Review of International Studies (1997), 23, 5–26 Copyright © British International Studies Association

The concept of security*

DAV I D A . B A L D W I N

Redefining ‘security’ has recently become something of a cottage industry.1 Most

such efforts, however, are more concerned with redefining the policy agendas of
nation-states than with the concept of security itself. Often, this takes the form of
proposals for giving high priority to such issues as human rights, economics, the
environment, drug traffic, epidemics, crime, or social injustice, in addition to the
traditional concern with security from external military threats. Such proposals are
usually buttressed with a mixture of normative arguments about which values of
which people or groups of people should be protected, and empirical arguments as
to the nature and magnitude of threats to those values. Relatively little attention is
devoted to conceptual issues as such. This article seeks to disentangle the concept of
security from these normative and empirical concerns, however legitimate they
may be.
Cloaking normative and empirical debate in conceptual rhetoric exaggerates the
conceptual differences between proponents of various security policies and impedes
scholarly communication. Are proponents of economic or environmental security
using a concept of security that is fundamentally different from that used by
Realists? Or are they simply emphasizing different aspects of a shared concept? Do
those who object to ‘privileging’ the nation-state rather than, say, the individual or
humanity share any conceptual views with students of ‘national security’? This
article attempts to identify common conceptual distinctions underlying various
conceptions of security.
Identifying the common elements in various conceptions of security is useful in at
least three ways: First, it facilitates asking the most basic question of social science,

* The author would like to thank the following scholars for helpful comments on previous versions of
this article: Richard Betts, Lea Brilmayer, Robert Jervis, Helen Milner, Jack Snyder, and Hendrik
Spruyt.
1
E.g. Lester Brown, Redefining National Security, Worldwatch Paper No. 14 (Washington, DC, 1977);
Jessica Tuchman Matthews, ‘Redefining Security’, Foreign Affairs, 68 (1989), pp. 162–77; Richard H.
Ullman, ‘Redefining Security’, International Security, 8 (1983), pp. 129–53; Joseph J. Romm, Defining
National Security (New York, 1993); J. Ann Tickner, ‘Re-visioning Security’, in Ken Booth and Steve
Smith (eds.), International Relations Theory Today (Oxford, 1995), pp. 175–97; Ken Booth, ‘Security
and Emancipation’, Review of International Studies, 17 (1991), pp. 313–26; Martin Shaw, ‘There Is No
Such Thing as Society: Beyond Individualism and Statism in International Security Studies’, Review
of International Studies, 19 (1993), pp. 159–75; John Peterson and Hugh Ward, ‘Coalitional
Instability and the New Multidimensional Politics of Security: A Rational Choice Argument for
US–EU Cooperation’, European Journal of International Relations, 1 (1995), pp. 131–56; ten articles
on security and security studies in Arms Control, 13, (1992), pp. 463–544; and Graham Allison and
Gregory F. Treverton (eds.), Rethinking America’s Security: Beyond Cold War to New World Order
(New York, 1992).

5
6 David A. Baldwin

‘Of what is this an instance?’.2 Second, it promotes rational policy analysis by

facilitating comparison of one type of security policy with another. And third, it
facilitates scholarly communication by establishing common ground between those
with disparate views. Perhaps scholars from different schools have more in common
than is generally acknowledged.3
In many ways the argument presented here was foreshadowed in the classic essay
by Arnold Wolfers entitled ‘ ‘‘National Security’’ as an Ambiguous Symbol’, pub-
lished more than forty years ago. Contrary to popular belief, Wolfers did not dismiss
the concept as meaningless or hopelessly ambiguous. He was, however, concerned
about the ambiguity of ‘national security’, as the following passage indicates:
It would be an exaggeration to claim that the symbol of national security is nothing but a
stimulus to semantic confusion, though closer analysis will show that if used without
specifications it leaves room for more confusion than sound political counsel or scientific
usage can afford.4
Wolfers’ ‘specifications’ refer not only to the concept of national security as a policy
objective but also to the means for its pursuit, i.e., national security policy. In the
discussion that follows, Wolfers’ specifications will be developed and set in the
context of more recent literature.
The discussion consists of seven parts. Sections 1, 2 and 3 provide background
with respect to the approach to conceptual analysis to be used, the neglect of con-
ceptual analysis by the field of security studies, and the possibility that conceptual
analysis is futile with respect to concepts like security that are alleged to be
‘essentially contested’. Section 4 develops a series of conceptual specifications that
facilitate analysis of security policy. Sections 5, 6 and 7 discuss the implications of
these specifications for determining the value of security, for the theory of
neorealism, and for the ‘new thinking’ about security.
Although this discussion is especially concerned with the security of nation-states,
most of the analysis is applicable, mutatis mutandis, to any level: individual, family,
society, state, international system, or humanity.

1. Conceptual analysis5

Conceptual analysis is not concerned with testing hypotheses or constructing

theories, though it is relevant to both. It is concerned with clarifying the meaning of
concepts. Some would dismiss such undertakings as ‘mere semantics’ or ‘pure
logomachy’. Without clear concepts, however, scholars are apt to talk past each
other, and policy-makers find it difficult to distinguish between alternative policies.
Felix E. Oppenheim has argued that ‘the elucidation of the language of political

2
I thank James Rosenau for this phrase.
3
See Kjell Goldmann, ‘Im Westen Nichts Neues: Seven International Relations Journals in 1972 and
1992’, European Journal of International Relations, 1 (1995), pp. 245–58.
4
Arnold Wolfers, ‘ ‘‘National Security’’ as an Ambiguous Symbol’, Political Science Quarterly, 67
(1952), p. 483.
5
This section draws on the discussion in David A. Baldwin, Paradoxes of Power (Oxford, 1989),
pp. 170–2.
Concept of security 7

science is by no means an idle exercise in semantics, but in many instances a most

effective way to solve substantive problems of research’.6
Conceptual explication attempts to specify the logical structure of given expressions: Taking
its departure from the customary meanings of the terms, explication aims at reducing the
limitations, ambiguities, and inconsistencies of their ordinary usage by propounding a
reinterpretation intended to enhance the clarity and precision of their meanings as well as
their ability to function in hypotheses and theories with explanatory and predictive force.7
The explication of concepts is subject to a set of criteria summarized by
Oppenheim: (1) Concepts should be operational in the broadest sense, although this
should not be interpreted as requiring quantification. (2) Concepts that establish
definitional connections with other terms are to be preferred. (3) Concepts that draw
attention to the theoretically important aspects of the subject matter that might
easily be overlooked are desirable. (4) Concepts should not preclude empirical
investigation by making true ‘by definition’ what should be open to empirical
inquiry. (5) Concepts should remain reasonably close to ordinary language.
‘Ordinary language’, however, does not necessarily mean the way most people would
define the term, but rather the ‘set of rules they implicitly follow when applying it to
a given situation’.8
It is important to be clear about the limits of this approach. Explicating the
concept of security does not provide empirical propositions, theories, or analytical
frameworks. Although clear concepts are useful for constructing propositions.
theories, and analytical frameworks, they are not a substitute for them.
This approach may be contrasted with those taken by Barry Buzan and Richard
Ullman. Although Buzan casts his discussion as an exploration of the concept of
security, his analysis intertwines conceptual analysis with empirical observations. For
example, Buzan presents plausible arguments for the empirical proposition that
security at the individual level is related to security at the level of the state and the
international system. His insistence that ‘security cannot be isolated for treatment at
any single level’, however, gives the impression that this is conceptually impossible
rather than simply an unwise research strategy. His justification for mixing con-
ceptual and empirical analysis is that ‘the search for a referent object of security
goes hand-in-hand with that for its necessary conditions’.9 This approach, however,

6
Felix E. Oppenheim, ‘The Language of Political Inquiry: Problems of Clarification’, in Fred I.
Greenstein and Nelson W. Polsby (eds.), Handbook of Political Science, Vol. I: Political Science: Scope
and Theory (Reading, MA, 1975), p. 284.
7
Carl G. Hempel, Fundamentals of Concept Formation in Empirical Science (Chicago, 1952), p. 12.
8
Oppenheim, ‘Language’, pp. 297–309. See also Felix E. Oppenheim, Political Concepts: A
Reconstruction (Chicago, 1981). For criticisms of this approach, see William E. Connolly, The Terms
of Political Discourse, 2nd edn (Princeton, 1983), and Richard E. Little, ‘Ideology and Change’, in
Barry Buzan and R. J. Barry Jones (eds.), Change and the Study of International Relations: the Evaded
Dimension (New York, 1981), pp. 30–45.
Regardless of whether one accepts the criteria suggested here, it is necessary to identify some
criteria for conceptual analysis. Barry Buzan’s contention that security is ‘weakly conceptualized’ and
‘underdeveloped’ would be more telling if he were to identify criteria for distinguishing between weak
and strong conceptualizations or between underdeveloped and fully developed concepts. Barry
Buzan, People, States and Fear: An Agenda for International Security Studies in the Post-Cold War
Era, 2nd edn (Boulder, CO, 1991), pp. 3–5.
9
Buzan, People, States, pp. 20–1, 26. Italics added. See also, Barry Buzan, ‘Peace, Power and Security:
Contending Concepts in the Study of International Relations’, Journal of Peace Research, 21 (1984),
pp. 109–25.
8 David A. Baldwin

risks conflating conceptual analysis with empirical observation. Understanding the

concept of security is a fundamentally different kind of intellectual exercise from
specifying the conditions under which security may be attained. Indeed, conceptual
clarification logically precedes the search for the necessary conditions of security,
because the identification of such conditions presupposes a concept of security.10
This failure to recognize the logical priority of conceptualization is also reflected
in Ullman’s observation that ‘one way of moving toward a more comprehensive
definition of security’ is to ask what one would be ‘willing to give up in order to
obtain more security’. Such a question, however, has little meaning until one first has
a concept of security. Likewise, his statement that ‘we may not realize what it [i.e.
security] is . . . until we are threatened with losing it’11 is difficult to comprehend. If
one has no concept of security, one cannot know whether one is threatened with
losing it or not. Inquiry into the opportunity costs of security is an excellent way to
determine the value of security, but it is no help at all in determining what security is.

2. Security as a neglected concept

It would be an exaggeration to say that conceptual analysis of security began and

ended with Wolfers’ article in 1952—but not much of one. The neglect of security as
a concept is reflected in various surveys of security affairs as an academic field. In
1965 one such study lamented that ‘thus far there have been very few attempts . . . to
define the concept of national security’.12 In 1973 Klaus Knorr began a survey of
the field by stating his intention to ‘deliberately bypass the semantic and definitional
problems generated by the term ‘‘National Security’’ ’.13 In 1975, Richard Smoke
observed that the field had ‘paid quite inadequate attention to the range of
meanings of ‘‘security’’ ’.14 In 1991, Buzan described security as ‘an underdeveloped
concept’ and noted the lack of ‘conceptual literature on security’ prior to the
1980s.15 Although Buzan sees some progress in the 1980s, there are still indicators of
neglect. Two recent surveys of security studies, for example, did not bother to define
security.16 And none of the eleven course syllabi described in Security Studies for the
1990s includes Wolfers’ seminal article on the concept of national security.17

10
‘The elaboration of hypotheses presupposes, logically, a conceptual framework in terms of which
clear hypotheses may be formulated’. Harold D. Lasswell and Abraham Kaplan, Power and Society:
A Framework for Political Inquiry (New Haven, CT, 1950), p. x.
11
Ullman, ‘Redefining Security’, pp. 130, 133.
12
P. G. Bock and Morton Berkowitz, ‘The Emerging Field of National Security’, World Politics, 19
(1966), p. 124.
13
Klaus Knorr, ‘National Security Studies: Scope and Structure of the Field’, in Frank N. Trager and
Philip S. Kronenberg (eds.), National Security and American Society: Theory, Process and Policy
(Lawrence, KS, 1973), p. 5.
14
Richard Smoke, ‘National Security Affairs’, in Fred I. Greenstein and Nelson W. Polsby (eds.),
Handbook of Political Science, Vol. 8: International Politics (Reading, MA, 1975), p. 259.
15
Buzan, People, States, pp. 3–4.
16
Stephen M. Walt, ‘The Renaissance of Security Studies’, International Studies Quarterly, 35 (1991),
pp. 211–39; and Joseph S. Nye, Jr and Sean M. Lynn-Jones, ‘International Security Studies: A Report
of a Conference on the State of the Field’, International Security, 12 (1988), pp. 5–27.
17
Richard Schultz, Roy Godson, and Ted Greenwood (eds.), Security Studies for the 1990s (New York,
1993). A recent forum on security in Arms Control, 13 (1992), including ten authors, never mentions
Wolfers’ article.
Concept of security 9

Considering the plethora of attempts to ‘redefine’ security since the end of the
Cold War18, one might question whether security should be described as a neglected
concept. Two reasons for doing so are compelling. First, security is an important
concept, which has been used to justify suspending civil liberties, making war, and
massively reallocating resources during the last fifty years. Despite the flurry of
recent works, it seems fair to describe security as a concept that received far less
scholarly attention than it deserved during that period. And second, most recent
works on security would not qualify as conceptual analysis in the sense described in
the previous section. Security has not received the serious attention accorded to the
concepts of justice, freedom, equality, obligation, representation, and power.19
Buzan suggests five possible explanations for the neglect of security.20 First, is the
difficulty of the concept. As Buzan admits, however, this concept is no more difficult
than other concepts. Second, is the apparent overlap between the concepts of
security and power. Since these are easily distinguishable concepts, however, one
would have expected such confusion to motivate scholars to clarify the differences.
Third, is the lack of interest in security by various critics of Realism. This, however,
does not explain why security specialists themselves neglected the concept. Fourth, is
that security scholars are too busy keeping up with new developments in technology
and policy. This, however, is more an indication that such scholars give low priority
to conceptual issues than an explanation for this lack of interest. And the fifth
explanation considered by Buzan is that policy-makers find the ambiguity of
‘national security’ useful, which does not explain why scholars have neglected the
concept.21 On balance, none of Buzan’s explanations is very convincing.
Paradoxical as it may seem, security has not been an important analytical concept
for most security studies scholars. During the Cold War, security studies was com-
posed mostly of scholars interested in military statecraft. If military force was
relevant to an issue, it was considered a security issue; and if military force was not
relevant, that issue was consigned to the category of low politics. Security has been a
banner to be flown, a label to be applied, but not a concept to be used by most
security studies specialists. Buzan’s puzzlement as to how a central concept
like security could be so ignored disappears with the realization that military force,
not security, has been the central concern of security studies.22

18
For an overview of these attempts, see Tickner, ‘Re-visioning Security’; and Emma Rothschild, ‘What
is Security?’, Daedalus, 124 (1995), pp. 53–98. See also references in n. 1 above.
19
Peter Digeser, ‘The Concept of Security’, paper delivered at the 1994 Annual Meeting of the
American Political Science Association, 14 September 1994. Unlike most recent works on security,
this paper seriously engages conceptual issues.
20
Buzan, People, States, pp. 7–11.
21
The only instance that has come to my attention of a scholar suggesting that the ambiguity of
security might be an asset is Buzan, ‘Peace, Power’, p. 111.
22
I have found few security specialists willing to defend the label ‘security studies’. The characteristic
response may be paraphrased as follows: ‘This is merely a semantic issue. Personally, I don’t care
whether the field is called security studies, military studies, or war studies. Security studies, however, is
widely used; and it seems to make the subject more acceptable in academia’. If security were really
the central focus of the field, one would expect a more vigorous defence of the label.
10 David A. Baldwin

3. Security as a contested concept

Some scholars have depicted security as an ‘essentially contested concept’.23 This

contention must be addressed before we proceed to analyse the concept of security,
for three reasons: First, there is some ambiguity as to what this means. Second,
security may not fulfil the requirements for classification as an ‘essentially contested
concept’. And third, even if security were to be so classified, the implications for
security studies may be incorrectly specified.24
Essentially contested concepts are said to be so value-laden that no amount of
argument or evidence can ever lead to agreement on a single version as the ‘correct
or standard use’.25 The stronger variants of this position lead to a radical sceptical
nihilism in which there are no grounds for preferring one conception of security to
another.26 Acceptance of this position would make the kind of conceptual analysis
undertaken here futile. There are, however, weaker forms of this position that allow
one to differentiate between better and worse conceptualizations, even though ulti-
mately none of the better conceptualizations can ever be said to be the best.27 Since
the analysis undertaken here purports only to improve on current usage, and not to
identify the single best usage, it is compatible with the weaker variant of the essen-
tial contestedness hypothesis.
It is not clear, however, that security should be classified as an essentially
contested concept. Of the several requirements for such a classification, two are
especially questionable with respect to the concept of security. In the first place, the
concept must be ‘appraisive in the sense that it signifies or accredits some kind of
valued achievement’.28 W. B. Gallie uses the concept of a ‘champion’ in sports to
illustrate the point, i.e., to label a team as champion is to say that it plays the game
better than other teams. Is the concept of security similar to the concept of a
champion? Neorealists seem to imply that it is. For them security is the most
important goal a state can have in the same way that winning a championship is
presumably the goal of all teams in Gallie’s example. Just as teams compete to be
champions, so states compete for security. And just as the champion is better at
playing the game than other teams, so states with more security than other states are

23
Buzan, People, States, and ‘Peace, Power’; and Little, ‘Ideology and Change’. For the original
formulation, see W. B. Gallie, ‘Essentially Contested Concepts’, Proceedings of the Aristotelian
Society, N.S., 56 (1956), pp. 167–98.
24
It should also be noted that the concept of an ‘essentially contested concept’ has itself been
contested. For references, see Christine Swanton, ‘On the ‘‘Essential Contestedness’’ of Political
Concepts’, Ethics, 95 (1985), pp. 811–27; Alasdair MacIntyre, ‘The Essential Contestability of Some
Social Concepts’, Ethics, 84 (1973), pp. 1–9; John N. Gray, ‘On the Contestability of Social and
Political Concepts’, Political Theory, 5 (1977), pp. 330–48; and Oppenheim, Political Concepts,
pp. 182–5.
25
Gallie, ‘Essentially Contested Concepts’, p. 168.
26
Gray, ‘On the Contestability’, p. 343; Swanton, ‘On the ‘‘Essential Contestedness’’ ’, pp. 813–14.
27
Swanton, ‘On the ‘‘Essential Contestedness’’ ’, pp. 813–14.
28
Gallie, ‘Essentially Contested Concepts’, p. 171. Not all value judgments are appraisive. Appraisal
presupposes an accepted set of criteria. Examples suggested by Oppenheim include ‘grading apples or
student papers, evaluating paintings in terms of their market value, [and] wine tasting’. Political
Concepts, pp. 170–6.
Concept of security 11

better at playing the neorealist version of the ‘game’ of international politics.29 From
the neorealist perspective, then, it is plausible to treat security as an appraisive
concept.
Wolfers, however, presents a different view of security. He contends that states
vary widely in the value they place on security and that some states may be so
dissatisfied with the status quo that they are more interested in acquiring new values
than in securing the values they have.30 From this perspective, saying that one state
has more security than another does not imply that one state is better than another
any more than saying that one state has more people or land area implies that one
state is better than another. For Wolfers international politics is not a ‘game’ in
which all states play by the same ‘rules’ and compete for the same ‘championship’.
Is security an appraisive concept? For neorealists, it may be. For others, such as
Wolfers, it is not. The purpose of this discussion is not to settle the issue, but only to
point out that this question is more difficult to answer than those who classify
security as an essentially contested concept imply.
A second requirement for classifying a concept as essentially contested—indeed,
the defining characteristic of such concepts—is that it must actually generate
vigorous disputes as to the nature of the concept and its applicability to various
cases. Gallie deliberately rules out policy disputes in ‘practical life’ that reflect con-
flicts of ‘interests, tastes, or attitudes’. These, he suggests, are more likely to involve
special pleading and rationalization than deep-seated philosophical disagreement.31
Thus, much of the contemporary public policy debate over whether to treat the
environment, budget deficits, crime or drug traffic as national security issues does
not qualify as serious conceptual debate by Gallie’s standards. For Gallie essential
contestedness implies more than that different parties use different versions of a
concept. Each party must recognize the contested nature of the concept it uses, and
each must engage in vigorous debate in defence of its particular conceptual view-
point.32 Yet the security studies literature, as the previous section pointed out, is
virtually bereft of serious conceptual debate. The neorealists may have a different
conception of security than Wolfers, but they do not debate his position; they ignore
it.33 Writers often fail to offer any definition of security. And if one is offered, it is
rarely accompanied by a discussion of reasons for preferring one definition rather
than others. This is hardly the kind of toe-to-toe conceptual combat envisioned by
Gallie with respect to such matters as what constitutes justice, democracy, or a good
Christian.

29
Cf. Kenneth N. Waltz, Theory of International Politics (Reading, MA, 1979), and ‘The Emerging
Structure of International Politics’, International Security, 18 (1993), pp. 44–79; and John J.
Mearsheimer, ‘Disorder Restored’, in Graham Allison and Gregory F. Treverton (eds.), Rethinking
America’s Security (New York, 1992), pp. 213–37.
30
Wolfers, ‘National Security’, p. 491–2.
31
Gallie, ‘Essentially Contested Concepts’, p. 169.
32
Ibid., p. 172.
33
In Waltz’s Theory, for example, security is posited as the principal goal of states; but little attention is
given to defining it or defending the definition against other conceptions of security. Wolfers is never
cited. What Tickner (‘Re-visioning Security’, p. 177) describes as ‘a fully fledged debate about the
meaning of security’ beginning in the 1980s is better characterized as a series of attacks on Realism
and neorealism. A debate implies that there are two sides. With the possible exception of Buzan, no
example of a Realist or neorealist engaging critics in serious conceptual debate has come to this
author’s attention. And Buzan cannot fairly be described as a defender of traditional Realist or
neorealist conceptions of security.
12 David A. Baldwin

Even if security were to be classified as an essentially contested concept, some of

the implications suggested by Buzan are questionable. One cannot use the desig-
nation of security as an essentially contested concept as an excuse for not
formulating one’s own conception of security as clearly and precisely as possible.
Indeed, the whole idea of an essentially contested concept is that various parties
purport to have a clearer and more precise understanding of the concept than
others. Yet Buzan explicitly disavows any intention of formulating a precise defini-
tion and suggests that to attempt to do so is to misunderstand the function of
essentially contested concepts in social science.34 ‘Such a conclusion’, as Ken Booth
points out, ‘is unsatisfying. If we cannot name it, can we ever hope to achieve it?’35
Another consequence Buzan attributes to the essential contestability of security is
a set of ‘contradictions latent within the concept itself’.36 It is not entirely clear what
this means, but such ‘contradictions’ seem to include those between the individual
and the state, between national and international security, between violent means
and peaceful ends, between blacks and whites in South Africa, between the Jews and
Nazi Germany, and so on. Indeed, Buzan’s assertion that the ‘principal security
contradiction’ for most states is between their own security and that of other states
suggests that the Cold War itself could be described as a ‘contradiction’ between the
security of the NATO allies and the Warsaw Pact countries.37 It is true, of course,
that the state’s pursuit of security for itself may conflict with the individual’s pursuit
of security; but this is an empirical fact rather than a conceptual problem. Most of
the phenomena designated by Buzan as conceptual ‘contradictions’ could more
fruitfully be called instances of empirically verifiable conflict between various actors
or policies.
In sum, the alleged essential contestedness of the concept of security represents a
challenge to the kind of conceptual analysis undertaken here only in its strong
variants. There are some grounds for questioning whether security ought to be
classified as an essentially contested concept at all. And even if it is so classified, the
implications may be misspecified. Insofar as the concept is actually contested this
does not seem to stem from ‘essential contestability’. Security is more appropriately
described as a confused or inadequately explicated concept than as an essentially
contested one.

4. Specifying the security problematique

National security, as Wolfers suggested, can be a dangerously ambiguous concept if

used without specification. The purpose of this section is to identify some
specifications that would facilitate analysing the rationality of security policy. The
discussion begins with specifications for defining security as a policy objective and
proceeds to specifications for defining policies for pursuing that objective.

34
Buzan, People, States, pp. 16, 374; and ‘Peace, Power’, p. 125.
35
Booth, ‘Security and Emancipation’, p. 317. On Buzan’s claim regarding the essential contestability of
security, see also Digeser, ‘Concept of Security’.
36
Buzan, People, States, pp. 1–2, 15, 364.
37
Ibid., p. 364.
Concept of security 13

The point of departure is Wolfers’ characterization of security as ‘the absence of

threats to acquired values’,38 which seems to capture the basic intuitive notion
underlying most uses of the term security. Since there is some ambiguity in the
phrase ‘absence of threats’, Wolfers’ phraseology will be reformulated as ‘a low
probability of damage to acquired values’. This does not significantly change
Wolfers’ meaning, and it allows for inclusion of events such as earthquakes, which
Ullman has argued should be considered ‘threats’ to security.39 The advantage of
this reformulation can be illustrated as follows: In response to threats of military
attack, states develop deterrence policies. Such policies are intended to provide
security by lowering the probability that the attack will occur. In response to the
‘threat’ of earthquakes, states adopt building codes. This does not affect the
probability of earthquakes, but it does lower the probability of damage to acquired
values. Thus the revised wording focuses on the preservation of acquired values and
not on the presence or absence of ‘threats’. With this reformulation, security in its
most general sense can be defined in terms of two specifications: Security for whom?
And security for which values?

Security for whom?

As Buzan rightly points out, a concept of security that fails to specify a ‘referent
object’ makes little sense.40 For Buzan, however, a simple specification, such as ‘the
state’ or ‘the individual’, does not suffice. Since there are many states and indivi-
duals, and since their security is interdependent, he argues that the ‘search for a
referent object of security’ must go ‘hand-in-hand with that for its necessary con-
ditions’.41 As noted above, however, this approach confuses concept specification
with empirical observation. For purposes of specifying the concept of security, a
wide range of answers to the question, ‘Security for whom?’ is acceptable: the
individual (some, most, or all individuals), the state (some, most, or all states), the
international system (some, most, or all international systems), etc. The choice
depends on the particular research question to be addressed.

Security for which values?

Individuals, states, and other social actors have many values. These may include
physical safety, economic welfare, autonomy, psychological well-being, and so on.
The concept of national security has traditionally included political independence
and territorial integrity as values to be protected; but other values are sometimes
added. The former American Secretary of Defense Harold Brown, for example,
includes the maintenance of ‘economic relations with the rest of the world on

38
Wolfers, ‘National Security’, p. 485.
39
Ullman, ‘Redefining Security’.
40
Buzan, People, States, p. 26.
41
Ibid.
14 David A. Baldwin

reasonable terms’ in his conception of national security.42 Failure to specify which

values are included in a concept of national security often generates confusion.
Wolfers distinguished between objective and subjective dimensions of security.43
His purpose was to allow for the possibility that states might overestimate or
underestimate the actual probability of damage to acquired values. In the former
case, reducing unjustified fears might be the objective of security policy; while in the
latter case, a state might perceive itself as secure when it was not. The definition
proposed above clearly includes the objective dimension, and the subjective dimen-
sion can be accommodated by designating ‘peace of mind’ or the ‘absence of fear’ as
values that can be specified. Whether one wants to do this, of course, depends on the
research task at hand.
It should be noted that specification of this dimension of security should not be in
terms of ‘vital interests’ or ‘core values’. For reasons to be discussed in the next
section, this prejudges the value of security as a policy objective, and thus prejudices
comparison of security with other policy objectives.
Although the two specifications above suffice to define the concept of security,
they provide little guidance for its pursuit. In order to make alternative security
policies comparable with each other and with policies for pursuing other goals, the
following specifications are also needed.

How much security?

Security, according to Wolfers, is a value ‘of which a nation can have more or less
and which it can aspire to have in greater or lesser measure’.44 Writing during the
same period as Wolfers, Bernard Brodie observed that not everyone views security as
a matter of degree. He cited as an example a statement by General Jacob L. Devers:
National security is a condition which cannot be qualified. We shall either be secure, or we
shall be insecure. We cannot have partial security. If we are only half secure, we are not secure
at all.45
Although Brodie, Wolfers, and others have criticized such views, the idea of security
as a matter of degree cannot be taken for granted.
Knorr has noted that treating national security threats as ‘matters of more or less
causes a lot of conceptual uneasiness’.46 And Buzan refers to similar difficulties:
The word itself implies an absolute condition—something is either secure or insecure—and
does not lend itself to the idea of a graded spectrum like that which fills the space between
hot and cold.47

42
Harold Brown, Thinking About National Security: Defense and Foreign Policy in a Dangerous World
(Boulder, CO, 1983), p. 4.
43
Wolfers, ‘National Security’, p. 485.
44
Ibid., p. 484.
45
Bernard Brodie, National Security Policy and Economic Stability, Yale Institute for International
Studies Memorandum No. 33 (New Haven, CT, 1950), p. 5.
46
‘Economic Interdependence and National Security’, in Klaus Knorr and Frank N. Trager (eds.),
Economic Issues and National Security (Lawrence, KS, 1977), p. 18.
47
Buzan, People, States, p. 18.
Concept of security 15

If this were true, it would be necessary to depart from common usage in defining
security as an analytical concept. This, however, does not appear to be the case. It is
quite common in ordinary language to speak of varying degrees of security.
One reason it is important to specify the degree of security a country has or seeks
is that absolute security is unattainable. Buzan recognizes this, but treats it as a
‘logical problem’ arising from ‘the essentially contested nature of security as a
concept’.48 If security is conceived of as a matter of degree, Buzan observes, ‘then
complicated and objectively unanswerable questions arise about how much security
is enough’.49 This, of course, is precisely why security should be so conceived. It is
not clear why such questions should be described as ‘objectively unanswerable’. They
are precisely the kind of questions that economists have been addressing for a long
time, i.e., how to allocate scarce resources among competing ends.50 Nor is there
anything peculiar about the unattainability of absolute security. As Herbert Simon
notes, the ‘attainment of objectives is always a matter of degree’.51
In a world in which scarce resources must be allocated among competing objec-
tives, none of which is completely attainable, one cannot escape from the question
‘How much is enough?’ and one should not try.

From what threats?

Those who use the term security usually have in mind particular kinds of threats.
Home security systems, for example, are usually directed at potential burglars; and
national security systems are often directed at other states. Since threats to acquired
values can arise from many sources, it is helpful if this dimension is clearly specified.
Vague references to the ‘Communist threat’ to national security during the Cold War
often failed to specify whether they referred to ideological threats, economic threats,
military threats, or some combination thereof, thus impeding rational debate of the
nature and magnitude of the threat. The concept of threat referred to in this
specification differs from that used by many students of international politics and
national strategy. Such scholars often use the term threat to refer to actions that
convey a conditional commitment to punish unless one’s demands are met.52 In
ordinary language, however, one often finds references to epidemics, floods, earth-
quakes, or droughts as ‘threats’ to acquired values. Ullman and others have argued

48
Ibid., p. 330.
49
Ibid.
50
Cf. Thomas C. Schelling, International Economics (Boston, MA, 1958), pp. 518–19; Alain C.
Enthoven and K. Wayne Smith, How Much Is Enough? (New York, 1971); Charles J. Hitch and
Roland N. McKean, The Economics of Defense in the Nuclear Age (Cambridge, MA, 1960); James R.
Schlesinger, The Political Economy of National Security (New York, 1960); and Thomas C. Schelling
and Malcolm Palmatier, ‘Economic Reasoning in National Defense’, in Alan A. Brown, Egon
Neuberger, and Malcolm Palmatier (eds.), Perspectives in Economics: Economists Look at their Fields
of Study (New York, 1971), pp. 143–59.
51
Herbert A. Simon, Administrative Behaviour, 3rd edn (New York, 1976), p. 177. On this point, see
also David A. Baldwin, Economic Statecraft (Princeton, 1985), p. 131.
52
On the concept of threats, see Baldwin, Paradoxes, pp. 45–81.
16 David A. Baldwin

that the concept of security should be expanded to include such phenomena.53 There
seems to be no reason not to use this more expansive concept of threats, especially
since it comports with common usage. Those who wish to refer to conditional
commitments to punish by social actors as security threats may make that clear
when specifying this dimension of security.

By what means?

Like wealth, the goal of security can be pursued by a wide variety of means. Wolfers
devotes considerable attention to making it clear that many different policies may
plausibly be adopted in the pursuit of security.
Specification of this dimension of security is especially important in discussions of
international politics. Since the publication of Wolfers’ article, ‘security studies’ has
emerged as a recognized subfield in international relations. The tendency of some
security studies scholars to define the subfield entirely in terms of ‘the threat, use,
and control of military force’54 can lead to confusion as to the means by which
security may be pursued. It can also prejudice discussion in favour of military
solutions to security problems.

At what cost?

The pursuit of security always involves costs, i.e., the sacrifice of other goals that
could have been pursued with the resources devoted to security. Specification of this
dimension of security policy is important because writers sometimes imply that costs
do not matter. One writer, for example, defines national security in terms of the
protection of core values, which he describes as ‘interests that are pursued not-
withstanding the costs incurred’.55 From the standpoint of a rational policy-maker,
however, there are no such interests. Costs always matter. Another writer asserts:
There is, in fact, no necessary conflict between the goal of maintaining a large and powerful
military establishment and other goals such as developing independence from Persian Gulf
oil, promoting self-sustaining development in poor countries . . . and promoting greater
public tranquility and a more healthful environment at home. All these objectives could be
achieved if the American people choose to allocate the resources to do so.56
Only the assumption of a cost-free world would eliminate the necessary conflict
among such goals as they compete for scarce resources. In thinking about security,
as in thinking about other policy goals, it is helpful to remember the TANSTAAFL
principle, i.e., ‘There ain’t no such thing as a free lunch’.57

53
Ullman, ‘Redefining Security’. See also, Allison and Treverton (eds.), Rethinking America’s Security.
54
Walt, ‘Renaissance’, p. 212. See also, Knorr, ‘National Security Studies’, p. 6; and Schultz et al. (eds.),
Security Studies, p. 2.
55
Melvyn P. Leffler, ‘National Security’, Journal of American History, 77 (1990), p. 145.
56
Ullman, ‘Redefining Security’, p. 132. Emphasis in original.
57
Edwin G. Dolan, TANSTAAFL (New York, 1971), p. 14.
Concept of security 17

Another way to imply that costs do not matter is by silence. During the last ten
years, neither the Realists/neorealists nor their critics have paid much attention to
costs. Although critics frequently state or imply that ‘too much’ is being spent on
armaments, this is usually treated as self-evident rather than requiring evidence and
argument.
Wolfers suggests an additional reason for specifying this dimension of security.
Arguing against those who would place national security policy beyond moral judg-
ment, he contends that the sacrifice of other values for the sake of security inevitably
makes such policies ‘a subject for moral judgment’.58 Given the crimes that have
been committed in the name of ‘national security’, this is a helpful reminder.

In what time period?

The most rational policies for security in the long run may differ greatly from those
for security in the short run. In the short run, a high fence, a fierce dog, and a big
gun may be useful ways to protect oneself from the neighbours. But in the long run,
it may be preferable to befriend them.59 Short-run security policies may also be in
conflict with long-run security policies.60

Summary

In response to Wolfers’ contention that specifications are needed in order to make

national security useful for ‘sound political counsel or scientific usage’,61 one could
specify security with respect to the actor whose values are to be secured, the values
concerned, the degree of security, the kinds of threats, the means for coping with
such threats, the costs of doing so, and the relevant time period.
The question remains, however: ‘How much specification is enough?’ Must all of
these dimensions be specified in detail every time one uses the concept of security?
Obviously not. Both the number of dimensions in need of specification and the
degree of specificity required will vary with the research task at hand. Each of the
dimensions can be specified in very broad or very narrow terms. Not all of the
dimensions need to specified all the time. For most purposes, however, meaningful
scientific communication would seem to require at least some indication of how
much security is being sought for which values of which actors with respect to which
threats. For purposes of systematic comparison of policy alternatives, the last three
specifications, i.e., means, costs, and time period, must be specified.
Although the dimensions of security can be specified very broadly, the utility of
the concept does not necessarily increase when this is done. For example, if security

58
Wolfers, ‘National Security’, pp. 498–9.
59
Cf. Kenneth E. Boulding, ‘Towards a Pure Theory of Threat Systems’, American Economic Review,
53 (1963), pp. 424–34.
60
See Robert A. Dahl and Charles E. Lindblom, Politics, Economics and Welfare (New York, 1953),
pp. 50–1.
61
Wolfers, ‘National Security’, p. 483.
18 David A. Baldwin

is specified in terms of threats to all acquired values of a state, it becomes almost

synonymous with national welfare or national interest and is virtually useless for
distinguishing among policy objectives.62

5. The value of security

Security is valued by individuals, families, states, and other actors. Security, however,
is not the only thing they value; and the pursuit of security necessitates the sacrifice
of other values. It is therefore necessary to ask how important is security relative to
other values. Three ways of answering this question will be discussed: (1) the prime
value approach, (2) the core value approach, (3) and the marginal value approach. It
will be argued that the marginal value approach is preferable to the other two.

The prime value approach

One way of determining the value of security is to ask what life would be like
without it. The most famous answer to this question is that by Thomas Hobbes to
the effect that life would be ‘solitary, poore, nasty, brutish, and short’.63 Such
reasoning has led many scholars to assert the ‘primacy’ of the goal of security.64 The
logic underlying this assertion is that security is a prerequisite for the enjoyment of
other values such as prosperity, freedom, or whatever.
The fallacy in this line of argument is exposed by asking the Hobbesian question
with respect to breathable air, potable water, salt, food, shelter or clothing. The
answer is roughly the same for each of these as it is for security; and a plausible case
for the ‘primacy’ of each can be made. This exercise, of course, merely underscores a
truth King Midas learned long ago, i.e., that the value of something—gold, security,
water, or whatever—is not an inherent quality of the good itself but rather a result
of external social conditions—supply and demand. The more gold one has, the less
value one is likely to place on an additional ounce; and the more security one has,
the less one is likely to value an increment of security.
To the extent that the prime value approach implies that security outranks other
values for all actors in all situations, it is both logically and empirically indefensible.
Logically, it is flawed because it provides no justification for limiting the allocation
of resources to security in a world where absolute security is unattainable. Em-
pirically it is flawed because it fails to comport with the way people actually behave.

62
‘Virtually’ rather than ‘totally’ useless because even the term ‘national interest’ distinguishes between
national interests and international or subnational interests. And even a very broad concept of
security distinguishes between protecting acquired values and attempts to acquire additional values.
63
The Leviathan (1651), Part I, Ch. XIII.
64
See Smoke, ‘National Security Affairs’, pp. 247–8; Mearsheimer, ‘Disorder’, pp. 221–2; Waltz, Theory,
p. 126; Joseph M. Grieco, Cooperation Among Nations: Europe, America, and Non-Tariff Barriers to
Trade (Ithaca, NY, 1990), p. 39; Robert G. Gilpin, ‘The Richness of the Tradition of Political
Realism’, in Robert O. Keohane (ed.), Neorealism and Its Critics (New York, 1986), p. 305; and
Lawrence Freedman, ‘The Concept of Security’, in Mary Hawkesworth and Maurice Kogan (eds.),
Encyclopedia of Government and Politics, vol. 2 (London, 1992), p. 730.
Concept of security 19

Prehistoric people may have lived in caves for security, but they did not remain there
all the time. Each time they ventured forth in pursuit of food, water or adventure,
they indicated a willingness to sacrifice the security of the cave for something they
presumably valued more. And in choosing places to live, settlers often forgo the
security of high mountain-tops in favour of less secure locations with more food or
water. Likewise, modern states do not allocate all of their resources to the pursuit of
security, even in wartime. Even the most beleaguered society allocates some of its
resources to providing food, clothing, and shelter for its population.
Even if ‘absolute’ security were a possibility, it is not obvious that people would
seek it. As Robert Dahl and Charles Lindblom observed long ago, ‘probably most
people do not really want ‘‘absolute’’ security, if such a state is imaginable; ‘‘opti-
mum’’ security would probably still leave an area of challenge, risk, doubt, danger,
hazard, and anxiety. Men are not lotus-eaters’.65

The core value approach

The core value approach allows for other values by asserting that security is one of
several important values. Although this approach mitigates the logical and empirical
difficulties associated with the prime value approach, it does not eliminate them.
One is still confronted with the need to justify the classification of some values as
core values and other values as non-core values. And if core values are always more
important than other values, this approach cannot justify allocating any resources
whatsoever to the pursuit of non-core values.

The marginal value approach

The marginal value approach is the only one that provides a solution to the resource
allocation problem. This approach is not based on any assertion about the value of
security to all actors in all situations. Instead, it is rooted in the assumption that the
law of diminishing marginal utility is as applicable to security as it is to other values.
Asserting the primacy of security is like asserting the primacy of water, food, or air.
A certain minimum amount of each is needed to sustain life, but this does not mean
that the value of a glass of water is the same for a person stranded in a desert and a
person drowning in a lake. As King Midas learned, the value of an increment of
something depends on how much of it one has.
According to the marginal value approach, security is only one of many policy
objectives competing for scarce resources and subject to the law of diminishing
returns. Thus, the value of an increment of national security to a country will vary
from one country to another and from one historical context to another, depending
65
Dahl and Lindblom, Politics, Economics, p. 50. Recent writers who have expressed similar doubts
about the value of security include: Barry Buzan, ‘Response to Kolodziej’, Arms Control, 13 (1992),
p. 484; James Der Derian, ‘The Value of Security: Hobbes, Marx, Nietzsche, and Baudrillard’, in
Ronnie Lipschutz (ed.), On Security (New York, 1995), pp. 24–45; and Ole Waever, ‘Securitization
and Desecuritization’, ibid., pp. 46–86.
20 David A. Baldwin

not only on how much security is needed but also on how much security the country
already has. Rational policy-makers will allocate resources to security only as long as
the marginal return is greater for security than for other uses of the resources.
There is nothing new about treating national security as one of many public
policy objectives competing for scarce resources and subject to diminishing returns.
Wolfers and his contemporaries used this approach, and defence economists have
long advocated it.66 Its neglect in recent writings on national security, however,
suggests the need to reiterate its importance.67
Critical theorists, feminist theorists, Realists, neorealists, liberals, Third World
theorists, and globalists all live in a world of scarce resources. In the end, all must
confront the question posed by Booth of ‘how many frigates to build’.68 Even paci-
fists, who answer ‘none’, must decide how to allocate resources among competing
non-military uses. The analytical tools of marginal utility analysis are available for
use by any or all of the schools mentioned above.
It is not always clear whether statements about the importance of security as a
goal are empirical observations or part of the definition of security. The ‘high
politics/low politics’ distinction, however, suggests that some scholars may be
making the value of security a matter of definition. Buzan, for example, includes in
security only those concerns that ‘merit the urgency of the ‘‘security’’ label’, thus
suggesting that urgency is part of his definition of security. And when he refers to
‘attempts to elevate particular economic issues onto the national security agenda’, he
seems to imply the inherent superiority of that agenda. Likewise, the intensity of the
threat seems to be a defining characteristic of security for Buzan.69
Ullman’s proposed definition of national security threats also includes elements
that prejudge the importance of security. Thus, he does not include all threats that
‘degrade the quality of life for the inhabitants of a state’, but only those that do so
‘drastically’ and quickly. And he does not include all threats that ‘narrow the range
of policy choices available to the state’, but only those that do so ‘significantly’.70
Both Buzan and Ullman seem to rule out the possibility of a minor or trivial
national security threat by conceptual fiat.
Policy advocates, of course, often try to win acceptance for their proposals by
declaring them to be ‘security issues’. Navies wanting frigates, educators wanting
scholarships, environmentalists wanting pollution controls, and so on are likely to
portray their respective causes as matters of ‘national security’. In this context the
declaration that something is a security issue is a way of asserting its importance.
66
E.g. Wolfers, ‘National Security’; Frederick S. Dunn, ‘The Present Course of International Relations
Research’, World Politics, 2 (1949), p. 94; Bernard Brodie, ‘Strategy as a Science’, World Politics, 1
(1949), pp. 467–88; Schelling, International Economics; Charles J. Hitch, ‘National Security Policy as a
Field for Economics Research’, World Politics, 12 (1960), pp. 434–52; and Schlesinger, Political
Economy.
‘It is peculiar to the training of an economist that he is continually aware of the need to optimize
rather than just to maximize, of the need to weight explicitly the value of more progress toward one
objective at the expense of progress toward another. By training, he is suspicious of any analysis that
singles out one conspicuous variable, some ‘‘dominant’’ feature, on which all attention is to be
focused, and which is to be maximized by putting arbitrary limits on the other variables’. Schelling
and Palmatier, ‘Economic Reasoning’, p. 148.
67
Buzan’s People, States contains only passing references to costs and no reference to diminishing
returns.
68
Booth, ‘Security and Emancipation’, p. 325.
69
Ibid., pp. 19, 131, 134. Emphasis added.
70
Ullman, ‘Redefining Security’, p. 133.
Concept of security 21

Thus one may argue that building urgency into the concept of security is a common
practice.71 If this practice is followed, however, the concept becomes useless for
rational policy analysis because the value of security relative to other goals will have
been conceptually prejudged.

6. Security and neorealism

The specifications of security presented here are also relevant to theorizing about
national security. No theory of international politics emphasizes security more than
neorealism, which posits it as the primary motivation of states. Given the
importance of security in neorealist analyses, they have devoted remarkably little
attention to explaining what security means. In an often quoted passage, Kenneth
Waltz observes:
In anarchy, security is the highest end. Only if survival is assured can states seek such other
goals as tranquility, profit, and power.72
This passage represents a simplification for the purpose of building a theory. Such
simplifications are permissible up to the point that they, to paraphrase Wolfers, leave
room for more confusion than scientific usage can afford. When dealing with a
particularly slippery concept like national security, a lack of specifications can be
especially worrisome. The specifications outlined earlier serve as a useful checklist
for deciding whether Waltz’s simplification goes too far.
The equation of security with survival provides little or no guidance with respect
to how to answer the question: ‘Survival of which values?’ To say that states strive to
ensure their own survival does not tell one very much. This is especially true for
Waltz, who defined states in terms of the functions they perform, including the
making and enforcement of laws, defence against external attack, and the provision
of food, clothing, housing, transportation, and other amenities consumed by the
citizens.73 If all of these functions are included as part of the acquired values that
define security, the concept becomes so broad that it loses its utility for distin-
guishing among policy goals. It might rule out subnational or international security
interests as well as acquisitive and self-destructive ones, but it rules out little else.
With respect to the question of the degree of security to which states aspire,
Waltz’s answer is: enough to assure survival. But this answer begs the question of
how much assurance is enough. Completely assured survival is a goal that can be
approximated but never attained. Regardless of what policies states adopt, there is
always some chance of survival and thus some assurance of security. The crucial
question is not whether security is ‘assured’, but rather, ‘How much assurance is
enough?’.74
The cost of security also receives little attention in neorealist theory. The passage
quoted above does not mention the possibility of diminishing marginal returns to
security policy, but it allows for them by implying that there is some (unspecified)
71
For a strong defence of this approach, see Waever, ‘Securitization and Desecuritization’.
72
Waltz, Theory, p. 126.
73
Ibid., p. 96.
74
This assumes that neorealists are treating ‘assurance’ of security as a matter of degree rather than as
a dichotomous variable. This is a point on which neorealists are not always as clear as one might
wish.
22 David A. Baldwin

level of assured survival that would justify shifting resources to the pursuit of other
goals.75 Waltz’s comparison of the goal of profits for a firm with the goal of security
for a state, however, raises questions about the treatment of security costs.76 It makes
no sense to describe firms as forgoing an increment of profit because the marginal
costs outweigh the marginal benefits, since profits are defined in terms of net
revenues. Thus, economic theory portrays firms as always seeking more profits. By
contrast, it makes a great deal of sense to describe states as forgoing an increment of
security because the marginal costs outweigh the marginal benefits. And any
political theory that portrays states as always seeking more security would be
seriously misleading.
Whether neorealist theory provides enough specification of the concepts of
security and security policy is ultimately a matter of judgment and cannot be
reduced to a simple formula. In making this judgment, however, it would be prudent
to consider its treatment of values to be protected, the degree of security to be
sought, and the costs to be incurred.
There is one additional aspect of the neorealist treatment of security that should
also be considered. Do neorealists view security as a zero-sum concept in the sense
that more security for one actor (unit) means less for another? When states are
described as ‘competing’ with one another for security, such a conception seems to
be implied.77 This suggests that the ‘winner’ of such a competition could be a state
surrounded by insecure states. The question of whether insecure neighbours are
good neighbours, however, should be carefully considered.
There are, of course, situations in which one state’s efforts to increase its security
reduce the security of other states—the well-known ‘security dilemma’—and any
concept of security that did not allow one to describe such situations would be
seriously defective. Not everything states do to enhance their security, however, takes
this form. Thus, a concept of security that required all security relations to be
described in zero-sum terms would be equally defective. The concept of security
explicated in previous sections of this article allows for the security dilemma, but it
does not make it a conceptual necessity.78

7. New security concepts?

The last decade has witnessed an outpouring of attempts to rethink the security
problematique. Whatever the merits of this literature as an aid to coping with the
75
Although Waltz models his theory after microeconomic theory, his treatment of security makes little
use of marginal analysis, which is one of the central ideas of microeconomic theory.
76
Waltz, Theory, pp. 90–2.
77
Cf. Mearsheimer, ‘Disorder’; Waltz, Theory, and ‘Emerging Structure’. One might object to the
contention that competition implies a zero-sum relationship by reference to mixed-motive games. This
objection, however, conflates competition and conflict. The concept of competition implies a special
type of conflict in which the parties play the same game in pursuit of the same goal. Competition
implies winners and losers, but in mixed-motive games each player wins or loses in terms of his own
value system. Thomas C. Schelling, The Strategy of Conflict (Cambridge, MA, 1960), p. 4. In such
games, everyone can gain (‘win’), and everyone can lose. This cannot happen in competitions.
78
Although it is sometimes suggested that the concept of national, as opposed to international, security
blinds one to the security dilemma, this was clearly not the case with Wolfers. He not only discusses
it, but also identifies an early version in Jeremy Bentham’s Principles of International Law, Essay IV.
‘National Security’, pp. 494–5.
Concept of security 23

post-Cold War world, it has added little to our understanding of the concept of
security. Emma Rothschild argues that many of the ‘new ideas’ about security have
eighteenth-century antecedents.79 It has been argued here that the basic conceptual
tools for rethinking security have been available at least since the publication of
Wolfers’ article in 1952.
The multidimensionality of security is not a new discovery. Wolfers pointed out
the need for specification with respect to which values to protect, from which threats,
by what means, and at what cost. The dimensions of security have not changed with
the end of the Cold War, but the substantive specifications of these dimensions that
were appropriate during the Cold War are likely to differ from those appropriate for
the 1990s. Economic security, environmental security, identity security, social
security, and military security are different forms of security, not fundamentally
different concepts. Each can be specified in terms of the dimensions discussed above.
Changing world circumstances and new issues do not necessarily require new
concepts. Voting power, military power, economic power, and persuasive power are
different forms of the same social phenomenon, i.e., power. The adjectives indicate
the differences, while the noun draws attention to the similarities. Both are
important.
Conceptualizing security at levels other than the nation-state is also not new.
Although Wolfers focused on national security, he acknowledged that security could
be discussed on higher and/or lower levels as well. And a book published the year
after Wolfers’ article still provides one of the most penetrating and useful accounts
of security in many forms at many levels.80
Although the approach to security presented here might seem to be incompatible
with the literature on identity politics and security,81 this incompatibility should not
be exaggerated. Individuals and nation-states are sometimes insecure about their
identities, and they sometimes adopt policies to cope with this insecurity. Indivi-
duals, for example, may consult a psychiatrist; and nation-states may revise their
immigration laws. Either situation could be described by the analytical scheme
offered above.82
In sum, to the extent that the new thinking about security focuses on conceptual
issues rather than empirical or normative issues, not much is new. Most of the ‘new
ideas’ about security can be accommodated by the conceptual framework elucidated
by Wolfers in 1952. The United Nations Secretary-General recently called for a
‘conceptual breakthrough’ which goes ‘beyond armed territorial security’ to include
‘the security of people in their homes, jobs and communities’.83 It may well be that
the world needs a theoretical breakthrough that provides a better understanding of
the post-Cold War world, a normative breakthrough that expands the notion of a
moral community, an empirical breakthrough that facilitates recognition of in-
creased interdependence, and a political breakthrough that strengthens the will to
pursue an expanded security agenda. But none of these requires a conceptual
breakthrough that goes beyond the specifications identified by Wolfers.

79
Rothschild, ‘What is Security?’
80
Dahl and Lindblom, Politics, Economics, esp. pp. 49–54.
81
See, for example, David Campbell, Writing Security: United States Foreign Policy and the Politics of
Identity (Minneapolis, MN, 1992). See also Digeser’s cogent critique of Campbell (‘Concept of Security’).
82
For an earlier discussion of identity politics, see Dahl and Lindblom, Politics, Economics.
83
Quoted in Rothschild, ‘What is Security?’, p. 56.
24 David A. Baldwin

8. Conclusion

Despite widespread use of ‘security’ by scholars and politicians during the last forty
years, not much attention has been devoted to explicating the concept. Although the
concept of power has generated a veritable mountain of explicative literature,84 the
comparable literature on security is more of a molehill. Although some scholars
contend that this is due to the essential contestability of security, it is probably more
accurate to describe the concept of security as insufficiently explicated than as
essentially contested. This essay has attempted to explicate the concept of security
broadly enough for use at any level, but with special reference to the nation-state.
The purpose is to define security as a policy objective distinguishable from others.
Since security competes with other goals for scarce resources, it must be
distinguishable from, yet comparable with, such goals. This requires that the relative
importance of security be left open rather than built into the concept in terms of
‘vital interests’ or ‘core values’. In 1952 Wolfers argued that specifications were
needed in order to make the concept of national security useful for ‘sound political
counsel or scientific usage’. It is especially important to reiterate and clarify such
specifications in the aftermath of the Cold War. Since much of the current public
policy debate focuses on whether and how to reallocate resources from security to
other policy objectives, it is more important than ever to have a concept of security
that facilitates comparisons of the value of security with that of other goals.85
Oppenheim’s criteria for evaluating scientific concepts may be applied to the
concept of security explicated above.

(1) Operationalization

The multiple dimensions of security discussed above will not be easy to

operationalize. They are not reducible to a simple formula, but each is
operationalizable in ‘principle’.
When I say ‘in principle’ I mean only that no data are demanded by the definition that we
cannot imagine securing with combinations of known techniques of observation and
measurement. The observations may be exceedingly difficult but they are not inherently
impossible: they don’t defy the laws of nature as we understand them.86

(2) Definitional connections

Unlike ‘power’, the concept of security easily connects with a verb. A variety of
values can be secured by a variety of means. Also, the use of adjectives permits
reference to many different kinds of security, e.g., economic security, environmental
security, military security, social security, physical security, identity security,
84
For references, see Baldwin, Paradoxes.
85
Both Tickner (‘Re-visioning Security’) and Booth (‘Security and Emancipation’) have described
pressures to revise the concept of security as stemming partially from concerns about excessive
defence spending.
86
Robert A. Dahl, ‘The Concept of Power’, Behavioural Science, 2 (1957), p. 214.
Concept of security 25

emotional security, and so on. This family of terms provides the security analyst
with a useful vocabulary without undermining the basic intuitive notion of security.

(3) Factual connections

The specifications recommended above direct attention to a number of theoretically

important and policy-relevant aspects of the subject matter that might easily be
overlooked. These include the facts that the values to be secured are variable, the
degree of security sought is variable, the potential threats to security are multiple,
the means by which security may be pursued are many, the costs of security are
inescapable, and the time period matters.

(4) Not precluding empirical investigation

The specifications discussed here do not preclude empirical investigation by making

true ‘by definition’ what had better be left open to empirical inquiry. For example,
the importance of security as a policy objective is not built into the concept by
including ‘vital interests’ or ‘core values’ in the definition. Also, the means by which
security may be pursued are not confined to the ‘threat, use, and control of military
force’,87 as some definitions of security studies seem to imply. Both the importance
of security as a policy objective and the means most appropriate for its pursuit are
matters best left open to empirical inquiry. It should also be noted that the question
of whether domestic threats to national security are more important than foreign
threats is left open.

(5) Ordinary language

None of the specifications suggested above deviates unnecessarily from ordinary

usage. It might be argued that common usage tends to equate national security
issues with important issues. As one study put it, ‘everyone agrees that ‘‘security
issues’’ are important and deserving of national prominence and financial sup-
port’.88 To the extent that this is true, the approach suggested here represents a
necessary departure from ordinary language. If national security issues are defined as
important, attempts to compare them with other issues will be prejudiced from the
start.
There is no shortage of labels to substitute for ‘security’ in referring to issues of
extraordinary importance, e.g., urgent issues, important issues, vital issues, core
issues, high priority issues, etc.
It is sometimes stated or implied that the conceptual problems of ‘national
security’ can be eliminated or greatly mitigated simply by substituting ‘international’

87
Walt, ‘Renaissance’, p. 212.
88
Schultz et al. (eds.), Security Studies, p. 1.
26 David A. Baldwin

or global’ for ‘national’ security.89 Although such adjectival shifts are meaningful
and useful for some purposes, they are not substitutes for the specifications
suggested by Wolfers. They pertain primarily to the first specification, ‘Security for
whom?’. The ambiguities arising from failure to specify the other dimensions are as
applicable to international or global security as they are to national security.
Although it is often asserted that international security, unlike national security,
denotes the interdependence of nation-states with respect to their security relations,
the logic of such an assertion is unclear. No matter which adjective is used, the con-
cept of security explicated here implies nothing whatsoever about the degree of
interdependence among states with respect to their security relations. This matter is
better left to empirical investigation and should not be built into the concept of
security. Those who believe that states are mutually dependent in their security
relations should make the case with evidence and argument rather than by
definition. Nor can it be argued that the concept of national security blinds one to
the security dilemma or to the more general interdependence of states in their
security relations. Wolfers explicitly recognized both matters without the help of the
concept of international security.90
National security has figured prominently in academic and political discussions of
foreign policy and international politics since the end of World War II. Usually, the
specifications suggested by Wolfers have been ignored. No social science concept has
been more abused and misused than national security. If the concept is to be
salvaged for use in policy analysis or theory construction, specifications of the sort
advocated here seem to be necessary. To argue that they are necessary, however, is
not to say that they would be sufficient. Careless use and abuse of the concept may
have already rendered it useless for everyone but the politicians.

89
See, for example, Joseph S. Nye, Jr, ‘The Contribution of Strategic Studies: Future Challenges’,
Adelphi Paper No. 235 (1989), p. 23; Nye and Lynn-Jones, ‘International Security Studies’, p. 7;
Haftendorn, ‘Security Puzzle’, pp. 5–11; and Richard Schultz, ‘Introduction to International
Security’, in Schultz et al. (eds.), Security Studies for the 1990s, pp. 45–6.
90
Wolfers, ‘National Security’, pp. 494–6.