Agency Theory

A New Model of Civil-Military Relations for Africa?


In this paper I assess a new approach to Civil-Military relations, Peter Feaver’s ‘Agency Theory’. After demonstrating that this theory offers important advances against the standard approaches to the topic exemplified by Huntington and Janowitz, I then turn to a consideration of the applicability of the theory in the African context. Feaver himself is pessimistic about the value of his theory in a context where there is not an established culture of submission to civilian rule among military forces, and where coups are a real danger. I argue, however, that even under these conditions Agency Theory remains extremely valuable as an analytic tool. I argue further that Feaver has not recognised the potential benefits that regional organisations offer to civilian principals in their goal of ensuring military obedience, even where a tradition of military professionalism does not exist.


Despite trendy and dramatic academic pronouncements of the decline and fall of the state as the key player on the global stage, states continue doggedly on. In those few countries where the structures of the state have, in fact, collapsed (such as Somalia), the state has not been replaced by some novel new system, but has instead regressed to chaos and anarchy. States, then, remain important.

Perhaps the defining feature of the state is that of its monopoly over the legitimate use of force, as famously expressed by Max Weber in his landmark 1918 speech Politik als Beruf.1 In contemporary times, with the rise of notions such as the doctrine of humanitarian intervention, it has also been increasingly asserted that the legitimacy of the state’s use of force is dependent on responsible civilian oversight of those state arms (such as the police and the military) that exercise that monopoly on behalf of the state. This is not the place to debate the legitimacy of undemocratic states in the international system. What can be confidently asserted is that the democratic control of armed forces, usually described in terms of ‘civil-military relations’, is an essential feature of democratic governance and an important element in the prevention of internal armed conflict. In a democratic state it is the people, through their properly elected representatives, who are to rule over every aspect of the public life of the nation, including (perhaps especially) the application of force. Military leaders and institutions must therefore be the servants, not the rulers, of the state. This feature of democratic states, however, contains within it a fundamental paradox – the very institution that is created to protect the state, the military, has the power to become its greatest threat. As Richard Kohn points out, ‘The military is, by necessity, among the least democratic institutions in human experience; martial customs and procedures clash by nature with individual freedom and civil liberty, the highest values in democratic societies’ (Kohn 1997:141).

In today’s world, the challenge of civilian control over the military has two distinct, though interconnected, aspects. At the most basic level, the challenge is that of preventing military forces from taking control of the nation’s political life, whether by means of a coup d’état or by way of less obvious, but no less pernicious, pressures on civilian government. This is made all the more difficult by the fact that it must be achieved without weakening the military to the extent that it faces serious defeat on the battlefield, thereby placing the state and its citizens at considerable risk. This fundamental challenge is the one that is felt most keenly by the young democracies of the world. In the older democracies, where civilian control of the military is an established feature of the political terrain, the challenge is less dramatic, but no less difficult to address. Here, as Kohn puts it, ‘the test is whether civilians can exercise supremacy in military policy and decision making – that is, frame the alternatives and define the discussion, as well as make the final choice. When the military enjoys great prestige, possesses advanced bureaucratic skills, believes that its ability to fulfil its mission may be at risk, or comes to doubt the civilian leadership, civilians can face great obstacles in exercising their authority’ (Kohn 1997:141).

Despite the importance of civil-military relations, theoretical understandings of this question have advanced surprisingly slowly in recent decades. Most academics and practitioners whose work touches on this issue still take as given the theory of Samuel P. Huntington, published nearly fifty years ago as The Soldier and the State: The Theory and Politics of Civil-Military Relations (Huntington 1957). The only other academic figure to achieve a significant profile in this regard during the preceding fifty years is the late Morris Janowitz, whose landmark book, The Professional Soldier: A Social and Political Portrait (Janowitz 1960), explored the sociological connections between the military and the society it serves. Recently, however, a number of new works have emerged on the question of civil-military relations, that threaten (or promise) to break the near-monopoly that Huntington and Janowitz have long held on this field. One of those important recent works is Michael Desch’s 1999 book, Civilian Control of the Military: The Changing Security Environment (Desch 1999), which expounds the thesis that civil-military relations are deeply affected by the level of threat in the security environment, and that civilian control is at its weakest when the threat is at its lowest. More recently, we have seen the emergence of an impressive new theory, ‘agency theory’, which has been developed by Peter D. Feaver of Duke University, and which is comprehensively laid out in his recent book Armed Servants: Agency, Oversight, and Civil-Military Relations (Feaver 2003).

The purpose of this paper is to assess whether Feaver’s agency theory offers any important advances over the traditional Huntington/Janowitz ‘professionalism’ approach, and, in particular, whether agency theory shows potential for the specific circumstances of Africa’s new democracies. That this is important is made clear by Jakkie Cilliers (1995:39) when he writes that:

Few states in southern Africa have ‘mature’ military forces or an indigenous professional military culture that accepts the supremacy of civilian and parliamentary authority over the military. Developing a military culture that can coexist with civilian supremacy and representative government is a task that should focus the minds of both Africans and their friends from the industrialized countries, and which should lead to joint and realistic strategies.

Of course, sound civil-military relations and a salutary professional military culture will not by themselves secure national or regional stability across southern Africa, but they do represent crucial building blocks for the future. Thus the nature of armed forces and civil-military relations in southern Africa merits careful attention.

In what follows I shall begin by giving a brief account of the traditional view, whereafter I shall present a comprehensive account of agency theory. I will then offer an assessment of the general advantages or disadvantages of agency theory over its older competitors, followed by an analysis of its performance in the specific context of democracy in Africa. As we will see, Feaver himself raises questions as to the applicability of his theory to states where a real threat of a coup d’état exists. I will argue, however, that agency theory is able to deal with this issue.

Huntington and Janowitz

At the heart of both Huntington’s and Janowitz’s analysis of the military and its relationship to the polity, is the notion of professionalism. For both, a profession is defined in a fairly conventional manner, as an occupation which has highly specialised characteristics in the areas of expertise, responsibility and corporateness. Both also restrict the membership of the society of military professionals to those who belong to the officer corps. In terms of civil-military relations, Huntington takes the notion of professionalism to be at the heart of what he calls ‘objective civilian control’ of the military (Huntington 1957:83–85 and passim). In more recent times Huntington (1995:9-10) helpfully summarises the features of objective civilian control as follows:

This involves: 1) a high level of military professionalism and recognition by military officers of the limits of their professional competence; 2) the effective subordination of the military to the civilian political leaders who make the basic decisions on foreign and military policy; 3) the recognition and acceptance by that leadership of an area of professional competence and autonomy for the military; and 4) as a result, the minimization of military intervention in politics and of political intervention in the military.

The key consequence of this professionalism-driven model is, in Huntington’s account, that it provides a way to weaken the military politically (by keeping it out of political matters) while at the same time allowing it to be strong militarily, thereby ensuring both civilian control and military effectiveness (Huntington 1957:83-85).

Because his focus is more sociological than political, in many respects Janowitz’s approach is not a competitor but rather a complement to Huntington’s theory. Perhaps the main difference between these theorists’ views regarding civilian control of the military relates to their views of professional autonomy. Huntington argued for a strict separation between the values of the military profession and those of liberal civil society, and believed that the imposition or infusion of liberal values into the military would undermine its military effectiveness. Writing as he was in the context of the US participation in the Cold War, and with this belief about liberal values in mind, Huntington saw a real crisis looming on the horizon. He believed that in a time of war the only way a liberal society can meet a serious threat is by suspending its liberal values for the period of that war. In the case of the Cold War, however, which he recognised as being likely to be a long-term feature of international relations, he believed the only hope for survival was for liberal values to be jettisoned altogether, and for the US to become a conservative republic: ‘The requisite for military security is a shift in basic American values from liberalism to conservatism. Only an environment which is sympathetically conservative will permit American military leaders to combine the political power which society thrusts upon them with the military professionalism without which society cannot endure’ (Huntington 1957:464).

In contrast to this view Janowitz (1960) argued that the distinction between war and peace was becoming increasingly difficult to draw in the modern world, and that military forces were increasingly becoming ‘constabulary forces’ rather than the traditional warfighters of old. As a result, he argued, the professional soldier, while remaining in important ways distinct from his civilian counterparts, must become politically aware, and must of essence therefore absorb many of the values of the society he serves. This is sometimes described as ‘subjective’ civilian control (see for example Williams 1995). He also described how the rise of bureaucratisation and the increasing dependence on technology in the military produced additional constraints on the autonomy of the military professional.

What is common to Huntington and Janowitz is that for both of them civilian control of the military is assured by the military’s professionalism, defined in terms of voluntary submission to civilian authorities – in Huntington’s case, submission based on the incentive of civilian non-interference in the military realm; and in Janowitz’ case, submission based on shared values. In this respect, then, Peter Feaver seems correct when he writes that ‘The tradition inspired by Morris Janowitz provides an important counterweight to Huntington, but on the crucial question of how civilian institutions control military institutions on a day-to-day basis the Janowitzean school does not represent a significant alternative to Huntington’ (Feaver 2003:2). In what follows I will outline Feaver’s own theory, which does indeed offer a refreshingly significant alternative to Huntington’s approach.

Feaver’s Agency Theory

Feaver makes it very clear from the start of his book that of the two aspects of civil-military relations outlined at the beginning of this paper – the fundamental issue of establishing and maintaining civilian control over the might of the military, on the one hand, and the day-to-day control over policy decisions on the other – it is the latter at which his agency theory is directed. To reiterate, this is not the issue of preventing the military from taking control of all the affairs of the state, but rather the question of ensuring civilian control over those affairs of the state that directly or indirectly affect the military, such as, for example, the defence budget. This remains, however, a central problem for the democratic state. For though it is not its physical coercive power that causes the military to be a challenge in this regard, the military nonetheless also wields power in other forms – its claim to special expertise in military matters, the general prestige of the military, and so on. The practical dangers of the military exercising too much power in this respect include the possibility that the military may drain the state of its resources, even to the point of collapse, or that it might seek out wars that are contrary to the interests and will of the society it serves (Feaver 2003:5). Perhaps more important than these practical considerations is the principle of democratic governance – that in a democracy the citizenry by definition retains the right to decide, through their elected representatives, on all matters of state, even on matters in which experts (such as the military2) may have greater expertise. As Feaver neatly puts it, ‘In a democracy, civilians have the right to be wrong’ (Feaver 2003:65).

Feaver’s model of civil-military relations is explicitly developed as an alternative to Huntington’s approach, which he considers to be the dominant paradigm. Janowitz is set aside because, ‘when it comes to understanding the day-to-day political management of the military, the Janowitzean approach does not differ from the Huntingtonian on any fundamental issue’ (Feaver 2003:9). The central difference between Feaver and Huntington is that, while not declaring these irrelevant, Feaver’s focus is not on nonmaterial determinants of behaviour (such as beliefs, norms and identity), but on material factors. For Huntington (and Janowitz for that matter) the central variable in civil-military relations is one of identity, the identity of the military officer as a professional.

Against this, Feaver builds a model that draws on principal-agent theory, a framework widely used in economic and political analysis. Its goal, as the name suggests, is to address problems of agency, particularly between actors in a position of superiority or authority (principals) and their subordinates (agents). The classic case is perhaps the employer-employee relationship. In such cases the goal of principal-agent theory is to address the problem of how the employer ensures that the employee does what is required of her, or in other terms, how the employer ensures that the employee is ‘working’ rather than ‘shirking’. Feaver argues that the principal-agent relationship can be easily applied to the issue of civil-military relations, and that this can be viewed as ‘an interesting special case of the general problem of agency’ (Feaver 2003:12-13). Because this ‘special case’ has features that are unique and not broadly applicable to other principal-agent relationships, Feaver coins the term ‘agency theory’ to describe it (Feaver 2003:55).

Let the military advise against foolish adventures, even advising strenuously when circumstances demand. But let the military execute those orders faithfully. The republic would be better served even by foolish working than by enlightened shirking.’

At the heart of agency theory is the idea that civil-military relations is essentially a form of strategic interaction between civilian masters (principals) and their military servants (agents). In that strategic interaction civilians choose methods by which to monitor the military. What methods are chosen depends on what expectations the civilians have about the degree to which the military will submit to their authority. Submission or obedience is, in Feaver’s terminology, ‘working’, while rebellion or refusal to obey is ‘shirking’. ‘The military decides whether to obey in this way, based on military expectations of whether shirking will be detected and, if so, whether civilians will punish them for it. These expectations are a function of overlap between the preferences of the civilian and the military players, and the political strength of the actors’ (Feaver 2003:3).

It’s worth pausing here to consider more closely what Feaver means by the terms ‘working’ and ‘shirking’. ‘Working’ is relatively unproblematic – an agent is working when she is diligently pursuing the tasks assigned to her by her superior. In the case of the military, the military is working when it diligently seeks to fulfil the wishes of its civilian overseers. ‘Shirking’, on the other hand, requires more exploration. In the everyday sense, shirking is simply failing to work, and is often associated with laziness and general inactivity. While this may well sometimes apply to the military, it is not however the central meaning of the term as used in agency theory. For the military may be vigorously pursuing military and/or policy goals, but it will still be shirking if those goals do not correspond with the desires of the civilian principal.

‘What civilians want’ is of course a complex and multidimensional issue in the context of civil-military relations, far more so than in the traditional economic applications of principal-agent theory. Feaver points out that, in structural terms, the desires of the civilian principal can be viewed as two-fold: firstly, civilians want to be protected from external enemies, and, secondly, they want to retain political control over the military. This two-fold distinction can be unpacked further as follows:

The functional goal includes the following:

  1. whether the military is doing what civilians asked it to do, to include instances when civilians have expressed a preference on both the ‘what’ and the ‘how’ of any given action;
  2. whether the military is working to the fullest extent of its duty to do what the civilians asked it to do;
  3. whether the military is competent (measured by some reasonableness standard) to do what civilians asked it to do.

The relational goal can be broken down into the following:

  1. whether the civilian is the one who is making key policy decisions (i.e. no de facto or de jure coup) and whether those decisions are substantive rather than nominal;
  2. whether the civilian is the one who decides which decisions civilians should make and which decisions can be left to the military;
  3. whether the military is avoiding any behaviour that undermines civilian supremacy in the long run even if it is fulfilling civilian functional orders (Feaver 2003:61).

In the civil-military context the concept of shirking is further complicated by the fact that one of the essential roles the military plays is that of advisor to the civilian principal. Here the problem is akin to the familiar one of taking advice from one’s mechanic about what repairs are necessary to one’s car. Given that the average person lacks the expertise to know what is wrong with his car when it breaks down, he is dependent on the mechanic to advise him on the best course of action. However, the mechanic has a strong pecuniary interest in suggesting the most expensive course of action, regardless of whether that is in the car-owner’s (the principal’s) interests. It is thus generally believed that an honest mechanic is hard to find. In the same way there is an incentive for the military to exaggerate threat levels and the cost of operations in cases where they wish to avoid carrying out missions desired by the civilian principal, or where there is a desire to minimise the risks inherent in the operation by applying overwhelming force.

In principal-agent theory terms, the above-described problem of getting the agent to work in the desired manner is called the ‘moral hazard problem’. Feaver points out that in the general literature on principal-agent theory, there are two distinct opinions in this regard. On the one hand, there are those who contend that the best way to ensure that the agent is working is by applying the best available monitoring system, whether that includes intrusive or non-intrusive instruments, or both. On the other hand, there are those who believe that monitoring is inefficient, and that the superior approach is to implement measures aimed at adjusting the agent’s preferences to increasingly coincide with those of the principal. Feaver’s own agency theory takes neither side in this regard, but instead draws on insights from both. Importantly, Feaver (2003:56) adds ‘a further consideration that is surprisingly absent from existing principal-agent treatments: how agent behavior is a function of their expectation that they will be punished if their failure to work is discovered; traditional principal-agent treatments assume punishment is automatic but…I argue…that assumption must be relaxed when analyzing civil-military relations’.

Envisaged in this way, civil-military relations are viewed as a game of strategic interaction in which each side attempts to achieve outcomes that maximally promote that side’s interests. This is clearly a significantly different approach to that favoured by Huntington and Janowitz, where nonmaterial factors such as identity and moral commitments arguably play the central role. These factors are not, however, irrelevant to agency theory. Instead, Feaver argues, agency theory provides a framework of analysis against which the influence of these factors may be measured and assessed. In particular they can be understood in terms of the attempt to seek convergence between the preferences of the civilian principal and the military agents. Thus Feaver points out that the military honour ethic ‘can work to mute the impulse to shirk for military agents, even when other factors…indicate they should. Traditional civil-military relations theory relies extensively on honor (also called the “ethic of subordination” and “professionalism”) to explain civilian control. It can be incorporated in a principal-agent framework, but is recognized as just one of the factors shaping the military’s preferences’ (Feaver 2003:64).

The preference for honour is one of three preferences that the military agent is assumed to hold by agency theory. Another is the preference for specific policy outcomes. This is different to the usual principal-agent relationship, where the agent generally has no interest in which economic policy is pursued by the principal. The military agent, on the other hand, has a preference for policies that do not needlessly risk his life. In addition, the military agent has a preference for policies that give overwhelming supremacy on the battle-field (Feaver 2003:63). The last basic military preference is one for maximal autonomy, which translates in large part into a desire for the minimisation of civilian interference in military affairs (Feaver 2003:64). All of these preferences can lead the military to attempt to influence policy in ways that undermine civilian control. In terms of democratic governance, this is pernicious even when such interference leads to better security arrangements than would otherwise have been achieved. Dealing with this is made all the more difficult by the fact that the military agent carries a particular moral status – her willingness to make the ultimate sacrifice for her country acts in some sense as a moral counterweight to the civilian principal’s political competence. As a consequence, ‘the moral ambiguity of the relationship bolsters the hand of a military agent should he choose to resist civilian direction’ (Feaver 2003:71-72).

The other central problem for civil-military relations, in terms of Feaver’s principal-agent derived theory, is the ‘adverse selection problem’. This is the problem facing the principal in selecting which agent to contract with to undertake the required task. The agent has a strong incentive to portray herself as being far more diligent than she is, in order to ensure that she is contracted, which complicates the principal’s task of selecting the best possible agent for the job. Of course in civil-military terms the issue here is not generally one of deciding whether to contract the military to undertake some activity, say offensive combat operations, rather than contracting some other body to fulfil the task.3 Instead, the task here is one of leadership selection – which potential senior officers are most likely to lead the military to work rather than to shirk?

The special nature of the military context gives this problem a unique twist. Feaver seems right to point out, for example, that it is ‘at least plausible’ that the sort of personality that is advantageous on the battlefield is by nature problematic in terms of the principal-agent relationship (Feaver 2003:72). Indeed, as we have seen, this is one of the central reasons that Huntington stresses a sharp differentiation of the civilian and military spheres.

So described, Feaver’s agency theory offers a very useful descriptive framework by which to understand civil-military relations. In the next section we turn to a consideration of the main advantages that arise from the application of agency theory.

Agency Theory: General Advantages

At its most basic, the notion of professionalism as Huntington uses it raises questions of circularity. For the defining feature of military professionalism, in Huntington’s account, is that of voluntary submission to civilian authority: ‘A highly professional officer corps stands ready to carry out the wishes of any civilian group which secures legitimate authority within the state’ (Huntington 1957:74). This obviously raises questions about definitional circularity – objective civilian control is defined in terms of military professionalism, while military professionalism is defined as voluntary submission by the military to civilian control. Perhaps, however, this circularity is avoided when one takes into account that the voluntariness of military submission to civilian control is incentive-driven – in exchange the military is given near free reign in its own arena of professional expertise (see Huntington’s point 4 above). Following this reading, this aspect of Huntington’s theory approximates agency theory, as we shall see below. Another problem with Huntington’s approach, as Kohn points out, is that ‘while “objective” civilian control might minimize military involvement in politics, it also decreases civilian control over military affairs’ (Kohn 1997:143).

The most obvious advantage of agency theory is that it introduces material factors into the account, something that is clearly missing from the theories proffered by Huntington and Janowitz, while at the same time not excluding the nonmaterial factors (like identity and moral commitments) highlighted by these theorists. These matters are demonstrated convincingly by Feaver in the central chapters of Armed Servants, where he shows that agency theory fits far more closely with real-world civil-military relations – the cases of Cold War and post-Cold War civil-military relations – than its main rival, Huntington’s theory of ‘objective civilian control’. There is little benefit in rehashing Feaver’s entire analysis here. One example will suffice to illustrate this point. Huntington’s prediction, writing as he was in 1957, was that the only way for the US to ‘win’ the Cold War would be for US society to move closer to the conservative values of the military, and become a conservative republic. History, of course, shows that this is not what happened. ‘On the contrary, the evidence shows that American society as a whole almost certainly became ever more individualistic and more antistatist than it was when Huntington warned of the dangers of liberalism in 1957. The Cold War coincided with the flowering of the civil rights movement, the expansion of the New Deal social welfare system, and a dramatic enlargement in court-protected individual rights, all extraordinary extensions of classical liberalism’s reach in American society’ (Feaver 2003:26-28).

By contrast, agency theory offers both a means of accurately measuring civilian control of the military, and mechanisms for addressing the constantly-threatened imbalance in the civil-military relationship. These control and monitoring mechanisms address the issue that is at the heart of both the moral hazard and the adverse selection problems – information – by providing means whereby the principal can cause the agent to reveal the necessary information, or else by adjusting the incentives on offer to the agent in ways that can give the principal confidence that the agents’ preferences are in line with those held by the principal.

Among the central means of monitoring singled out by Feaver are the following (Feaver 2003:76-85):

  1. Restricting the scope of delegation to the military. Devising strategy, drawing up operational plans, directing the equipping and provisioning of the military, and setting rules of engagement are among the means civilians can employ to achieve this.
  2. Contractual incentives. This less intrusive approach ‘builds in’ incentives to the contract between the principal and the agent. In economic terms profit sharing is an obvious example. Obviously no direct equivalent to economic profit exists in the civil-military relationship.4 Feaver points out, however, that the military’s desire for autonomy might well fit here. ‘Since monitoring mechanisms vary in their degree of intrusiveness, and assuming that the military prefers less intrusive means, civilians have a powerful incentive with which to influence military behavior: offer to use less intrusive means to monitor military agents. Indeed, this is how traditional civil-military relations theory treats autonomy’ (Feaver 2003:78).
  3. Screening and selection mechanisms. Such mechanisms (referred to as ‘accession policy’, in the military context) enable civilians to address the adverse selection problem, and also provides (in a relatively non-intrusive manner) information that can help the civilian in predicting the future behaviour of the military agents thereby selected.
  4. ‘Fire alarms’. These are third parties who have an interest in the behaviour of the agent in question, and who therefore monitor the agent. In the civil-military context it is primarily defence-orientated think tanks and the media that play this role, but other less formal groups (such as draftees in a conscription-based military) can play the same role. Inter-service rivalry can also function in this way.
  5. ‘Police patrols’. This describes principal-initiated investigations of the agent, and is a significantly more intrusive form of monitoring than those already mentioned. ‘Police patrols include regularized audits and intrusive reporting requirements designed to turn up evidence of agent wrong-doing and, through regularized inspection, to deter moral hazard. Public investigative hearings and specific mandated reports are staples of congressional oversight and represent one of the more visible avenues of political control’ (Feaver 2003:84).
  6. Revocation of delegated authority. Where necessary, civilian principals retain the option of withdrawing authority that has previously been delegated to the military. This can either be a complete withdrawal (by, for example, re-delegating a particular area of responsibility to a competing arm of the state, such as the police or intelligence community), or a partial withdrawal, in which the civilian agency involved takes on greater powers over, for example, the planning of military operations. This is the most intrusive form of monitoring.

These monitoring mechanisms are general features of most accounts of the principal-agent relationship, though obviously their specific application to the civil-military context is new. What sets agency theory apart from other principal-agent theories, however, is the close attention that is given to punishment mechanisms. Perhaps not surprisingly, the general principal-agent literature tends to take punishment more or less for granted. But this is clearly significantly more difficult in the context of civil-military relations, due to the unique coercive power that is incumbent in military forces. As Feaver points out, history is replete with examples of coups that have been triggered by an attempt to punish the military for some or other indiscretion. Thus, the first question here is whether or not civilian principals can in fact punish the military. Feaver expressly sidesteps this question, arguing that it is not relevant in a discussion of agency theory, which takes as given a relationship in which the military understands itself in the role of the agent, subservient to its civilian masters (Feaver 2003:89). Nonetheless, Feaver makes his own view of the prerequisites for punishment mechanisms in the civil-military context very clear, when he writes that ‘the power to punish rests on a normative foundation – that is, the willingness of the military to be punished – and this normative foundation is thus a prerequisite for democratic civil-military relations. It exists in the United States and other advanced democracies but not necessarily in all countries. As discussed in the concluding chapter, this may limit the applicability of the agency model to other countries’ (Feaver 2003:90). This is an important qualification of agency theory, one to which we shall return in the next section

Where punishment is a possibility, there are, Feaver argues, a wide range of punishment options available to civilians, which generally fall into the following five categories (Feaver 2003:90-93):

  1. Intrusive monitoring. As described above.
  2. Budget cuts and withdrawal of privileges. Given that one of the most likely forms of shirking involves offering inflated estimates of threats and costs in order to boost the military’s share of the national budget, this is often a particularly appropriate punishment mechanism.
  3. ‘Forced detachment from the military’. This is the military equivalent of ‘firing’ a shirking agent. The US military’s ‘up-or-out’ career path system, in which military personnel who do not achieve promotion within a set time-limit are discharged from the service (and thereby lose out on the considerable financial rewards of continuing on in the military), is one mechanism of this kind. Another option available in the US military system is that of forcing a disgraced officer to retire at a rank lower than that achieved, which again results in a significant financial loss.5
  4. Military justice. Military law, such as that encapsulated in the Uniform Code of Military Justice, provides for a range of punishments (including dismissal and imprisonment) that can be applied to transgressing military agents.
  5. Extralegal civilian action. ‘This is a miscellaneous category of actions ranging in severity from private oral rebukes all the way to the infamous Stalinist purges against the Soviet military in the 1930s in which thousands of officers were shot for suspected disloyalty to the Soviet regime. An intermediate form might be a situation in which the military advisor is publicly reprimanded or denied access to the civilian leader because that leader has lost confidence in him’ (Feaver 2003:93).

In addition to these five categories of punishment, Feaver muses that, under some circumstances, war itself might be understood as a form of punishment: ‘War, of course, inflicts hardships on the military, including the possibility of the ultimate sanction, loss of life, and one might consider war as a form of punishment. If war performance is itself a function of civil-military relations, then battlefield defeat could be a form of punishment experienced by the military (and, less directly, by the civilians as well) for adopting sub-optimal civil-military arrangements’ (Feaver 2003:94). He concludes, however, that war is too rare an occurrence for this to be considered a meaningful punishment mechanism.

These mechanisms for monitoring and punishment represent a clear advantage for agency theory over its rivals, which offer no framework of this kind. In terms of analysis, the practical value of agency theory is clear. Feaver is correct when he writes that ‘[a]gency theory cues us to look for certain things and to ask certain questions in a case study and thereby illuminates the give and take of day-to-day civil-military relations in ways that a straightforward journalistic account might miss’ (Feaver 2003:234). Furthermore, agency theory offers a way of understanding civil-military relations that extends beyond the traditional concern with coups to the everyday strategic interactions between the military agent and the civilian principal. The presumption that the military will desire to shirk – where this is understood as more than simply disobeying orders – is an important one, and emphasises the need for civilians to be constantly vigilant. In addition, agency theory offers a means of systematically analysing historical examples of civil-military relations, which Feaver himself does very thoroughly in his analysis of Cold War and Post-Cold War US civil-military relations.

It is also clear that agency theory, as outlined by Feaver, offers scope for further development, and can incorporate insights from other theories (such as Huntington’s) without changing its basic structure. This flexibility is important in the context of today’s rapidly changing security environment, in which military forces are increasingly deployed in unfamiliar roles, with potentially unexpected results for civil-military relations. Related to this is another advantage of agency theory, namely that it ‘preserves the civilian-military distinction – the sine qua non of all civil-military theory – but without reliance on an ideal-type division of labor. And it preserves the military subordination conception essential to democratic theory, without assuming military obedience’ (Feaver 2003:10).

Clearly then, there are good reasons for considering agency theory to be among the most useful frameworks for the analysis of civil-military relations – at least in the day-to-day strategic interactions between the military agent and the civilian principal – currently in existence. An important question remains however. Given Feaver’s own restrictions of the scope of agency theory to mature democracies where there is a normative presumption of civil control of the military, how applicable is it to the more fragile democracies of the world? It is to this question we now turn.

Agency Theory and the African Context

Because Feaver’s focus is on US civil-military relations, he is careful to emphasise that his conclusions regarding the applicability of agency theory can only be seen as applicable in mature democracies like the US, where the submission of the military to civilian authority is a normative presumption. That disclaimer in place, Feaver also recognises that the obvious next step for agency theory research is to examine whether it is applicable in less stable democracies, where the threat of coups is a real one. Although Feaver doesn’t attempt any kind of systematic answer to this question, it seems that his intuitions lead him in different directions. On the one hand, he stresses the connection between agency theory and the normative presumption of civilian oversight mentioned above (Feaver 2003:90). On the other hand, despite his laudable conservatism in not extrapolating results beyond the scope of his investigation, he is unable to completely restrain a cautious optimism on this point: ‘Agency theory may even make some contributions to the study of civil-military relations in countries where the threat of coups is real. After all, a coup represents the ultimate in shirking – reversing the principal-agent relationship so that the old agent (the military) becomes the new principal (the dictator). Pathological civil-military interactions within the agency framework could end up in a coup’ (Feaver 2003:293).

Despite this flash of optimism, Feaver immediately expresses his belief that if agency theory were to be applied to ‘coup-ridden’ states it would require significant modification. It is on this point that I find myself for the first time in disagreement with Feaver. For the essence of the principal-agent relationship is that it is a ‘game’ of strategic interaction between two parties, where each party is presumed to be seeking dominance. Thus, as long as both parties exist (i.e. where civilian government is real rather than just a front for the military), then there will be the ‘game’ of strategic interaction between the two, and therefore there is a good prima facie reason to believe that agency theory will be applicable. The central question is whether, under such conditions, civilian monitoring and punishment is possible. Even where it is not, measuring such cases against agency theory enables a clear and objective judgement about civil-military relations in that particular state to be made.

It is also arguable that, even in states where there is no presumption that the military will submit to civilian oversight, and where civilian governments have few resources with which to apply coercive power over military agents, these civilian governments can still put into place both monitoring and punishment mechanisms, by means of its involvement in regional organisations. In the African context organisations like the African Union (AU) and the Southern African Development Community (SADC) offer a prime example. Such organisations include principles and commitments that have a direct impact on military subservience to civilian governments. Take for example Operation Boleas, the 1998 SADC intervention in Lesotho, which was a response to a suspected coup. The justification for that intervention was explicitly an appeal to a clear SADC principle rejecting the military overthrow of civilian governments. The AU has shown a similar commitment to this principle – in recent times, for example, Mauritania has had its membership of the AU revoked as a result of a military coup in 2005. Furthermore, the AU has in its toolbox the African Peer Review Mechanism which seems to give ample scope for the monitoring of civil-military relations. Punishments available to address transgressions of accepted norms of military submission to civilian authority could range in severity from military intervention to oust the coup plotters (as in the case of Lesotho), sanctions, and diplomatic censure – where coups have actually taken place – to more limited measures that affect only the military directly, for less serious forms of shirking. Examples of these limited measures could include ejection from prestigious regional military arrangements such as the AU standby battalions, withdrawal of training and other assistance, expulsion of officers of the offending military from military colleges in other countries, and the like.

Admittedly, the action by the AU against Mauritania did not lead to the reversal of the coup, and the voluntary status of the African Peer Review Mechanism means that it has relatively little weight. But focusing on the lack of impact of such measures in recent history in Africa misses the point. The point is that such actions are conceptually possible, and in some cases have been actualised. Punishing coups can be done, but how much punishment results, and how much of a deterrent there is against coups as a result, depends on how seriously civilian principals take the threat and how much they are prepared to invest in the deterrent.

Thus there seems good reason to reject Feaver’s claim that ‘agency theory is only applicable in those settings where the military conceives of itself as the agent of the civilian; crucial to that conception is a recognition of the civilian’s right to sanction, and hence an explicit commitment to submit to sanctions’ (Feaver 2003:89). More important than this is the possibility of monitoring and punishment mechanisms, even in cases where the military’s commitment to principles of civilian oversight are limited or even non-existent, and as we have seen this is a real possibility.

A further advantage of agency theory in the context of weak democracies like those common in Africa lies in its value with regard to attempts to ‘democratise’ military forces that have traditionally existed as part of totalitarian regimes. Democratisation programmes, given the Huntingtonian orthodoxy, have tended to focus on inculcating a professional ethos among the officers of the military in question. While there is undoubtedly value in such training (which, as we have seen above, can be incorporated into the theoretical matrix of agency theory), this can only be seen as a long-term strategy, for such values take time to become embedded into the ethos of a military force. In the shorter term, a clear grasp of the monitoring/punishment relationship described by agency theory can allow for immediate implementation of practical mechanisms for the oversight of the military. The presumption that the military will desire to shirk is a particularly useful one in such contexts.

It seems, therefore, that there are good reasons to believe that a more comprehensive investigation than can be attempted here will show that agency theory is just as applicable in the context of the new and emerging democracies of Africa as it is in more established democracies. It is clear, also, that the agency theory represents a major advance over the Huntingtonian orthodoxy in understanding and managing civil-military relations.


  1. Cilliers, Jakkie K. 1995. Security and Transition in South Africa. Journal of Democracy, 6(4) pp. 35-49.
  2. Cohen, Eliot A. 2002. Supreme Command: Soldiers, Statesmen and Leadership in Wartime. New York NY: The Free Press.
  3. Desch, Michael 1999. Civilian Control of the Military: The Changing Security Environment. Baltimore MD: Johns Hopkins University Press.
  4. Feaver, Peter D. 2003. Armed Servants: Agency, Oversight, and Civil-Military Relations. Cambridge MA: Harvard University Press.
  5. Howe, Herbert M. 2001. Ambiguous Order: Military Forces in African States. Boulder CO: Lynne Rienner.
  6. Huntington, Samuel P. 1957. The Soldier and the State: The Theory and Politics of Civil-Military Relations. Cambridge MA: Harvard University Press.
  7. Huntington, Samuel P. 1995. Reforming Civil-Military Relations. Journal of Democracy, 6(4), pp. 9-17.
  8. Janowitz, Morris 1960. The Professional Soldier: A Social and Political Portrait. New York NY: Macmillan.
  9. Kohn, Richard H. 1997. How Democracies Control the Military. Journal of Democracy, 8(4), pp. 140-153.
  10. Williams, John Allen. 1995. The International Image of the Military Professional. African Security Review, 4(5). <>, accessed 8 June 2006.


  1. ‘Politics as a Vocation’. An English translation can be found online at <> (accessed 7 June 2006).
  2. As Feaver (2003:300) points out, there is some evidence that the general presumption that the military is more likely to be correct than civilians on questions of national security is not necessarily the case. In this, Feaver points particularly to Eliot A. Cohen’s important work Supreme Command (Cohen 2002). Feaver (2003:302) comments as follows: ‘History shows that the military is not as “right” in civil-military disputes as the military triumphalists might suppose. But even when the military is right, democratic theory intervenes and insists that it submit to the civilian leadership that the polity has chosen. Let civilian voters punish civilian leaders for wrong decisions.
  3. I say that this is not ‘generally’ the question at hand. It strikes me, however, that this is precisely the relevant question when decisions are made by civilian governments whether to employ military forces or so-called ‘private military companies’ to undertake a particular task. Agency theory seems to offer a very useful framework with which to assess the legitimacy or otherwise of state employment of private military companies. I will address this issue in a forthcoming paper.
  4. At least, not if corruption is disallowed. Some commentators have argued that something like this ‘profit-sharing’ relationship between the military and the state is a major factor in some African contexts, such as recent interventions in the DRC by countries such as Angola and Zimbabwe (see, for example, Howe 2001:88-92).
  5. A recent example of this is the case of Janis Karpinski, who was the Brigadier-General in command of the Abu Ghraib prison, but who was demoted and dismissed from the service at the rank of Colonel.

This Issue


  • Richard Kamidza
  • Jannie Malan

The Case against Taylor’s Asylum

A Review of Nigeria's Domestic and International Legal Obligations

  • Mba Chidi Nmaju

The Dilemma of Restorative Justice when ‘All Are Guilty’

A Case Study of the Conflicts in the Niger Delta Region of Nigeria

  • Austin Onuoha

Integrated Development Planning in South Africa

Lessons for International Peacebuilding?

  • Sybert Liebenberg
  • Elsona van Huyssteen
  • Richard Gueli

Alternative Dispute Resolution (ADR) in the Workplace

The South African Experience

  • Hanneli Bendeman