1: Conceptual and analytical framework: Relational power
Power has been conceptualized in various ways, and famous definitions have typically encountered criticism and invoked debate. The term has therefore not only been called an ‘essentially messy concept’, but also an ‘essentially contested’ one, implying that it would be inherently impossible for people of different ideological persuasions to agree on one definition (e.g. Connolly [1974] 1983; Lukes 1974:26; Gray 1983; Ball 1988:80).1 Such contestability, or at least lack of consensus, is evident in the large number of cross-disciplinary anthologies on the subject (e.g. Bell et al. 1969; Lukes 1986; Wartenberg 1992; Scott 1994), and from the numerous distinctions over which scholars have argued:
Is power a property or a relationship? Is it potential or actual, a capacity or the exercise of a capacity? By whom, or what, is it possessed or exercised: by agents (individuals or collective?) or by structures or systems? Over whom or upon what is it exercised: agents (individual or collective?) or structures or systems? Is it, by definition, intentional, or can its exercise be partly intended or unintended? Must it be (wholly or partly) effective? What kinds of outcomes does it produce: does it modify interests, options, preferences, policies, or behaviour? Is it a relation which is reflexive or irreflexive, transitive or intransitive, complete or incomplete? Is it asymmetrical? Does exercising power by some reduce the power of others? (Is it a zero-sum concept?) Or can its exercise maintain or increase the total of power? Is it demonic or benign? Must it rest on or employ force or coercion, or the threat of sanctions or deprivations? (And, if so, what balance of costs and rewards must there be between the parties for power to exist?) Does the concept only apply where there is conflict of some kind, or resistance? If so, must the conflict be manifest, or may it be latent: must it be between revealed preferences or can it involve real interests (however defined)? Is it a behavioural concept, and, if so, in what sense? Is it a causal concept?
(Lukes 1991:83–4, emphasis in original)
Surveying ways to define and use power in the academic post-Cold War Japanese foreign policy discourse, the Introduction may have appeared to take sides in such debates. Yet, the underlying aim of the chapter was pragmatic, namely, to evaluate the extent to which different ideas of power have added to a coherent understanding of Japan’s foreign policy.
The present chapter is also no contribution to the contestability debate. It is not concerned with uncovering a superior definition (e.g. Lukes 1974: 9, 30–1; cf. Barry 1989:305) or the true nucleus of power (e.g. Morriss 1987). Nor is it an attempt to understand the term’s diversification and contestation by writing conceptual history (e.g. Ball 1988). Instead, the aim of this chapter is simply to introduce Lukes’ relational concept of power, slightly adjust it to make it clearer and better suited to foreign policy analysis (cf. Weber 1968:20), and then to elaborate on some of its methodological and theoretical implications. Since it is the first time that a concept of power explicitly connected to the debates spurred by Lukes’ three-dimensional view is used in this kind of analytical setting,2 and since the ‘three faces debate’ ‘has failed as a methodological agenda for empirical research’ (Isaac 1992:53),3 the quest for appropriate methods is still largely new territory. The chapter thus aims to accomplish a general improvement of a concept of power, which is not taken to be a general one itself. As such, the chapter remains unaffected by the Wittgensteinian conclusion that Dahl, Bachrach and Baratz, Lukes and others have perverted the lexical meaning of the term so severely that ‘we usually talk more sense [of “power”] in the pub than in the seminar’ (Morriss 1987:1), and that ‘those who have spent longest puzzling over the term seem to make least sense of it in their writings’ (ibid.). Lukes’ little book and the ‘three faces debate’ have simply received too much attention to be discarded that easily. Finally, regardless of the Oxford English Dictionary’s stipulations (ibid.: 9–10; cf. Isaac 1992:45), the meaning assigned to words change, and Lukes, Bachrach and Baratz, and Dahl and others have contributed to such a change in the social sciences (Ball 1988:86–91).
The next section first presents the ‘three dimensions of power’ in their own right. With Lukes and his predecessors as the starting point, the ensuing section then turns to the question how to conceptualize power in this study. It does so by scrutinizing a number of discussions raised in connection with the three-dimensional definition. Two other topics are then developed: ‘Power and statecraft’ and ‘Power and interests’. The former section addresses some implications of a relational concept of power for the relationship between power and capability. The latter gives a brief background to the role of interests in political analysis, and then proceeds to present an interpretation of Lukes’ notion of ‘real interests’ in classical liberal terms. Next, a method for relational power is developed incrementally. This endeavor sets out from the proposition that a reconstructive and interpretative methodology outlined in a spirit of process-tracing and intentional modes of analysis goes well with the relational concept, and the last section of the chapter proceeds to spell out the exact questions of ‘relational power analysis’. The resultant conceptual framework can be understood as the major point of view, or ‘means of exposition’
(Darstel-lungsmittel) of this study (Weber 1949:78, emphasis in original). This is what Weber calls an ‘ideal type’ (Idealtypus), meaning a ‘purely ideal limiting concept with which the real situation or action is compared and surveyed for the explication of certain of its significant components’ (ibid.: 93, emphasis in original; cf. Weber 1968:20).4 In this study, the term ‘ideal type’ is thus used to signify concepts, which are instrumental to understanding or interpreting historical phenomena.
Three dimensions of power
In this section, the three dimensions of power are briefly presented in their own right: Dahl’s conception symbolizes the first dimension, Bachrach and Baratz represent the second, and Lukes represents the third.
First dimension: a pluralist concept of power
Robert Dahl has defined power in various ways. However, most coherent with his approach at large, power is defined as
a successful attempt by A to get a to do something he would not otherwise do.
(1957:292, cf. ibid.: 290)
With the preferred research procedures depending largely on available data, the one-dimensional view is operationally pragmatic. Yet, ideal as well as proximate observations center on overt decision-making or bargaining in directly observable disagreements or conflicts concerning key (as opposed to ‘routine’) issues (in the shape of policy alternatives). The key question is who participates, who gains and who prevails. Hence, actors who, against the will of others, successfully initiate, oppose, veto or alter alternatives in concrete agenda setting or decision-making are deemed most powerful (in those settings). In short, power is analyzed by way of reconstructing actual behavior (and thereby, it is believed, preferences) as it appears in documents, interviews, news articles, etc. (1958:36–41; cf. Lukes 1974:11–15).
Second dimension: a reformist concept of power
Although it emerged as a critique of Dahl, Bachrach and Baratz’ idea of power does not deviate considerably from that of their predecessor. However, it is operationalized both more narrowly and more broadly than his concept. Bachrach and Baratz’ approach is narrower in the sense that power is more exclusively defined: A power relation exists only when (1) there is a conflict over values, interests or courses of action between A and B; (2) B complies with A’s wishes; and (3) B does so out of fear of being deprived by A of a value which B regards more highly than those which would have been achieved by non-compliance (Bachrach and Baratz 1963: 98). The two-dimensional view is broader to the extent that it does not confine power phenomena to the publicly observable political debate. Power is also exercised by A over B through A’s determination of what particular issues and people be allowed into the debate as such:5 ‘if issues are prevented from arising, so too may actors be prevented from acting’ (Gaventa 1980:9). Hence, Bachrach and Baratz additionally stress the importance of analyzing non-decision-making,6 potential issues and covert conflicts. Even if actors are divided over decision-making, a ‘mobilization of bias’ turns non-decision-making into a consensus-prone activity of status-quo defenders (1962:88). It is thus necessary to analyze both the issues that make up the agenda and the potential ones that do not (cf. Majone 1989). However, methodologically Bachrach and Baratz do not really diverge from Dahl. Non-decisions (or ‘omissions’) are depicted as readily observable species of ‘decisions not to act’, and they are discerned as the origin of overt and covert grievances—the latter existing outside of the political system. However, the two-dimensional view ‘does not go so far as to include how power may affect conceptions of grievances themselves’ (Gaventa 1980:10–11)—an objection that spurred Lukes’ three-dimensional view.
Third dimension: a radical concept of power
The ‘radical view’ thus evolved as the result of Lukes’ criticism that the one- and two-dimensional views are biased towards the political systems that they investigate; that they neglect the groups and issues that have been shut out of ‘politics’. Lukes’ own definition of power reads:
A exercises power over B when A affects B in a manner contrary to B’s interests.
(1974:34)
One characteristic of this view is the way in which it relates power to the notion of ‘interests’ (ibid.: 24–5, 33–5; cf. Connolly 1983:104). In short, the core of Lukes’ three-dimensional view is that power can be exerted without an explicit conflict of interest between A and B. This is to say that A exerts power over B by means of control, manipulation and authority, without B, or even A, being aware of that relationship. Whenever A exerts power over B unconsciously or unintentionally,7 observable conflicts are absent, and issues become potential rather than actual. By arguing that seemingly consensual relations may embody latent conflicts of interest, i.e. inconsistency between A’s interests and the ‘real interests’ of B, Lukes partly disassociates his conception from the behaviorist ontology of his predecessors. This approach to interests is further discussed in the section ‘Power and interests’ below.
Relational power: a conceptual analysis
This section presents Lukes’ conceptualization in greater detail while examining some objections that have been raised against it. In the process, an increasingly operational concept of power evolves piecemeal. Themes not exhausted here are brought up again in later sections.
The Lukesian merger
The three dimensions of power are clearly distinct, and yet the three-dimensional view is generally interpreted to incorporate its predecessors (Lukes 1974:26–7; Gaventa 1987:22, 50; Isaac 1992:39), thus forming what could be called a ‘Lukesian merger’.8 The three dimensions do indeed have some traits in common. They first and foremost share the same underlying definition of power (Lukes 1974:27), namely Lukes’ three-dimensional one (cf. ibid.: 30). They also share an emphasis on contextuality. This is the position that scope (the aspect of B affected by A, for example in terms of issues), domain (the boundaries of B, for example with regard to time and space), weight (the probability that A affects B), means and cost (for both A and B) matter and should be clarified (cf. Baldwin 2002:178).9 Finally, all three views take for granted that the most relevant units of analysis are actors rather than structures. However, there is no consensus in regard to the last point—a topic that is further elaborated next.
Actors v. structures in power analysis
Annica Kronsell’s interpretation of Dahl, Bachrach and Baratz, and in particular of Lukes (1997:19–39), completely diverges from mine and from that of many others (e.g. Beronius 1986:42–3; Ball 1988:98; Isaac 1992:39, 41). She describes them as disinterested in power relationships, and calls Lukes a structuralist (1997:23). This view, which is shared by a number of scholars (e.g. Gill and Law 1988:73–4; Hindess 1996:81; Frølund Thomsen 2001:54), is supported by the following remark:
the bias of the system is not sustained simply by a series of individually chosen acts, but also, most importantly, by the socially structured and culturally patterned behaviour of groups, and practices of institutions, which may indeed be manifested by individuals’ inaction…
[C]ollectives and organisations are made up of individuals—but the power they exercise cannot be simply conceptualised in terms of individuals’ decisions or behaviour.
(Lukes 1974:21–2; cf. ibid.: 24)
The introduction of A’s unintended exertion of power over B is also taken as evidence that Lukes has a structure-oriented concept in mind.
Yet, despite the acknowledgment that there is a ‘structural’ aspect not just to Lukes’ three-dimensional view, but also to Bachrach and Baratz’ two-dimensional one (Guzzini 1993:462–3),10 Stefano Guzzini associates Lukes—‘for whom power becomes attached to personal autonomy and the moral discourse of freedom and justice’ (ibid.: 470)—with a critique against ‘structural reductionism’:
By not sufficiently stressing the fundamental agent reference of power, the criticism runs, the concept of power becomes either synonymous with structural constraint, thus rendering structural power a contradiction in terms, or else it becomes a rather amorphous all-encompassing concept like social control.
(Ibid.: 469)
Guzzini himself believes that ‘it is important that systematic bias be part of any power analysis’ (2000b: 63, emphasis in original), but without being ‘collapsed into the concept of power’ (ibid. emphasis in original). He argues that such phenomena had better be understood in terms of ‘governance’, defined as ‘the capacity of intersubjective practices to effect’ (1993: 471). This idea is almost completely disentangled from specific human agents; governance is merely passing through them, ‘reproduced and realized via practices, habits, dispositions, and sometimes even through th...