Trustees of Princeton University

The Security Dilemma Revisited

Author(s): Charles L. Glaser
Reviewed work(s):
Source: World Politics, Vol. 50, No. 1, Fiftieth Anniversary Special Issue (Oct., 1997), pp. 171201
Published by: Cambridge University Press
Stable URL: http://www.jstor.org/stable/25054031 .
Accessed: 12/06/2012 01:17
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Cambridge University Press and Trustees of Princeton University are collaborating with JSTOR to digitize,
preserve and extend access to World Politics.

http://www.jstor.org

THE SECURITY DILEMMA

REVISITED
By CHARLES L. GLASER*

Jerviss

article

under

"Cooperation

the Security

re
in international
is among the most
important works
ar
two
lations of the past few decades.
In it, Jervis develops
essential
is
he
the
the key to
that
dilemma
First,
guments.
security
explains
states
in an anarchic
with
how
international
fun
system
understanding

Dilemma"
ROBERT

damentally compatible goals still end up in competition and atwar. The

a state tries
means
exists when
security dilemma
"many of the
by which
to increase its
the
of
decrease
others."1
It
the
security
security
provides
for what Jervis termed
the "spiral model,"
which
rational foundation
describes

how

can fuel
competition
explains that the magnitude

curity
on

two variables:

ferentiation.3

As

ness of
cooperative

states

between

the offense-defense
a result,
states

time. Although
that does not vary,

and

that are seeking only se

and strain political
relations.2 Second, Jervis
and nature of the security dilemma
depend

interaction

the

balance

and offense-defense
across

can vary
the security dilemma
exist in a condition
of international

dif
space

anarchy
can be
variation
in the attractive
significant
a
or
the prospects
for achieving
means,
competitive
there

high level of security, and the probability of war.

*
For comments on earlier drafts Iwould like to thank Lynn Eden, Colin Elman, Matt Evangelista,
and Steve
Jim Fearon, Lloyd Gruber, Chaim Kaufmann, Andy Kydd, Joseph Lepgold, Jim Morrow,
Walt.
1
World Politics 30 (January 1978), 169.
under the Security Dilemma,"
Robert Jervis, "Cooperation
see John H. Herz, "Idealist Internationalism
and the
For earlier discussions of the security dilemma,
identifies the basic concept, but does not
Security Dilemma," World Politics 2 (January 1950), which
Relations (London: Collins,
1951).
develop it extensively; and Herbert Butterfield, History and Human
2
in International Politics (Princeton: Princeton University
Robert Jervis, Perception andMisperception
Press, 1976), chap. 3, esp. 62-76; these pages provide a more thorough discussion of the basic workings
For an early discussion
of the security dilemma than does "Cooperation under the Security Dilemma."
see J. David
and the Armament-Tension
of this type of interaction,
Singer, "Threat-Perception
Dilemma," Journal of 'Conflict Resolution 2 (March 1958); for a recent discussion, see Charles L. Glaser,
and Refining
the Spiral and Deterrence
of Military
"Political Consequences
Strategy: Expanding

Models," WorldPolitics44 (July 1992).

3

similar arguments are developed by George Quester, Offense and Defense in the International
Many
to
1977). See also Marion William
Boggs, Attempts
Define and Limit 'Ag
System (New York: Wiley,
of Missouri,
and Strategy (Columbia: University
1941); and Marlies Ter Borg, "Re
gressive" Armament
of 1932," Journal of Peace Research 29, no. 2 (1992).
ducing Offensive
Capabilities: The Attempt

WorldPolitics 50 (October 1997), 171-201

172

WORLD POLITICS

These arguments laid the foundation for amajor debate within real
ism,which iswidely considered the dominant paradigm in international
Jervis s analysis rests on the same basic assumptions?anar
of goals?that
structural
underlie
chy and fundamental
compatibility
relations. As

re

alism, it should be viewed as part of this overall tradition. The broad

are
of his argument
different from the stan
implications
dramatically
dard structural-realist
since the latter does not focus
analysis, however,
on the
a
envisions
dilemma
and
security
consistently
consequently
more
and
world.4
the
debate over
competitive
dangerous
Although
these formulations
of
of structural realism has evolved since publication
to
matter
is
be
the
resolved.5
Jervis's article,
yet
can also
We
and impact
the importance
of security
appreciate
dilemma
have

and offense-defense

now

most
policy,
sources

employed

important

these

questions

the
including
of moderation

constraints,8

alliance

that scholars
by recognizing
to
address
many of the
arguments
effectively
of international
relations theory and security
arguments

effectiveness
in Soviet
behavior,9

of deterrence
policy,7
military

the severity
doctrine,10

and

reassurance,6
of relative gains
imperial

expan

4
The major statement of the standard structural-realist
analysis is Kenneth N. Waltz,
Theory of In
ternational Politics (Reading, Mass.: Addison-Wesley,
the State
1979); also important are idem, Man,
and War (New York: Columbia University Press, 1959), esp. chaps. 6,7; and idem, "Reflections on The
toMy Critics,"
in Robert O. Keohane,
ed., Neorealism and Its
ory of International Politics: A Response
Critics (New York: Columbia University
Press, 1986).
5
as
See Charles L. Glaser, "Realists as Optimists:
Cooperation
Self-Help," International Security 19
and Stephen Van Evera, Causes ofWar, vol. 1, The Structure of Power and the Roots
(Winter 1994-95);
(Ithaca, N.Y.: Cornell University
Press, forthcoming).
ofWar
6
not framed in terms of the security dilemma, see, for
Although
example, Richard Ned Lebow and
Janice Gross Stein, "Beyond Deterrence," Journal of SocialIssues 43 (Winter 1987). See also Stein, "De
terrence and Reassurance,"
et al., Behavior,
in Philip E. Tetlock
Society and Nuclear War, vol. 2 (New
to the se
York: Oxford University
Press, 1991), esp. 17; this essay gives somewhat greater prominence
curity dilemma. On the more general question of resolving political rivalries, see Sean Lynn-Jones,
in International Politics" (Ph.D.
between Adversaries
Accommodation
"Rivalry and Rapprochement:
in process).
diss., Harvard University,
7
et al. (fn. 6),
Matthew
in Soviet Security Policy," inTedock
"Sources of Moderation
Evangelista,
and Disarmament
in the 1950s," World Pol
esp. 290-96; and idem, "Cooperation Theory
Negotiations
see Deborah Welch
itics 42 (July 1990). On the Soviet use of concessions,
Larson, "Crisis Prevention
41 (Winter 1987).
and the Austria State Treaty," International Organization
8
in International Relations Theory," American Politi
Robert Powell, "Absolute and Relative Gains
cal Science Review 85 (December
1991); Powell casts the argument in terms of the costs of war, not in
terms of the offense-defense
balance. And see Glaser
(fn. 5), 79.
9
see
On balancing versus bandwagoning,
(Ithaca, N.Y.:
Stephen M. Walt, The Origins ofAlliances
see Thomas
Cornell University
On the tightness of alliances,
Press, 1987), esp. 24-25,165-67.
J.
Christensen
and Jack Snyder, "Chain Gangs and Passed Bucks: Predicting Alliance Patterns inMul
44
tipolarity," International Organization
"Perceptions and
(Spring 1990); and Thomas J. Christensen,
in Europe, 1865-1940,"
International Organization
is James D.
Allies
51 (Winter 1997); disagreeing
47
"Arms versus Allies: Trade-offs
in the Search for Security," International
Morrow,
Organization
see Karl Mueller,
"Patterns of Alliance: Align
(Spring 1993). On the alliance choices of small powers,
ment Balancing
and Stability in Eastern Europe," Security Studies 5 (Autumn 1995).
10
Barry R. Posen, The Sources ofMilitary Doctrine: France, Britain and Germany between the World
Wars (Ithaca, N.Y.: Cornell University
236-39;
Press, 1984), esp. 67, 74, 221-22,
Jack Snyder, The

SECURITYDILEMMA REVISITED
sion,11

revolution

and war,12

ethnic

conflict,13

173

conventional

arms

con

nuclear
and arms
trol,14 U.S.
policy
of
the escalatory
conventional
dangers

control,15 nuclear proliferation,16

war,17 U.S. grand strategy,18 and
and policies
for preserving
it.19 Lim

for peace in
the prospects
Europe
an examination
itations of space unfortunately
here of the
preclude
are
in
which
dilemma
and
offense-defense
ways
arguments
security

used in this literature.

The

first sections

discuss work

of this article

that has added

Jervis's basic arguments,

recapitulate
to these arguments,
and offer clarifications

and further extensions. Although the security dilemma is referred to

quite frequendy, relatively little effort has been devoted to examining
its core logic, some of which was left incomplete by Jervis himself. The
and how the security dilemma
gaps concern whether
important
actors.
in
between
rational
I explore three ways
operates
Consequendy,
a state's efforts
a
to increase
its security when
which
security
facing
most

dilemma

can, without
sirable outcomes.

states

suffering

misperceptions,

generate

unde

see Scott D.
Press, 1984); for disagreements,
Ideology of the Offensive (Ithaca, N.Y.: Cornell University
and Instability," International Security 11 (Fall 1986); the corre
Sagan, "1914 Revisited: Allies, Offense
and Elizabeth Kier,
spondence between Sagan and Snyder, International Security 11 (Winter 1986-87);
"Culture and Military Doctrine:
France between theWars," International Security 19 (Spring 1995).
11
(Ithaca, N.Y.: Cornell
Jack Snyder, Myths
of Empire: Domestic Politics and International Ambition
Press, 1991), 21-26.
University
12
Press, 1996), esp. 33-45,
Stephen M. Walt, Revolutions and War (Ithaca, N.Y.: Cornell University
334-38.
13
in Ethnic Conflict,"
Survival 35 (Spring
Barry R. Posen, "The Security Dilemma
to Ethnic Civil Wars," International
Chaim Kaufmann,
"Possible and Impossible Solutions

1993);

and

Security 20

1996).
(Spring
14

Conventional
Forces: Soviet Proposals andWestern
Jack Snyder, "Limiting Offensive
Options,"
see also
International Security 12 (Spring 1988), esp. 67-71;
Stephen Duane Biddle, "The Determi
nants of Offensiveness
in Conventional
Land Warfare"
and Defensiveness
(Ph.D. diss., Harvard Uni
the basic approach of offense-defense
theory, although apparendy not direcdy
versity, 1992). Following
influenced by it, is the nonprovocative
defense literature. See, for example, the special issue of the Bul
letin of theAtomic Scientists A4 (September 1988); and Anders Boserup and Robert Neild, The Founda
tions ofDefensive Defense (New York: St. Martins,
1990).
15
Nuclear Policy (Princeton: Princeton University Press, 1990).
Charles L. Glaser, Analyzing
Strategic
16
Shai Feldman, Israeli Nuclear Deterrence: A Strategy for the 1980s (New York: Columbia University
Press, 1982).
17
and NATO's Northern
Flank," International
Barry R. Posen, "Inadvertent Nuclear War? Escalation
Security 7 (Fall 1982); and idem, Inadvertent Escalation: Conventional War and Nuclear Risks (Ithaca,
a different
N.Y.: Cornell University
is Jeffrey W. Legro, "Military
Press, 1992). Offering
explanation
inWorld War II," International Security 18 (Spring 1994).
Culture and Inadvertent Escalation
18
"The Case for Finite Containment:
U.S. Grand Strategy," Interna
Analyzing
Stephen M. Walt,
tional Security 14 (Summer 1989), esp. 22-30.
19
Stephen Van Evera, "Primed for Peace: Europe after the Cold War," International
Security 15
A Demand
for Behav
(Winter 1990-91),
esp. 11-17; Ted Hopf,
"Managing Soviet Disintegration:
International
and Clifford A.
ioral Regimes,"
1992); Charles A. Kupchan
Security 17 (Summer
Security, and the Future of Europe," International Security 16 (Summer
Kupchan, "Concerts, Collective
1991), esp. 133-37; and Charles L. Glaser, "Why NATO Is Still Best: Future Security Structures for
Europe,"

International

Security 18 (Summer

1993),

esp. 26-33,

38-47.

WORLD POLITICS

174
The

section

following

that two additional

argues

ex

variables?the

tent of the adversary's greed (that is,motives beyond security) and of

the

adversary's

unit-level

of the

knowledge

state s motives?influence

the magnitude of the security dilemma. Thus, whether a theory posits

only

security

seekers

or instead

posits

some

greedy

states

is a pivotal

choice. In the latter case, the role of the security dilemma is diminished
and

competitive
policies
on unit-level
states can
rely

are more

to avoid conflict. Whether

likely
can have
information
about others' motives

a state to be secure when

it
equally
important
enabling
implications,
would
in some cases supports more coop
otherwise
be insecure, which
ones.
erative policies
and in other cases more competitive
By consider
one
these
also
the
into
dilemma
variables,
ing
security
integrates
broader debates over international
relations
and
theory
security policy.

These variables, for example, establish the divide between the spiral
model andwhat Jervis termed the "deterrencemodel," which applies to
secure

states
greedy
The final section

and offense-defense
greedy

states, not

and therefore
addresses

rejects the security dilemma.

basic criticisms
of the security dilemma

theory, including (1) the empirical claim that

the security

are the main

dilemma,

source of interna

tional conflict; (2) the security dilemma does not really exist, because its
or because
are
over
its constraints
always
logic is flawed,
or because
secu
states construct
whelmed
other
the
considerations,
by
can choose not to; and (3) offense-defense
dilemma
and
therefore
rity
I argue that the greedy-states
criticism poses a serious
theory is flawed.
internal

challenge

but that the others

The

are based

on

Security Dilemma: How Does

Undesirable
Outcomes?

the security dilemma

Jervis defines
a state tries to increase
which
by

or flawed

incomplete

as a situation

analysis.

It Lead to
in which

"the means

its security decrease

the security of
not
This
others."
definition
make
clear why
the security
by itself does
states value their own
is a
dUemma
however:
if
problem,
security but
an
not the
action
would
that
makes
one's
adver
of
others, why
security
secure
reason
most
less
be
bad?
is
The
obvious
that
the
sary
necessarily

adversary is likely to react to having its security reduced. And by the

same

reaction will
in turn reduce the state's secu
logic the adversary's
not this action-reaction
But
does
leave the
process
rity.
why
simply
state's security unchanged,
since the adversary's
reaction could j'ust off
set the state's action?
This
versary

section

identifies

less secure

three distinct

can be
self-defeating,

ways

inwhich

leaving

one's ad
making
the state worse off than

SECURITYDILEMMA REVISITED

175

before its initial action: (1) by setting inmotion a process that reduces
the state's own military capability, that is, its ability to perform military
missions;

(2) by increasing

the value

the adversary

places

on

expansion,

which makes it harder to deter; and (3) by simply wasting money. Al

though Jervis touches on each of these possibilities, he does not fully
the first two could

how

Thus,
happen without
misperceptions.
stems from
that
the
"heart
of
the
dilemma"
security
although
"the anarchic context of international
relations," Jervis leaves some key
to
is intended
discussion
The
arguments
following
underdeveloped.20

explain

he stresses

close

some
states

why

a brief discussion
section ends with
gaps. This
cannot avoid these undesirable
outcomes.
sometimes
of these

of

By further developing these arguments I do not mean to imply that

or not states suffer from
are
sig
misperceptions
unimportant. Whether
a rational state would
we need to
nificant misperceptions,
how
analyze
act ifwe

are to determine

how much

of the conditions

it faces

product
come when

misperceptions

Reducing

the

The

could

is really the
insecurity
worse
relations be

do occur.

State's Military

idea that arming

of a state's
and how much

reduce

Capability
capabilities

might

seem

self-contra

dictory, but the issue is one of terminology. I use "military capability" to

not
to
to the state's
refer specifically
ability
perform military missions,
an
to the size of its forces or its total
action-re
assets.21
Thus,
military
action process will
leave the state with more military
assets, but it may
its adversary.
reduce its military
capability vis-?-vis
its
when
the state arms, it makes
of the security dilemma,
secure
to defend
itself.
the
less
ability
adversary's
by reducing
adversary
arms in order to restore its
The adversary
then buys additional
military
nevertheless
Because

capability.22At first glance, itmight appear that the net effect of this
be to leave both countries' military
process would
forces would
since
additional
each
country's
unchanged,
the other's.23

action-reaction
bilities
offset

20
Jervis (fn. 2), 76.
21
On the advantages

capa
simply

of this usage of "capability," see Glenn H. Snyder, "Process Variables inNeo

realist Theory," Security Studies 5 (Spring 1996), 180-83.
22
the logic of the se
For the sake of simplicity, I will focus on the decision to buy arms. However,
to take territory and to acquire allies to increase
curity dilemma ismore general, including the decision
an action-reaction
into part of a buffer zone
process could be expansion
security. Regarding
territory,
that leads one's adversary to expand into the remainder of the zone.
23
Jervis (fn. 2), 64, says that such an action-reaction
process results in reduced security because
too little ... too little be
"when states seek the ability to defend themselves,
they get too much and
cause others, being menaced, will increase their own arms and so reduce the first state s security." How
its initial
ever, this explains only why the state's security is reduced relative to the situation following
to the prior military status quo.
buildup, but not why it should be reduced relative

176

WORLD POLITICS

In fact, however, when

the action-reaction
process shifts the offense
a
or an in
defense
is
the
result
instead
decrease
balance,
change?a
a
new
state
crease?in
If the
weapons
military
capabilities.
deploys
that
favors
if
its
offensive
missions
and
system
adversary
responds by

deploying this system, the country's ability to defend itself will be re

secure than before this round of
leaving it less
arming. MIRVed
are
to be this
considered
of
offensive
innovation,
type
usually

duced,
missiles

reducing the ease with which

could meet
ploys
matches

the United States and the Soviet Union

their

for deterrence.
If, however,
requirements
an innovation
that favors defensive
missions
and
the net result would

it, then

the state de

its adversary
in the state's capa

be an increase

bility to defend itself and an increase in its security. The precision

guided munitions (PGMs) deployed on the Central Front during the
cold war
The
with

appear to have been such an innovation.24

can also vary
to
and defensive missions
ability
perform offensive
force size.25 Thus,
action-reaction
in larger
that
result
processes

forces (as distinguished from different types of forces) can increase or

decrease
increases

the state's military

for defense.
capability
in the size of conventional
ground forces

For

example,
equal
can result in an in

crease in a state's ability to defend, by enabling it to increase the density

of forces deployed along the front. Similarly, equal increases in the size
of nuclear

can

forces

increase

both

countries'

retaliatory
capabilities,
In such cases, an action

their deterrent

thereby enhancing
capabilities.
reaction process
increases security.
of the possibility
that arms
Appreciation
both

countries'

competition
Jervis's work

can reduce
on the secu

military
capabilities
precedes
In fact, the complementary
observation
that adversaries
rity dilemma.
can have a mutual
arms
interest in reciprocating
restraint is one of the
core

insights

of modern

arms control

theory, which

was

developed

in the

late 1950s and the early 1960s.26 Setting this insight within the broader
framework of the security dilemma, however, highlights an important
issue that arms control
the

intrinsic

goals

to
theory tends
gloss over?the
of the states involved. The
security

of
compatibility
dilemma makes

explicit the possibility that both states in the competition are interested
only
24

in security,

John J.Mearsheimer,

chap. 7.
25
This

and

it deepens

Conventional

our

Deterrence

understanding

(Ithaca, N.Y.:

of how

Cornell

this compe

University

Press,

1983),

is discussed in Chaim Kaufmann

and Charles Glaser, "What Is the Offense-De
possibility
fense Balance and Can We Measure
It?" International
(forthcoming).
Security
26
See, for example, Thomas C. Schelling and Morton H. Halperin,
Strategy and Arms Control(New
York: Twentieth
Century Fund, 1961), esp. 1-2.

177

SECURITYDILEMMA REVISITED
can arise.

tition

In addition,

Jervis

lays the groundwork

work by framing the problem of cooperation

Dilemma

and Stag Hunt

for cooperation

recent

in terms of Prisoners'

matrices

and exploring
how the
in the relative size of coun

game-theory

vary with changes

In more recent years,
170-83).27
(pp.

prospects
tries'
payoffs

for more

theorists

cooperation

have used game theory to establish amore rigorous foundation for the
core findings of arms control theory; they frequendy use the repeated
Prisoners'

arms

to model

Dilemma

competition.28

Making One's Adversary

Less Secure, Which
Value It Places on Expansion
the

danger

of the

the

Increases

insecurity

adversary's

an
can reduce the state's own
a
security in
adversary's
security
Reducing
second way?by
the value the adversary places on expansion,
increasing
an
more
to deter.
insecure
it
harder
Making
adversary
thereby making
will

often

increase
pansion

increase

its interest

can often
since expansion
in expansion,
a more
insecure adversary will find ex

security. For example,

more desirable when
it can provide more

secure borders,

strate

gic depth, or control of resources that are valuable for building military
capabilities.29
Consequendy,
net result
ity, the
the state will

enjoy

a state's
military
arming increases
capabil
in its security. On the one hand,
be a reduction
the enhanced
deterrent
and defense
pro
capabilities

even when
could

vided by its improved military capability. On the other hand, because

the adversary

is now harder

to deter,

itmay

not be deterred

en
by these

hanced capabilities, even if itwould previously have been deterred by

less effective

military
capabilities.
states that can achieve
Thus,
military
answer
is no general
seek them. There

advantages
to whether

should
sustainable

not

always

military

27
that can lead to arms competition
For earlier use of game theory to explore the different motives
see Thomas C. Schelling,
Pro
"A Framework
for the Evaluation
of Arms-Control
and cooperation,
contribution
of
104 (Summer 1975). While
remaining positive about the potential
posals," Daedulus
in "Realism, Game Theory,
and Co
formulations,
Jervis explores their shortcomings
game-theoretic
World Politics 40 (April 1988).
operation,"
28
The Evolution
Robert Axelrod,
1984). For a focus on
(New York: Basic Books,
of Cooperation
see George W. Downs, David M. Rocke, and Randolph M. Siverson,
"Arms
questions of arming,
in Kenneth A. Oye, ed., Cooperation under Anarchy
Races and Cooperation,"
(Princeton: Princeton
Press, 1986).
University
29
as
se
Jervis (fh. 1), 168-69. Although
Jervis identifies this phenomenon
being separate from the
it as an integral part of the security dilemma. On the resource value of ter
curity dilemma, I describe
International Security 18 (Fall 1993); and idem,
"The Spoils of Conquest,"
ritory, see Peter Liberman,
Societies (Princeton: Princeton University
Does
of Occupied Industrial
Conquest Pay? The Exploitation
Press, 1996).

WORLD POLITICS

178
advantages
crease the

that
state's

find

trade-off

leave one's

that

adversary
but policy

security,
sometimes

more

less

secure will

increase

or de

this
that grapple with
analyses
to
or restraint
is
preferable
cooperation
In the current debate over NATO expansion,

competitive
policies.
into Central
for example,
argue that even though expansion
opponents
increase
it
would
also
increase
NATO's
military
capability,
Europe would
invasion of its neighbors
Russian
insecurity and therefore make Russian
more
likely.30

ACTIONSTHAT REDUCEAN ADVERSARY'SSECURITY:

SIGNALING
OFMOTIVES
can make
state's military
buildup
more
ways. First, and
straightforward,

its adversary
the state's

less

secure

buildup

in two

can reduce

the adversary's ability to defend itself. Even if the adversary matches

in its mil
the net result could be a reduction
initial buildup,
as described
is
that
It
the
above.
also possible
adversary
itary capability,

the state's

will end up less capable of defense because it is unable to match or

counter

the state's buildup.

a state's
Second,
military

buildup

can

the adversary's beliefs

that the state is in
the adversary
change

the state's motives,

convincing
more
than it had previously
believed. More
specif
dangerous
herendy
assessment
increase the adversary's
of
could
the
state's
ically,
buildup
reasons
to
it is motivated
the
desire
for
the extent to which
by
expand
about

reaction
are

seeking

more

term
I will
type of action
greed.32 This
in which
countries
that
of the spiral model,
are
their
motives
that
conclude
adversary's
only security
than previously
believed.33 Although
that is, greedier,

security,31 which
is the focus
process

than

other

malign,
more
can make
spirals
misperceptions
stresses that both the security dilemma
tional
30

intense, Jervis, as already noted,

have a ra
and the spiral model

foundation.34

Survival
37
E. Brown,
"The Flawed Logic of NATO Expansion,"
See, for example, Michael
a security dilemma-based
argument against nuclear superiority, see Glaser (fh. 15),
(Spring 1995). For
5.
chap.
31
The adversary could also become less secure if it concludes that the state places a higher value on
state harder to deter. For
a
security or demands
higher level of security, both of which could make the
on the
assessment of the state's greed.
adversary's
simplicity, I focus
to
321 use the term "greedy" because states can be motivated
expand for two fundamentally different
more common terms
are blurred
and greed?which
types of reasons?security
by the
"expansionist" and
states can be defined in terms of greed and security seeking; see Glaser (fh. 2),
"aggressive." Four types of
501-3. In referring to states as greedy, I do not mean to imply that they do not also seek security.
33
inAlliance Pol
Jervis (fh. 2), chap. 3, esp. 62-76. See also Glenn Snyder, "The Security Dilemma
can generate re
itics," World Politics 36 (July 1984), 468-70;
Snyder argues that firm alliance policies
actions that are comparable to a spiral generated by an arms buildup.
34
In addition to the passage cited in fn. 20, see, for example, Jervis (fn. 2), 62.

SECURITYDILEMMA REVISITED
Nonetheless,
a

179

the possibility of a rational spiral presents something of

the arms policy of a rational state that seeks only

on
its
sound inferences,
that
security
adversary, based
logically
more
was
state
answer
the
is
than
is
believed? The
dangerous
previously
not obvious
for the following
reasons, the adversary will un
because,
not
derstand
that an arms buUdup may be motivated
by security,
greed.
state
not
its motives,
that
the
does
know
the
will
adversary
Recognizing
How

puzzle.

could

convince

the state's desire for adequate

defense
which
appreciate
capabilities,
could require more or
armed forces. In addition,
appreciating
improved
that forces that the
the security dilemma,
the adversary will understand
state
could reduce
its own security. A
requires for increased
security
to expect a pure secu
therefore
have reasons
adversary will
a
arms
to
in
and consequently
engage
threatening
rity seeker
buildup
will not automatically
conclude
that such a buildup
reflects greedy
motives.
rational

then is, when

does a state's military
buildup
signal
question
motives?
claim
is
A
that
structural
common,
incorrect,
greedy
although
about the motives
of oth
theories do not allow states to know anything
The

the possibility
But this claim overlooks
ers, since they are unobservable.
can communicate
because
that certain
actions
valuable
information
they

a
to be taken
state and a pure se
by
greedy
equally likely
a state launches
a
seeker. Therefore,
for example, when
military

are not

curity

a
buildup that ismore likely to be taken by a greedy state than by pure
an
that is making
sound inferences will up
security seeker,
adversary
date its assessment
of the state's motives,
that the state is
concluding

more likely than previously believed to be greedy.35

Two types of military buildups could help with this kind of differen
tiation.
forces:
beyond

size
type, different
types of states prefer different
more
a
to add forces
is
than
seeker
security
likely
greedy
for adequate
defense
of its territory.
Even
those
required

In the first
a

state

though extra forces would provide some additional capability to defend,

a state

see less value in these forces

only in security would
to pay for
state and therefore would
be less willing
greedy
states
In
different
them.
the second
type of buildup,
types of
prefer
a state has a choice between
forces. For example, when
forces that add
to offensive
missions
and defensive
ones, a
roughly
military
equally
a
to choose
state is more
the
than
seeker
type of
security
greedy
likely
interested

than would

35
On signaling, see Robert Jervis, The Logic of Images in International Relations
(Princeton:
ton University
Press, 1970); and James D. Fearon, "Threats to Use Force: The Role of Cosdy
in International Crises" (Ph.D. diss., University
of California, Berkeley, 1992).

Prince
Signals

WORLD POLITICS

180

state sees both greater

its offense.36 The
greedy
improve
not
in
value in offense and sometimes
less value
others, since
provoking
it anticipates
conflict anyway.
forces

that

The
sound

that makes
is, can an adversary
logically
about the
reach incorrect
conclusions

that remains

question
inferences

nevertheless

state's goals?37 For

the interaction

between

pure

to pro

seekers

security

duce the increasing fear and insecurity explained by a rational spiral

model,

the answer must

quired
This

a
to generate
spiral.
of
interaction
type

be yes. Otherwise,
can occur

would

be re

are uncertain

about

misperceptions
if the states

a
the size or type of forces required to maintain
given level of security.
case
some pure se
inwhich
consider
the simple
this uncertainty,
Given
a lower level of forces and some
be satisfied with
curity seekers would

with a higher level, but all greedy stateswould require a higher level.A
pure security seeker that builds to the higher levelwill then convince its
adversary

that

it is more

to be
greedy,

likely

since

only

some

security

seekers but all greedy states would build to this level. Similar interac
tions can occur

if there

about whether

is uncertainty

security

seekers

re

quire offensive capabilities and if there is uncertainty about the level of

security that security
A spiral can also

is adequate.38

believe

seekers
result

from

an

of its motives.

state's understanding

about the
uncertainty
adversary's
the adversary
For example, when

believes that the state believes there is only a small probability that the
that the state does not fear it, the ad
is greedy and, therefore,
adversary
for
is largely unnecessary
state's
that
the
will
conclude
versary
buildup

security and therefore that the state is likely to be motivated by greed. If

initial estimate
the adversary's
is too low, then this increase

of the probability
in its assessment

that the state

is fearful

that the state

is greedy

will be too large, resulting in a spiral. Kydd's formal analysis of this in

teraction

shows

that under

awide

range of conditions

a rational

adver

sarywill find the state's buildup to be provocative and that updating of

beliefs

is sensitive

to

prior

expectations

about

the state's motives.39

36
is an oversimplification,
This
however, since a pure security seeker might want the capability to
take territory for a variety of reasons. See fh. 57 below for qualifications.
37
states can
Jervis identifies the types of buildups, he does not really explain how rational
Although
notes that states often cannot accurately infer motives from others'
spiral; Jervis (fn. 1), 199-201. He
to assume the worst; Jervis (fh. 2), 64-65. This would clearly
military forces and therefore they tend
additional information
is logically flawed, since without
generate increased hostility, but the judgment
states should not assume the worst in the face
should remain unchanged. Moreover,
states' assessments
since this can support policies that are too competitive/provocative.
of uncertainty,
38
see Jervis (fn. 1),
of related issues concerning
For discussion
subjective security requirements,
174-76.
39
and the Spiral Model," World Politics 49 (April 1997). Kydd notes
Andrew Kydd, "Game Theory
the extreme cases in which bias leads the adversary to
(p. 373) that Jervis explores this dynamic for

SECURITYDILEMMA REVISITED

181

The logic of these signaling arguments alsoworks in the opposite di

state can sometimes

rection?a
to reduce

the adversary's

concern

use restraint
about

in

building
its greediness.40

forces
military
A greedy state

wants to mislead its adversaries into believing that it is interested only

to pursue
in security,
since its adversaries would
then be more
likely
that leave them vulnerable. Given
these incentives
for a greedy
policies
state to

misrepresent

its motives,

security

seeker

can communicate

in

formation about itsmotives only by adopting a policy that would be

for it than for a greedy
on the conditions
Depending

less cosdy

state.

states can try to communi

they face,
via three types of military
their benign
intentions
policies.41 First,
arms control
that limit both countries' current or future abil
agreements
cate

a lack of
a
communicate
offensive missions
greed, since
perform
a pure secu
state sees greater value in offensive missions
than
does
greedy
a state may be able to
rity seeker. Second,
adopt unilateral defense, choos
a
even if its
to protect
its country with
defensive
doctrine,
ing
adversary
one. When
to pursue an offensive
continues
offense has the advantage,
ity

to

its security via unilateral

defense will require the state to
are
its adversary, which
reinforces
the message
that its motives
outspend
a state can exercise unilateral
restraint, that is, reduce its
benign. Finally,

maintaining

military capability below what itwould choose for adequate deterrence

and defense
unilateral

were

it not considering
the effects of signaling.
set in motion
is primarily
restraint
tactical?to

One use of
a process of
re
unilateral

if not reciprocated,
restraint.42 Even
however,
reciprocated
straint can succeed by communicating
that the state is not greedy and is
to
relations. Thus,
the state's ability to
committed
although
improving
is reduced, the net effect can be an increase in its security. There
the state's re
the danger that the adversary will misinterpret
is, however,
a
a
in
lack
this case, re
of
lack
of
resolve
instead
straint, seeing
greed;
state.
to
the
The
dual
straint encourages
the adversary
dangers
challenge
states re
of their resolve make
of military
shortfalls and underestimates
restraint.
of unilateral
luctant to pursue ambitious
policies

defend

secure. For pure security seekers, however, all that is necessary to gener
some
that the state is secure. See also
is
for
the
adversary to believe there is
possibility
spiral
and David M. Rocke, Tacit Bargaining, Arms Races, and Arms Control (Ann Arbor
George W. Downs
of Michigan
Press, 1990), chap. 4.
University
40
a formal treatment.
Kydd (fn. 39) provides
41
Glaser
(fh. 5), 67-70; and idem (fh. 2), 526-33; for related points, see Downs, Rocke, and Siver
son (fn. 28). On reassurance more generally, see Stein (fn. 6); on the inhibiting effect of appearing
in Crisis Bargaining," International
with Linkage
weak, see James D. Morrow,
"Signaling Difficulties
Studies Quarterly 36 (June 1996).
42
to
This approach, often referred to as GRIT, was developed by Charles E. Osgood, An Alternative
of Illinois, 1962).
War or Surrender (Urbana: University
assume
ate a

the state is definitely

WORLD POLITICS

182
MISPERCEPTIONS

In contrast to the rational updating we have considered so far, Jervis

to states'
biases in contributing
the role of psychological
recent work has fo
assessments
of others' actions.43 More

also discusses
hostile

overly
cused

on bureaucratic

sources

and domestic

political

as alternative

processes

of bias.
do not understand

who

Statesmen

the security

dilemma

and

there

fore do a poor job of appreciating the choices faced by their adversary

will

infer

incorrectly

that

the adversary's

reflects

buildup

greedy

mo

tives.A state is likely tomake errors that build on each other: a key ini
is for a state to assume

tial mistake

that others

it is interested

know

only

in security; the state is then likely to assume that others will not be
threatened by its buildup. Consequendy, the state is inclined to see the
arms

adversary's

as a

buildup

sign of greed, when

in fact the adversary

is building in response to the state's buildup. This is a distorted form of

the rational

spiral, described
when
about
motives;
tainty
cure and
will
competition
rational

be more

intense

than

is predicted

by

dilemma.

security

Attribution

is driven entirely by uncer

above, which
states will be more
inse
this bias prevails,

theory offers a psychological explanation for this type of

make
of interpret
the mistake
reasoning.44 Leaders
commonly
even
terms of their
in
countries
the
behavior
of
other
ing
goals/motives,

flawed

though they understand their own behavior differendy, in terms of the

situation

they face. As
face a security

countries
as

others
have

acted

Recent

a result,
dilemma
even

greed,
reflecting
the same way.
work on misperceptions

leaders

fail

to

and therefore
though
provides

pure

appreciate
interpret
security

alternative

that other
the actions

of

seeker might

explanations,

lo

cating these analytic flaws at the level of the state instead of at the level
of the

individual.

and domestic

This

work

uses

theories
to

of organizational
behavior
states often exaggerate
explain why
are inclined
to exaggerate
the of

dynamics
political
adversary's
hostility.45 Militaries
inten
of the adversary's forces and to impute malign
fensive potential
are
even
of
when
the
the
tions,
purposes
ambiguous.
adversary's forces
an

43
in "Cooperation under the Security Dilemma"
receives far less discussion
(but see
Misperception
than in Jervis (fh. 2), 67-76.
pp. 181-83)
44
On attribution theory, see Jervis (fn. 2), esp. 35-48; Deborah Welch
Larson, Origins of Contain
ment: A
and
Press, 1985), esp. 34-42;
(Princeton: Princeton
University
Psychological Explanation
Press, 1996).
Jonathan Mercer, Reputation and International Politics (Ithaca, N.Y.: Cornell University
45
see
For an emphasis on organizational
Stephen Van Evera, "Causes ofWar" (Ph.D.
perspectives,
of California, Berkeley, 1984), pt. 2; and Snyder (fn. 10). Snyder (fh. 11) emphasizes
diss., University
domestic political dynamics.

SECURITYDILEMMA REVISITED

183

Powerfvd interest groups that would benefit from military competition

or

that exag
often advance
strategic arguments
expansion
self-serving
or
motives.
the
threat
the
Fur
gerate
posed by
adversary's
capabilities
are
states
to create
about the unthreatening
inclined
thermore,
myths
nature

of their own behavior,

which

makes

adversaries

appear

still more

dangerous.46
a state's
In sum,
rational updating
(or re
military
buildup
through
to alter its assessment
of the state's mo
straint) can lead an adversary
on
states are inclined to do
cautions
tives. Research
that
misperceptions

a poor job of updating, with a bias toward exaggerating the hostility of

states will
they suffer from these biases,
severe than it
is.
is more
dilemma
actually

others. When
security

Simply Wastes

act as

though

the

Money

secure can be self

third way inwhich
efforts to make oneself more
two cases, the
is
the
first
Unlike
money.
defeating
by simply wasting
a
state would
not end up less secure for
having pursued
policy that ini
its adversary
less secure. In this case, the action-reaction
tially made
as we have seen,
not
the
state's military
does
reduce
process
capability;

The

could leave the state's ability to defend

itself
processes
a
increas
Nor does the action-reaction
generate
spiral of
unchanged.
views
which
is
since
of
the
motives,
adversary's
possible
ingly negative
not all
motives.
information
about
Neverthe
military
buildups
provide
action-reaction

results
less, the security dilemma
no
more
state less prosperous,
yet

in self-defeating
secure.47

Why Can't States Cooperate

Less Desirable
Outcomes?
When
would

and engaging
arming
be self-defeating,
why

undesirable
under

efforts

that leave the

to Avoid

in competitive
cannot rational

more
policies
generally
states cooperate
to avoid

of international
relations
Jervis's formulation
raises this question
since
Hunt
cooper
starkly,
Stag

outcomes?

anarchy

as a

ation would then be the best option for both countries.48

46
Van Evera

(fn. 45), chap. 8; and idem, "Why States Believe Foolish Ideas: Non-Self-Evaluation
Political Sci
of the American
and Society" (Paper presented at the annual meeting
by Government
ence Association,
D.C.,
1988).
Washington,
47
is
resources; he argues (p. 462) that alliance formation
Snyder (fh. 33), 461, emphasizes wasted
similar to arming, in that all states would be better off remaining outside an alliance; action and reac
tion nevertheless
generate alliance blocks that are cosdy but fail to increase security.
48
and related games, see Kenneth A. Oye, "Explaining Cooperation
under Anar
On Stag Hunt
chy: Hypotheses

and Strategies,"

inOye

(fn. 28).

184

WORLD POLITICS
lies in appreciating
that states are uncertain
confidence
that others are pure
motives,
lacking
means
motives
about
that states cannot be
seekers. Uncertainty

core of the answer

The
about

their adversaries'

security
confident
tainty

are in a
they
Stag Hunt,
the type of game can make

that

about

even

if they

are. This

uncer
the state's

competition/arming

best option. Jervis takes a different view, identifying different sources of

in a
and irrational be
when
competition
Stag Hunt?misperceptions
as
havior
and
irrational
behavior
misperceptions
(p. 168). Important
can be, however,
to the basic
these are additions
From
the per
story.49
on
it places
the implica
spective of structural realism and the emphasis
tions of anarchy,
its understanding

its motives
about the adversary?both
and
uncertainty
of one's own motives?is
the fundamental
ingredient

for understanding
model
game-theory

This
competition.50
of how uncertainty

subsection
about

the prospects
for productive
cooperation.
to
two
It is usefid
by
considering
why
begin
be in a Stag Hunt.51
the one hand, a pure
On

describes

the adversary

a
simple
could re

duce

security

security
of not being attacked has no reason to expand:
it controls
the territory
and does not see instrumental

confident
with

ditional

territory,

because

including

how

it does

not fear attack.

It could

seekers might
seeker that is
it is satisfied
value
prefer

in ad
the ter

ritorial status quo (cc) to unopposed expansion (DC) for a variety of

reasons,
because
Because

or
to govern
the additional
territory
costly it is
an
norm
state
violates
international
that the
values.

expansion
the state values

its own

territory,

it could

prefer fighting

to pro

tect its territory (dd) to allowing its adversary to expand unopposed

seekers have these preferences
and this is
(CD). If two pure security
are
a
common
in
then they
and both countries
knowledge,
Stag Hunt
to cooperate.
choose
should optimally
its adversary
is a
hand, a state that is unsure whether
a
situation.
seeker
faces
A
different
pure security
very
greedy adversary
to the territorial
status quo; it has
will prefer
expansion
unopposed
or
Even
if the state's own preference
Prisoners' Dilemma
preferences.
its
in
remains
of
this
adver
light
dering
preferred option
unchanged,52
is
behavior
competition.
sary's expected
On

the other

49
see Downs, Rocke, and Siverson (fn. 28), 134-37.
For discussion of their impact on a Stag Hunt,
50
In addition, a type of uncertainty
that is not explored in the text is also important?uncertainty
an
about whether
adversary will become greedier. This uncertainty cannot be eliminated because lead
ers cannot bind themselves and their successors to current goals; see Jervis (fh. 1), 168.
51
can be devel
over territory, a
focuses on competition
my discussion
Although
parallel analysis
arms
for
oped
competition.
52
If facing a greedy adversary, however, a pure security seeker would now see instrumental value
state would
in expansion,
if this would
increase its security. The
then have Prisoners' Dilemma

SECURITYDILEMMA REVISITED

185

If the state is unsure about the type of adversary it faces, then it is

uncertain about which game it is in.The state should therefore weigh
its payoffs for cooperation and competition by its estimate of whether
the
is greedy or a pure security seeker.53
is more
adversary
Competition
attractive
the higher
is
the state's estimate
that the adversary
greedy.

Offense-Defense

Variables: Variation
Security Dilemma

in the

is his
of how the
Jervis's second major contribution
explanation
magni
on two variables?the
tude and nature of the security dilemma
depend
balance and offense-defense
differentiation.
offense-defense
In consid
ering

how

these variables

influence

he generates

a vari

including most prominently hypotheses about the

ety of hypotheses,

the prospects

for competition,

pressures

state behavior,
for

international

cooperation,

and the probability of war.

Jervis defines

the offense-defense

in terms of the ease of tak

balance
ease

at
of holding
ing territory
compared
territory when
tacked:
the advantage
of defense
increases with
the ease of holding
a way of
then proposes
the relative
territory
measuring
(p. 187). He
ease of offense
and defense:
"Does the state have to spend more or less
than one dollar on defensive
forces to offset each dollar spent by the
with

the

other side on forces that could be used to attack?" (p. 188).

The
fense

as the offense-de
decreases
severity of the security dilemma
defense
balance
shifts toward greater defense
advantage. When

a status quo
in
the advantage,
the forces deployed
power will
by
more
its security
than they decrease
the adversary's
security. Both
states will achieve reasonable
levels of security from action and reaction

has

crease

cycles,

and arms

races will

peter out. When

next
"aggression will be

great,
sufficiendy
international
anarchy
By contrast, when

the advantage
to

of defense

is

thus rendering

impossible,
(p. 187).
relatively unimportant''
it is
offense
has the advantage,
for
impossible
arms
size to enjoy high levels of security simultaneously;

states of
equal
be intense

races will

because

when

one

country

adds forces,

its adver

In addition, the state would see a higher payoff for war,

instead of Stag Hunt preferences.
or of leaving the adversary relatively weaker, which
some prospect of successful
expansion
makes competition more attractive.
on the
to this formulation,
Some of my game-theory
grounds that payoffs should
colleagues object
a
be fixed and not vary with the type of adversary. A more adequate formulation
requires multiperiod
game.
S3
the relative size of payoffs matters. Jervis devotes much of "Cooperation under the
Consequendy,
to
variables are among
Security Dilemma"
exploring factors that influence payoffs. Offense-defense
the most important and are discussed in the following section.
preferences
ifwar held

186

WORLD POLITICS

sary will
Jervis

to make

have

argues

that offense

to restore its
to defend.54
larger addition
ability
war more
a
makes
for
advantage
likely
variety

of reasons. (1)War will be quick and decisive and therefore profitable,

so greedy states will find war more attractive. (2)
Following the basic
logic of the security dilemma,
more valuable, which
pansion

ing security. (3) The

which

advantage,

attacks

emptive
will be

states will
makes

be more

war more

ex
insecure, making
to
states
attractive
seek

advantage of striking first grows with offense

the probability
via pre
of crises escalating
wars
are
arms
races
and accidents.55
(4) Because
likely,

still more

increases

their danger. Offense

increasing
advantage
states to form alliances
forcing
diplomacy,
during peace
too
to allow the effective
for
time, since wars will be decided
quickly
once
mation
of alliances
has started.56
fighting
also

intense,

influences

The

second

extent

to which

weapons

that influences
the security dilemma
is the
key variable
are differentiated,
and defense
offense
that is, "whether

and policies

that protect

the state also

provide

the capability

for attack" (p. 199). Offense-defense differentiation has the potential

virtually to eliminate the security dilemma: if completely differentiated,
a country

can then
are usefid
its
deploy forces that
only for protecting
not
to
reduce its adversary's
does
defend
itself.
territory, which
ability
differentiation
enables a country to signal its
offense-defense
Moreover,
a
to take
type, since only
country that wants
territory will buy forces that
have offensive potential.57 Pure security seekers can therefore reduce con
cern about whether
in turn increases
they harbor greedy motives, which
their own security. The differentiation
of offense and defense makes pos
that ban weapons
sible arms control agreements
that are usefid for of
fensive missions,
54
For related analysis,

thereby
seeMalcolm

increasing
W.

Hoag,

both

countries'

"On Stability

ability

in Deterrent

to defend.58

Races," World Politics

13

1961).
(July
55

and accidents, see Schelling and

On the relationship between first-strike advantages, preemption,
(fn. 26), 14-16.
Halperin
56
See also Quester
(fh. 3), 105-6.
57
As Jervis (fh. 1) notes, this is an overstatement
(pp. 201-2). A pure security seeker might buy of
fense for a number of reasons: (1) if offense has a great advantage over defense; (2) if the state has ex
tended deterrence
commitments;
(3) because offense may be necessary to regain territory lost at the
can enhance deterrence. On extended
a
beginning of war, and (4) because the threat of counteroffense
see
Van
W.
Defense
and
Is Offense
Best?"
deterrence,
Evera,
"Offense,
Strategy: When
Stephen
at the annual meeting
Political Science Association,
of the American
(Paper presented
Chicago,
Sep
see Samuel P. Huntington,
tember 1987). On counteroffense,
"Convention Deterrence
and Conven
in Europe," International Security 8 (Winter 1983-84);
and Barry R. Posen, "Crisis
tional Retaliation
120 (Winter 1991).
Arms Control," Daedulus
Stability and Conventional
58
is implicit in the distinction between qualitative and quantitative
The concept of differentiation
arms control, with the former relying on differentiation.
This concept lies at the core of modern arms
control theory; see, for example, Thomas C. Schelling, "What Went Wrong with Arms Control," For
eign Affairs 64 (Winter 1985-86).

security

dilemma

In contrast, when
offense
to
forces
try buying
protect

its territory

revisited

and defense

187

are not differentiated,

only choose forces

can

a coun
that re

duce its adversary's ability to defend. Signaling becomes much harder

and riskier, because pure
same
types of forces.

security

seekers

states will

and greedy

buy

the

This stock of hypotheses has been supplemented by Van Evera, who

on how

adds hypotheses
courages

styles

offense

of diplomacy

fense advantage heightens

advantage
that increase

war

fuels preventive
the probability

and en

of war. Of

the significance of shifts in power, which

to use fait
for preventive war. States are more
likely
tactics when
offense has the
dis
because winning
accompli
advantage
more
is
is
when
which
scarce,
putes
encourages
security
important
states to overlook
States negotiate
less and
negotiable
compromises.
increases

incentives

less successfiilly,
because
makes
offense
it more
advantage
important
more
that agreements
be carefiilly balanced, which makes
negotiations
and
increases
because
offense
the
of vi
difficult,
advantage
advantages
which makes
riskier. In addition,
of
agreements,
agreements
olating
states more
secretive,
advantage makes
can increase their
forces
military
vulnerability.
fense

since

information

Secrecy

about

in turn increases

the probability of war by fueling miscalculations of military capabilities

and of states'

interests.59

The deductive strength of this body of offense-defense

hypotheses

but recent work has questioned

the rela
largely unchallenged,
war.
offense
and the frequency
of
Because
tionship between
advantage
the risks of war could be greater for the attacker when
offense has the

has gone

attackers
should face countervailing
advantage,
potential
them more
make
when
cautious,
especially
considering
war
which
could sometimes
make
less likely.60
Van Evera
hypotheses,

has performed
even
although

the most
these

are

extensive

pressures

that

large wars,

tests of offense-defense

preliminary.61

Focusing

on

Europe

since 1798, he finds strong support for the theory's basic hypothesis:
war

ismore

likely when

tage. Examining

offense

has

(or is perceived

the FirstWorld War

to have)

the advan

in detail, Van Evera finds sup

59
Van Evera (fn. 5), chap. 5.
60
49 (Summer
forWar," International
James D. Fearon, "Rationalist Explanations
Organization
1995), 402-3. And see the following papers prepared for the annual meeting of the International Stud
iesAssociation,
Balance andWar since 1648";
Chicago, February 1995: Fearon, "The Offense-Defense
of Offense-Defense
and Charles Glaser and Chaim Kaufmann,
"Inspecting the Foundations
Theory:
Can They Bear theWeight?"
61
Van Evera (fh. 5), chaps. 5,6; and idem, "The Cult of the Offensive
and the Origins
of the First
War," International
"Polarity, the Offense-De
Security 9 (Summer 1984). See also Ted Hopf,
Political Science Review 85 (June 1991); and Fearon (fn. 60, Febru
fense Balance, andWar,"American
ary 1995), who argues that the frequency of war in broad historical periods runs counter to standard
offense-defense
predictions.

World

188

WORLD POLITICS

port for the broad range of hypotheses about the dangers generated by
offense advantage.62Many of these findings about the FirstWorld War
have been challenged by Trachtenberg, who is especially critical of as
sertions
sion,

about
and

the role of offense

casts

also

incentives

preventive

doubt

on

in bringing

Offense-Defense
Theory
Realism
Structural

in fueling German
expan
of first-strike
and
importance

dominance
the
about

the war.63

and the Divide within

It is surprising that Jervis's analysis is often not considered part of the

since his discussion
of the security dilemma
family,
as
fundamental
does structural realism?
assumptions
an
of international
seek security and live under the condition

structural-realist

rests on the same

that states

archy.64In part, this is because Jervis did not explicidy frame his analysis
a critical
theory. Nevertheless,
by proposing
improve
a
from
within
structural
Jervis's analysis poses
major
challenge
toWaltz's
in turn
widely
accepted version of the theory, which

in terms of realist
ment,65
realism

transforms

the standard

predictions

of structural

realism.66

This improvement is to shift the focus of the theory from power to

military capability, specifically to the ability to carry out military mis
sions. Focusing
a state's
ability
force effectively,

on
is the
military missions
preferable
approach because
on
to achieve
its goals depends
its ability to use military
that is, to accomplish
the military missions
that are nec

a
essary to achieve security and possibly nonsecurity
goals. For example,
on the
can deter, and if
that
its
state's security depends
forces
probability
to focus on mil
defeat, an adversary's offensive. The decision
necessary
in offense-defense
vari
itary capabilities
essentially
requires bringing
in combination
it is these variables
with power, not power
ables, because
a
to
that
influence
alone,
country's ability
perform military missions.67
62

in 1914," in Robert Jervis et al.,

See also Jack L. Snyder, "Perceptions of the Security Dilemma
and Deterrence
Press, 1985).
(Baltimore: Johns Hopkins University
Psychology
63
Marc Trachtenberg,
Press, 1991), chap. 2,
History and Strategy (Princeton: Princeton University
(Cam
esp. 64-72. See also David Kaiser, Politics and War: European Conflict from Philip II toHitler
Interest? Reflections
Press, 1990), 324; and idem, "Deterrence or National
bridge: Harvard University
on the
in
"Militarization
and Diplomacy
Origins ofWar," Orbis 30 (Spring 1986); David Stevenson,
before 1914," International
Europe
Security 22 (Summer 1997).
64
the similarity, since Waltz
for example, assumes that states
This slighdy overstates
(fn. 4,1979),
assumes that
seek at least security but may have other goals as well (p. 126), while Jervis's formulation
states are pure security seekers. This difference
is not problematic,
however.
65
under the Security Dilemma"
Although
precedes Theory of International Politics by
"Cooperation
a year, much ofWaltz's
in Fred Green
is available in "Theory of International Relations,"
argument
stein and Nelson
Polsby, eds., The Handbook
of Political Science (Reading, Mass: Addison-Wesley,
1975). Waltz
(fn. 4,1979) makes only passing reference to the security dilemma (pp. 186-87).
66
and Van Evera (fn. 5).
See Glaser (fn. 5), which presents additional reasons for modification;
67
vari
Jervis (fn. 1) does not overlook power, he is explicit that both power and offense-defense
ables matter (p. 187).

SECURITYDILEMMA REVISITED

189

Focusing on military mission capabilities identifies possibilities over

looked

and/or

by theories

contradicted

on

power to ex
a
remains
anarchy
in
leads to variation

that focus

solely

state

behavior.
Most
plain
generally,
although
variation
in the offense-defense
variables
constant,
or
to which
states should compete
the extent
cooperate
of war. There

probability

are many

more

and

differences.

specific

in the

For exam

ple, small (weak) countries may be able to defend themselves effectively

balance
against large (powerful) countries if the offense-defense
favors

strongly

defense,
If offense

states.

weapons

as it does

in a world

and defense

of

advanced

are differentiable,

states

nuclear
should

often find that significant military cooperation is feasible. These pre

dictions

and many
tural realism.

others

run counter

to standard

power-based

struc

more
between
structural
realists have become
disagreements
the
"offensive
labels
realists"
and
"defensive
realists"
sharply defined,
have come to be used to
favor the standard,
distinguish
analysts who
from
those
who
believe
that
structural
realism
competitive
predictions
As

variation

greater

predicts

in countries'
to focus

is whether
disagreement
and, therefore, on offense-defense
offensive
realists do not address whether
vantage,

but

power-based

behavior.68

on power
variables.

or on
Since

offense

At

the heart

of the

military
capabilities
on power,
they focus
an ad
has
generally

that competitive
they do conclude
policies flow from their
formulation
of structural
realism. Defensive
realists are

commonly said to believe that security is plentifid, because defense has

the advantage.
is not a central
claim of analysts who
This,
however,
on offense-defense
in
its validity hinges on
and
the
end
variables,
assessments
of the offense-defense
balance. For this reason I
empirical

focus
have

the alternative

suggested

label of "contingent

realism."69

Two Additional
Variables: Greed and Unit-Level
of Motives
Knowledge
In addition to the two offense-defense variables that Jervis highlights,
there

are two additional

dilemma:

security
the adversary's

variables

the extent

unit-level

of the

knowledge

that

influence

of the
the magnitude
extent of
and
the
greed

adversary's
about the state's motives,

in par

ticular, knowledge gleaned by studying the innerworkings of the state

68
See, for example, Snyder (fn. 11), who uses "aggressive" instead of "offensive" (pp. 10-13); and
Sean M. Lynn-Jones
and Steven E. Miller,
E. Brown, Lynn-Jones,
and Miller,
"Preface," inMichael
eds., The Perils ofAnarchy (Cambridge: MIT Press, 1995), xi.
69
See Glaser
(fn. 5).

WORLD POLITICS

190
and

its leaders. These

appropriately,
curity dilemma.

do not

are not
play

structural

variables
and therefore,
quite
in Jervis's discussion
of the se
these variables
is also valuable
it
because

a central

Considering
us
place the security dilemma
helps
ternational
relations
theory.

role

within

still broader

over

debates

in

Greed
In a world
basic
there

of pure

puzzle?that
is competition

security seekers, the security dilemma

even when
states have
compatible,
and conflict.

In a world

with

one

the puzzle
the incompatibility
largely disappears:
a
for competition
provides
straightforward
explanation
not
The
does
become
security dilemma
unimportant
states,

helps
benign
or more

solve a
goals,

greedy
of states' goals
and conflict.70
in aworld

with

states can also be insecure.71

because
states, however,
greedy
greedy
a
true
state
it
is
This
also
is
that
when
knows
that its adversary
said,
more
im
becomes
greedy, maintaining
adequate military
capabilities
to
and
sensitive
the
becomes
less
portant
being
adversary's
insecurity
the importance
of the security dilemma
both
Furthermore,
important.
on how
for explaining
and for avoiding
international
conflict depends
is: all else being equal, the security dilemma
is of
greedy one's adversary
is greedier.
less significance
when
the state's adversary
The
of greedy states has been
importance
that argues that structural realism exaggerates

the focus

of recent work

the importance

of the se

curity dilemma by trying to explain international politics primarily in

terms

states
between
that seek only
security.
that
this
"status
bias"
leaves
neorealism
unable to
argues
quo
to
in modern
"most great-power
behavior
history." According

of

Schweller
explain
Schweller,

the

interaction

its
realist theory more effective
requires broadening
making
to include at least some states that are
motives
states'
about
assumptions
in nonsecurity
that is, that are greedy.72 As I ex
interested
expansion,
are
conclusions
somewhat
because
exaggerated,
plain below, Schweller's
he believes
explanatory

that greedy states rob the security dUemma of all

incorrecdy
in
his basic point is sound: differences
value. Nevertheless,

70
seekers as well as greedy states?
Another
basic puzzle still exists, however why states?security
do not compromise
instead of incurring the cost of fighting; see Fearon (fh. 60, Summer 1995).
71
For related points, see Robert Jervis, "Arms Control,
Stability, and Causes ofWar," Political Science
a
L. Schweller argues otherwise,
holding that when
Quarterly 108 (Summer 1993), 244-45. Randall
state exists there is no security dilemma; see Schweller, "Neorealisrr?s Status-Quo
Bias: What
greedy
Security Dilemma?"
Security Studies 5 (Spring 1996).
72
about the relative importance of greed and inse
Schweller
(fn. 71), quote at 106. Disagreement
formulations
offered by realists, see, for
curity is long standing. For a comparison of the contending
Discord and Collaboration: Essays on International Politics (Baltimore: Johns
example, Arnold Wolfers,
Press, 1962), esp. 83-84.
Hopkins University

SECURITYDILEMMA REVISITED
states'

matter

because

191

can lead states to choose

they
sometimes
choose
cooperation

different poli
a
seeker
will
under
cies;
when,
security
a
state will choose
stan
the same conditions,
The
greedy
competition.
dard structural-realist
this point by mistakenly
pre
analysis obscures
in
that
seekers
will
pursue competitive
dicting
security
general
policies.
realism
the importance
of
suggests
By contrast,
defensive/contingent
a
motives
that under
range of conditions
by explaining
security seekers
goals

should find cooperation to be their preferred policy and, therefore, that

cases

in these

may reflect motives

competition
beyond
security.
that the explanatory
Given
decreases
power of the security dilemma
are
when
adversaries
the
that
follow
from
greedier,
policy prescriptions

imagining conflict driven by the security dilemma become at best less

and at worst

helpfid

potentially

disastrous.

For

example,

arms control

will tend to be riskier because it requires forgoing the opportunity to

to this risk, unilateral
In addition
by competing.
the
state's
motives
becomes
still
designed
security
signal
more
a
more
to
since
dangerous,
greedy adversary may be
likely
exploit
its military
defense
instead
advantages.73 Unilateral
defense?building
communicate

resolve

to

restraint

of offense?is
are more
fensive

more
likely

option

Although

likely
to conclude

makes

to weaken

deterrence,
that the defender's

the risks of war

states
because greedier
lack of a retaliatory
of

acceptable.74

the security dilemma becomes less important in aworld

variables
still play a role in
states, offense-defense
explaining
on
most
the
security seekers,
focusing
interesting
security
concern how offense
dilemma
increases
mutual
explanations
advantage
creates conflicts of interest. In contrast, when
and
fo
insecurity
thereby
on
are
most
offense-defense
variables
cusing
greedy states,
important for
war.
to
of
the
the benefits
At least according
standard argu
explaining

with

greedy
war. When

offense
still increases
that probability
of war, but in
ment,
advantage
are greater and/or be
this instance because
of winning
the prospects
cause its costs are lower, which makes
states harder.
deterring
greedy

Unit-Level
The

Knowledge

of the

State's Motives

second variable to be considered is the adversary's knowledge of

the state's motives.

uncertainty
73
Restraint

The

about whether

is driven
security dilemma
by the adversary's
secu
the state is in fact motivated
purely by

may not be ill-advised, however, since for a greedy insecure adversary the reduction in its
the net effect re
insecurity could still offset the reduction in the state's military capabilities. Assessing
more
and a detailed model of interaction. By contrast, unilateral concessions
quires
specific assumptions
are
as described
always misguided when one is facing a secure greedy state,
by the deterrence modeL
74
see fn. 57.
On the deterrent value of counteroffensive
capabilities,

WORLD POLITICS

192

concerns.
If the adversary were certain that it faced a pure security
rity
then
the state would
face a gready reduced
seeker,
security dilemma.75

The adversary would understand that military buildups reflected the

state's insecurity
rather than its greed. And military
advantages would be
far less
since
the
would
know
that the state did not
adversary
threatening,
want
to attack. As a result, in contrast to the
of the security
predictions
and
would
dilemma,
arming
gaining military
advantages
usually increase
the state's security, not reduce it, and at worst would waste money.
A key assumption
realism guarantees,
in combination
of structural
with

the security dilemma,

others' motives.

tain about

on

not

the

internal

that states will

states do

posits
of other

in which

of the policies

that would

theory
characteristics

uncer

be at least somewhat
aworld

The

states?for
rely
example,
or economic
their type of political
divine
their motives.
In
system?to
states
to
actions
motives.
their
adversaries'
infer
their
stead,
interpret
can
from acting in
The
however,
prevent adversaries
security dilemma,
above, when
ways that would
entirely clarify their motives. As discussed
states face a
security

dilemma,

many

necessary military

capabilities will provide ambiguous

about

For example,

their motives.

tirely differentiated,

when

offense

provide

information

and defense

are not en

the military policies adopted by a pure security

also be taken by an adversary that was motivated

seeker might
partly by
offense
has a large advantage,
seekers
pure security
greed; and, when
to
and
defensive
forces
Under
these
be
unable
afford
may
strategies.
states to
would
conditions,
uncertainty
require
deploy forces
reducing
is their only option for signal
since
this
that are militarily
inadequate,
are
that these policies
motives.
States will often conclude
ing benign
more

dangerous

than alternatives

that do not

reduce

uncertainty

about

theirmotives but that do provide better military capabilities.

If, however,
that structural

states

can

rely
allows,

on

sources

of

information

beyond

those

to reduce
uncertainty
they may be able
The
dilemma.
result can be a
the
security
thereby mitigate
that democ
the argument
shift to more
policies. Consider
cooperative
were
are
to
not
If this
the case,
have greedy motives.
racies
believed
further

realism

and

then amilitary buildup launched by a democracy would be less alarm

ing

to its adversaries

than a similar

buildup

launched

by

an authoritar

75
seem to eliminate the security dilemma,
at first glance this might
this need not be the
Although
war to increase its security; consequendy,
case. A state motivated
the
entirely by security might choose
the security dilemma would re
stated insecurity should lead the adversary to fear it.Thus, eliminating
not only that the state was a pure security seeker but also that the state
quire that the adversary know
are overstated.
some
did not fear it. Consequendy,
points in this paragraph

security

dilemma

revisited

a result, the
faces
democracy
regime.76 As
and interactions
democracies
between
dilemma;77
instead
mocratic
intense
of
peace
competition,
create a severe
conditions
security dilemma.
ian

Unit-level

information

more

that mitigates
policies. This

193
a less
could

severe

security
in a de

result

even when

structural

the security dilemma

occurs when
a state

can also

remains
competitive
of an adversary's motives
but believes
unit-level
information
to
its
that it is a pure security seeker. Com
adversary
appreciate
now become more
since they do not suffer
desirable,
policies

support
uncertain
enables
petitive

from a key shortcoming identified by the security dilemma; that is, they

do not
This

signal malign motives.

line of argument
plays

a central

role

in the "deterrence

model,"

which rejects the security dilemma completely, albeit implicidy, by as

that

suming
Combining

the adversary
knows
the state is a pure security
seeker.78
this with
the assumption
is greedy, the
that the adversary

deterrence model calls for highly competitive policies andwarns against

the dangers of restraint and concessions.
in
For example,
the
describing
war
cold
between
the
United
States
and
the
Soviet
Union,
competition
the deterrence model
held that the Soviets were bent on
for
expansion

entirely greedy reasons and knew that they had nothing to fear from the
United

States.

a
In short, examining
of key nonstructural
variables
high
couple
on certain
the
fact
that
the
role
of
the
dilemma
lights
security
depends
basic theoretical
realism is built on assumptions
Structural
assumptions.
a
that guarantee
central role for the security dilemma.
If, however, we
start from different
or the sources of
about states' motives
assumptions
information
security

about

dilemma

state's motives,
can decrease.

the

importance

and

severity

of the

76
In this spirit, see Bruce Bueno de Mesquita
and David Lalman, War and Reason: Domestic and In
ternational Imperatives (New Haven: Yale University
Press, 1992), chap. 5. A different argument is that
the openness
that characterizes mature democracies
results in domestic debate that cannot be manip
ulated simply to deceive an adversary, and therefore provides valuable information about motives.
See
Andrew Kydd, "Signaling and Structural Realism" (Manuscript
1996); and Kenneth A. Schultz, "Do
mestic Political Competition
and Bargaining
in International Crises" (Ph.D. diss., Stanford University,
1996). Schultz develops this type of argument for crisis interactions.
77
and the Escalation
Arguing
along these lines is James D. Fearon, "Domestic Political Audiences
a
of International Disputes," American Political Science Review 88
(September 1994), 587. If, however,
case and it is not, then we get the kind of
described
democracy believes this to be the
misperception
above: the democracy believes that it doe not face a security dilemma and therefore does not moderate
its building; its adversary, not confident that the democracy
is a pure security seeker, then responds to
the buildup; and the democracy
then increases its estimate of the adversary's greediness because it be
lieves that its adversary knew that a response was unnecessary
its security.
for maintaining
78
See Jervis (fn. 2), chap. 3; and Glaser (fn. 2).

WORLD POLITICS

194

of the Security Dilemma

Criticisms
Theory
Offense-Defense
1:Greedy
Critique
Are the Problem

States, Not

the

Security

and

Dilemma,

As discussed above, the importance of the security dilemma for both

one or more
and
decreases when
of the major
explanation
prediction
a
state.
is
for
Critics
this theo
offer
powers
support
greedy
empirical
war
source
in many
of competition
and
retical observation?that
the
cases has been
not
have
Critics
key
greed,
insecurity.
argued that the
the cold war, certain major
War,
is
liance behavior
better explained
by focusing
their insecurity.

First World

For example, Glynn

of the First World War

criticizes

al
and much
empires,
on states'
on
than
greed

as

those analyses
flawed
fundamentally
on the
offense ad
security dilemma,
arms
that they miss
and the
races, arguing
com
ambitions.
The
military
hegemonic

that focus

vantage, German
insecurity,
the real problem?Germany's

petition could not be slowed by negotiation

cause was

"the very

nature

of German

in the same

because its underlying

ambitions

and of the German

antagonism
spirit that "the political
is
via arms control...
for alleviation

regime."79 Gray argues

the objective
that generates
need
arms control must fail."80 This
is not an isolated
the very reason why
view. Much
of the rationale for competing
with
the Soviet Union
dur

war rested on the basic argument

that Soviet greed, not in
ing the cold
was the root
a
in Jervis's description
of
security,
point captured
problem,
common
occurrence
the deterrence
that
the
model.81
Schweller
argues
of greedy

states
that

explains
is, join

bandwagon,
the gains of expansion.82

Although
is obviously

states
in which
the frequency
of alliances
to
status quo and share
the
together
change

resolving the empirical question raised by these criticisms

beyond

the scope of this article,

two

general

points

deserve

79
Patrick Glynn
focuses on Van Events arguments; see Glynn, Closing Pandoras Box: Arms Races,
at 21. For
Arms Control and the
History of the Cold War (New York: Basic Books, 1992), chap. 1, quote
a more nuanced but at least
see
(fn. 63), chap. 2, esp.
Trachtenberg
partially sympathetic discussion,
49-57.
80
Press of Kansas, 1993), 174. A
Colin S. Gray, Weapons Dont Make War (Lawrence: University
similar theme runs through idem, House of Cards:
Arms Control Must Fail (Ithaca, N.Y.: Cornell
Why
Press, 1992).
University
81
see
Jervis (fn. 2), chap. 3. Douglas
Seay reviews the debate among Soviet specialists;
Seay "What
inTheir Foreign, Military,
Are the Soviets' Objectives
and Arms Control Policies?" in Lynn Eden and
Steven E. Miller,
eds., Nuclear Arguments
(Ithaca, N.Y.: Cornell
82
Randall L. Schweller, "Bandwagoning
for Profit: Bringing
national Security 19 (Summer 1994).

Press, 1989).
University
the Revisionist
State Back

In," Inter

SECURITYDILEMMA REVISITED
to be
emphasized.
adversaries' motives
makers

decision

above, we

noted

arguments
also that the most

that security-dilemma
and offense
in a world with
but
apply
greedy states,
and satisfying aspects of these arguments

remember
to

continue

penetrating
less important.

become

2:The

Critique
the

states may face irreducible

about
First,
uncertainty
or adversaries may have mixed motives;
either way,
as
should strive for a robust mix of policies.83
Second,
should

defense

security

195

Security

dilemma

Dilemma

is logically

Does Not

Exist

flawed

Schweller argues that the logic of the security dilemma is internally

the assumption
of structural realism that states are pure se
no
unless states are un
security dilemma,
curity seekers, there should be
He argues that this is
certain of other's motives/intentions.
problematic,

flawed. Given

not real,"
is then
the security dilemma
ualways apparent,
but instead the re
and because conflicts of interest are then not genuine,
which
"violates realism's most basic tenet."84
sult of misunderstanding,
are off the mark because
to
the
criticisms
These
they fail
appreciate
because

both

core assump
role that uncertainty
plays in structural realism. A
states will be at least
is that under most
conditions
tion of the theory
states are
uncertain
about others' current motives:
somewhat
imagined
as black boxes that
no information
about
internal
differences,
provide
of their international
except for the observable
outputs
policy choices.
central

above, these outputs will often not eliminate

uncertainty
of a
motives.
from the perspective
the adversary's
Therefore,
or
not
is
the
structural theory, this uncertainty
real,
product of
imagined
a
real security dilemma,
As a result, the state faces
misunderstanding.
in the beginning
sections of this article.
for the reasons elaborated

As

discussed

about

of uncertainty
the combination
formulation
In the structural-realist
can
between
and anarchy
pure security seek
generate
incompatibility
ers. States'
but they reflect the incom
interests are genuine,
conflicting
not ends.85
of
means,
patibility

DILEMMA
AREUNINHIBITEDBYTHE SECURITY
SECURITYSEEKERS
Some

structural

ers. Mearsheimer,
relative

power

realists
for

argue
example,

position_The

that security seekers are power maximiz

their
says that states "aim to maximize
reason

is simple:

83
Jervis (fh. 2), 112-13; Glaser (fh. 2), 505-6.
84
Schweller
(fn. 71), 117-20, quotes at 117 and 118.
85
Jervis (fn. 2) deals with precisely this issue (pp. 75-76).

the greater

the military

196

WORLD POLITICS

one has over other states, the more secure it is."86 States maxi
advantage
mize relative power "in order to maintain
the means
for self-defense."87

Although Mearsheimer does not mention the security dilemma, he

is arguing implicidy that it does not exist or at least that it should never
states. Recall
that one of the basic arguments
of the security
can be
is that military
advantages
by making
self-defeating:
can make
one's adversary more
the ad
insecure, a military
advantage
a
to
net
harder
with
the
in the
reduction
effect
deter,
versary
being
as
a
state's security. States should maximize
rule
power
general
only if
never
occurs.
interaction
this self-defeating
More
in certain
specifically,
cases a
to
to com
maximize
its
would
have
power
country attempting
in
offensive
the
alternative
of ac
pete
thereby forgoing
capabilities,
constrain

dilemma

cepting parity in defensive capabilities. A security seekerwould usually

be more
How

secure
are we

in this
parity than engaging
accepting
competition.
to understand Mearsheimer's
assertion? One
possibility

is that he simply rejects the logic of the security dilemma but has not
spelled out why. A second possibility is that he ismaking an unstated
states face conditions
under which
the coop
judgment?that
empirical
erative possibilities
identified by security-dilemma
and offense-defense
are too
be the case when
the security
arguments
dangerous. This might
it turns out that even then arms racing is
is severe, although
a state's best
not
In any event, Mearsheimer
does not
option.
clearly
but notes only that states
claim that states face such a severe condition

dilemma

some offensive
"possess
vides neither deductive

military
capability." In short, Mearsheimer
nor
for his claim.
rationales
empirical

pro

THE SECURITY
DILEMMAISCONSTRUCTED(ORNOT) BY STATES
are not
or na
argues that "security dilemmas
given by anarchy
of intersub
"is a social structure composed
ture";88 a security dilemma
states are so distrustfid
in
which
that
they make
jective understandings
worst-case
each
others'
intentions."89
Because
the
about
assumptions

Wendt

86
"The False Promise of International
Institutions," International Security 19
John J.Mearsheimer,
seeWaltz
structural realists reject this position;
11-12. Other prominent
(fn. 4,
(Winter 1994-95),

1979), 118,126,127.

87
"Back to the Future: Instability in Europe after the Cold War," International
John J.Mearsheimer,
"Realism and Domestic
Politics: A Review
Security 15 (Summer 1990), 12. See also Fareed Zakaria,
Essay," International
Security 17 (Summer 1992), 190-96.
88
of Power Poli
States Make of Ir.The Social Construction
Alexander Wendt,
"Anarchy IsWhat
46 (Spring 1992), quote at 407, and also 401-2; and idem, "Con
tics," International
Organization
1995), 77. For a very different
Politics," International
structing International
Security 20 (Summer
focuses on the offense-defense
constructivist
balance, instead of the security dilemma,
critique, which
see Kier (fn. 10).
89
73.
Wendt
(fn. 88,1995),

197

SECURITYDILEMMA REVISITED
is created

dilemma

security

by states'
reassurance?that

states

interactions,
will avoid

can choose

creating it. Accord

these possibilities
for avoiding
security
of states' interests and security in ways
assume
states
that guarantee
For example,
competition.
they
identify
act on the basis of worst
with
others'
and
should
negatively
security
case
is to
Because
his objective
the logic of anar
assumptions.
clarify
in terms of
frames
his critique
structural
chy, Wendt
third-image

policies?for
example,
to
realists overlook
ing Wendt,
dilemmas
conceive
because
they

structural realists is not rooted in

theory; thus, his disagreement with

over which

disagreements

ismost

level of analysis

important.90

This disagreement between Wendt and realists may appear to be

simply one of terminology: Wendt is using "security dilemma" to de
scribe

the results

of states'

interaction, whereas
Jervis and the literature
to
refer to a situation created by
"security dilemma"

use

he has spawned
the material
conditions

states,

facing

such as geography

technology. By redefining well-established

has created
More
Wendt
petitive

and prevailing

terminology, then,Wendt

confusion.
are the
significant
the extent to which

important
exaggerates
policies

and,

substantive

the extent

therefore,

First,
disagreements.
realism calls for com

structural
to which

it leads

to security

dilemmas, as he defines them. As discussed above, although offensive

realists

system requires states to pursue com

realists who place the greatest
impor
realists?believe
dilemma?defensive/contingent

the international

believe

the

structural

petitive
policies,
tance on the
security
that under a range of conditions
problem
themselves

states

is that

the assumptions
either controversial

should

that Wendt
or

Part

cooperate.

of the

to realism

assigns
flawed.
Instead

are

of seeing
realists envision

logically
with
the
security of others,
negatively
identifying
as indifferent
as
to others'
states as
therefore
and
security, except
egoists
it directly
influences
their own security.91 The
dilemma
then
security
states

explains

why

egoists might
in other
increases

being equal,
s claim
rity.Wendt
on worst-case

see a

that

is, all else

one's own secu

interaction;
positive
increase
states' security

states should base their

believe
policies
is
is
also
this position
(even though
wrong
assumptions
some
to
to
realists and
realists might
attributed
offensive
try

commonly
defend
it). The

that realists

core

logic

of the

security

dilemma

makes

clear

that

90
72.
Wendt
396; see also idem (fn. 88,1995),
(fn. 88,1992),
91
This said, some realists have argued otherwise.
See, for example, Joseph M. Grieco, Cooperation
is because they have conflated
(Ithaca, N.Y.: Cornell University
Press, 1990). This
among Nations
means and ends. See Charles L. Glaser, "Correspondence: When
Cumulative
Relative Gains Matter,"
International Security 21 (Spring 1997). See also Robert
can Political Science Review 87 (March 1993), 127.

Powell,

"Guns, Butter,

and Anarchy," Ameri

198

WORLD POLITICS

can be
since forces that are larger or
self-defeating,
can lead to reduced
than necessary
military
threatening
capabili
ties and/or negative
In fact,
spirals in beliefs about adversaries' motives.
states
Wendt's
how
motives
of
about
the
description
facing uncertainty
to the realist
of others should interact?his
alternative
prescription?is
worst-case

analysis

more

essentially the signaling behavior envisioned by realistswho emphasize

the role of the security dilemma.92
states can
in
Second,
change "the intersubjective
arguing that
constitutes
the
that
their practices,
edge
system" by changing

knowl
Wendt

implicidy rejects the existence of (standard/material) security dilem

can sometimes
is that a security dilemma
it
make
problem
states to
for
their
motives.
I
have
(As
stressed,
risky
signal
benign
on the
this is not always true, since the risks
however,
severity
depend

mas.93 The
too

of the security dilemma, which varies.) Because Wendt

dilemma

believes the se

it
of states, he overlooks
the constraints
curity
can
too
is
about the general ability
impose and consequendy
optimistic
to avoid
of states to change their practices
and use signaling
competi
states may be able to over
tive relations.94 As I have already
explained,
come these constraints
on unit-level
of others'
by relying
knowledge
motives.
is cast

is the creation

is not what Wendt

This, however,
in terms of structural/third-image

Critique

3: Offense-Defense

Theory

of offensive
indistinguishability
undermines the theory

is

arguing,

since his critique

analysis.

Is Flawed

and defensive

weapons

cannot be
and defense
since
argue that offense
distinguished
can
all
be
used
for
both
and
offense
defense.
There
weapons
virtually
cannot be measured
the balance
and the policy
fore, they continue,

Critics

guidance
sheimer

of offense-defense
argues,

for example,

cannot be
Mear
theory
implemented.95
that determining
the offense-defense
bal

ance is problematic because "it is very difficult to distinguish between

offensive

and defensive

weapons."96

92
Wendt
404-5.
(fn. 88,1992),
93
Ibid., 407.
94
turns to the existence of
406. Wendt
See especially Wendt
(fn. 88,1992),
predatory (greedy) states
a
to
realm (pp. 407-9). However,
the weaker structural-re
explain how anarchy becomes
competitive
in combination with a security dilemma,
is sufficient.
alist assumption of uncertainty
about motives,
95
This section draws on Kaufmann
and Glaser
addresses a number of additional
(fn. 25), which
and Its Critics," Security Studies 4
"Offense-Defense
criticisms, as does Sean M. Lynn-Jones,
Theory
(Summer 1995). Jack S. Levy provides some of the early criticisms; see Levy, "The Offensive/Defen
sive Balance of Military Technology:
28 (June 1984).
Quarterly
96
Mearsheimer
(fn. 86), 23.

A Theoretical

and Historical

Analysis,"

International

Studies

SECURITYDILEMMA REVISITED
line of criticism,

This

however,

reflects

two fundamental

199
misunder

standings. First, whether offense and defense are distinguishable does

not affect our
to assess the offense-defense
balance. To access the
ability
we
start
that
offense-defense
attacker
and de
balance,
by assuming
fender deploy
the weapons
that best enable them to achieve
their re
missions.

spective
defender

in the
may or may not result
of the same types of weapons.
the balance
then requires
measuring
This

some

attacker

and

Either

way,
deploying
a
given these forces,
performing
net
kind of
the ability of the attacker's forces to
assessment?analyzing
defeat the defender's
forces.
the size of the attacker's forces to
Adjusting
determine
how large they must be to succeed provides
the information
cost
to
that determines
the
ratio of offense
defense.97
in

Second,

on whether

focusing

a weapon

can be used

by both

the

attacker and the defender, critics misunderstand distinguishability; they

should

focus

balance.

For

instead

on how

example,

the weapon
influences
the offense-defense
concludes
that "weapons may
Huntington

be

usefully differentiated in a variety of ways, but the offense/defense distinc

states cannot
tion is not one of them."98
Consequendy,
signal by forgoing
arms control to
certain types of weapons,
and they cannot use qualitative
shift the offense-defense
critics are mistaken,
balance. These
however,

because distinguishability does not depend onwhether both attacker and

defender would deploy the weapon. Distinguishability
should be de
fined

by comparative

net assessment,

that

is, by comparing

the offense

defense balance when both sides deploy the weapon with the balance
when neither deploys it. If deploying theweapon shifts the balance to
ward

offense

(defense),

then

the weapon

can be classified

as offensive

(defensive), and states will be able to implement the policy prescrip

tions that depend on the distinguishability of offense and defense.
PERCEPTIONS OF THE BALANCE, NOT THE BALANCE, ARE KEY
not the balance
of the offense-defense,
argue that perceptions
war
arms
assum
determine
states'
decisions
for
and
itself,
racing. Even
states have the
measure
to
that
and skill required
the of
ing
knowledge

Critics

or other biases
balance accurately, political, psychological,
can still
serious
the "objective"
generate
misperceptions.
Consequendy,
is not useful for
balance
offense-defense
states'
behavior.99
predicting
fense-defense

97
For a different response, see also Lynn-Jones
(fn. 95), 674-77.
98
Samuel P. Huntington,
"U.S. Defense
Strategy: The Strategic Innovations of the Reagan Years,"
in Joseph Kruzel, ta., American Defense Annual,
1987-1988
Books,
(Lexington, Mass.: Lexington
28.
1987), 36. See also Gray (fn. 80,1993),
99
See, for example, Levy (fn. 95), 222. For related discussion of this criticism, see Lynn-Jones
(fn.
95), 677-82.

WORLD POLITICS

200

this criticism is valid, it does not reduce the value of of

Although

the
offense-defense
theories,
theory. Like all structural
are mediated
structure
must
of
the
effects
that
ory
through
recognize
or not. Nevertheless,
accurate
whether
offense
states' perceptions,
to enable us to
defense
predict states' behavior given
theory is necessary
In addition,
balance.
of the offense-defense
their actual perception
fense-defense

when

misperceptions
as a,baseline
necessary

the offense-defense

do occur,

against
and their consequences.

misperceptions

Research
article

This

to assess

which

leave

should

remains

balance
the seriousness

of the

Agenda
under

that "Cooperation

little doubt

the Se

curity Dilemma" has both established the importance of key arguments

about

the security

large and

and offense-defense

dilemma

impressive

literature. What

work

a
theory and spawned
remains to be done? I sug

gest beginning at leastwith the following areas.

versus

Security

variety
offense-defense

of angles,

Greed
a

From

we

have

seen

and
the security dilemma
when
and insightful
important
that

are less
arguments
secure
states,
greedy states,
greedy
especially
that helps determine
tional conflict. Research

are at the heart

of interna

and inten
the frequency
states
In
valuable.
be
therefore
would
addition,
quite
sity of greedy
concern
for theory building:
about greedy states poses a basic challenge
a reasonable
theories
that deal adequately
be to develop
goal would
as well
states'
of
and varying
the mixture
with
motives?greed
intensity
as
security.

Empirical
Given

Testing

of Offense-Defense

Hypotheses

their importance, the fiill range of offense-defense

warrants

hypotheses

can be made
contributions
testing. Valuable
empirical
states assess their
in
whether
(1) that examines
security

further

by research

terms of power ormilitary capability, (2) that explores how the balance
be measured,
cases.
to
specific
should

Caution
As

further

and

(3) that systematically

applies

these measures

versus Competition
empirical

testing

proceeds,

the time

is ripe to explore more

thoroughly the deductive strength of offense-defense hypotheses.Why

do countervailing

considerations

not moderate

the

impact

of offense

SECURITYDILEMMA REVISITED
advantage?

For

example,

if offense

advantage

means

201
as

that attackers

well as defenders could lose big in amajor war, then why does offense
advantage

not

sometimes

states more

make

arms races become more

likely?100 If
take greater risks to stop them? And

Rational
The
on

versus Biased

and war

then why

do states not

less

States

and predictive
explanatory
to which
states
the extent

and political biases. Maybe

diction
that relations will

dangerous,
so on.

cautious

value
suffer

of the security

dilemma

depends

from

bureaucratic,
psychological,
most
the
pre
important,
theory's optimistic
be cooperative
when
the security dilemma
is

mild depends on states accurately perceiving the conditions they face. If

are inclined

to exaggerate

to
the advantages
of offense,
ignore
or
to
overlook
others'
restraint,
security dilemma,
then the opportunity
created by objective
conditions
will be squan
a
amount
dered. Although
substantial
of research has focused on mis
states

that others

perceptions,
important

face

answers

to critical

have misperceptions

remain wide
open. How
questions
been in fueling competition
and war?

Are states likely to be better at avoiding flawed policies in the future?

Given these topics, the already large body of work that builds on
and offense-defense
Jervis's security-dilemma
to grow in size and
continue
importance.

100
See fn. 60 for work

that has already raised this possibility.

arguments

is likely

to