Ƶ

An open access publication of the Ƶ
Fall 2017

Building Security Forces & Stabilizing Nations: The Problem of Agency

Author
Stephen D. Biddle
View PDF
Abstract

After fifteen years of war in Afghanistan and Iraq, many now see “small-footprint” security force assistance (SFA) – training, advising, and equipping allied militaries – as an alternative to large U.S. ground-force commitments to stabilize weak states. SFA, however, confronts challenges of interest misalignment between the United States and its typical partners. The resulting agency losses often limit SFA’s real ability to improve partners’ military effectiveness. For SFA, small footprints usually mean small payoffs.

STEPHEN BIDDLE is Professor of Political Science and International Affairs at George Washington University. He is the author of Military Power: Explaining Victory and Defeat in Modern Battle (2004) and has published widely on national security and defense policy.

Security force assistance (SFA) – training, advising, and equipping allied militaries – is an increasingly common U.S. response to threats emanating from weak states. Many Americans have grown tired of large U.S. land wars in such places after more than ten years of continuous conflicts in Afghanistan and Iraq involving as many as 160,000 U.S. troops. Yet the world remains a violent place, and the United States has interests in a number of unstable parts of the world. For many, SFA offers a means to secure such real but limited interests without the massive U.S. ground commitments of the last fifteen years. In fact, “small-footprint” SFA has become a major pillar of U.S. national security policy.

Yet its actual military efficacy has been little studied. This essay thus presents a systematic analysis of SFA’s ability to improve allies’ military effectiveness.

My central finding is that effective SFA is much more elusive in practice than often assumed, and less viable as a substitute for large unilateral troop deployments. For the United States in particular, the achievable upper bound is normally modest, and even this is possible only if U.S. policy is intrusive and conditional, which it rarely is. This is because SFA is best understood as a principal-agent problem, and one whose structural conditions promote large agency losses for the SFA provider. That is, the conditions under which the United States provides SFA commonly involve large interest misalignments between the provider (the principal) and the recipient (the agent), difficult monitoring challenges, and difficult conditions for enforcement: a combination that typically leaves principals with limited real leverage and that promotes inefficiency in aid provision. To overcome these challenges requires atypical interest alignment between the United States and its SFA partner, a larger U.S. footprint than many would prefer, intrusive U.S. policies designed to monitor its ally’s behavior and enable strict conditionality in aid provision, or ideally all of the above. These conditions are not impossible, but the combination has not been a common feature of U.S. security force assistance in the modern era. Nor is it likely to become so in the future: in principle, U.S. policy-makers can design SFA programs to be intrusive and conditional, but it is much harder to create political interest alignment, and this is often absent.1

Principal-agent (PA) theory comprises a body of ideas originally developed by economists to explain interactions between parties to a contract and subsequently generalized and adapted to a wide range of situations in which one actor (the principal) delegates authority to another (the agent) to carry out actions on its behalf. In political science, it has been applied to explain interactions between elected officials and bureaucrats, legislators and committees, civil authorities and the military, domestic agencies and multinational organizations, or guerillas and state patrons, among many others.2

At their root, all such delegation decisions, and thus all of PA theory, are cost-saving strategies. They enable principals to undertake manufacturing, home repair, regulation, legislation, or national defense at a lower cost than doing it themselves. But in exchange, the act of delegation creates problems. In particular, the Գ貹’s interests always differ from the Գ’s to some degree: homeowners want tireless work at low cost but carpenters want high wages for lighter work; civilians want interservice cooperation and low defense budgets, officers want generous funding for their own service and its priorities. Principals can try to overcome this interest asymmetry and impose their preferences through conditionality (paying only when satisfactory work is complete or cutting budgets for services that decline to cooperate) or other enforcement means. But enforcement requires monitoring to know whether and how well the agent is performing, and agents typically know more about their efforts and circumstances than principals do. To overcome this information asymmetry, principals must spend resources to gather data on the agent and its work. Yet the more the principal spends on monitoring, the more expensive the project becomes and the less well the arrangement satisfies the original purpose of reducing cost. Payment, moreover, is a promise of future benefit if the agent “works” (serves the Գ貹’s interests), whereas enforcement is a threat of future sanction if the agent “shirks” (serves the Գ’s self-interest instead); effectiveness in either role turns on the Գ貹’s credibility. Principals must reassure agents of their promises, but the more reassurance they provide the less credible their threat of sanctions becomes, and vice versa: a principal whose commitment to support the agent is unshakable encourages the agent to take advantage and shirk with less fear of penalty. Moral hazard on some scale is thus inevitable in all PA transactions. These problems of interest asymmetry, information asymmetry, and moral hazard thus impose an inherent agency loss, or divergence between the outcome the principal seeks and the outcome the principal obtains: delegation to an agent can reduce costs, but it typically produces imperfect performance to some degree, and often the greater the cost saving, the more imperfect the performance.3

Security force assistance is a classic PA problem. In SFA, the United States is the principal, the ally receiving the aid is the agent, and the Գ貹’s aim is to meet a threat to American security more cheaply than by sending a large U.S. ground force to do the job directly. As with any other PA problem, SFA is thus subject to agency loss as a consequence of interest asymmetry, information asymmetry, and moral hazard; unfortunately, the particular circumstances of SFA promote agency losses that are much larger than many SFA advocates expect.

Large interest asymmetries, for example, are ubiquitous in U.S. SFA. Of course, no two states ever have identical interests. This is true even for close allies like the United States and Great Britain: during World War II, divergent U.S. and British interests led to tension over the priority placed on campaigns in Southern Europe and North Africa, for example, where British postwar geopolitical and colonial interests conflicted with America’s.4 U.S. SFA, moreover, is rarely provided to allies as close as Britain. The top fifteen recipients of U.S. SFA between 1980 and 2009 have included Pakistan, which provides safe haven for Al Qaeda’s global headquarters and for Taliban militants who have killed thousands of U.S. soldiers in Afghanistan; Sudan, which has been accused of widespread ethnic cleansing against its non-Arab minority; four of the top seven state sources of foreign fighters for ISIL; and Afghanistan, which ranks fourth on Transparency International’s list of the world’s most corrupt states (placing behind only Somalia, a top-twenty-five recipient of U.S. SFA, Sudan, a top-fifteen recipient, and North Korea).5

In fact, this is a systematic phenomenon. If we use UN voting patterns as a proxy for interest alignment, then there is a statistically significant negative correlation between U.S.-partner interest alignment and U.S. SFA provision: the closer the interest alignment, the less likely the United States is to provide military aid.6 We see a similar relationship if we consider corruption: a state’s rank on the Transparency International list of most corrupt states correlates directly with its rank on the list of U.S. SFA recipients, with an ability to reject the null hypothesis of no relationship at the 0.1 level.7

This relationship is not an accident. The United States rarely gives SFA to Switzerland or Canada because they do not need it; the states that need it are rarely governed as effectively as Switzerland or Canada.8 And the governance problems that give rise to the U.S. interest in SFA often simultaneously promote interest divergence between the United States and its partner.

Regional instability, terrorist infrastructure, and humanitarian crises – the kinds of real-but-limited threats to U.S. interests that SFA is often meant to address – are strongly associated with weak states and corrupt, unrepresentative, clientelist regimes. In such states, political order often requires what Douglass North, John Wallis, and Barry Weingast have called a “double balance,” wherein the distribution of economic spoils matches the distribution of power among potentially violent elites.9 Regimes that allow the internal balance of power to misalign with the balance of rents risk violent overthrow, and in such systems, the threat of violence from armed elites within the state apparatus often exceeds the real threat from foreign enemies, international terrorists, or antigovernment insurgents. Rational leaders of such states thus cannot treat their militaries as disinterested defenders of the state against foreign enemies; the armed forces are natural rivals and potential threats. Order under such conditions thus requires regimes to undertake some mixture of appeasement, mutual implication, and enfeeblement toward their own militaries. Appeasement strategies buy off potential rivals with economic spoils proportional to the rivals’ real power; for armed forces with ready access to violence, this can create an officer class accustomed to economic privilege as the price of obedience, with little incentive to pursue disinterested expertise. Mutual implication encourages loyalty by implicating officers in criminal or unethical regime behavior, tying officers’ fate to the regime’s. Enfeeblement shifts the internal balance of power by deliberately weakening armed forces’ ability to seize power or intimidate rivals. For example, many such regimes create multiple, overlapping lines of military command, discourage lateral communication among officers, create redundant security organizations, and replace foreign-trained military technocrats with reliable political loyalists.10 Foreign military aid (such as U.S. SFA) is often welcome in such settings (especially when it takes the form of financial transfers or gifts of equipment), but not for the purposes the providers often assume; instead, regimes typically see such aid as a form of largesse, an additional source of benefits to be distributed to buy political loyalty.11 More broadly, under the conditions common among U.S. SFA recipients, the regime’s interests are typically focused less on external enemies than on internal threats from rival elites, and especially the state military itself, which is often seen as a threat at least equal to that of foreign enemies.

By contrast, U.S. interests in such states typically focus on external threats, and especially transnational terrorists or aspiring regional hegemons.12 U.S. SFA is commonly intended to strengthen partner militaries’ ability to meet these ostensibly common threats by improving the partners’ military proficiency. But whereas Americans often assume that these external dangers threaten the partner as well as the United States, and that strengthening the partner military will therefore serve both parties’ interests, this is often mistaken. In fact, the kind of powerful, politically independent, technically proficient, noncorrupt military the United States seeks is often seen by the partner state as a far greater threat to their self-interest than foreign invasion or terrorist infiltration. Increased military capability destabilizes the internal balance of power; diminished cronyism and corruption weakens the regime’s ability to control the empowered officers. The result is a commonplace and major divergence in U.S. and partner interests that derives from the very issues that created the demand for U.S. SFA in the first place.

The monitoring and enforcement strategies normally employed to mitigate interest asymmetries in PA relationships, moreover, face systematic barriers in SFA. As a cost-reduction strategy, SFA’s whole purpose is to limit the U.S. “footprint”: that is, its presence on the ground in the partner country. Hence, by design, there will be few U.S. monitors in the country to observe the partner’s behavior. And partners are adept at using U.S. aid to pursue their own interests rather than their provider’s, employing techniques that are very hard for a handful of U.S. monitors to detect. Financial and material aid are fungible: even if the nominal assistance goes to professional military purposes, this can displace state funding that can then be redirected to political allies as rents, leaving the host military no more effective than before. Training can be used as a status reward for reliable loyalists, rather than a means of improving technical proficiency. Material aid can be diverted onto the black market. Aid money transferred to the state treasury can be laundered and directed to other purposes. To detect such abuses requires intrusive, labor-intensive monitoring of a nominal ally’s behavior, and often a sustained presence by enough U.S. personnel to thwart partner concealment. In other settings, principals can often rely on monitoring via independent reporting from the press, from domestic rivals of the agent, or from routine overseers such as auditors or oversight agencies;13 in SFA, by contrast, press freedom in the recipient state is often minimal, domestic rivals are often either repressed or complicit, and the only trustworthy auditors would be the U.S. personnel whose presence the United States is trying to minimize. The lighter the U.S. footprint, the harder effective monitoring becomes.

(In commercial PA relationships, principals can combat information asymmetries by paying agents based on outcomes rather than monitoring behavior directly: if the agent delivers a satisfactory product, the principal pays, and vice versa, whether the principal can observe the Գ’s level of effort or not. In SFA, however, outcome-based monitoring faces major causal attribution challenges: if the agent fails in combat, is this because the agent is shirking or because war is uncertain and outcomes are influenced by a host of exogenous variables beyond the Գ’s control? To overcome information asymmetries in SFA thus requires direct monitoring of the Գ’s behavior.)14

Monitoring, moreover, is useless without enforcement, which normally means conditionality: a credible U.S. threat to withdraw aid from allies who misuse it. For SFA, however, conditionality is often very hard to implement in practice. In the economics literature, conditionality is often proposed as a means of mitigating moral hazard: agents will not exploit their information advantages by shirking if principals can condition their payments on successful completion of the work. Yet conditionality is subject to moral hazard problems itself, and these loom particularly large for SFA.

Conditionality involves two promises of future action: a promise to withhold payments if the agent shirks, and a promise to pay if the agent works. Because both are promises of future action, credibility is always an issue. But the credibility of the threat and the credibility of the promise are in tension. The more forcefully the United States threatens an ally with aid withdrawal in the event of shirking, the more a rational ally will doubt the U.S. promise to follow through with its commitment if the ally works. When a U.S. administration threatens an ally with aid withdrawal, this often undermines U.S. domestic support for the ally (as has been the case with Pakistan, for example). From the ally’s perspective, why risk domestic instability by forcing reform on an unwilling military for the sake of an American patron whose commitment to your survival is so contingent and domestically controversial? How does the ally know that, if the result is a coup or internal schism, the Americans will save them, when U.S. polls show American indifference to their fate in the aftermath of a U.S. campaign of public pressure on your regime? Threats of conditionality thus create a problem of moral hazard on the Գ貹’s part: once the allied regime has reformed as the principal wanted and has accepted the associated internal risks, the apparently indifferent Americans may pocket the benefits to U.S. interests but then walk away and withhold critical assistance in the event of internal crisis.

Conversely, the more the U.S. principal seeks to reassure the agent that U.S. promises are good and aid will be forthcoming if only the agent accepts the internal risks of professionalizing its military, the greater the risk of moral hazard in the other direction. To build U.S. domestic support for aid, administrations often frame the ally as vital to U.S. national security; a credible promise of aid is normally built on a foundation of American assurance – both to the ally and to the U.S. public and Congress – that the ally’s survival is essential to American self-interest. The more forceful these assurances, the more a rational ally will doubt the accompanying U.S. threat to halt aid if the ally shirks. From the ally’s perspective, why risk domestic instability by forcing reform on an unwilling military when the external threat such reform is meant to confront will presumably be met by the Americans on your behalf anyway? Promises and reassurance thus create a problem of moral hazard on the Գ’s part: they encourage the agent to shirk on reforms, trading ineffectiveness against external enemies for internal stability in the belief that American aid will continue anyway and that American arms will ultimately save them if the external threat proves greater than expected.

And because conditionality requires both a credible threat and a credible promise, it is very hard in practice to overcome both problems of moral hazard at once. Success with one tends to undermine success with the other; efforts to balance the two run the risk that neither the threat nor the promise is fully credible. Conditionality in SFA thus poses a dual-commitment problem: it is difficult for the agent to credibly commit itself to work and not shirk if the principal “pays” the agent, but it is also difficult for the principal to credibly commit itself to pay the agent if the agent works.

This problem is compounded, moreover, if the agent has access to multiple principals and can threaten each with defection to the other if aid is withheld. For U.S. SFA to Iraq, for example, the Iraqi agent can respond to U.S. threats and conditions by turning instead to Iran for aid, and can use the opposite threat to reduce Iranian leverage in turn. The net result is a complex set of challenges that must be overcome for conditionality to be effective in SFA.

In domestic commerce, by contrast, contracts are enforceable by law. Legal costs give rise to agency loss even here, but the availability of legal recourse gives conditionality by contract provision a degree of inherent credibility. In SFA, there is no meaningful legal authority to enforce conditionality, hence the moral hazards inherent in delegation loom larger.

The net result in SFA is major agency loss much of the time. Agents whose interests often focus on domestic power balancing commonly use U.S. aid not to work by professionalizing their militaries, as the United States prefers, but to shirk by reinforcing clientelism. Limited U.S. monitoring often provides only ambiguous evidence of such shirking, and conditionality to enforce U.S. preferences on the use of aid is often undermined by moral hazard, rather than mitigating it. In the end, U.S. aid has much less ability to improve partners’ real military effectiveness than the scale of U.S. assistance would suggest.

This is not to say that aid is irrelevant (or adverse) to the partner’s military performance; even poorly used aid can be better than none at all. And the theory above suggests that the scale of agency loss, while often large, will vary with local conditions. As PA theory implies, agency loss is proportional to the degree of interest misalignment between the principal and the agent: where U.S. interests are more closely aligned with the partner’s, we can expect greater improvement in partner military effectiveness per dollar of SFA expenditure. PA theory also implies that the greater the Գ貹’s investment in monitoring and the more conditional the aid provision, the smaller the agency loss. Hence we can expect that where the United States monitors more intrusively and conditions aid more credibly, we should see greater military impact per dollar of SFA expenditure. The analysis above suggests that close interest alignment, intrusive monitoring, and credible conditionality will be rare for U.S. SFA, but where observed, these unusual conditions should promote greater improvements in the partner’s military than in more typical cases.

An illustrative example of these dynamics at work is the Second Iraq War. From 2003 to 2011, the United States invested over $25 billion in the Iraqi Security Forces (ISF), devoted tens of thousands of U.S. personnel to training and advising Iraqi forces, and, by 2007, deployed more than one hundred thousand other U.S. troops to provide security until the ISF could take over.15 Yet the Iraqi military that emerged from this immense effort collapsed in June 2014 when challenged by Islamic State fighters in Mosul. How could such a scale of assistance have failed to produce an ally who could defend its country against a militant group with only a fraction of its nominal strength?

The answer lies in a major interest divergence between the U.S. principal and its Iraqi agent. The U.S. and Iraqi governments had two very different visions for the ISF. The United States wanted a technically proficient force capable of defending all sects’ interests and focused on counterinsurgency warfare against both Sunni insurgents and Shiite militias. By contrast, the Iraqi Ibrahim al-Jaafari and Nouri al-Maliki regimes were focused on preserving their position in a mostly intra-Shiite struggle for political power in which the ISF was seen as a potentially decisive arbiter in a potentially lethal contest. For the regime’s purposes, a politically disinterested technocratic military of the kind the Americans sought would have been a danger, not an asset: not only would Jaafari or Maliki have been unable to ensure such officers’ personal loyalty in internal political jockeying, but both men would be likely to see American-trained technocrats as a kind of Trojan horse, a tool of American influence and interference that might undermine the consolidation of power in Jaafari’s or Maliki’s office. By consistently elevating sectarian loyalists over those more professionally inclined, the Iraqi government created strong incentives for members of the military to learn only those skills required to be a good loyalist militia, which does not include the ability to conduct modern, large-scale combat operations.16 By cultivating deliberate corruption in the officer corps, the regime created a financial incentive for military cooperation, and by turning a blind eye to death squad activity by government forces, the regime tied the complicit officers to its own fate.17 The results created an ISF whose performance was largely insensitive to U.S. aid and training: Americans could provide weapons and teach tactics, but a corrupt, politicized officer corps could neither absorb the training nor generate the combat motivation needed to persuade troops to risk their lives on behalf of such a project. As a result, the ISF never gained the ability to independently plan and conduct even medium-scale combat operations effectively. And when U.S. leverage diminished with the progressive withdrawal of U.S. combat forces, regime incentives that had been an important brake on military proficiency all along now had free reign with even less U.S. interference. Particularly when the violence began to wind down after 2007 and the number of U.S. troops on the ground began to shrink, Maliki began to systematically replace the few apolitical officers the United States had managed to install.18 Realistic training became less frequent and corruption even more common, the combination of which thoroughly undermined the SFA program Americans had invested in so heavily.

El Salvador, by contrast, is often presented as an example of SFA’s ability to substitute for large U.S. troop deployments. Between 1979 and 1982, a combination of $5 billion in U.S. aid and a small contingent of under two hundred American advisors helped the Salvadoran government survive the FMLN (Farabundo Martí National Liberation Front) insurgency. Without U.S. SFA, the government could well have fallen.

Yet even here the problem of interest misalignment and agency loss was serious. The Salvadoran regime shared the U.S. goal of preventing its overthrow, but sharply opposed U.S. pressure for military professionalization, economic reform, and political participation as means to this end, and were much less committed to ending the war than were the Americans. Salvadoran agrarian elites had relied for generations on an internal balance in which a handful of wealthy families shared rents from a sharply unequal economy that they controlled via repressive governance and a security apparatus that was both organized along semifeudal family lines to ensure its loyalty and bound to the regime by complicity in violence against political activists.19 American proposals for economic reforms that would undermine the financial basis of the traditional elite’s power thus posed existential threats to them, as did U.S.-advocated military professionalization that would weaken plutocratic control. For the ruling oligarchy, the system of economic and social privilege it enforced and the intraelite balance this created was thus at least as important as defeating the insurgency; in fact, for them, the counterinsurgency campaign was chiefly a means to preserving their wealth and influence, and the regime preferred to terrorize opponents rather than accept what they saw as self-defeating reforms.20 Once U.S. military aid had blunted the FMLN’s early hopes of toppling the government, these interest divergences between the U.S. principal and its Salvadoran agent made further progress increasingly difficult, and the war lapsed into a long stalemate that resolved only when the Cold War ended and mutual exhaustion enabled a negotiated compromise settlement. The net result was a real – but limited – payoff for SFA, even in an example that many see as its strongest case in point.

Such cases show the limits on SFA effectiveness under many conditions. Better performance is not impossible, but it requires circumstances that have been rare in practice. The Korean War offers an opportunity to observe such unusual circumstances and their effects.

When North Korea invaded the South in June 1950, the United States rapidly expanded a small prewar assistance mission into a force of almost three thousand advisors plus the equivalent of almost $1 billion in today’s dollars in annual military aid, with weapons and equipment sufficient for fifty thousand men. This was coupled with unusual intrusiveness and conditionality. The United States insisted on assuming command of ROKA (Republic of Korea Army) forces in 1950, and used its advisors in part as a fact-gathering agency for the U.S. command by reporting on Korean unit behavior and capabilities.21 U.S. advisors were given control of ROKA units’ budgets and were expected to oversee expenditures to ensure against black-market diversion of funds.22 The U.S. command took control of the ROKA’s personnel policy from early 1951, preventing old factions from operating and allowing young, competent officers to assume leadership positions. These young leaders adopted American military practices and reinforced the new emphasis on professionalism and meritocracy.23 U.S. leaders threatened withdrawal of weapons and support from underperforming ROKA units unless the Koreans demonstrated leadership and training worthy of that support; in 1953, American negotiators threatened total U.S. withdrawal if Korean President Syngman Rhee refused to accept American preferences on armistice talks.24

This program produced an unusual scale of improvement in allied military performance. Whereas the ROKA had offered minimal resistance to the initial invasion and scored systematically poorly in early U.S. advisor assessments, by 1953, ROKA battalions fought in coherent units, gave ground only when necessary, counterattacked with skill and motivation, and had proven themselves able to fight even Chinese regulars to a stalemate in head-to-head combat. Assessments from U.S. advisors reflected this improvement: reports from 1952 – 1953 frequently commented on the ROKA’s increased competence, noting that Republic of Korea soldiers had “showed improvement in every field of military endeavor.”25 By January 1953, ROKA units occupied 59 percent of the front line, met 87 percent of the enemy’s probes and attacks, and inflicted 50 percent of the enemy casualties.26

The results suggest that SFA thus can catalyze important improvements in recipients’ military effectiveness. But this does not happen simply because the patron provides resources. The ROKA had received nontrivial aid and training prior to the invasion, yet showed little ability to use it competently in the field until the military crisis of 1950 created appropriate incentives. The North Korean invasion and the ROKA’s chaotic retreat to the Pusan perimeter posed an existential crisis for Rhee: hostile conquest now posed a more immediate threat than internal violence, and his personal interests now aligned with the Americans’ in an urgent need to defeat an external enemy. With incentives aligned, U.S. aid became a powerful tool for improving allied effectiveness. But even then, interest alignment was not so perfect as to remove any potential for agency loss and inefficiency: aggressive monitoring and credible conditionality were needed to eliminate holdover corruption and limit subsequent backsliding into clientelist behavior. The Korean case shows that where conditions are conducive, agency losses in SFA can be mitigated – but it also shows how difficult that can be to accomplish in practice.

SFA is best understood as a principal-agent problem wherein agency losses will often be high. Major interest asymmetries are the norm. Monitoring is difficult and costly. Conditionality must overcome credibility dilemmas that can be managed but never wholly eliminated. These challenges normally preclude big payoffs from modest aid, and even large investments commonly yield disappointing results. SFA thus faces major challenges as a solution to the twenty-first-century dilemma of weak states posing real but limited threats to U.S. interests.

This does not make SFA useless, however. As the Korean case shows, U.S. and allied interests will sometimes align in ways that reduce agency losses, especially if U.S. policy is intrusive and conditional. Such alignment is rare, but when it happens, it offers an opportunity for efficient aid that makes a real military difference.

And even inefficient aid with serious agency losses can sometimes be worthwhile. More training and equipment is usually better than less, so SFA will typically improve recipient capability at least somewhat. If little is needed, then SFA may suffice. In El Salvador, U.S. SFA never produced an ESAF (Armed Forces of El Salvador) that could actually win the war, but it could at least avert defeat and sustain a grinding stalemate until exogenous events eventually enabled a settlement. Though many hoped for more, this was better than the alternative. In Iraq and Syria today, SFA is unlikely to truly defeat ISIL, but it can help drive a weak opponent back underground even if it cannot enable U.S. allies to stabilize populations who distrust them.27 If the mission is simply to contain ISIL rather than defeat it, then even an inefficient SFA effort with limited payoff could still suffice in a less demanding role. The less one asks, the better the odds that SFA can provide it.

It may also be possible to improve SFA implementation in ways that make it more effective in the future. Partly this means choosing one’s battles carefully: more Koreas and fewer Iraq-scale interest misalignments would certainly improve the prognosis.

SFA policies should also be more attentive to the recipients’ political interests and incentives. The policy debate tends to assume an apolitical capacity-building model for SFA in which military resources translate into military power in a straightforward way: the more training and equipment the United States provides, the better the ally’s effectiveness. If the ally is underperforming, the natural implication is to provide more aid. By contrast, a PA approach highlights the ally’s political interests as central for SFA. Hence, policies designed to realign the ally’s interests and create incentives to work and not shirk are essential. This approach is inherently political, and can often be highly coercive. The whole point of conditionality in PA theory is to manipulate allies’ incentive structures in ways that encourage them to work and not shirk; in a PA approach, if an ally is underperforming, the best response will often be to reduce assistance, not increase it.28

A more political understanding of SFA might also emphasize elite special forces, rather than regular conventional soldiers, both as providers and as recipients of assistance. As SFA providers, special forces can offer language skills, cultural awareness, and intelligence-gathering skills to serve as more-effective monitors of partner behavior, as a more-conditional PA approach to assistance requires. As SFA recipients, partners’ special forces are by definition small units whose very size makes them less destabilizing to the internal political balance in the host government. In the Philippines after 2001 and in South Vietnam in the 1960s and 1970s, host governments were more willing to tolerate professionalization for small special forces units than for their mass regular military, enabling more-efficient training with smaller agency losses per soldier trained.29

But while SFA can help if done properly under the right conditions, there are important limits on its utility: much of the time, conditions will not be suitable. In particular, many recipient regimes fear internal rivals within the governing elite more than they fear the external threats the United States typically focuses on. For much of the U.S. experience in Iraq, this hamstrung SFA effectiveness, as it did in Afghanistan and in a range of cases from Vietnam to Mali to Nigeria to Pakistan.30 Such regimes are disproportionately likely to be candidates for U.S. SFA and, in these contexts, the United States rarely has the leverage it needs for major military improvements: when allies see existential risks in reform, even the sweetest carrots and strongest sticks available are unlikely to outweigh such incentives. More training and more equipment will not simply solve the problem in such cases and yield a capable, professional military. Apolitical capacity-building that ignores underlying interest asymmetries is subject to large agency losses and can at times make things worse by fueling the corruption and clientelism that undermines effectiveness.

Even so, SFA is still cheaper than deploying one hundred thousand soldiers. In a world of imperfect options, “enabling partners” may be the least imperfect for a given contingency. But SFA’s real costs and risks are easy to underestimate, and its military benefits have often been oversold. Overuse is thus a real danger: SFA can help, but only rarely will modest investments in training and equipment provide major improvements in effectiveness. And overdependence has real costs: ground force reductions may be necessary, but an SFA alternative does not make them free of risk. Under many conditions and for many purposes, a small military payoff is the most one can expect from a small SFA footprint.

ENDNOTES

1 The discussion below draws from the analysis in Stephen Biddle, Julia MacDonald, and Ryan Baker, “Small Footprint, Small Payoff: The Military Effectiveness of Security Force Assistance,” Journal of Strategic Studies, published online April 12, 2017, http://www.tandfonline.com/doi/full/10.1080/01402390.2017.1307745, and appears by permission. The analysis is based upon work supported by the U.S. Department of the Navy, Office of Naval Research, under Award No. N00014-14-1-0843.

2 For example, see John Ferejohn and Charles Shipan, “Congressional Influence on Bureaucracy,” Journal of Law, Economics, and Organization 6 (1990): 1–43; Roderick Kiewiet and Mathew McCubbins, The Logic of Delegation: Congressional Parties and the Appropriations Process (Chicago: University of Chicago Press, 1991); Daniel L. Nielson and Michael Tierney, “Delegation to International Organizations: Agency Theory and World Bank Environmental Reform,” International Organization 57 (2) (2003): 241–276; George W. Downs and David M. Rocke, “Conflict, Agency, and Gambling for Resurrection: The Principal-Agent Problem Goes to War,” American Journal of Political Science 38 (2) (May 1994): 362–380; Peter D. Feaver, Armed Servants: Agency, Oversight, and Civil-Military Relations (Cambridge, Mass.: Harvard University Press, 2003); and Idean Salehyan, “The Delegation of War to Rebel Organizations,” Journal of Conflict Resolution 54 (3) (2010): 493–515.

3 For more detailed treatments, see Jean-Jacques Laffont and David Martimort, The Theory of Incentives: The Principal-Agent Model (Princeton, N.J.: Princeton University Press, 2002); and Gary J. Miller, “The Political Evolution of Principal-Agent Models,” Annual Review of Political Science 8 (2005): 203–225.

4 Michael J. Lyons, World War II: A Short History, 3rd ed. (Upper Saddle River, N.J.: Prentice Hall, 1999), 185–188; and Rick Atkinson, An Army at Dawn: The War in North Africa, 1942–1943 (New York: Holt, 2002), 11–14.

5 Based on cumulative spending on Foreign Military Financing, the Military Assistance Program, “Section 1206,” International Military Education and Training, and Excess Defense Articles. The data are a subset of those used in Michael McNerney, Angela O’Mahony, Thomas S. Szayna, et al., Assessing Security Cooperation as a Preventive Tool (Santa Monica, Calif.: RAND Corporation, 2014), and were derived from United States Agency for International Development, The GreenBook, . See also Stephen Watts, Jason H. Campbell, Patrick B. Johnston, et al., Countering Others’ Insurgencies: Understanding U.S. Small-Footprint Interventions in Local Context (Santa Monica, Calif.: RAND Corporation, 2014), 173. On Pakistan’s support for the Afghan Taliban, see K. Alan Kronstadt, Pakistan-U.S. Relations (Washington, D.C.: U.S. Congressional Research Service, 2012), 4, 19. On ethnic cleansing in Sudan, see Ted Dagne, Sudan: The Crisis in Darfur and Status of the North-South Peace Agreement (Washington, D.C.: Congressional Research Service, June 15, 2011), 23–24. On the nationality of foreign fighters in ISIL, see Aaron Y. Zelin, ICSR Insight: Up to 11,000 Foreign Fighters in Syria; Steep Rise among Western Europeans, International Center for the Study of Radicalisation and Political Violence, December 17, 2013, . (Assuming Yelin’s “high estimate”; his “low estimate” implies four of the top six states would be among the top fifteen recipients of U.S. SFA.) On Afghanistan, see Transparency International, Corruption Perceptions Index 2014, .

6 Patricia L. Sullivan, Brock F. Tessman, and Xiaojun Li, “U.S. Military Aid and Recipient State Cooperation,” Foreign Policy Analysis 7 (2011): 275–294.

7 The Spearman’s Rho is -0.14 with a p-value of 0.09. This calculation assumes that the fifteen countries in NATO at the end of the Cold War received essentially no U.S. SFA from 2000 to 2010. SFA data are derived from McNerney et al., Assessing Security Cooperation as a Preventive Tool. Corruption was measured using the Transparency International Corruption Perceptions Index, available at http://www.transparency.org/research/cpi/.

8 For a more extensive discussion, see Daniel Byman, “Friends Like These: Counterinsurgency and the War on Terrorism,” International Security 31 (2) (Fall 2006): 99–109.

9 Douglass North, John Wallis, and Barry Weingast, Violence and Social Orders: A Conceptual Framework for Interpreting Recorded Human History (New York: Cambridge University Press, 2009), 20. For a similar analysis, see Stephen David, Choosing Sides: Alignment and Realignment in the Third World (Baltimore: Johns Hopkins University Press, 1991).

10 For example, see Stephen Biddle and Robert Zirkle, “Technology, Civil-Military Relations, and Warfare in the Developing World,” Journal of Strategic Studies 19 (2) (June 1996): 171–212; Risa Brooks, Political-Military Relations and the Stability of Arab Regimes (New York: Oxford University Press, 1998); James Quinlivan, “Coup-Proofing: Its Practice and Consequences in the Middle East,” International Security 24 (2) (1999): 131–165; Caitlin Talmadge, The Dictator’s Army: Battlefield Effectiveness in Authoritarian Regimes (Ithaca: Cornell University Press, 2015): 15–18; and Eliot Cohen, “Distant Battles: Modern War in the Third World,” International Security 10 (4) (Spring 1986): 143–171.

11 In fact, under such conditions, U.S. aid can actually make things worse by fueling corruption. On this dynamic, see Sarah Chayes, Thieves of State: Why Corruption Threatens Global Security (New York: W. W. Norton, 2016); and Vanda Felbab-Brown, Aspiration and Ambivalence: Strategies and Realities of Counterinsurgency and State-Building in Afghanistan (Washington, D.C.: Brookings Institution, 2012).

12 Andrew Boutton, “U.S. Foreign Aid, Interstate Rivalry, and Incentives for Counterterrorism Cooperation,” Journal of Peace Research 51 (6) (2014), argues that U.S. aid recipients prefer to arm against external rivals, whereas the United States prefers the opposite. Often, however, the recipients’ nominal interest in arming against foreign rivals actually serves the internal interest of placating a domestic military that benefits from this, as in Pakistan. See Azeem Ibrahim, “How America is Funding Corruption in Pakistan,” Foreign Policy, August 11, 2009; and Ayesha Siddiqa, Military Inc.: Inside Pakistan’s Military Economy (London: Pluto Press, 2007). My usage of internal and external is relative to the government itself: internal threats are those posed to government actors by other factions within the government (such as the military or other rival elites); external threats are those outside it (such as insurgents or neighboring states).

13 See Feaver, Armed Servants, chap. 3.

14 For a more detailed discussion, see ibid.

15 Amy Belasco, The Cost of Iraq, Afghanistan, and Other Global War on Terror Operations since 9/11, RL33110 (Washington, D.C.: Congressional Research Service, 2014), 62.

16 On ISF cronyism, see Catherine Dale, Operation Iraqi Freedom: Strategies, Approaches, Results, and Issues for Congress, RL34387 (Washington, D.C.: Congressional Research Service, 2009), 101.

17 On ISF corruption, see Learning from Iraq: A Final Report from the Special Inspector General for Iraq Reconstruction (Washington, D.C.: U.S. Government Printing Office, 2013), 95, 99. On death squad activity, see Michael R. Gordon and General Bernard E. Trainor, The Endgame: The Inside Story of the Struggle for Iraq, from George W. Bush to Barack Obama (New York: Vintage Books, 2012), 146–148, 185–187, 367, 542–543; and James A. Baker III and Lee H. Hamilton, The Iraq Study Group Report (New York: Vintage Books, 2006), 9–10.

18 Kenneth M. Pollack, “Iraq Military Situation Report,” Up-Front, Brookings Institution, June 14, 2014, http://www.brookings.edu/blogs/up-front/posts/2014/06/14-iraq-military-situation-pollack; and Yasir Abbas and Dan Trombly, “Inside the Collapse of the Iraqi Army’s 2nd Division,” War on the Rocks, July 1, 2014, http://warontherocks.com/2014/07/inside-the-collapse-of-the-iraqi-armys-2nd-division/.

19 David H. Ucko, “Counterinsurgency in El Salvador: The Lessons and Limits of the Indirect Approach,” Small Wars & Insurgencies 24 (4) (2013): 673–674; Mark Peceny and William D. Stanley, “Counterinsurgency in El Salvador,” Politics and Society 38 (1) (2010): 72–73; Benjamin C. Schwarz, American Counterinsurgency Doctrine and El Salvador (Santa Monica, Calif.: RAND Corporation, 1991): 59–62; and U.S. Central Intelligence Agency, El Salvador: Managing the Military, April 1, 1988, accession no. EL00233, Digital National Security Archive, George Washington University (DNSA).

20 Peceny and Stanley, “Counterinsurgency in El Salvador,” 73–75, 81–82; and Schwarz, American Counterinsurgency Doctrine, 45–49.

21 HQ Eighth U.S. Army Korea (EUSAK), Special Problems in the Korean Conflict, U.S. Army Military History Research Collection (Carlisle, Penn.: Army War College, 1952), 8.

22 Jongnam Na, Making Cold War Soldiers: The Americanization of the South Korean Army, 1945–1955 (Ph.D. diss., University of North Carolina at Chapel Hill, 2006), 65–66.

23 Ibid., 141–142.

24Bryan Gibby, The Will to Win: American Military Advisors in Korea 1946–53 (Montgomery: University of Alabama Press, 2012), 173–174.

25 Korean Military Advisory Group (KMAG) Command Report, December 1952, Section I, 1, as quoted in Gibby, Will to Win.

26 Ibid.

27For example, see Erin Banco, “ISIS Unable to Break through Shiite Militias Around Baghdad, Wages Urban Warfare with Suicide Attacks,” International Business Times, October 16, 2014.

28See also Walter C. Ladwig III, “Influencing Clients in Counterinsurgency: U.S. Involvement in El Salvador’s Civil War, 1979–92,” International Security 41 (1) (Summer 2016): 144.

29On the Philippines case, see Linda Robinson, Patrick B. Johnston, and Gillian S. Oak, U.S. Special Operations Forces in the Philippines, 2001–2014 (Santa Monica, Calif.: RAND Corporation, 2016); and Thomas Lum and Ben Dolven, The Republic of the Philippines and U.S. Interests–2014 (Washington D.C.: Congressional Research Service, 2014), esp. 2–4. On the South Vietnamese case, see Talmadge, The Dictator’s Army, 71–138.

30 See Seth Jones, Olga Oliker, Peter Chalk, et al., Securing Tyrants or Fostering Reform? U.S. Internal Security Assistance to Repressive and Transitioning Regimes (Santa Monica, Calif.: RAND Corporation, 2006); William Rosenau, U.S. Internal Security Assistance to South Vietnam (Santa Monica, Calif.: RAND Corporation, 2007); Jordan Olmstead, “Fixing America’s Aid to Afghanistan and Pakistan,” The Diplomat, December 23, 2014; Gordon Adams and Richard Sokolsky, “Governance and Security Sector Assistance: The Missing Link–Parts I and II,” Lawfare, July 12 and 19, 2015; and Michael Shank and Cassidy Regan, “Aid Gone Awry in Africa,” U.S. News and World Report, July 19, 2013.