First, influencing others through financial support becomes more efficient the longer the agents’ time horizon, which is represented by a larger value of the discount factor δ in the model, since commitment problems may be ameliorated for more patient actors. Paradoxically, an insurance function of foreign assistance can create the false impression that aid conditionality has failed. One implication of this is that a positive association between a country’s aid receipts and counterterrorism effort may indicate a lack of strategic foresight on the part of the security partners. Second, another empirical lesson to be drawn from the model is that, to promote the efficiency of her influence attempt, the donor must inevitably allow a recipient whose security priorities do not match her own to siphon off some aid funds.

I now elucidate the plausibility of the key propositions derived from the model. As “both Pakistan and Yemen refrained from committing a high level of resources or effort for a sustained period of time but still avoided major punitive action by the United States” (Tankel 2018, 97), I examine these two similar US attempts to prompt more frequent and vigorous military offensives against al-Qaeda strongholds following September 11. The findings reveal that the conditional assistance programme for Islamabad was implemented more efficiently over time than the one for Sanaa. This is reflected in the increasingly negative relationship between Pakistan’s aid receipts and counterterrorism effort.

US–Pakistan

Since 9/11, Pakistan was enlisted as a front state in Washington’s fight against al-Qaeda terrorists in exchange for US-provided security assistance. However, the security partners’ conflicting strategic objectives and divergent understandings of the threat environment considerably impeded more effective counterterrorism collaboration. Islamabad did not share Washington’s main preoccupation with dismantling al-Qaeda. Rather, Pakistan’s prime interest lay in striking a more favourable military balance of power with its traditional enemy India. As a result, US-Pakistani relations were initially transactional (Tankel 2018, 133, 144f.).

The largest single source of US security assistance to Pakistan was the Coalition Support Fund (CSF). “Because CSFs go directly to the Ministry of Finance, where they can be used as the government wishes, […] CSFs serve as little more than a bribe to secure Pakistan’s continued participation in the global war on terror. […In fact,] much of the American money […] has been diverted to help finance weapons systems designed to counter India, not Al Qaeda [. …Therefore, the] CSF appears to be used as a quid pro quo for Pakistani support of U.S. goals” (Fair et al. 2010, 160).

Although Pakistan’s counterterrorism effort on behalf of US objectives is difficult to gauge, it can be judged by the number of arrests and killings of alleged or actual (high-ranking) terrorists. “Such a body count approach is appealing because it provides a concrete measure of success and failure” (Byman 2003). To accommodate American concerns about moral hazard, US authorities thus implicitly conditioned foreign assistance on the apprehension and elimination of (predominantly leading al-Qaeda) extremists (Dongen 2011, 362; Puri 2012, 38f.; Tankel 2018, 102).

The level of counterterrorism effort that Islamabad can afford to exert depends on its state capacity, which is ideally conceptualised as the amount of financial resources a recipient can muster without foreign aid. However, such a financial indicator remains elusive in the case of Pakistan, since, among other reasons, the country’s economy is dominated by the Pakistani army and the revenues from its business operations are extremely opaque (Siddiqa 2017). Instead, the following case illustration employs a readily available but rather coarse and imprecise indicator of Pakistan’s state capacity: the country’s annual economic growth rate (see Faridi et al. 2021, 183ff.), ¹² which reflects the ease with which the government can raise money to fund its security objectives in a given fiscal year.

The remaining section traces the evolution of US-Pakistani counterterrorism collaboration over the course of four periods: first, from 9/11 until the mid-2000s; second, to the global financial crisis of 2007–8; third, to the early 2010s; and fourthly, for most of the remaining decade. The beginning of the new millennium was marred by substantial economic challenges for Pakistan, which contributed to a relatively steep downturn in its annual growth rate and deprived the country of access to private capital markets (Faridi et al. 2021, 183). Regardless of massive injections of US military assistance during this low-capacity phase, Islamabad’s limited financial resources posed an obstacle to mounting sophisticated anti-terror operations.

Pakistan’s obsession with India detracted from the country’s mountainous tribal areas in the west bordering Afghanistan, which have been a jihadist safe haven where Islamabad’s security forces have struggled to project power. Especially during economic slumps, Islamabad repeatedly felt obliged to deploy its scarce military resources and reinforcements to face off against Indian armed forces in preparation for a possible war. However, such military buildup along the eastern border with India curtailed Islamabad’s ability to amass a troop size large enough to successfully execute a counterterrorism campaign against al-Qaeda cadres. Pakistan’s security establishment regarded military spending on maintaining conflict-readiness or even fighting on two fronts as unsustainable (Jaffrelot 2018). Notwithstanding this (‘mis’)allocation of troops, the influence bought by American aid led to the incarceration and extradition of numerous al-Qaeda members in the early years after September 11, including several high-value targets (Puri 2012, 39, 46; see also Tankel 2018, 140).

An economic upswing would permit Islamabad to field additional security personnel and equipment along the Afghan border as well as procure armaments more suited for counterinsurgency rather than conventional warfare (Smith 2007, 14). However, even though the second period ushered in surging state capacity in Pakistan, its modest counterterrorism effort of the early years tapered off, together with a drop in US aid payments (Tankel 2018, 141). ¹³ The association between the volume of American aid deliveries to Pakistan and its performance as an ally in the War on Terror was positive. Ultimately, the US had been “paying a high price for getting just enough in return to keep it relevant and interested in continuing the aid” (Zaidi 2011, 16).

It became evident that the enticement of Washington’s pledge to provide security support proved insufficient to elicit a satisfactory level of unrelenting counterterrorism effort from Islamabad. This held true even in periods of Pakistan’s high state capacity, which would actually give the government the necessary financial leeway to enhance its counterterrorism measures. “It was clear that U.S. policy toward Pakistan was not working. Something had to change. Pakistani leaders had complained for years that the United States pursued a transactional relationship with their country while forging a strategic one with India” (Tankel 2018, 142f.). They also voiced anxieties that Washington might abandon their country once its security goals in Afghanistan were accomplished, as it allegedly did in the 1990s when the Soviet Union’s withdrawal from Afghanistan and its eventual demise diminished Pakistan’s geopolitical importance in the containment of communism (Ali 2019, 47; Fair et al. 2010, 148ff.). ¹⁴ For this reason, the Obama administration aimed to augment the efficiency of US aid as a counterterrorism tool by offering Pakistan a strategic rather than a myopic relationship. In particular, US officials strove to “reassure Pakistan of Washington’s long-term commitments to its military needs and help bolster its anti-insurgent efforts” (Schmitt and Sanger, 2010).

With Pakistan’s economy hit hard by the global financial crisis, which marked the beginning of the third period, the renewed deterioration in state capacity contributed to Islamabad’s recurrent reluctance to redeploy large portions of the military along the western frontier with Afghanistan, opting instead to station the bulk of the army on the eastern front to engage India in the (unlikely) event a war breaks out between the two countries (Fair et al. 2010, 167; Rashid 2008). At this decisive juncture, Washington chose not only to ramp up its security assistance to buttress the government in Islamabad, but also sought to demonstrate its steadfast commitment to Pakistan’s security. To this end, the US Congress passed bipartisan legislation in 2009 authorising the appropriation of vast amounts of foreign assistance to Pakistan beyond 2014, when the US was expected to withdraw from Afghanistan. “Independent analysts viewed the legislation as a landmark expression of the U.S. Administration’s and Congress’s intent to provide significant, long-term support for its Pakistani allies” (Epstein and Kronstadt 2013, 13; see, for instance, Ali 2019, 47). Ultimately, the US administration presented Pakistan with a credible prospect to shore up its conventional military capabilities vis-à-vis India, especially at times when its national budget was excessively strained by sluggish economic growth (Tankel 2018, 143ff.; Puri 2012, 39ff.). In practice, Washington acquiesced to Islamabad’s use of some of its counterterrorism funds to balance against India (Fair et al. 2010, 161).

In fact, this American initiative to provide a tangible quasi-insurance against supposedly existential risks arising from Pakistan’s faltering economy and worsening fiscal position was instrumental in increasing Islamabad’s readiness to exert more intensive and lasting counterterrorism effort in the ensuing fourth period of high state capacity in the country. Against the background of restoring macroeconomic stability and a strong upturn in growth rates for most of the remaining decade, Pakistani decision-makers felt confident enough to continue receiving security support from their American counterparts to actually decide to thin out the army’s presence on the Indian front and instead concentrate considerable segments of the freed-up security forces on the border with Afghanistan. Furthermore, Pakistan’s security services mounted a series of well-resourced and comprehensive military incursions into both the rural regions of the northwest as well as the urban areas of the country, from which they had previously abstained. Albeit being less extensive than US policy-makers had hoped, these accelerated anti-jihadist activities yielded vital counterterrorism gains and thus earned recognition in Washington (Lalwani 2015; Tankel 2018, 150ff.; see also Rehman et al. 2017). ¹⁵

Remarkably, in the midst of Pakistan’s economic recovery and the concomitant budgetary consolidation, Islamabad stepped up its efforts to disrupt and degrade terrorist networks against the backdrop of dwindling US aid flows (Manson and Bokhari 2018), rendering the correlation between the scope of American security assistance acquired and the scale of counterterrorism effort undertaken negative. In accordance with the model’s predictions for high-capacity periods under optimality, Pakistan’s strengthened counterterrorism measures were largely self-financed from domestic income sources (Ali 2019, 52). In sum, although Washington and Islamabad remain fundamentally at odds on security issues and unable to commit, the regular transfer of US military aid that is responsive to the Pakistani government’s ability to draw on other sources of revenue ensured the country’s ongoing fight against terrorism.

US–Yemen

That the Pakistani efforts to defeat al-Qaeda had been at least partially worthwhile is evidenced by the fact that “most of its members are believed to have moved to Yemen expecting a more hospitable” operating environment (Mumtaz 2010, 4). Similar to al-Qaeda’s modus operandi in Pakistan’s tribal badlands subsequent to 9/11, al-Qaeda-aligned fighters who fled Pakistan have established sanctuaries in the rural areas of Yemen’s rugged mountains, over which the central government in Sanaa lacks firm control (Baron et al. 2017; Byman 2008, 23). This led Washington to rank the group as the second most formidable and urgent jihadist danger to US security interests after the core al-Qaeda organisation in Pakistan (Boucek 2009, 22). For the Yemeni regime, however, its divergent security paradigm meant that indigenous security concerns took precedence over the menace posed by al-Qaeda (Rugh 2010). I confine the following analysis to one important opposition group within the country that had imperilled Yemeni president Saleh’s position of power since launching an armed revolt against the state in the 2000s: the so-called Houthi clan also known as the Houthis.

Once Yemen became a major battleground in the American War on Terror, the Bush administration harnessed the lure of military assistance to persuade Saleh to capture or neutralise (high-value) terrorists affiliated with the local branch of al-Qaeda (Byman 2008, 37) without US troops having to directly engage jihadists (Baron et al. 2017). Washington’s offer of security support was eagerly accepted by the Yemeni regime, since it had few other substantial sources of government funding (Boucek 2009, 9; Rugh 2010, 113) and the country’s precarious political and security context restricted foreign investment (Whalley and Wadhah, 2023). Yemen’s fragile cohesion may have broken down if Sanaa had run out of money to bankroll its security apparatus.

After all, “it is the lack of sufficient resources that undermines the president’s grip on power: his ability to buy political support […] and to deal effectively with the conflicts in the country depends on the scant finances at his disposal” (Hellmich 2012, 624f.). Yemen’s chief source of domestic revenue during the time under review was its oil and gas sector (Whalley and Wadhah, 2023), constituting a suitable indicator of the country’s state capacity. ¹⁶

The remaining section examines US-Yemeni counterterrorism collaboration dynamics across three periods: first, from late 2001 to the middle of the decade; second, during the global financial crisis; and third, up until 2012. The initial period in the early 2000s unfolded amid Yemen’s peak state capacity. Irrespective of the ballooning expenses for the acquisition, recruitment and deployment of military materiel and forces to clamp down on jihadist cells (Baron et al. 2017; Knights 2010), the national budget was kept afloat by soaring hydrocarbon rents, enabling Sanaa’s security forces to detain or execute all relevant terrorists at the time (Rugh 2010, 110; Johnsen 2013, 90, 111). By 2004, the al-Qaeda network in Yemen had apparently collapsed, leading Washington (erroneously) to believe that there was no need to sustain Sanaa’s largely successful counterterrorism campaign. The short-lived American interest in Yemen waned dramatically (Tankel 2018, 206, 217). Once Washington estimated that al-Qaeda had been defeated in the country, it “effectively abandoned Yemen in 2005” by pulling or curtailing security support programmes (Hellmich 2012, 626). It was also these abrupt cessations of US relief measures that contributed to al-Qaeda’s ability to regenerate itself (Johnsen 2013, 176ff.). The US resumed its engagement once the group’s unanticipated resurgence in 2006 clarified to Washington that al-Qaeda in Yemen had never been close to defeat.

In the second period, Yemen’s state capacity began to plummet. The precipitous fall in international oil prices in the wake of the global financial crisis placed an enormous burden on state finances. The regime in Sanaa encountered increasing difficulties in repelling the Houthi insurrection (Freeman 2009, 1008ff.), which consumed much of its limited attention, security assets and national budget, including US counterterrorism aid. As a consequence, Yemen’s armed forces were overstretched to adequately address the imminent threat emanating from al-Qaeda elements within the country (Boucek 2009, 14f.; Mumtaz 2010, 3ff.). Saleh requested further military support from Washington (Johnsen 2013, 184), since he thought he would be rewarded for his cooperation against al-Qaeda. But his entreaties were denied, mainly because US officials suspected the Yemeni leader of planning to use aid money to fight the Houthis (Baron et al. 2017; Tankel 2018, 236). Neither the Bush nor the Obama administration showed any inclination to sponsor Sanaa’s ambition to repress internal unrest with force (Knickmeyer 2010), and even went to great lengths to implement safeguards against the diversion of American funds (Scahill 2013, 131; Schmitt and Shane 2010).

However, the US failure to help rehabilitate Yemen’s severely depleted budget hampered the establishment of a consistent and durable counterterrorism collaboration between the two countries. Since Saleh had been given the impression that Washington was a fickle security partner that would ultimately abandon him again, ¹⁷ his initially cooperative posture began to lessen (Scahill 2013, 280; Tankel 2018, 222). As aid deliveries to help cope with the exigent challenges the Yemeni regime was facing domestically were called into question, Saleh became increasingly reluctant to dedicate scarce financial and military resources to the tough and unpopular fight against jihadists (Tankel 2018, 54, 225).

This opportunistic security cooperation remained unsatisfactory throughout the third period, which lasted from 2009 until Saleh was ousted in 2012. During this time, Yemen’s crude oil output saw a gradual yet inexorable contraction. Nevertheless, Sanaa’s state capacity oscillated between high and low, in tandem with its hydrocarbon revenues, which were episodically boosted by high world market prices for oil. US donations remained predominantly crisis-driven, geared towards averting the most acute perils from transnational terrorism on an ad hoc basis, but the release of military assistance was unresponsive to Yemen’s financial woes stemming from its internal power struggles (Baron et al. 2017; Mumtaz 2010, 5). Without sustainable sources of funding, the country’s “security forces sometimes took the fight to […al-Qaeda], but not with the frequency or skills necessary to keep it on the defensive” (Tankel 2018, 225; see also Byman 2008, 22). The Yemeni “government was committed to doing just enough to keep aid flowing. […] Washington could threaten to withhold or cut off aid, but this might further erode stability in Yemen and lead it to curtail its already limited counterterrorism operations” (Tankel 2018, 224f.). ¹⁸ As a result, the association between American aid transfers and Yemen’s counterterrorism performance was positive. In other words, a noticeable smoothing of Sanaa’s security expenditures over time was absent. Although the sporadic effectiveness of US assistance manifested itself in occasional reports of surging numbers of al-Qaeda operatives declared dead or in custody (Hellmich 2012, 621), its pro-cyclical provision rendered it comparatively inefficient as an anti-terror instrument as time went by (Tankel 2018, 206ff.).

Once again, it was not until the Obama administration took office in 2009 that the US sought to secure steadier cooperation from Sanaa by trying to remedy the intermittent and unpredictable nature of American financing. Embracing an explicitly more forward-looking approach to counterterrorism, Washington leveraged the promise of permanently elevated security support at a moment when Yemen’s state capacity was under grave strain (Tankel 2018, 226). Unfortunately, this policy initiative came too late. Saleh resigned in 2012 in the wake of the ‘Arab Spring’ uprisings. With the outbreak of violent turmoil throughout Yemen, oil and gas production came to a standstill, the country disintegrated politically, and finally descended into civil war in 2015.

Brief Discussion

The Obama administration’s attempts to put US counterterrorism collaborations with the regimes in Islamabad and Sanaa on more sustainable footings, as well as to dispel the image of American unreliability, were more successful in Pakistan than in Yemen. Washington’s intervention aimed to reinforce the parties’ future orientation, which gave rise to equilibrium behaviour whereby Islamabad intensified its counterterrorism effort in high-capacity periods in exchange for the opportunity to tap US aid funds in low-capacity phases.

Two aspects appear to be decisive for the discrepancy in efficiency between both influence attempts. First, a more institutionalised form of cooperation was established with Islamabad. Washington’s disengagement from Yemen stands in stark contrast to the more formalised and enduring security relationship it built with Pakistan since 2001. For instance, the Yemeni government did not receive reimbursements through CSF allowances for costs incurred in counterinsurgency operations (Fair et al. 2010, 150ff.). ¹⁹

Second, Washington also adopted a more permissive stance towards Islamabad than Sanaa regarding the redirection of its foreign aid. Whereas US policy-makers proved unwilling to extend American counterterrorism subsidies to Sanaa to quell domestic upheavals, Washington exhibited strikingly less reluctance to cross-finance Islamabad’s interstate feud. This salient asymmetry may thus be ascribed to the disparate sources of risk to which the Pakistani and Yemeni regimes felt exposed: external and internal. Indeed, Scahill (2013, 280), for example, reports that a senior US government official “told Saleh it would be against the law to give him military support against the Houthis because the United States considered the Houthis a ‘domestic insurgency’”.

In sum, the evidence suggests that Yemen was not provided with an American quasi-insurance policy against an aggravating threat environment comparable to that afforded to Pakistan. However, if Washington aims to both outsource much of its counterterrorism campaign and improve the return on its security investments, American decision-makers should be prepared to tolerate some periodic rerouting of US military and financial assets. As Kagan and Harnisch, 2010 “have found repeatedly in similar situations around the world (particularly in Pakistan), local governments will not focus on terrorist groups that primarily threaten the U.S. or their neighbors at the expense of security challenges that threaten them directly. A strategy that attempts to pressure or bribe them to go after our enemies is likely to fail. […President Saleh] is an unpalatable partner, and we don’t want to be drawn into Yemen’s internal conflicts more than necessary. But he is the only partner we have in Yemen. If we want him to take our side in the fight against al Qaeda, we have to take his side in the fight against the Houthis. […] This strategy will require helping Yemen defeat the Houthi insurgency” (see also Baron et al. 2017).

Conclusion

As a strategy for mitigating transnational terrorist dangers on foreign soil while minimising American troop involvement, Washington enlisted the help of external proxy forces in exchange for conditional aid. Nevertheless, critics of the US foreign assistance programme point out the marked jump in funding for security partners, despite a notable decline in counterterrorism effort (e.g. Fair et al. 2010, 160). In general, many observers doubt the effectiveness of aid conditionality given policy disagreements between donors and recipients and their inability to commit. As Krasner (2012, 96) contends, analysts often conclude that “the United States is better off treating [a counterterrorism partner such as] Pakistan as a hostile power than continuing to spend and get nothing in return.” In particular, most sceptics agree that the US “must prevent the diversion of assistance to bloody internal conflicts such as the […Yemeni] government clash with the Houthis” (Knights 2010, 2; see also Boucek 2010, 2).

However, having illuminated the dynamic properties of conditional aid partnerships in light of recipients’ fluctuating revenue streams, this paper’s formal model uncovers that an inverse association between foreign assistance and policy concessions could obscure an actually efficient exercise of influence. The mechanism behind this finding is the insurance function of foreign aid. Since financial shortages limit a recipient’s capability and willingness to carry out the donor’s bidding, the efficacy of conditional aid is contingent upon his state capacity. An ‘insurance coverage’ provides reliable access to supplementary security support during phases of sudden fiscal distress. This renders the donor’s financial backing sufficiently attractive overall as to give the recipient a strong impetus to devote substantial domestic resources to advance the donor’s security imperatives during periods of fiscal abundance. The resulting smoothing of the recipient’s security spending ensures that continuous pressure is maintained on jihadist networks, which is necessary to keep the terrorist threat at bay.

The model’s insights are corroborated by empirical assessments and yield a twofold conclusion: First, donors could extract more from recipients by cultivating longer-term assistance partnerships and, counterintuitively, accepting a certain degree of aid reallocation. Second, decision-makers should temper their expectations about the influence that conditional aid can bring to bear. The delivery of financial assistance alone is rarely adequate to modify a recipient’s perception of his threat environment or alter his strategic calculus. It is therefore difficult and expensive “to incentivize a partner to conduct comprehensive counterterrorism operations if it does not consider the target of these operations to be a threat” (Tankel 2018, 22).

Future research could take into account a recipient’s capacity for storing wealth or the presence of renegotiation-proof punishments. The accumulation of financial assets would counteract the aid sanction threats used to enforce redistributive measures. Furthermore, the imposition of a permanent aid cut-off may not be credible, insofar as it is not immune to renegotiation. By enacting such a penalty, the donor would forego all future gains from the surplus internal to the risk-sharing opportunities that prompted the parties to collaborate in the first place. In other words, the donor would be better off relaxing a total embargo on foreign assistance during the punitive phase. Therefore, incorporating the possibility of savings or renegotiation-proofness constraints would probably dilute the equilibrium intensity of aid conditionality.

Supplemental Material

Supplemental Material - Foreign Aid and Policy Concessions: Counterterrorism Assistance as Insurance

Supplemental Material for Foreign Aid and Policy Concessions: Counterterrorism Assistance as Insurance by Boris Brekhov in Journal of Conflict Resolution.