Tracking the rise of United States foreign military training: IMTAD-USA,a new dataset and research agenda

Abstract

Training other countries’ armed forces is a go-to foreign policy tool for the United States and other states. A growing literature explores the effects of military training, but researchers lack detailed data on training activities. To assess the origins and consequences of military training, as well as changing patterns over time, this project provides a new, global dataset of US foreign military training. This article describes the scope of the data along with the variables collected, coding procedures, and spatial and temporal patterns. We demonstrate the added value of the data in their much greater coverage of training activities, showing differences from both existing datasets and aggregate foreign military aid data. Reanalyzing prior research findings linking US foreign military training to the risk of coups d’état in recipient states, we find that this effect is limited to a single US program representing a small fraction of overall US training activities. The data show comprehensively how the United States attempts to influence partner military forces in a wide variety of ways and suggest new avenues of research.

Keywords

aid coups military training security assistance security cooperation United States

Introduction

Training other countries’ armed forces is an increasingly important foreign policy tool for the United States and other states. Between 1999 and 2016, across 34 different programs, the USA trained some 2,395,272 trainees from virtually every country in the world, peaking at 292,753 in 2008. Iraq and Afghanistan accounted for over half of these trainees, but even leaving these two countries aside, the total figure was 971,054, with as many as 78,722 individuals in a single year (2016). The United States spent some $14.8 billion worldwide on its training efforts and sold training worth another $4.9 billion, leaving aside the larger expenses of equipping and paying client forces in countries like Iraq and Afghanistan. The increase in training activities appears to reflect a concern for state building in a context of transnational security threats like terrorism, insurgency, and drug trafficking. It also comes as the USA, in the wake of Afghanistan and Iraq, seeks to limit its own direct costs by offloading them to local partners. As opposed to other elements of security assistance, training focuses on human capital. In doing so, in theory it addresses a critical element of contemporary military power (Talmadge, 2015). But it impacts not just military capability, but how recipient militaries choose to use that capacity – a key problem of civil–military relations and governance (Atkinson, 2006; Savage & Caverley, 2017). By influencing recipient armed forces’ norms and practices, it aims to improve local partners’ military capacity while avoiding the dangers of simply sending arms or cash.

To assess the origins and consequences of security assistance, as well as changing patterns over time, we provide a new, global dataset of US foreign military training efforts: the International Military Training Activities Database-USA (IMTAD-USA). This dataset goes beyond existing work by adopting a global scope and by covering the full array of US training programs, including key details such as: objectives; activities such as train-and-equip; the location of training; and characteristics of the forces trained. These complementary variables allow researchers to go beyond the raw training inputs to examine specific features that potentially lead to different outcomes. In other words, IMTAD-USA is a dataset on a key foreign policy tool that has both comprehensive scope and considerable depth.

These data document a remarkable increase in training over the past two decades, and efforts directed towards a wider variety of objectives. We trace these developments and discuss their larger implications, not least the fact that the expansion over the past two decades suggests a highly general, everyday tool of international hierarchy. The overall portrait is consistent with the USA seeking to maintain influence and confront security problems throughout the world while limiting its commitments by standing up other countries’ armed forces.

Yet training is controversial. We discuss one important debate in reanalyzing Savage & Caverley’s (2017) finding that training through the International Military Education & Training (IMET) program increases coup risk. While we validate this result for IMET specifically, our comprehensive dataset shows that there is little evidence of an overall relationship between US training and coup risk. However, the ubiquity of training suggests the potential for worldwide political impacts and for the broad diffusion of norms. Whether such norm transmission has in fact occurred, whether training is effective, how the USA allocates training to partner military forces, why recipient governments enter such partnerships, and the implications of the scale and bureaucratic complexity of training activities for accountability, are important questions for future research.

The IMTAD-USA dataset

The dataset covers US foreign military training from 1999 to 2016, building on the annual Department of Defense and Department of State Foreign Military Training Report (FMTR), submitted annually to Congress since 1999, covering worldwide training with the exception of NATO allies, Japan, Australia and New Zealand (see e.g. US DOS & DOD, 2019).¹ The FMTR provides the most comprehensive information on US training in terms of costs and the number of trainees (McNerney et al., 2014: 44). We go well beyond the FMTR data, however, relying on other government documents and secondary sources to supplement missing information and reconcile discrepancies. Above all, FMTR only covers the Afghanistan Security Forces Fund (ASFF) as of 2016, and never covers the Iraq Security Forces Fund (ISFF), but these are the two largest US training activities by a significant margin. IMTAD-USA includes data for these programs from a variety of official documents.

Defining military training

Different government agencies variously refer to training in terms of military assistance, security assistance, security cooperation, building partner capacity, and train-and-equip. For the purposes of the FMTR, and thus for our dataset, military training is defined as:

formal or informal instruction of foreign students in the United States or overseas by officers or employees of the United States, contract technicians, contractors (including instruction at civilian institutions), or by correspondence courses, technical, educational, or information publications and media of all kinds, training aids, orientation, and military advice to foreign military units and forces. (1961 Foreign Assistance Act, Section 644(n), 22 USC. 2403(n))

Unit of analysis, coding procedure and variables

The most disaggregated unit of analysis in the dataset is the program-country-year: characteristics of US training programs as they operate in a given country in a given year. Almost all of the 34 programs in the dataset operate across multiple countries, and most recipient countries have training activities operating simultaneously through a variety of programs. This most disaggregated level includes data on the number of trainees and the expense of training. Where unclassified, we include binary variables for whether particular types of forces received training, with indicators for armed forces, police, others, and ‘counterweight’ forces. For the latter, often used to defend regimes, we link with De Bruin’s (2021) State Security Forces dataset, which defines counterweights as armed forces under state control but outside the regular army, deployed with access to the capital city. Finally, this level includes a variable for the location of training activities, whether in the United States, the recipient country, or elsewhere.

At the level of the program we coded binary variables for overall program objectives including military performance, counterterrorism, counternarcotics, human rights, gender, civil–military relations, democratization, and good governance. To gather information on program objectives, we consulted program descriptions in FMTR, Defense Security Cooperation Agency budget estimates, and program-specific documents, complemented by secondary sources. A sample coding, as well as a full list of sources, is available in the Online appendix.

IMTAD-USA’s variables permit us to analyze the mechanisms through which training influences politics and security. Coding program objectives allows for an explicit measure of US efforts at military norms transmission, for instance. Whether training occurs in the United States, the recipient country, or a third country potentially shapes the avenues available for both norms transmission and the formation of transnational networks. Finally, the nature of forces trained can shape civil–military relations – favoring coup-proofing forces for example (De Bruin, 2021).

Limitations

The FMTR source data have several limitations. First, the reports do not systematically identify the use of military contractors. Second, FMTR includes total budgets and trainees for both classified and unclassified activities, but excludes additional details on classified activities and omits training by the Central Intelligence Agency. This creates biases. Considerable US assistance goes under the radar in order to provide plausible deniability for training the security forces of unsavory regimes, or to limit escalation in adversarial relationships (McManus & Yarhi-Milo, 2017). IMTAD-USA may therefore omit training activities in authoritarian states or in strategically sensitive areas. Analysts using these data may wish to introduce controls for regime type or for the type of strategic situation in which the United States has an interest in limiting escalation. These variables would then have to be interpreted as reflecting, in part at least, an interest in conducting training overtly or covertly.

Third, training objectives can vary across countries, which we do not capture with these data. The training objective data were instead uniformly applied to each country where a specific program was active for a given year. This simplification is necessary given the large number of countries where the USA carries out foreign military training activities. Future work could attempt to disaggregate country-specific training objectives.

Comparison with existing data

With the exception of classified data, and within the limits of sometimes contradictory and incomplete reporting, our dataset contains as full coverage as possible of US training activities. Past work has focused on military training and security assistance in the contexts of specific countries, such as Somalia (Williams, 2019), Iraq (Biddle, Macdonald & Baker, 2018), or Colombia (Dube & Naidu, 2015), or region-specific activities such as the School of the Americas (Gill, 2004; Scharpf, 2020). While these intensive programs merit close attention, they do not represent the ongoing, networked nature of much foreign military training and its truly global reach. Other studies comparing global training efforts have generally focused on single programs, notably IMET (Atkinson, 2006; Savage & Caverley, 2017) and officer exchanges (Ruby & Gibler, 2010). US training activities, however, go well beyond these programs. Martinez Machain (2021) and Omelicheva, Carter & Campbell (2017) cover nearly the full array of programs using the FMTR, but exclude Afghanistan (ASFF) Iraq (ISFF), by far the two largest programs, and lack IMTAD’s complementary data.

Another data source is the Security Assistance Monitor (SAM), which researchers have used to explore military aid (Boutton, 2021).² Covering training and arms transfers, SAM’s training component also uses FMTR data. However, unlike IMTAD-USA, SAM includes neither spending related to training alone, nor complementary information concerning training objectives, locations and types of recipients. Similarly, Security Force Monitor uses FMTR, digitizing it to provide raw, uncoded data about individual trainings, including course names and specific locations.³ However, unlike IMTAD-USA, it does not include spending and trainees for many classified training activities or training objectives.

Figure 1.

Total military training activities. Training expenses (millions of constant 2012 dollars); number of trainees (individuals).

Training correlates with other forms of US assistance but is significantly distinct. Comparing IMTAD-USA to USAID (2020) data, but excluding Iraq and Afghanistan, there is a correlation between training expenditures⁴ and total economic aid (0.334) and with total military aid (0.290). The correlation is stronger with major arms transfers (0.498) according to SIPRI (2020). Moreover, there is a pronounced difference when examining trainees versus expenditures: the number of trainees correlates at only 0.246 with economic aid, 0.098 with military aid, and 0.177 with major arms transfers. Other data fail to capture the scope of training in terms of the number of military personnel who pass through it. There is, in other words, a key divergence in the types of security assistance the USA offers in different places and times.

Trends and patterns in US training activities

The data reveal important patterns. The first is the expansion of training activities over the past two decades (Figure 1). This is most obvious in Iraq and Afghanistan, but the increase is general. Outside these two countries, there was a threefold rise from an annual average of $305 million in 1999–2001 (about half in aid, half in sales) to $672 million in 2014–2016, 40% in aid, 60% in sales (in constant 2012 dollars). There was, however, a real decline in the 2004–2006 period as training ramped up in Iraq and Afghanistan. Training is also an increasingly

Figure 2.

Training expenditures as a percentage of total US military aid expenditures over time

significant share of US military aid over time (Figure 2), though this is largely accounted for by Afghanistan alone. Thus, while the focus on Iraq and Afghanistan is justified, training is a much more general policy.

Second, while the geography of US training has changed, the web of training relationships is truly global (Figure 3). Almost all countries experienced some US training since 1999. The exceptions, visible in the maps, are problematic adversaries such as Myanmar, Cuba and North Korea. Others closely track changes in bilateral relations; for example, note the decline in training in Russia, Venezuela or Zimbabwe, or the increase in Libya. Focusing on regional patterns, the Middle East,

Figure 3.

Geographic scope of US training activities, by nominal expenses

Latin America and key allies in East Asia have been constant areas of focus. In contrast, training in other regions fluctuates, with Africa experiencing large increases with the war on terror.

A related pattern contrasts expenditures with trainees (Figure 4). The difference is striking in sub-Saharan Africa, a region with modest US spending relative to the number of trainees. This suggests a focus on building broad military capacity in the rank and file, with whole units participating in major exercises, in contrast with an officer focus for other regions. In other words, training operates differently from place to place.

These regional trends mask the high degree of concentration of training activities in certain countries – and notably the yawning gap between Afghanistan, Iraq, and the rest (Tables I and II). Key US allies in the Middle East figure prominently in spending. The picture changes, however, when examining the top recipients by number of trainees. Colombia, with extensive focus under Plan Colombia, is a third

Figure 4.

Regional distribution of US training over time, excluding Iraq and Afghanistan for clarity

Table I.

Top ten recipients in training spending, 1999–2016

	Aid			Sales
	Country	Spending (in nominal US$)		Country	Spending (in nominal US$)
1.	Afghanistan	8,773,157,454	1.	Saudi Arabia	1,221,458,007
2.	Iraq	1,885,700,424	2.	Singapore	1,187,673,741
3.	Colombia	378,176,179	3.	UAE	514,044,007
4.	Egypt	245,470,436	4.	Iraq	359,815,522
5.	Uganda	197,348,005	5.	Taiwan	259,749,360
6.	Israel	182,149,502	6.	Kuwait	197,905,118
7.	Mexico	152,758,211	7.	India	155,776,604
8.	Pakistan	146,650,170	8.	South Korea	125,946,118
9.	Jordan	129,116,397	9.	Afghanistan	103,057,113
10.	Philippines	129,958,465	10.	Georgia	93,218,881

Table II.

Top 10 recipients in individuals trained, 1999–2016

	Aid			Sales
	Country	Trainees		Country	Trainees
1.	Afghanistan	947,546	1.	Saudi Arabia	24,890
2.	Iraq	473,546	2.	Singapore	20,272
3.	Colombia	97,286	3.	Taiwan	14,754
4.	Nigeria	39,051	4.	South Korea	12,806
5.	Burundi	36,248	5.	UAE	7,853
6.	Uganda	35,026	6.	Colombia	7,463
7.	Mexico	27,388	7.	Georgia	6,367
8.	Ghana	25,979	8.	Kuwait	5,768
9.	Rwanda	23,660	9.	Israel	4,478
10.	Benin	21,248	10.	Lebanon	3,415

outlier beyond Afghanistan and Iraq. Otherwise, this list mirrors the discrepancy in Figure 4 between spending inputs and trainee outputs. Finally, the difference between aid and sales unsurprisingly reflects income to a large degree.

Ultimately, then, training is both broad and deep: it is something the USA does both globally and in critical cases of counterinsurgency and state building. The focus on Iraq and Afghanistan should not obscure the ubiquity of US training activities, and we should be wary of generalizing findings from these large-scale missions to the wider web of relationships.

Broad geographic scope is matched by administrative complexity. A recent RAND report refers to the

Figure 5.

Spending on training programs, by program objective (excluding ASFF and ISFF), millions of constant 2012 dollars

‘complex patchwork’ of security cooperation since 2001, identifying over 160 different Congressional statutes authorizing security cooperation between the USA and its foreign partners (Thaler et al., 2016). A recipient will typically receive training through multiple programs. This diversity in part reflects a wide array of objectives for training.

However, this administrative complexity has several ramifications. First, US personnel often stitch together different sources of funding, with myriad administrative procedures, reporting requirements, deadlines and objectives. This makes tracing the full extent of activities difficult. Consequently, the transparency provided by the FMTR is not as useful as it might otherwise be, with implications for oversight and accountability. Second, it also makes the administration of programs less effective, augmenting transaction costs and continuity problems.

We code each program’s objectives under eight broad headings (accounting for how programs change objectives over time). On average, each program focuses on between two and three objectives. Figure 5 shows how objectives have changed over time, with total funding for programs that include each objective. Importantly, because programs can have multiple objectives, these figures are not mutually exclusive, and they do not indicate, within a given program, how much money was spent on specific objectives (e.g. counterterrorism vs. governance). Figure 5 also excludes ASFF and ISFF, since these outliers throw off a general assessment of objectives. Military performance was, unsurprisingly, the most-funded objective. However, most other objectives increase in prominence over time, with training becoming more multidimensional.

One caveat follows from this administrative complexity. The ways US officials navigate this system mean potential gaps between what the program says it does and what it actually does. For example, as hybrid and cyber warfare increased in prominence, US officers stretched existing counterterrorism programs to cover security cooperation in this domain, rather than waiting for formal Congressional authorization (Thaler et al., 2016: 12–16). This suggests inherent limitations to IMTAD-USA’s coding of objectives: authorizations in Washington may not match the practical purposes on the ground.

Empirical application: A new perspective on military training and coups d’état

Underscoring how IMTAD-USA’s more extensive coverage of US training can significantly alter existing findings, we revisit Savage & Caverley’s (2017) analysis of the relationship between training and coups. With data on IMET (1970–2009) and the Counterterrorism Fellowship Program (CTFP) (2002–2009),⁵ the authors find that ‘any US FMT [foreign military training] corresponds to a doubling of the probability of a military-backed coup attempt in the recipient country’ (Savage & Caverley, 2017: 543). They further speculate that because IMET focuses on liberal norms of human rights and civilian control, ‘IMET trainees are therefore the population where we are most likely to see the effect of norms (i.e. the easy case). If we discover more coups in countries with a large number of IMET trainees, this relationship will likely be stronger in less-scrutinized programs with less focus on liberal values’ (2017: 548). Despite conceding that the focus on IMET is a limitation in their empirical analysis, they ‘find a robust relationship between US training of foreign militaries and military-backed coup attempts’, explaining this in terms of the human capital training vests in the military and the ways this changes the balance of power between the military and the regime (Savage & Caverley, 2017: 553). Frequent academic and media references to this work also cite it as indicating that US training provokes coups (e.g. Henke, 2019: 165; Dieng, 2019: 493, Economist, 2019).

By applying IMTAD-USA to this puzzle, we can replicate the findings concerning IMET and assess whether coup risk generalizes to the full array of programs. We reanalyze their time-series cross-section logit

Table III.

Replication of Savage & Caverley (2017) with IMTAD-USA: US military training and coups

	Logit coefficient estimate	Standard error
IMET, binary	1.044*	(0.502)
IMET, log total students	0.959*	(0.428)
IMET, log total real expenditures	1.010*	(0.493)
All programs except sales, binary	0.707	(0.680)
All programs except sales, log total students	0.476	(0.428)
All programs except sales, log total real expenditures	0.675	(0.523)
RCSS, binary	−0.359	(0.447)
RCSS, log total students	−0.125	(0.467)
RCSS, log total real expenditures	−0.245	(0.408)

DV in all models = military coup occurred in a given country in a given year.

* p <0.05.

Each coefficient estimate comes from a separate logit model with the same control variables (not shown). N = 1,579. Real expenditures are, as with Savage & Caverley (2017), in constant 2005 dollars.

models for coup incidence, with the same control variables and statistical procedures (Table III). It is important to note that the time period is much shorter in our data since the FMTR reports do not start until 1999, whereas Savage & Caverley (2017) draw on the Defense Security Cooperation Agency Historical Facts Book, which provides numbers of trainees and expenditures for IMET for 1970 onward. However, the first three rows of Table III replicate their finding about the increased coup risks associated with IMET in recipient countries for this overlapping 1999–2009 time period, using our IMTAD data.⁶ Any differences that emerge are not merely a consequence of different timeframes.

Indeed, there are differences. In extending the analysis to encompass the other non-sales programs outside of ISFF and ASFF as well (rows 4–6), the result disappears. IMET represents 30% of expenditures and 13% of trainees when also leaving aside sales of training (and 14% and 11% when including sales). It is also unrepresentative of the total; the correlation coefficients between IMET and non-IMET aid programs are, for expenditures, 0.248; for trainees, 0.227 (and 0.090 and 0.222 when including sales).⁷ IMET is focused on junior and senior officers, unlike programs training rank-and-file soldiers. But we also find a null result when examining the Regional Centers for Security Studies (RCSS) (rows 7–9), a ‘most similar’ comparison. This program, like IMET, focuses on officers and seeks explicitly to cultivate leaders and develop US military networks abroad, and should therefore impact civil–military relations via similar mechanisms. Conversely, it differs from IMET insofar as RCSS focuses on senior rather than junior officers and conducts seminars and short courses on broad strategic issues, while IMET supports a much wider variety of activities. In sum, IMTAD-USA’s comprehensive coverage suggests that while Savage & Caverley are correct to point to the dangers of IMET, their hypothesis that this effect is likely to be even stronger in other programs appears unsupported – at least for those programs for which we have publicly-available data.

Discussion and avenues of further research

Beyond implications for existing research, IMTAD-USA opens up new lines of inquiry. First, the scale of the phenomenon suggests a transmission belt for ideas and capabilities that shape US power and global security. The US military is the most important agent in diffusing a particular global ‘military culture’ shaping norms and ideas (Farrell, 2005) and technologies of counter-rebellion (Kalyvas & Balcells, 2010). What are the consequences of transmitting particular forms of knowledge with specific objectives (Savage & Scharpf, 2020)? Does human rights training, for example, foster compliance? Similar questions concern civil–military relations, anticorruption efforts, gender training and other normative objectives. Relatedly, US security assistance shapes the capabilities of security forces and state institutions. To what extent and how does US training boost counterterrorism capabilities (Bapat, 2011), change counterinsurgency outcomes (Lyall & Wilson, 2009), deter civil war onset (Cunningham, 2016), or impact post-conflict conditions (Sullivan, Blanken & Rice, 2020)? Recent case-study work is almost unanimously negative in assessing the effectiveness of large-scale training missions, past (Ladwig, 2017; Karlin, 2018) and present (Biddle, Macdonald & Baker, 2018; Reno, 2018; Jowell, 2018; Matisek, 2018). But we know little about the impacts of smaller-scale, routinized activities outside the context of active insurgency. The dataset can provide both a macro-level take on these questions and guide case selection for micro-level analysis. Research could, in particular, apply these data to other dependent variables, such as repression (Sullivan, Blanken & Rice, 2020) and instability (Boutton, 2021).

A second avenue of research concerns the ways training manifests security hierarchy. The global reach of security and economic hierarchies controlled by the United States is, as Lake observes, ‘the most striking fact about the pattern of authority in the modern world system’ (2009: 82). Do training relationships reinforce a US-centric military system? Do they increase US influence and access to recipient states (Blankenship & Miles Joyce, 2020; Martinez Machain, 2021)? Do they impact how the USA mobilizes military coalitions (Henke, 2019)? Training, after all, is a way of creating military power (Brooks & Stanley, 2007) and signaling support for recipients (McManus & Yarhi-Milo, 2017). In light of the enormous expense of maintaining US military presence in Iraq and Afghanistan, training relationships represent an attempt to outsource hegemony and square commitments with capabilities.

This raises a third set of questions about whether training accomplishes this objective. Do partners do what the USA wants them to do? Does training help to contain security challenges at lower expense than direct intervention? And does it do so with fewer agency problems than other forms of security cooperation, such as financing and arms transfers (Sullivan, Tessman & Li, 2011)? As researchers disaggregate forms of external support in civil war (Sawyer, Cunningham & Reed, 2017; Sullivan & Karreth, 2019), how does knowledge- and human-centric training compare with material security assistance (e.g. McNerny et al., 2014)? Training may be subject to similar principal-agent problems as other forms of security assistance, with recipients using it for their own ends (Byman, 2006; Ladwig, 2017; Savage & Caverley, 2017; Biddle, Macdonald & Baker, 2018). In addition, as the Vietnam, Iraq and Afghanistan cases illustrate, training can sometimes express waning US commitment, sending a perverse signal that the USA cannot be relied upon in the long term.

Fourth, we know little about the foreign military training of other states and organizations. Despite increased Russian and Chinese activity, we know little about their training and how it impacts alignments, access and influence. Much training occurs in multilateral contexts, most notably through NATO in Afghanistan, with the European Union and African Union also playing important roles. Divergent providers of training may lead to divergent outcomes, even when the trainers are allies. Grewal (2019), for example, has found that Tunisian officers differ in their attitudes about civil–military relations depending on whether they trained in the United States or in France. The USA frequently works through private military companies whose role merits further research, as do regional powers such as Turkey and the Gulf states who are actively training partners in places like Somalia and Libya.

Fifth, the attractiveness of ‘working by, with, and through’ local partners suggests that US foreign military training is likely to continue. But why do some countries rather than others receive training, why does the USA work with particular units in partner states, and why does training takes particular forms? How well do past security relationships, common threats, cost sensitivity, and concerns for democracy and human rights explain who gets training, in what domains, and how much? Conversely: Why and how do governments choose to partner with the USA (Scharpf, 2020), or designate particular units to receive training?

Finally, there are important questions about accountability and democratic control against the backdrop of a liberal international order under intense pressure. The deaths of four US Special Forces trainers and five of their Nigerien counterparts in an ambush in Niger in October 2017, or the Trump administration’s politicization of security assistance to Ukraine, represent important controversies. How does training influence democracy and civil–military relations in the USA? By compiling data on US training and making it publicly available, we hope to further research into the normative implications of security assistance.

Footnotes

Replication data

The dataset and Online appendix, including the codebook, can be found at .

Acknowledgments

Sincere thanks to Marianne Bergeron, Ines Butler, Laurie Chartrand, Emmanuel Méthot-Jean, Raphaël Scarborough, and Ludwine Tchatat. We kindly thank participants at the Workshop on Foreign Military Training Activities and the Inter-University Seminar on Armed Forces and Society for advice and feedback.

Funding

This research was supported by the Social Sciences and Humanities Research Council of Canada, with a supplement from the Department of National Defence, Canada (award no. 435-2017-0726).

ORCID iDs

Theodore McLauchlin

Simon Pierre Boulanger Martel

Notes

References

Atkinson

Carol

(2006) Constructivist implications of material power: Military engagement and the socialization of states, 1972–2000. International Studies Quarterly 50(3): 509–537.

Bapat

Navin A

(2011) Transnational terrorism, US military aid, and the incentive to misrepresent. Journal of Peace Research 48(3): 303–318.

Biddle

Stephen

Macdonald

Julia

Baker

Ryan

(2018) Small footprint, small payoff: The military effectiveness of security force assistance. Journal of Strategic Studies 41(1–2): 89–142.

Blankenship

Brian

Miles Joyce

Renanah

(2020) Purchasing power: US overseas defense spending and military statecraft. Journal of Conflict Resolution 64(2–3): 545–573.

Boutton

Andrew

(2021) Military aid, regime vulnerability, and the escalation of political violence. British Journal of Political Science 51(2): 507–525.

Brooks

Risa

Stanley

Elizabeth

(2007) Creating Military Power. Stanford, CA: Stanford University Press.

Byman

Daniel

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

Cunningham

David

(2016) Preventing civil war: How the potential for international intervention can deter conflict onset. World Politics 68(2): 307–340.

De Bruin

Erica

(2021) Mapping coercive institutions: The state security forces dataset, 1960–2010. Journal of Peace Research 58(2): 315–325.

10.

Dieng

Moda

(2019) The multi-national joint task force and the G5 Sahel Joint Force: The limits of military capacity-building efforts. Contemporary Security Policy 40(4): 481–501.

11.

Dube

Oeindrila

Naidu

Suresh

(2015) Bases, bullets, and ballots: The effect of US military aid on political conflict in Colombia. Journal of Politics 77(1): 249–267.

12.

Economist (2019) Friendly fire: A shooting puts the spotlight on military training for allies. The Economist 12 December (https://www.economist.com/united-states/2019/12/12/a-shooting-puts-the-spotlight-on-military-training-for-allies).

13.

Farrell

Theo

(2005) World culture and military power. Security Studies 14(3): 448–488.

14.

Gill

Lesley

(2004) The School of the Americas: Military Training and Political Violence in the Americas. Durham, NC: Duke University Press.

15.

Grewal

Sharan

(2019) Military defection during localized protests: The case of Tataouine. International Studies Quarterly 63(2): 259–269.

16.

Henke

Marina E

(2019) Constructing Allied Cooperation: Diplomacy, Payments and Power in Multilateral Military Coalitions. Ithaca, NY: Cornell University Press.

17.

Jowell

Marco

(2018) The unintended consequences of foreign military assistance in Africa: An analysis of peacekeeping training in Kenya. Journal of Eastern African Studies 12(1): 102–119.

18.

Kalyvas

Stathis

Balcells

Laia

(2010) International system and technologies of rebellion: How the end of the Cold War shaped internal conflict. American Political Science Review 104(3): 415–429.

19.

Karlin

Mara

(2018) Building Militaries in Fragile States: Challenges for the United States. Philadelphia, PA: University of Pennsylvania Press.

20.

Ladwig III

Walter C

(2017) The Forgotten Front: Patron–Client Relationships in Counterinsurgency. Cambridge: Cambridge University Press.

21.

Lake

David

(2009) Hierarchy in International Relations. Ithaca, NY: Cornell University Press.

22.

Lyall

Jason

Wilson

Isaiah

(2009) Rage against the machines: Explaining outcomes in counterinsurgency wars. International Organization 63(1): 67–106.

23.

Martinez Machain

Carla

(2021) Exporting influence: US military training as soft power. Journal of Conflict Resolution 65(3): 313–341.

24.

Matisek

Jahara

(2018) The crisis of American military assistance: Strategic dithering and Fabergé Egg armies. Defense & Security Analysis 34(3): 267–290.

25.

McManus

Roseanne

Yarhi-Milo

Keren

(2017) The logic of “offstage” signaling: Domestic politics, regime type, and major power-protégé relations. International Organization 71(4): 701–733.

26.

McNerney

Michael

O’Mahony

Angela

Szayna

Thomas S

Eaton

Derek

Baxter

Caroline

Clarke

Colin P

Cutrufello

Emma

McGee

Michael

Peterson

Heather

Payne

Leslie Adrienne

Trenkov-Wermuth

Calin

(2014) Assessing Security Cooperation as a Preventive Tool. Santa Monica, CA: RAND.

27.

Omelicheva

Mariya

Carter

Brittnee

Campbell

Luke B

(2017) Military aid and human rights: Assessing the impact of US security assistance programs. Political Science Quarterly 132(1): 119–145.

28.

Reno

William

(2018) The politics of security assistance in the horn of Africa. Defence Studies 18(4): 498–513.

29.

Ruby

Tomislav Z

Gibler

Douglas

(2010) US professional military education and democratization abroad. European Journal of International Relations 16(3): 339–364.

30.

Savage

Jesse Dillon

Caverley

Jonathan D

(2017) When human capital threatens the Capitol: Foreign aid in the form of military training and coups. Journal of Peace Research 54(4): 542–557.

31.

Savage

Jesse Dillon

Scharpf

Adam

(2020) Security assistance and political militarization. Working paper, Trinity College Dublin.

32.

Sawyer

Katherine

Cunningham

Kathleen Gallagher

Reed

William

(2017) The role of external support in civil war termination. Journal of Conflict Resolution 61(6): 1174–1202.

33.

Scharpf

Adam

(2020) Why governments have their troops trained abroad: Evidence from Latin America. International Studies Quarterly 64(3): 734–747.

34.

SIPRI – Stockholm International Peace Research Institute (2020) SIPRI arms transfers database (https://www.sipri.org/databases/armstransfers).

35.

Sullivan

Patricia

Karreth

Johannes

(2019) Strategies and tactics in armed conflict: How governments and foreign interveners respond to insurgent threats. Journal of Conflict Resolution 63(9): 2207–2232.

36.

Sullivan

Patricia

Blanken

Leo

Rice

Ian

(2020) Arming the peace: Foreign security assistance and human rights conditions in post-conflict countries. Defence and Peace Economics 31(2): 177–200.

37.

Sullivan

Patricia L

Tessman

Brock F

Xiaojun

(2011) US military aid and recipient state cooperation. Foreign Policy Analysis 7(3): 275–294.

38.

Talmadge

Caitlin

(2015) The Dictator’s Army: Battlefield Effectiveness in Authoritarian Regimes. Ithaca, NY: Cornell University Press.

39.

Thaler

David

McNerney

Michael J

Grill

Beth

Marquis

Jefferson P

Kadlec

Amanda

(2016) From Patchwork to Framework: A Review of Title 10 Authorities for Security Cooperation. Santa Monica, CA: RAND Corporation.

40.

USAID – United States Agency for International Development (2020) Foreign aid explorer (https://explorer.usaid.gov/data).

41.

US DOS & DOD – United States Department of Defense & Department of State (2019) Foreign military training report, fiscal years 2018 and 2019 (https://www.state.gov/wp-content/uploads/2019/12/FMT_Volume-I_FY2018_2019.pdf).

42.

Weidmann

Nils B

Kuse

Doreen

Gleditsch

Kristian Skrede

(2010) The geography of the international system: The CShapes dataset. International Interactions 36(1): 86–106.

43.

Williams

Paul

(2019) Fighting for Peace in Somalia: A History and Analysis of the African Union Mission (AMISOM), 2007–2017. Oxford: Oxford University Press.