State-induced Security Dilemma

Maintaining the Tragedy

ALAN COLLINS

ABSTRACT
The security dilemma, first coined by John Herz in 1950, has come to
explain why states that do not seek to harm one another can still end
up in competition and war. Its very essence is one of tragedy. The
security dilemma has been expounded since by a number of authors,
and in this article I examine one such development; Jack Snyder’s defi-
nition of the security dilemma, being applicable where one state
requires the insecurity of another. I label this a state-induced security
dilemma. I argue that in order for this to be an accurate expansion of
the security dilemma, the essence of tragedy must be retained. I suggest
that with some refinement Snyder’s definition does extend the applica-
tion of the security dilemma and raises important questions regarding
the dilemma’s mitigation and escape. I have used the relationship
between China and Taiwan to elucidate the utility of the state-induced
security dilemma.
Keywords: benign; China; deterrence; hegemony; Herbert Butterfield;
Jack Snyder; John Herz; revisionist; Robert Jervis; security dilemma;
spiral; status quo; Taiwan

[I]t is one of the tragic implications of the security dilemma that mutual fear
of what initially may never have existed may subsequently bring about
exactly that which is feared most. (Herz, 1966: 241)

The term ‘security dilemma’ is a relatively new one, coined first by the
American political scientist John Herz in ‘Idealist Internationalism and the
Security Dilemma’ published in 1950. Herz is not, however, the only ‘grand-
father’ of the subject since writing on the other side of the Atlantic Ocean
was the British historian Herbert Butterfield. Butterfield, in Christianity
and History (1949) and History and Human Relations (1951), was to refer
to the security dilemma as Hobbesian Fear. Both authors highlighted the
essential essence of the security dilemma; it’s a tragedy. Since the 1950s, the
security dilemma has been expounded by a number of authors, most
notably Robert Jervis and Charles Glaser. It has become a key feature in
the offence–defence debate within Realism, and although it emerges from
the Realist camp it is not wedded to Realism. Alexander Wendt has used it

Cooperation and Conflict: Journal of the Nordic International Studies Association

Vol. 39(1): 27–44. Copyright ©2004 NISA www.ps.au.dk/NISA
Sage Publications www.sagepublications.com
0010-8367. DOI: 10.1177/0010836704040833
28 COOPERATION AND CONFLICT 39(1)

in his constructivist writings and it has also appeared in the critical security
studies literature.1 Its most recent utilization has been in such diverse top-
ics as ethnic conflict in Europe (Posen, 1993; Kaufmann, 1996; Kaufman,
1996) and the international politics of East Asia (Christensen, 1999; Collins,
2000).
This expounding of the security dilemma has led to an array of prefixes.
We have structural security dilemmas, perceptual security dilemmas, imperi-
alist security dilemmas (Snyder, 1985), deliberate security dilemmas, in-
advertent security dilemmas (Wheeler and Booth, 1992), tight security
dilemmas, regular security dilemmas, loose security dilemmas (Roe, 2001),
deep security dilemmas (Jervis, 2001) and even knotty security dilemmas
(Christensen, 2002). To these we can add system-induced and state-induced
security dilemmas. Such a variety of different security dilemmas, while indi-
cating the health of the concept, also raises a concern. Does such wide
application, in so many different guises, mean that it has become ambigu-
ous and lost any meaning? If so, this would be a tragedy. In this article I
reinstate Butterfield’s and Herz’s tragedy and examine the security
dilemma’s applicability in scenarios where actors require the insecurity of
others for their own security.
The article is specifically intended to expound upon Jack Snyder’s imperi-
alist security dilemma. Introduced to the literature in 1985 the imperialist
security dilemma contains an actor that requires the insecurity of others
and, as such, represents an extension of Butterfield’s and Herz’s security
dilemma. In his 2001 synthesis of the security dilemma, Paul Roe refers to
this as a regular security dilemma, thus implying that it has become the stan-
dard utilization of the concept. I argue in this article that the imperialist/
regular security dilemma requires refinement in order to maintain the
essence of tragedy. This type of security dilemma I have labelled state-
induced in order to distinguish it from Butterfield’s and Herz’s traditional
security dilemma which, because it is driven by the anarchic nature of the
international system, can be considered system-induced.

Tragedy: the Essence of the Security Dilemma

The security dilemma describes a situation in which war can occur between
two or more participants where none of those involved desires such an out-
come. Thus, war can occur without the participation of a revisionist or rev-
olutionary state seeking to alter the status quo. Hence Butterfield’s claim
that:

[t]he greatest war in history could be produced without the intervention

of any great criminals who might be out to do deliberate harm in the world.
It could be produced between two Powers both of which were desperately
anxious to avoid a conflict of any sort. (Butterfield, 1951: 19–20)

The security dilemma is thus concerned with explaining a deterioration in

relations between status quo powers.
 COLLINS : STATE - INDUCED SECURITY DILEMMA 29

In explaining how such a tragedy could occur, Butterfield highlights a key

element to understanding the security dilemma; uncertainty. He notes the
uncertainty that statesmen have determining others’ intentions and focuses
upon the fear and suspicion such uncertainty causes. In his now famous
passage Butterfield writes:

It is the peculiar characteristic of the situation that I am describing … that you

yourself may vividly feel the terrible fear that you have of the other party, but
you cannot enter into the other man’s counter-fear, or even understand why
he should be particularly nervous. For you know that you yourself mean him
no harm, and that you want nothing from him save guarantees for your own
safety; and it is never possible for you to realise or remember properly that
since he cannot see the inside of your mind, he can never have the same assur-
ance of your intentions that you have. (1951: 21)

In essence, therefore, the uncertainty that statesmen have of each other’s

intentions leads to fear and suspicion that others may intend harm, and thus
a war can occur despite neither desiring such an outcome. The very char-
acter of the security dilemma is one of tragedy, since the incompatibility the
participants perceived to exist was illusory; the participants’ intent was
benign.
This critical element in the working of the security dilemma (benign
intent/illusory incompatibility) has been acknowledged by a number of
contemporary writers. Charles Glaser, in his 1997 review of the security
dilemma literature, writes, ‘the security dilemma is the key to understand-
ing how in an anarchic international system states with fundamentally com-
patible goals still end up in competition and war’ (p. 171). Randall
Schweller is quite emphatic on this point. He writes:

[When] the goal of one or more states is something other than mutual secur-
ity, conflict is not only apparent but real; and because it is real, the resulting
insecurity cannot be attributed to the security dilemma. States acquire more
arms not because they misperceive the security efforts of other benign states
but because aggressive states truly wish to harm them. (1996: 104)

Robert Jervis captures this illusory incompatibility when writing that if the
Cold War was a security dilemma then it ‘is a “no fault” argument. No one
was to blame and everyone was to blame’ (2001: 38). Finally, Paul Roe,
emphasizing the role of Butterfield, refers to a Butterfieldian security
dilemma where ‘the question of intentionality is fundamental — do the
actors involved have benign or malign intentions (emphasis in original)’
(2000: 378). While Roe is not incorrect to emphasize Butterfield when
thinking of the security dilemma as a tragedy, it is in Herz’s 1960s piece,
International Politics in the Atomic Age, that this is manifest most clearly.
Referring to Nazi Germany, Herz claims that the Second World War was
‘provoked by Hitler’s policy of world domination. It can hardly be main-
tained that it was a German security dilemma which lay at the heart of that
conflict, but rather one man’s, or one regime’s, ambition to master the
30 COOPERATION AND CONFLICT 39(1)

world’ (1966: 234–5). For Herz, the malign intent inherent in Germany’s
political ambition refutes the existence of the security dilemma. In
Butterfield’s terms, Hitler was a great criminal who was deliberately seek-
ing to do harm.
In view of the importance of the actors’ benign intent, or at least a com-
patibility of their goals and ambitions that allows for mutual security, the
security dilemma appears to be of value only in examining relations
between status quo powers (security-seekers). The existence of a revision-
ist (power-seeker), such as Nazi Germany, would make it redundant.
However, this division of states into non-expansionist status quo powers
and aggressive revisionists requires clarification. Both Glaser and Jervis
note that security-seekers may see expansion as necessary for their security.
Glaser argues expansion could become attractive to status quo powers ‘if
acquiring additional territory would provide a buffer zone against invasion
or additional resources for defence and/or would deny those resources to
the defender’ (1992: 504). Jervis concurs by noting that when statesmen per-
ceive that offensive capabilities provide a greater degree of security than
their defensive counterparts:

[The] security dilemma is at its most vicious [because] … commitments, strat-

egy, or technology dictate that the only route to security lies through expan-
sion. Status quo powers must act like aggressors; the fact that they would
gladly agree to forego the opportunity for expansion in return for guarantees
for their security has no implications for their behaviour. (1978: 187)

In other words, status quo powers see their security requiring the insecurity
of others, and this includes expanding to take their territory. This is Paul
Roe’s regular security dilemma and it also matches Jack Snyder’s imperial-
ist security dilemma. We return to status quo powers requiring others to be
insecure, and what this means for the tragedy of the security dilemma, when
state-induced security dilemmas are noted below. Before this, and indeed
because it is hugely important in addressing this issue of status quo powers
that require the insecurity of others, it is necessary to examine Robert
Jervis’s deterrence and spiral models of international relations. These were
introduced to the literature in his highly regarded monograph, Perception
and Misperception in International Politics (1976).

Deterrence and Spiral Model

The difference between the deterrence and spiral model centres on the
adversaries’ intentions. In the deterrence model the adversaries’ intentions
are malign, and it is only by deterring the adversary that decision-makers
can protect their state. This model is therefore marked by the perception
that resolve on all matters, regardless of their intrinsic value, is necessary in
dealing with the adversary, since any sign of weakness will be exploited.
In contrast, the spiral model emphasizes the setting in which statesmen
conduct state relations. In this model the anarchic nature of international
 COLLINS : STATE - INDUCED SECURITY DILEMMA 31

relations compels statesmen to provide for their own security. In so doing,

as Jervis points out, a problem arises:

When states seek the ability to defend themselves, they get too much and too
little — too much because they gain the ability to carry out aggression; too lit-
tle because others, being menaced, will increase their own arms and so reduce
the first state’s security. (1976: 64)

The spiral model is clearly relevant to the security dilemma. Uncertain of

others’ intent, statesmen pursue the prudent option of procuring arms, yet
since others can witness this acquisition and feel menaced by it, they too
respond in order to safeguard their security. None of the actors intends
to harm the other, yet they have begun to view each other with suspicion
and fear. Their prudent/rational actions, tragically, have created a sense of
insecurity rather than security. Since it is the anarchic nature of the inter-
national system (self-help behaviour, uncertainty) that generates the secur-
ity dilemma, this traditional understanding of the concept can be referred
to as system-induced.
Unlike the deterrence model where competitive policies indicate resolve
and deter the aggressor, in the spiral model they compound the problem by
reinforcing the image of the other as an aggressor. Echoing Butterfield’s
famous line concerning the inability of statesmen to enter into the other’s
counter-fear, Jervis claims, ‘statesmen … rarely … consider seriously the
possibility that such a policy will increase the danger of war instead of less-
ening it’ (1982: 360). Therefore, although no harm is intended the states-
men’s uncertainty leads them to pursue prudent policies, but since these are
the cause of the rising tension and suspicion they are also self-defeating or
paradoxical policies. If statesmen only knew that the other was benign,
a means of mitigation and possible escape could be found. Although
not easy, Jervis does note that the security dilemma can be mitigated. He
writes, ‘empathy and skilful statesmanship can reduce this danger’ (1978:
212).
Why skilful? Recognizing that a security dilemma is in operation is par-
ticularly difficult because a similar action–reaction of events can also be
witnessed in the deterrence model. However, Jervis is right to note that
while this may resemble the operation of the spiral model, the malign intent
of one of the participants ensures that paradoxical policies are not being
pursued and thus it is not a security dilemma. He writes:

The aggressor, of course, is hostile because its expansion is blocked, but this
does not develop the unfounded fear that the status quo power is menacing
its existence. It may increase its arms because it sees that its foreign policy
aims have outrun its military strength, and the increase of arms and tensions
can continue for several cycles as each side matches the other’s belligerence.
But this process resembles that explained by the spiral model only super-
ficially. It is completely rational. Each side is willing to pay a high price to gain
its objectives and, having failed in its initial attempt to win a cheap victory, is
merely acting on its unchanged beliefs about the value of the issue at stake.
The heightening of the conflict does not represent, as it does in the spiral
32 COOPERATION AND CONFLICT 39(1)

theory, the creation of illusory incompatibility, but only the real incompatibil-
ity that was there from the beginning. Thus the spiral explanation of the pro-
cess is not correct, and an attempt to apply spiral prescriptions would not
have the intended effects. (1976: 80)

Spiral prescriptions, such as confidence-building measures, are means of

lessening statesmen’s uncertainty by aiding accurate threat assessment. The
implication is that with reduced uncertainty the security dilemma can be
mitigated or even escaped as statesmen become more confident about the
others’ benign intentions. Since such prescriptions are inappropriate in the
deterrence model, Jervis is concurring with Herz’s analysis of Nazi
Germany: the security dilemma is not at the heart of all conflicts. Indeed,
Jervis is explicit in noting that too much emphasis on the security dilemma
can lead to the false conclusion that ‘security, rather than expansion, is
the prime goal of most states’ (1976: 75). Therefore, even if a statesman was
sensitive to the security dilemma, the heightening of tension in a state’s
relations with others, manifest by an arms race, for example, does not mean
that the security dilemma is necessarily the cause. Indeed, to initiate spiral
prescriptions would be extremely dangerous, since such actions would
undermine the state’s security if its adversary harboured malign intent.
Thus, even statesmen who are sensitive to the security dilemma, a rarity in
itself, may find they cannot avoid the heightening of tensions and a deteri-
oration of relations. This helps to explain the following claim by Barry
Posen:

Often statesmen do not recognise that [the security dilemma] exists: they do
not empathise with their neighbours; they are unaware that their own actions
can seem threatening. Often it does not matter if they know of this problem.
The nature of their situation compels them to take the steps they do. (1993: 28)

The suggestion that statesmen are compelled to act may be too determinis-
tic, but the implication that even statesmen who are sensitive to the security
dilemma may be unable to avoid it is probably correct. After all, even if the
other is benign, there is no guarantee it will remain so.

State-Induced Security Dilemma

In 1985, Jack Snyder introduced an imperialist security dilemma to the liter-

ature. He wrote that, ‘[i]n a security dilemma the adversary’s malign inten-
tions’, and in so doing appeared to be challenging the orthodox view of the
security dilemma (p. 155). Indeed, he defines the security dilemma as ‘a
situation in which each state believed that its security required the insecur-
ity of others’ (emphasis added) (p. 153). The requirement that others be
insecure and the existence of malign intent seem at odds with Butterfield’s
and Herz’s conception of the security dilemma as a tragedy. Snyder intro-
duces to the literature the structural, perceptual and imperialist security
dilemmas; it is the last-mentioned which is problematic.
 COLLINS : STATE - INDUCED SECURITY DILEMMA 33

In his imperialist dilemma, Snyder introduces a revisionist state that from

the outset seeks a goal that would require the target state to forfeit a valu-
able asset (territory, sovereignty). Snyder asserts that the imperialist state
would become fearful of the status quo power and explains this using a sim-
ilar scenario to that noted earlier by Jervis. Snyder writes:

In order to achieve its expansive … goals, the aspiring imperialist state devel-
ops offensive military forces for the purpose of conquest or intimidation.
When resistance is met, a testing of will and capabilities ensues. An arms race
occurs as the imperialist and its opponent both try to prove that they have the
capability to achieve their … aims. Crises are staged as the states test each
other’s willingness to risk war rather than retreat. (1985: 165)

Snyder suggests that if either state believes that it might be taken advantage
of during a period of temporary weakness (a window of opportunity), it
may initiate a preventive or pre-emptive war. That is, the revisionist state
does believe the status quo power intends to do it harm and therefore it
arms not only to acquire its original goal, but also to protect itself.
In this scenario, since both are prepared to wage a preventive or pre-
emptive war, both harbour malign intent. According to Snyder the revision-
ist’s malign intent remains because it mitigates a compromise that is
mutually beneficial — ‘mutual security guarantees [are] difficult to devise
because they may rule out the kinds of capabilities that the imperialist
needs for its campaign of limited expansion’ — while it appears as though
the target state also sees aggression as the only means of achieving security
(Snyder, 1985: 166). The incompatibility between the states is real, since
they both intend the other harm. However, the existence of incompatibility
is not enough for the situation to be regarded as a security dilemma; the
incompatibility must be illusory. If one state, or both, actually intends to
harm the other and is being foiled by the other state’s policies, then those
policies are not paradoxical. Jervis’s claim noted earlier — that while this
type of arms race may resemble a spiral model, it is actually the workings
of the deterrence model — is right. The states in Snyder’s imperialist
dilemma may be uncertain of each other’s intent but their subsequent fear
and suspicion of each other is well founded.
This rejection of Snyder’s imperialist security dilemma also has implica-
tions for Jervis’s and Glaser’s notion that the security dilemma is relevant
to those situations where a status quo power, for defensive reasons,
embarks upon an expansionist policy. If decision-makers determine that the
best means of achieving security for their state lies in taking neighbouring
territory, whether to establish a buffer zone or because offence is perceived
to be in the ascendancy, then how different is this ‘status quo power’ from
a revisionist? Hitler defended his expansionist policy in terms of German
security requirements (recovery of lost territory, need for living space).
Whatever the motivations behind the expansionist policy, the target state is
faced with a real threat to its territory. This is not a spiral model; this is a
deterrence model. If Snyder’s imperialist dilemma does not appear to be an
accurate expansion of the security dilemma’s application, however, the
34 COOPERATION AND CONFLICT 39(1)

same cannot be said for his definition. Requiring the insecurity of others
enables the security dilemma to be applied to status quo powers that con-
ceive of their security as requiring the insecurity of their neighbours, but do
not seek to change the current status quo. This creates a more complex
version of the security dilemma (state-induced) and one that has particular
significance for the question of mitigating or escaping the dilemma.
In their synthesis of the literature, Nicholas Wheeler and Ken Booth
identify Snyder’s version of the security dilemma and distinguish it from the
traditional (system-induced) version by labelling Snyder’s deliberate and
the traditional inadvertent. Their deliberate version consists of two exam-
ples. The second involves a revisionist state and suffers the same problems
as Snyder’s imperialist version (Collins, 1997: ch. 2; Wiseman, 2002: 18). It
is the first example, however, that provides an exciting expansion for the
security dilemma’s applicability.
Here the security dilemma arises when a status quo power pursues a
deliberately aggressive policy vis-à-vis its neighbours in order to intimidate
them. The aim is not to overthrow the existing status quo, but rather to con-
solidate it by making others too frightened to challenge it. This aggressive
policy is not a precursor to war; instead it is designed to provide the state
with security by requiring others to feel insecure. The state is seeking a posi-
tion of hegemony and the security dilemma can arise in this situation
because the other states are unlikely to be able to distinguish this approach
from a revisionist state that harbours malign intent. Indeed, given how
aggressive the hegemon appears, they are much more likely to interpret its
actions as hostile. If they respond to the hegemon’s attempt to achieve
dominance by enhancing their own power, then the increase in absolute
power that the hegemon acquires will not translate into an increase in rel-
ative power. The other states’ actions are likely to be interpreted as aggres-
sive since the hegemon knows it does not intend them harm, but their
actions suggest that they are not satisfied with the current status quo. That
is, the hegemon and its neighbours are unable to enter into each other’s
counter-fear and are pursuing paradoxical policies that are leading each to
assume the worst. They have fallen victim to a state-induced security
dilemma which differs from the traditional, system-induced variant because
in the latter none of the states seek security at each other’s expense; they
just perceive it this way.
A good example of a state-induced security dilemma is China’s relation-
ship with Taiwan. For the policy-makers in Beijing, Taiwan is a province of
the People’s Republic of China (PRC), and thus if Taiwan was to declare
itself a sovereign state this would amount to an act of secession. Since its
creation in 1949 the PRC has sought reunification with Taiwan, as indeed it
has with other lost territories from the ‘Century of Shame’, such as Hong
Kong. The communist regime has tied its legitimacy to the recapture of
these territories and has made it abundantly clear that it will resort to force
if Taiwan declares itself an independent state. Since the end of the Cold War
the PRC has been engaged in a military build-up and has begun to acquire
the type of offensive forces necessary to invade Taiwan. These forces do not
represent an idle threat, as evidenced by Beijing’s sabre-rattling during the
 COLLINS : STATE - INDUCED SECURITY DILEMMA 35

TABLE 1
Security Dilemmas

System-induced State-induced Imperialist

Uncertainty arises from
anarchy. This is also the
spiral model
Neither actor intends
harm but is interested in
maintaining the status
quo
Mitigation and escape
possible because
incompatibility is illusory
Solution lies in pursuing
conciliatory policies
Uncertainty arises from Uncertainty arises from
hegemon’s requirement aggressor’s threatening
that others are insecure behaviour
Neither actor intends to Aggressor is a revisionist
change the status quo, but actor and intends to
status quo determined by change status quo
hegemon
Mitigation and escape Mitigation and escape not
possible by altering the possible through
hegemon’s perception of concessions or threats
what its security
requirement entails
Solution lies in combining This is not a security
threats and concessions dilemma and war is likely
to alter the hegemon’s
requirement that others
are insecure without
challenging its dominant
position

Taiwanese presidential elections of 1996 and 2000. It is not, though, the aim
of the PRC to use these military forces to attack Taiwan, occupy the island
and bring about a forced reunification. Reunification is the goal, but the
preference is for a soft landing brought about through the steady and incre-
mental development of economic linkages between the PRC and Taiwan
(Ross, 2002: 71). The acquisition of military capabilities by the PRC is there-
fore not intended to alter the status quo, but rather to preserve it by mak-
ing Taiwan too frightened to change it by declaring itself an independent
sovereign power.
The security dilemma arises because the United States has responded to
the PRC’s military build-up by strengthening its commitment to Taiwan.
The US was a supplier of military equipment to Taiwan before the end of
the Cold War, but in the last decade Washington’s commitment to the
defence of the island has become more tangible. During the 1996 Taiwanese
presidential election the US responded to Chinese military exercises by
deploying two aircraft carriers to the Taiwan Strait. The US also strength-
ened its defence ties with Japan in 1997, which included extending the range
of Japanese logistical support for US action. This was interpreted in Beijing
as extending the US–Japanese defence agreement to include Taiwan.
China’s military build-up is therefore not only designed to intimidate
Taiwan, but also to deter US and possibly US–Japanese involvement.
36 COOPERATION AND CONFLICT 39(1)

A security dilemma thus arises as both sides become suspicious of

the motivations behind each other’s increasing military might, yet neither
the US nor the PRC is seeking to change the status quo. It can be con-
sidered a state-induced security dilemma because the PRC’s sense of
security requires Taiwanese and US recognition of the PRC’s hegemonic
status. It requires the Taiwanese to accept a subordinate position and
the US to recognize the PRC’s claim on Taiwan. Hence the importance
Beijing attaches to Taiwan’s reaffirmation of the ‘One China’ policy
and America’s recognition of the three no’s. The three no’s are, first, no sup-
port for Taiwanese independence; second, no support for two Chinas, or
one-China, one-Taiwan; and third, no support for Taiwanese admission
to international organizations. We return to the Taiwan case below to deter-
mine if this security dilemma is operating at high pitch, or whether it is
mitigated.
For Snyder the key point concerning one state requiring others to be
insecure is the effect this has on the means of mitigating or escaping the
security dilemma. In the system-induced security dilemma the states
involved can mitigate the dilemma by reassuring each other about their
benign intent. In the state-induced security dilemma where one state
requires others to be insecure, it is unlikely that the other states, even if they
assumed the state had benign intent, would accept such vulnerability. It is
this problem that led Snyder to assert that, ‘[i]f threats could not resolve the
security dilemma, neither could concessions’, because, ‘[a]ppeasing the
security fears of one’s neighbour would have entailed accepting one’s own
insecurity’ (1985: 154–5). Thus, according to Snyder, the hegemon’s ‘inten-
tions can be made more benign by another state only by changing those
circumstances — or the [hegemon’s] assessment of them — and not by the
use of threats or concessions’ (1985: 155). This creates a more complex
approach to mitigation or escape than is required in the system-induced
security dilemma.

Mitigating and Escaping the Security Dilemma

Although Herz considers the security dilemma to be all-pervasive, operating

between individuals as well as states, he considers that in the international
system, because there is no sovereign, ‘the vicious circle of the power and
security dilemma works here with more drastic force than in any other field’
(1951: 200). It is the belief that this anarchy is a permanent feature of inter-
national relations that explains why, according to Jervis, ‘[t]he security
dilemma cannot be abolished, it can only be ameliorated’ (1982: 378). How-
ever, identifying anarchy and the self-help world it creates as the raison
d’être of the security dilemma has also led writers on the subject to suggest
means of lessening its effects (mitigation) and even nullifying it (escape).2
In the early 1980s Barry Buzan introduced the concept of a mature
anarchy. He suggested that it was possible to consider a spectrum along
which anarchy matures, or becomes more tamed, the more orderly it
becomes. He wrote that a mature anarchy:
 COLLINS : STATE - INDUCED SECURITY DILEMMA 37

[would] be a highly ordered and stable system in which states would enjoy a
great deal of security deriving both from their own inner strength and matur-
ity, and from the strength of the institutionalised norms regulating relations
among them. (1991: 177)

At the mature end of the spectrum, such norms would include the avoid-
ance of force in the settlement of disputes, thus equating a mature anarchy
with Karl Deutsch’s security community (Deutsch, 1957). This is important
for the security dilemma, since in a security community, the defining feature
of which is that force is not considered an option for the settlement of dis-
putes, the security dilemma can no longer operate. Also writing in the early
1980s, Jervis uses regime theory to claim that if states were able to abide by
norms constraining their behaviour — to become part of a security regime
— these norms would make statesmen less uncertain of each other’s inten-
tions (1982). Such a security regime would lie towards the mature end of
Buzan’s spectrum, though not at the extreme, and would indicate that
the security dilemma’s detrimental effects were mitigated. In other words,
mitigation and escape are possible by reducing uncertainty, which in turn
reveals to statesmen the others’ benign intent, thereby enabling their self-
defeating or paradoxical policies to be halted. Escape is therefore depen-
dent upon anarchy being subject to change.
In what has become a valuable addition to the literature, Alexander
Wendt has argued that states can construct the society in which they exist.
His argument is that ‘[s]elf-help and power politics are institutions, not
essential features of anarchy. ‘Anarchy is what states make of it (emphasis in
original)’ (1992: 395). He considers self-help to be a process within inter-
national relations as opposed to its structure. If statesmen, through a pro-
cess of learning from interactions with others, are capable of realizing that
the other is not a threat, then, in Butterfield’s terms, they will be able to
enter into each other’s counter-fear and reassure one another about their
intentions. Using the example of Mikhail Gorbachev’s ‘New Thinking’
policy, Wendt claims that the USSR was able to alter its own image, its
image of the West and, while he was in the process of writing, was altering
the West’s image of the USSR. What is particularly intriguing with Wendt’s
assertion is that the process can go beyond mitigating the statesmen’s fear
to actually escaping it. Thus, while he recognizes that statesmen can create
norms of behaviour via cooperation and as a result grow more confident in
predicting others’ intentions and actions, he suggests this process can go
further. He argues that in the process of cooperating, by not defecting on
these norms, statesmen are creating new and communal interests and com-
mitments. These collective interests will be resistant to change, not because
there is a penalty for the state that defects, but because all the actors gain
from its continued observance. What statesmen perceive as necessary for
their state’s security has changed. They have acquired mutual or common
interests.
The idea of statesmen acquiring new and shared interests becomes even
more significant when applied to the state-induced security dilemma. Since
the hegemonic state requires others to be insecure, this would, assuming the
38 COOPERATION AND CONFLICT 39(1)

others are not prepared to accept such insecurity, prevent them from solv-
ing the security dilemma through reassurance of their intentions. The hege-
mon’s goal is simply not compatible with those of the other states and thus
unilateral action is unlikely to be reciprocated. Hence Snyder’s claim that
‘[a]lthough the security dilemma [is] in some sense a spiral process, it [is]
not a spiral that could be unwound by the concessionary policies that spiral
theorists usually advocate’ (1985: 155). Hence mitigation/escape is only
possible by changing the circumstance or the hegemon’s understanding of
the circumstance. Wendt’s focus on states changing the anarchy in which
they conduct relations by acquiring common interests takes on added sig-
nificance because altering the hegemon’s assessment of what is required for
security becomes essential. Over a period of time, cooperation with the
hegemon could alter how the hegemon identifies its relationship to the
other states and therefore mitigate or even escape the security dilemma.
The key to mitigating and ultimately escaping the security dilemma lies
in reducing the uncertainty statesmen have about their neighbours’ inten-
tions. This can be achieved by a steady improvement in state relations —
the maturing of anarchy — via the establishment of recognized codes of
behaviour. Such codes constrain the action of the hegemon, thus reducing
the fears of the smaller powers, while also codifying the current status quo
and thereby reassuring the hegemon that its position is not going to be chal-
lenged. This does, though, entail the smaller powers accepting their
subordinate position. Returning to the Taiwan case, the complexity of miti-
gating a state-induced security dilemma can be easily discerned.
Escaping this security dilemma would entail Taiwan’s acceptance of the
‘One China’ policy and America’s recognition of the three no’s. The three
no’s were noted above; they essentially amount to the US not supporting or
encouraging Taiwan to declare independence. In other words, the accept-
ance of both of these Chinese phrases would amount to recognizing China’s
hegemony. Since Taiwan does acknowledge that there is one China and the
US administration does recognize the three no’s, it might seem that the
security dilemma is escaped. However, this is far from true, and instead
the security dilemma’s operation is at times mitigated but at other times it
operates at a high pitch. The reason why it has not been escaped is that both
Taiwan and the US interpret these Chinese phrases in different ways to the
PRC.
With regard to the ‘One China’ policy, throughout the Cold War the PRC
and Taiwan were in agreement that there was only one China; they just
happened to differ over who was the legitimate representative of China. In
the late 1980s, though, this began to change. Although both the PRC and
Taiwan continued to acknowledge the ‘One China’ principle, their inter-
pretation of its meaning differed. For the PRC it means that Taiwan is sub-
ordinate to Beijing but that it can have a considerable degree of autonomy
in what is referred to as the ‘one country, two systems’ formula currently in
operation in Hong Kong. For Taiwan the principle of ‘One China’ reflects
the shared history and culture the island has with the mainland, but it does
not prevent Taiwan from acting as an ‘equal’ and ‘independent’ state. This
difference of interpretation has created difficulties and a rise in the security
 COLLINS : STATE - INDUCED SECURITY DILEMMA 39

dilemma’s intensity. During the 1990s, under the leadership of Lee Teng-
hui, the public statements from Taiwan were increasingly assertive. Lee
spoke of ‘state-to-state’ relations across the Taiwan Strait and talks on
reunification stalled. The relationship between the PRC and Taiwan deteri-
orated because Taiwan appeared to be challenging the status quo. China
made acceptance of its definition of ‘One China’ a precondition for contin-
ued reunification talks, and in 2000 the situation appeared to take a turn
for the worse for Beijing when Chen Shui-bian, the leader of the pro-
independence party, won the Taiwanese presidential election.
The situation has also been aggravated by the current US administration
expanding arms sales to Taiwan, which includes supporting Taiwanese
submarine acquisition. The US had previously not supported submarine
acquisition. The talk is of the US being eager to help modernize Taiwan’s
military forces and even including Taiwan in theatre missile defence. Far
from keeping within the spirit of the three no’s, the fear in Beijing is that
these actions might embolden Chen Shui-bian to embark upon revisionist,
secessionist policies. The US acknowledgement of the three no’s does not,
therefore, preclude it from supplying Taiwan with advanced military equip-
ment. The security dilemma can thus operate at a high pitch, as both Taiwan
and the US appear to thwart China’s security requirement — that they
accept Chinese hegemony over Taiwan. Yet, the security dilemma does not
always operate at a high pitch, and indeed its operation is at times miti-
gated. Given the nature of this security dilemma — that it requires
Taiwanese subordination to the PRC — how can it be mitigated?
Although Snyder says that neither concessions nor threats can unwind
the spiral, the answer to mitigating a state-induced security dilemma may
be found in not implementing either of these separately, but combining
them.3 In the case of the US, this would mean maintaining both a sufficient
commitment to Taiwan that it deters Chinese adventurism but not so much
that it encourages the Taiwanese to declare independence. For the
Taiwanese, the threat aspect also entails deterring Chinese adventurism by
maintaining a sufficient military capability and close links with the US. The
concession aspect, though, is the intriguing one, since it ultimately entails
engaging the PRC without making substantive progress on reunification. It
is what Steven Goldstein refers to as ‘pseudo-engagement’. He writes, ‘the
purpose is to buy time and security on the assumptions that mutual goals
are basically irreconcilable, but that talk and some contact is better than
conflict’ (1999: 79). It is thus more about the ‘process of diplomacy than its
substance’, and ‘it merely seeks to evade or, at least, to ease the pressure on
Taiwan from the mainland’s demands without satisfying them’ (pp. 79–80).
Despite the arrival of a pro-independence president, Taiwan has not broken
with this approach. Indeed, the attitude of the Taiwanese electorate has
been to favour the status quo and not be either pro-independence or pro-
unification. Since pseudo-engagement evades resolving the problem,
Goldstein regards it as ultimately flawed. However, because pseudo-
engagement maintains the current status quo and does not provoke China,
it helps to mitigate the security dilemma. This means that at times when
Taiwanese leaders are more likely to comment upon the island’s status, such
40 COOPERATION AND CONFLICT 39(1)

as during elections, the tension will rise, but afterwards relations will settle
again. However, this type of engagement will not escape the security
dilemma. This would entail either Taiwan accepting the PRC’s definition of
the ‘One China’ policy or the Beijing regime de-coupling the issue of reuni-
fication from its own legitimacy. Neither of these outcomes appears likely
at present.
It could be argued that perhaps escape could be achieved through an
entirely different approach. Writing in the late 1970s, Jervis introduced the
offence/defence debate to the security dilemma in his seminal piece,
Cooperation under the Security Dilemma (1978). Since then a number of
authors have given this aspect of the security dilemma a prominent position
in the literature.4 While escape might be possible via cooperation, achieving
such cooperation amongst egoists is fraught with difficulties. This does not
make such an approach impossible, but it does make an approach that
could escape the security dilemma without the need for cooperation very
intriguing. It is precisely this prospect that Jervis raises when he writes, ‘[i]t
would … be possible for states to escape from the security dilemma with-
out developing the sorts of cooperative understandings that help amelio-
rate political conflicts across a broad range of issues’ (1982: 374).
For Jervis the key lies in the types of military forces statesmen procure
and the prevailing belief of how to wage war. He claims that ‘when offen-
sive and defensive postures are different, much of the uncertainty about the
other’s intentions that contribute to the security dilemma is removed’
(1978: 201). In essence, his argument is that where offensive force postures
exist, and offensive actions are perceived to hold the advantage in war, the
security dilemma will operate at its highest pitch. Conversely, where
defence is perceived to be the stronger form of warfare and forces are
deployed in defensive formations, the security dilemma is mitigated, poss-
ibly even escaped. In this latter scenario an increase in defensive weapon
acquisitions will not be seen as threatening, since they will provide only a
limited capability for conducting an invasion. Since status quo powers
will not require offensive weapons, the procurement of such weapons will
indicate an aggressive actor. Thus, rather than seeking cooperation with a
perceived foe, Jervis claims that where defence is stronger and distinguish-
able from offence,

[t]he differentiation between offensive and defensive systems permits a way

out of the security dilemma … There is no reason for a status quo power to
be tempted to procure offensive forces, and aggressors give notice of their
intentions by the posture they adopt. (1978: 214)

It is this prospect of mitigating or even escaping the security dilemma via

defensive force postures that makes non-offensive defence relevant to the
security dilemma. Hence Ken Booth’s assertion that strategies ‘of non-
provocative defence … seek to deal with the military problem which lies at
the heart of the security dilemma’ (1991: 20).
While the offence–defence equation is useful in the system-induced
security dilemma, this approach to lessening a security dilemma is not likely
 COLLINS : STATE - INDUCED SECURITY DILEMMA 41

to be effective in the state-induced version. In the latter, the hegemon will

need to procure and deploy forces that are capable of meeting a challenge
to the status quo, wherever that challenge might arise within the area of
hegemony. This will entail procuring and deploying power-projection
weapon systems, which are by their very nature offensive. Thus, even when
defence is perceived to be in the ascendancy, the hegemon will continue to
deploy offensive forces, since it will be unable to maintain the status quo by
adopting a non-offensive or non-provocative defence posture. The hege-
mon’s requirement that its own sense of security is dependent upon others
feeling insecure robs the offence–defence equation of its utility in mitigat-
ing/escaping this type of security dilemma.

Conclusion

The essence of the security dilemma is that it is a tragedy. Two or more

actors, both desperately seeking to avoid a conflict, make that outcome
more likely by the actions they take. The security dilemma has evolved
quite considerably since Herz introduced it back in 1950, and we now have
quite an array of prefixes circulating in the literature. It would, though, be a
tragedy in itself if the security dilemma were to lose its tragic essence and
become a phrase used to explain any deterioration in state relations. This is
particularly evident in those circumstances where one state perceives that
it requires the insecurity of others for its own security. This does represent
a development for the security dilemma and one to be welcomed. However,
if that requirement includes the taking of a valuable asset from the target
state, then the scenario becomes a deterrence model and the security
dilemma/spiral model no longer applies. For a security dilemma to arise in
a situation where a state’s security requirement is driving it, that is for it to
be state-induced, then to remain true to Herz’s and Butterfield’s tragedy
the maintenance of the status quo must be the goal. With this the goal of
both parties, the action-reaction dynamic that takes place is a spiral model.
The intriguing aspect with the state-induced security dilemma concerns
its mitigation and/or escape. Although the hegemon and the other states
wish to maintain the status quo, it is a status quo determined by the
hegemon. Thus, escape lies in the hegemon either giving up its position of
hegemony or altering its assessment of what its hegemony entails. This need
for a security requirement to alter makes mitigating a state-induced secur-
ity dilemma more complex than mitigating a system-induced security
dilemma. In the case of Chinese relations with Taiwan, mitigation occurs
through a combination of concessions and threats. The former through
‘pseudo-engagement’ and the latter through close defence ties to the US.
Mitigation is possible because, although ultimately the PRC wants Taiwan
to be an integral part of China, Beijing prefers the current status quo to the
distinct probability of conflict if it forced reunification (the threat aspect).
The concession part is Taiwanese sensitivity towards the PRC’s security
requirements by not declaring independence and indeed engaging with the
PRC. This has not, though, led to escaping the security dilemma, because
42 COOPERATION AND CONFLICT 39(1)

they have not reached a common interpretation of what ‘One China’

means, and to do this either the PRC or Taiwan will need to alter its posi-
tion. If the PRC does this, then it would have altered its interpretation of
what its hegemony entails.
In this article, China’s relationship with Taiwan has acted as the case
study to utilize the state-induced security dilemma. Its applicability, though,
is not limited to just this one case; elsewhere it has been used to explain
Chinese actions in the South China Sea (Collins, 2000). Indeed, since the
state-induced security dilemma involves states with dissimilar power, it
enables the security dilemma to be applied more widely than the traditional
concept. Whether it is relevant to other cases ultimately lies in whether one
or more actors are revisionists, or great criminals in Butterfield’s terms, in
which case it is not applicable. It is, though, relevant where one of the states
wishes to maintain the status quo but requires its neighbours to be insecure.
In such a scenario, a security dilemma can arise that is caused by the state’s
security requirement rather than just the anarchic nature of the inter-
national system. This, state-induced security dilemma, can still be thought of
as a tragedy, since although relations can be quite tense, none of the partic-
ipants involved seek conflict.

Notes

This is a revised version of a paper presented at a panel in honour of John Herz at

the 44th Annual Conference of the International Studies Association (ISA) in
Portland, Oregon, USA. I am grateful for the financial assistance provided by the
ISA and the British Academy which made it possible for me to attend the confer-
ence. I am also grateful for the suggestions made by the anonymous reviewers of
Cooperation and Conflict.

1. The security dilemma has been used to redirect security concerns away from
its Cold War obsession with military power, to understanding the development of
identity formation, as well as broadening the scope of security to encompass eco-
nomic, environmental and other concerns (Dalby, 1997: 12–18).
2. The precise role anarchy performs is a disputed topic in the literature. For
some, anarchy is not a cause of the security dilemma but rather a necessary con-
dition. That is, because anarchy promotes self-help behaviour and leaves states
uncertain of others’ intent, it creates propitious conditions for the security dilemma,
but is itself not a cause. Hence Alexander Wendt’s assertion that ‘[w]e do not begin
our relationship … in a security dilemma; security dilemmas are not given by anar-
chy or nature’ (1992: 407). Instead, it is the interaction of states and the perceptions
that statesmen gain from such contact that will determine if the security dilemma
occurs. In contrast, Robert Jervis argues, ‘[b]ut the heart of the security dilemma
argument is that an increase in one state’s security can make others less secure not
because of misperception or imagined hostility, but because of the anarchic context
of international relations’ (1976: 76). Whether a cause or a necessary condition, what
is clear is that its role is important.
3. Charles Glaser has also noted the importance of mixing competitive and coop-
erative policies when, in his example, confronting an insecure but greedy opponent.
Such an opponent, though, fits awkwardly within the security dilemma framework,
 COLLINS : STATE - INDUCED SECURITY DILEMMA 43

since its greed motivations make it a real threat to neighbouring powers (1992:
501–6).
4. The value of the offence/defence variable for the security dilemma has in cer-
tain instances gained such a prominent position that it has almost become synony-
mous with the security dilemma. This is particularly evident in Posen’s article
concerning ethnic conflict. While the ‘offensiveness’ of a force posture is an import-
ant factor in exacerbating a security dilemma, there are also important cognitive
factors, such as enemy imaging, zero-sum thinking and ethnocentrism, that play an
equally vital role. Reducing the security dilemma to the offence/defence variable is
highly misleading (1993).

References

Booth, Ken (1991) ‘The Interregnum: World Politics in Transition’, in Ken Booth
(ed.) New Thinking About Strategy and International Security, pp. 1–28. London:
Harper Collins.
Butterfield, Herbert (1949) Christianity and History. London: G. Bell and Sons Ltd.
Butterfield, Herbert (1951) History and Human Relations. London: Collins.
Buzan, Barry (1991) People, States and Fear: an Agenda for International Security
Studies in the Post-Cold War Era, 2nd edn. Hemel Hempstead: Harvester
Wheatsheaf.
Christensen, Thomas J. (1999) ‘China, the US–Japan Alliance, and the Security
Dilemma in East Asia’, International Security 23(4): 49–80.
Christensen, Thomas J. (2002) ‘The Contemporary Security Dilemma: Deterring a
Taiwan Conflict’, The Washington Quarterly 25(4): 7–21.
Collins, Alan (1997) The Security Dilemma and the End of the Cold War. Edinburgh:
Keele University Press.
Collins, Alan (2000) The Security Dilemmas of Southeast Asia. Basingstoke:
Macmillan.
Dalby, Simon (1997) ‘Contesting an Essential Concept: Reading the Dilemmas in
Contemporary Security Discourse’, in Keith Krause and Michael C. Williams
(eds) Critical Security Studies: Concepts and Cases, pp. 3–31. London: University
College of London Press.
Deutsch, Karl (1957) Political Community and the North Atlantic Area: International
Organization in the Light of Historic Experience. Princeton, NJ: Princeton
University Press.
Glaser, Charles L. (1992) ‘Political Consequences of Military Strategy: Expanding
and Refining the Spiral and Deterrence Models’, World Politics 44(4): 497–538.
Glaser, Charles L. (1997) ‘The Security Dilemma Revisited’, World Politics 50(1):
171–201.
Goldstein, Steven M. (1999) ‘Terms of Engagement: Taiwan’s Mainland Policy’, in
Alastair Iain Johnston and Robert S. Ross (eds) Engaging China: the Management
of an Emerging Power, pp. 57–86. London: Routledge.
Herz, John (1950) ‘Idealist Internationalism and the Security Dilemma’, World
Politics 2(2): 157–80.
Herz, John (1951) Political Realism and Political Idealism: A Study in Theories and
Realities. Chicago, IL: Chicago University Press.
Herz, John (1966) International Politics in the Atomic Age. New York: Columbia
University Press.
Jervis, Robert (1976) Perception and Misperception in International Politics.
Princeton, NJ: Princeton University Press.
44 COOPERATION AND CONFLICT 39(1)

Jervis, Robert (1978) ‘Cooperation under the Security Dilemma’, World Politics
40(1): 167–214.
Jervis, Robert (1982) ‘Security Regimes’, International Organization 36(2): 357–78.
Jervis, Robert (2001) ‘Was the Cold War a Security Dilemma?’, Journal of Cold War
Studies 3(1): 36–60.
Kaufman, Stuart J. (1996) ‘Spiralling to Ethnic War: Elites, Masses, and Moscow in
Moldova’s Civil War’, International Security 21(2): 108–38.
Kaufmann, Chaim (1996) ‘Possible and Impossible Solutions to Ethnic Civil Wars’,
International Security 20(4): 136–75.
Posen, Barry (1993) ‘The Security Dilemma and Ethnic Conflict’, Survival 35(1):
27–47.
Roe, Paul (2000) ‘Former Yugoslavia: The Security Dilemma That Never Was?’,
European Journal of International Relations 6(3): 373–93.
Roe, Paul (2001) ‘Actors’ Responsibility in “Tight”, “Regular” or “Loose” Security
Dilemmas’, Security Dialogue 32(1): 103–16.
Ross, Robert S. (2002) ‘Navigating the Taiwan Strait: Deterrence, Escalation
Dominance, and US-China Relations’, International Security 27(2): 48–85.
Schweller, Randall L. (1996) ‘Neorealism’s Status-Quo Bias: What Security
Dilemma?’ Security Studies 5(3): 90–121.
Snyder, Jack (1985) ‘Perceptions of the Security Dilemma in 1914’, in Robert Jervis,
Richard Ned Lebow and Janice Gross Stein (eds) Psychology and Deterrence,
pp. 153–79. Baltimore, MD: Johns Hopkins University Press.
Wendt, Alexander (1992) ‘Anarchy Is What States Make of It: the Social
Construction of Power Politics’, International Organization 46(2): 391–425.
Wheeler, Nicholas J. and Booth, Ken (1992) ‘The Security Dilemma’, in John Baylis
and N. J. Rengger (eds) Dilemmas of World Politics: International Issues in a
Changing World, pp. 29–60. Oxford: Clarendon Press.
Wiseman, Geoffrey (2002) Concepts of Non-Provocative Defence: Ideas and
Practices in International Security. Basingstoke: Palgrave.

ALAN COLLINS is a lecturer in the Department of Politics and

International Relations at the University of Wales Swansea. He was
previously a British Academy Post-Doctoral Fellow. He is the author of
The Security Dilemma and the End of the Cold War (Keele University
Press), The Security Dilemmas of Southeast Asia (Palgrave) and has
recently published an undergraduate text on Security and Southeast
Asia (Lynne Rienner).
Address: Department of Politics and International Relations,
University of Wales Swansea, Singleton Park, Swansea, SA2 8PP, Wales.
[email: a.collins@swansea.ac.uk]