The Security Dilemma in Alliance Politics

Transcription

Trustees of Princeton University
The Security Dilemma in Alliance Politics
Author(s): Glenn H. Snyder
Source: World Politics, Vol. 36, No. 4 (Jul., 1984), pp. 461-495
Published by: Cambridge University Press
Stable URL: http://www.jstor.org/stable/2010183 .
Accessed: 05/11/2014 22:57
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 [email protected].
.
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
This content downloaded from 163.239.216.77 on Wed, 5 Nov 2014 22:57:51 PM
All use subject to JSTOR Terms and Conditions
THE SECURITY DILEMMA IN
ALLIANCE POLITICS
By GLENN H. SNYDER*
THE
SECURITY
DILEMMA
A
CENTRAL concept in internationalrelationstheory,but stillone
of the most under-studiedempirically,is that of the "securitydilemma."' The term is generallyused to denote the self-defeatingaspect
of the quest for securityin an anarchic system.The theorysays that
even when no state has any desire to attack others,none can be sure
that others' intentionsare peaceful,or will remain so; hence each must
accumulate power for defense. Since no state can know that the power
accumulation of othersis defensivelymotivatedonly,each must assume
that it might be intended for attack. Consequently,each party'spower
incrementsare matched by the others,and all wind up with no more
securitythan when the vicious cyclebegan, along withthe costsincurred
in having acquired and having to maintain theirpower.
States accumulate power in many ways; the most prominentmethods
are by armament, territorialaggrandizement,and alliance formation.
The "supergame" of internationalsecuritymay thus be divided, for
analyticalpurposes, into threesubgames: the armamentsgame, the adversarygame, and thealliance game. Typically,discussionsofthe security
dilemma are illustratedby the armamentsgame: the arms race is seen
as the epitome of competitionfor illusorysecurity.Sometimes,at least
implicitly,the securitydilemma is seen operatingin the adversarygame
(competitionother than armaments),as in explanationsthat ascribe the
cold war to the United States and the Soviet Union misperceivingeach
other's "defensive" actions in Europe as "aggressive." Little attention
has so far been paid to the securitydilemma dynamics of the alliance
game. The presentessay is intendedto fillthisgap, and to explore some
* I wish to thank Robert Art,
David Goldfischer,Robert Jervis,Robert Osgood, Dean
Pruitt,JackSnyder,and Kenneth Waltz, forhelpfulcomments;and the Woodrow Wilson
InternationalCenter for Scholars and the Instituteof War and Peace Studies, Columbia
University,forfinancialsupportand officespace.
I The best recenttreatmentis RobertJervis,
"Cooperation under the SecurityDilemma,"
WorldPolitics30 (JanuaryI978), i67-2 I4. The conceptof thesecuritydilemma was originated
by JohnHerz, in his PoliticalRealismand PoliticalIdealism(Chicago: Universityof Chicago
Press, I95I).
462
WORLD POLITICS
interactionsbetween the alliance game and the adversarygame in multipolar and bipolar systems.
ALLIANCE
FORMATION
THE
PRIMARY
IN A MULTIPOLAR
ALLIANCE
SYSTEM:
DILEMMA
The securitydilemma in the alliance game has two phases: primary
and secondary.The primaryphase occurs during the processof alliance
formation,the secondaryone afteralliances have formed.
In a multipolarsystem(such as the one thatexisted before I945), the
primaryalliance dilemma among the major states follows the logic of
an N-person prisoner'sdilemma. Each state has two options: seek allies
or abstain from alliances. If all states are about equally strongand are
interestedonly in security,all are fairlywell offif all abstain,since each
has moderate securityagainst individual others,while alliances involve
various costs,such as reduced freedomof action,commitmentsto defend
the interestsof others,and so forth.Alliances will form,however, for
two reasons: (I) some states may not be satisfiedwith only moderate
security,and theycan increaseit substantiallyby allyingifothersabstain;
(2) some states,fearingthat otherswill not abstain,will ally in order to
avoid isolation or to preclude the partner from allying against them.
Once an alliance forms,a counter-alliancenecessarilyfollows,since there
is no way of knowing thatthe firstalliance is intendedonly fordefensive
purposes. The eventual resultis the divisionof the systeminto two rival
coalitions. This outcome is worse than all-around abstention because
each state has incurred the risks and burdens of alliance with little
improvementin its security.2
Figure I portraysthe primaryalliance securitydilemma. Although
it is cast in two-person form,it is understood that for player A, the
other player,B, means "all otherplayers,"and vice versa. The numbers
in the cells are ordinal, ranked from 4 (best) to I (worst). The first
number in each cell representsA's payoff,the second B's. The logical
This is, of course, an idealized model based on certain assumptionsfrom which the
empirical world will deviate more or less, fromtime to time. The basic assumptionsare
that: (i) no state is aggressive,but none can know the intentionsof others; (2) the states
are roughlyequal in militarystrength;and (3) militarytechnologyis such that thereis no
time to forma successfuldefensealliance afterwar begins. Uncertaintyabout the aims of
others is inherentin structuralanarchy. If a state clearly reveals itselfas expansionist,
as in the prisoners'dilemma
however,thealliance thatformsagainstit is not "self-defeating"
(securitydilemma) model. Or, if some statesare weaker than others,theirmotivesto ally
will be differentfromthe incentivesof the prisoner'sdilemma. The thirdassumptionhas
been valid since about i870. Before then (when the pace of warfare was slower), the
compulsion to ally in peacetime was much weaker than suggestedby the model. Despite
these qualificationsand possiblyothers,the model does capturesome essentialdynamicsof
multipolaralliance formationbetween i870 and I939.
2
ALLIANCE
FIGURE
THE
PRIMARY
DILEMMA
SECURITY
ALLIANCE
463
I
SECURITY
DILEMMA
B
ABSTAIN
ABSTAIN
All-around
abstention
3,3
A
ALLY
A's coalition
forms
B isolated
4,1
ALLY
B's coalition
forms
A isolated
"4
Two rival
coalitions
2, 2
forall players.The
outcome,two rivalcoalitions,is the second-worst
allsecond-best,
the
best,formingan alliancewhileothersdo not,and
becauseofuncertainty
cannotbe obtained,primarily
aroundabstention,
need to guard
about the intentionsof othersand the overwhelming
againsttheworstoutcome,thatof isolation.
This modelpredictsonlythatallianceswill form.It does notpredict
risks,and costsof an
who will align withwhom,or how thebenefits,
alliancewillbe dividedamongitsmembers.These matterstheoretically
are decidedby a processof bargainingin whichthe statescompetein
Each state
sharesofthealliance's"payoff."
each otherattractive
offering
has two principalaims in the bargaining:to be in the mostpowerful
These
coalition,and to maximizeitsshareofthealliance'snetbenefits.3
of the statein the alliancegame. If thesewere the
are the "interests"
at stake,the alliancebargainingprocesswould be comonlyinterests
pletelyindeterminatethatis, each statewould be equallyeligibleas
theallyor theadversaryof everyotherstate.
is reduced,thoughnot
In therealworld,however,theindeterminacy
from
thealliancegame
whichexistapart
eliminated,
byotherinterests,
and which predisposestatesto align with certainothersand against
3 "Being in" the most powerfulcoalition does not necessarilymean that states join the
most powerfulcoalition that is already in existence.Indeed, theywill more likelyjoin the
weaker one which then becomes the most powerful as a consequence of their joining,
because this gives them leverage to bargain for a maximum share of the alliance's payoff.
Thus the logic of N-person game theoryis consistentwith Waltz's argumentthat states
"balance" ratherthan "bandwagon." See Kenneth N. Waltz, TheoryofInternational
Politics
(Reading, Mass.: Addison Wesley, I979), I25-26.
464
WORLD
POLITICS
others. Here we must distinguishbetween "general" and "particular"
interests.General interestsstemfromtheanarchicstructureof the system
and the geographic position of the state. They include, for instance,a
state'sinterestin defendinga close neighbor,or in expansion to enhance
its security,or even more generally,in preservinga balance of power
in the system.Typical examples of general interestswould be England's
traditionalinterestin preservingthe independenceof the Low Countries
and in maintaininga balance of power on the Continent.Such interests
are "general" because they do not involve conflictsover specificissues
with specificother states,but will be defended or acted upon against all
comers.Since theyare valued chieflyfortheirpower and securitycontent,
I will occasionally referto them as "strategic" interests.Such general
strategicinterestsintroduceonly a modest amount of determinacyinto
the basic N-person model.
The indeterminacyis furtherreduced by the "particular" interestsof
states, which bring them into conflictor affinitywith specific other
states.These conflictsand commonalitiesmay have some power content
or theymay stem fromideological, ethnic,economic,or prestigevalues.
The Franco-English conflictsover Egypt in the i9th centuryand the
Austro-German ethnic affinityare examples.
Particular conflictsor affinitiesof interestestablisha tacit patternof
alignment,prior to or apart fromany overt alliance negotiations.That
is, stateswill expect to be supportedin some degree by those with whom
theyshare interestsand to be opposed by those with whom theyare in
conflict.To illustrate:in the decade of the i870s, Germany had a serious
conflictwith France over Alsace-Lorraine, but no significantconflicts
with any of the other major powers. Russia had conflictswith Austria
over the Balkans, and withEngland over theStraitsquestion and colonial
issues fromthe Middle East to China. England was on good termswith
Germany,Austria,and Italy,but was in conflictwith Russia and France
over colonial issues. Italy had territorialconflictswith Austria in Europe
and with France in Africa. The autocratic governmentsof Germany,
Russia, and Austria stood on ideological common ground, but were
ideologically antagonistictoward the three Western democracies. The
alignmentsimplied by these conflictsand affinitieswere: (i) on grounds
of territorialinterest: England and Austria versus Russia; England,
Germany,and Italy versus France; and France and Russia versus England; (2) on ideological grounds: Germany,Austria, and Russia versus
England, France, and Italy.
Such conflictsand alignments of interestand ideology establish a
ALLIANCE SECURITY DILEMMA
465
backgroundof relationships
againstwhichtheovertalliancebargaining
processtakesplace,and whichaffectthatprocessconsiderably,
predisposingthe systemtowardcertainalliancesand againstothers.These
relationships
may foreclosesome combinations
if the conflictis severe
enough;in othercases,theabsenceof conflict
betweensomepairsmay
make them naturalallies. More likely,however,theseconflictsand
affinities
will narrowtherangeof indeterminacy
ratherthaneliminate
it. Naturalpartnersmayfailto allybecauseone of themoverestimates
the other'sconflictswith thirdpartiesand triesto drive too hard a
bargain,as Germanydid in negotiations
withEnglandin i899 and i9oi.
And naturalopponentsmaybe able toovercometheirconflicts,
as France
of
and England did in I904. Technically,conflicts
and commonalities
particularinterestenterinto the bargainingprocessby reducingor
increasing
thetotalvalue,or "payoff,"
ofcertainalliances,thusreducing
or increasingthe likelihoodthattheywill form.For example,a state
withwhichone has a conflict
will appearas a morelikelyopponentin
war than otherstates;hence,an allianceagainstit will yieldgreater
valuethanan alliancewithit;thelatterwouldrequirea priorsettlement
of theconflict,
interests.
incurring
costsin theformof compromised
Thus, althoughan alliancebetweenFrance and Germanywas not
logicallyout of the questionbetweeni870 and I9I4, it was extremely
of theAlsace-Lorraine
unlikely:it requireda settlement
issue,a compromisethatwould have costone or bothpartiestoo muchcompared
to the deals theycould negotiatewithothers.Francecould get Russia
as a partner
at theminorcostofswallowingsomeideologicalrepugnance,
and she could get England by givingup a positionin Egyptthatwas
valuableonlyforprestigereasonsand could hardlybe maintainedfor
long.
The choiceof alliesis also influenced
bytheinternalpoliticalconfigurationsof statesapart fromthe generalideologicalpreferences
just
mentioned.For example,the Anglo-FrenchEntenteof I904 probably
would not have occurred,and certainly
would not have developedso
soon intoa quasi-allianceagainstGermany,if the Radical ratherthan
theImperialistwingof theBritishLiberalpartyhad been in chargeof
theforeignpolicymakingpostsin theCabinet.
To summarize:in a multipolarsystemthereis a generalincentiveto
allywithsomeotherstateor states,followingthelogicof theN-person
prisoner'sdilemma,or security
dilemma,thatis generatedbythestructureof the system.Who alignswithwhom resultsfroma bargaining
processthatis theoretically
indeterminate.
The indeterminacy
is reduced,
466
WORLD POLITICS
though not eliminated, by the prior interests,conflicts,and affinities
between states and their internalpolitical make-up.4
AFTER
THE
ALIGNMENTS
SECONDARY
ALLIANCE
FORM:
DILEMMA
Once alliances have begun to form, the alliance securitydilemma
takes on a differentcharacter.That is, having already "defected" in the
primarydilemma by choosing to make alliances, states move into the
second phase of the alliance dilemma, in which their choices are no
longer whetherto ally or not, but how firmlyto commit themselvesto
the proto-partnerand how much supportto give thatpartnerin specific
conflictinteractionswith the adversary.The horns of this secondary
dilemma may be described by the traditionallabels "cooperate" (C) and
"defect" (D), where cooperation means a stronggeneral commitment
and full support in specificadversaryconflicts,and defectionmeans a
weak commitmentand no supportin conflictswith the adversary.The
secondary alliance dilemma may or may not be a prisoner'sdilemma.
(Henceforthin thisdiscussion,the terms"alliance game" and "dilemma"
will referto the secondarygame ratherthan the primaryone.)
Each horn of the dilemma has both prospectivegood and prospective
bad consequences; and the "goods" and "bads" foreach alternativetend
to be the obverse of those of the other.In the alliance securitydilemma,
the principal "bads" are "abandonment" and "entrapment,"and the
principal "goods" are a reduction in the risks of being abandoned or
entrapped by the ally.5
In a multipolar system,alliances are never absolutelyfirm,whatever
the textof the writtenagreement;therefore,the fearof being abandoned
by one's ally is ever-present.Abandonment, in general, is "defection,"
but it may take a varietyof specificforms: the ally may realign with
the opponent; he may merelyde-align,abrogatingthe alliance contract;
he may fail to make good on his explicitcommitments;or he may fail
to provide supportin contingencieswhere supportis expected. (In both
of the lattertwo variants,thealliance remainsintact,but theexpectations
of support which underlie it are weakened.) Suspicion that the ally is
4Despite the importanceof internalpolitics,this referencewill be the only one in this
essay.For reasonsof theoreticalparsimonyand space limitations,theanalysisis based entirely
on what in recentlypopular academic terminologyis called the "rationalactor model," the
actors being states.
5The conceptsof abandonmentand entrapmentwere firstposited,I believe,by Michael
Mandelbaum, in The Nuclear Revolution:InternationalPoliticsBeforeand AfterHiroshima
(New York: Cambridge UniversityPress, i98i), chap. 6.
467
consideringrealignment
maygeneratean incentiveto realignpreemptively.
Entrapment means being dragged into a conflictover an ally's intereststhat one does not share, or shares only partially.The interestsof
allies are generallynot identical; to the extenttheyare shared,theymay
be valued in differentdegree. Entrapmentoccurs when one values the
preservationof the alliance more than the cost of fightingfor the ally's
interests.It is more likely to occur if the ally becomes intransigentin
disputeswithopponentsbecause of his confidencein one's support.Thus,
the greater one's dependence on the alliance and the strongerone's
commitmentto the ally, the higher the risk of entrapment.The risk
also varies with the ally's inherentdegree of recklessnessor aggressiveness.
The risks of abandonment and entrapmenttend to vary inversely:
reducing one tends to increase the other.Thus a "C" strategyof strong
commitmentto an ally reduces the risk of abandonment by reducing
his fear of abandonment; he is discouraged fromdefectingby his confidence in one's support. But this very support may encourage him to
excessiveboldness in disputesor criseswith the adversary,thus exposing
one to the risk of a war that one would not wish to fight.Conversely,
a "D" strategyof weak or vague commitment,or a record of failingto
support the ally in specificconflicts,tends to restrainthe ally and to
reduce the risk of entrapment;but it also increases the risk of abandonment by castingdoubt on one's loyalty,hence devaluing the alliance
for the ally. Thus, the resolutionof the alliance securitydilemma the
choice of strategy requires chieflya comparisonand trade-offbetween
the costs and risks of abandonment and entrapment.
There are certainother"goods" and "bads" thatenterinto thealliance
securitydilemma. A strategyof strong commitmentand support will
have the undesired effectof reducingone's bargainingleverage over the
ally.If he knows he can counton being supported,he is less influenceable.
Conversely,bargainingpower over the ally is enhanced to the extenthe
doubts one's commitmentbecause one can then make credible threats
of nonsupport.Alliance bargaining considerationsthus tend to favor a
strategyof weak or ambiguous commitment-a "D" strategyin the
alliance game. (Note that the opposite is the case in the adversarygame
where firmcommitmentsto defend one's intereststend to strengthen
bargaining power vis-a-visthe opponent.)
Another negative effectof strong commitmentis that it tends to
forecloseone's own options of realignment.Despite the general com-
468
WORLD POLITICS
pulsion to align in a multipolarsystem,statesusually want to keep their
commitmentstentativeor vague as long as possible-both to preserve
opportunitiesfor shiftingpartnersin case the presentone turnsout to
be unsatisfactory
and to maximize bargainingleverage over the current
partner by showing that they have alternatives.A strategyof weak
commitmenthas the desirableeffectof keeping alignmentoptionsopen.
Finally,a strongcommitmentto theally tendsto solidifytheadversary
alliance by increasing the degree of threat to it. A weak or tentative
commitmentreduces this effectand may even weaken or divide the
opposing alliance by preserving,for statesin that alliance, the apparent
option of realigningwith oneself.
INTERACTION
BETWEEN
ALLIANCE
AND
ADVERSARY
GAMES
Up to now, I have been discussing the alliance game or dilemma
more or less independentlyof adversaries. But allies are dealing with
theiradversariesat the same time theyare dealing with each other.The
alliance and adversarygames proceed simultaneouslyand complement
each other in various ways. Strategiesand tacticsin the alliance game
will have directeffectsin thatgame, but also side-effects
in the adversary
game-and vice versa. Strategychoices in either game must therefore
take account of both kinds of effects.
Table I presentsa compositesecuritydilemma thatcombinesthe two
games. Each of the strategypairs, I and II, shows an alliance strategy,
togetherwith its complementarystrategyin the adversarygame. (The
cooperative"C" strategiesand thedefecting"D" strategiesare shorthand
for a wide range of empirical variationsthat cannot be explored here.)
The two columns show the possible directconsequences of alliance and
adversarystrategiesin their respectivegames.
The column labeled "alliance game" simplysummarizes the previous
discussion. The column labeled "adversary game" shows the possible
"good" and "bad" effectsof either conciliationor firmnesstoward an
adversarywhen one is interestedonly in maintainingthe general status
quo. The securitydilemma arises from the state's uncertaintywhether
its adversaryhas far-reachingexpansionistaims or, like itself,is interested essentiallyin the preservation,or limitedmodification,of the status
quo. If the opponent is expansionist,a policy of firmnesspromises the
desirable effectsof deterringhim and enhancing one's own reputation
for resolve. However, if he is basically oriented toward the status quo,
a tough stance may provoke him, increase tension,and induce an "in-
ALLIANCE
SECURITY
TABLE
THE
COMPOSITE
SECURITY
469
DILEMMA
I
DILEMMA
IN A MULTIPOLAR
SYSTEM
PossibleConsequences
Strategies
Alliancegame
Adversary
game
GOODS
I.
I
ALLIANCE
2.
reduce risk of
abandonment
Enhance reputation
forloyalty
GOODS
I.
2.
Deter,or prevail
over, adversary
Enhance reputation
for resolve
C:
Support,
strengthenpcommitment
ADVERSARY
Reassureally,
D:
Stand firm.
BADS
Increase risk of
entrapment
2. Reduce bargaining
power over ally
3. Foreclose realignment
option
4. Solidifyadversary's
alliance
I.
BADS
I.
Provoke adversary;
increasetension;
insecurityspiral
GOODS
I.
GOODS
I.
Resolve conflict;
reduce tension
Increase bargaining
power over ally
3. Preserverealignment
option
II
ALLIANCE
Restrainally, reduce
risk of entrapment
2.
D:
Withhold support,
weaken commitment
4. Divide adversary's
alliance
BADS
ADVERSARY
C:
Conciliate
I.
2.
Increase risk of
abandonment
Reduce reputation
forloyalty
BADS
I.
2.
Encourage adversary
to stand firmer
Reduce reputation
forresolve
securityspiral" that is, a vicious circle of "unnecessary"power competitionbecause the adversaryinterpretsone's own firmnessas aggressiveness toward him. Conversely, a conciliatorypolicy may have the
desirable effectof resolving conflictand reducing tension with an essentiallynonaggressiveadversary; on the other hand, if the opponent
470
WORLD POLITICS
has expansionistgoals, conciliationmay encourage him to make further
demands in the belief that one lacks resolve.6
Strategiesin each game can have desirable or undesirableside effects
in the other.In making choices,a statemust thereforecalculate the sum
of directeffectsand side effects.To save space, I will discuss here only
the side effectson the alliance game of strategiesin the adversarygame,
and omit the obverseside effects.An adversary"D" strategyof firmness,
resistance,or coercion will tend to reassure an ally who doubts one's
loyalty,and reduce the risk of the ally's defectionor realignment.Opposed to this "good" are several "bads." Firmness toward the adversary
increases the risk of entrapmentby the ally, as the ally becomes intransigentthroughhis confidencein one's support.Firmnessin the adversary
game reducesbargainingpower over theallyin thealliance game because
it reduces the credibilityof a threatto withholdsupport.A tough stance
toward the adversaryalso tends to close off the option of realignment
with him. Toughness will also cause the adversaryto move closer to his
own allies, thus solidifyinghis alliance.
6 The adversarydilemma presentedhere is a secondarysecuritydilemma, analogous to
the secondaryalliance dilemma. In both cases, the primarysecuritydilemma, which is a
prisoner'sdilemma, has already been resolvedby mutual defection.That is, alliances have
formedand adversaries have adopted a general posture of power/security
rivalryin the
"DD" cell of the primarygame, which we mightalso call the international"supergame."
Once adversariesare in this cell they may be able to reduce their conflict;on the other
hand, theymay sink deeper into conflictand competition.That is, having protectedthemselves against the worst,theyare now able to consider whetherthey might not improve
their situationby conciliatingthe opponent-although they must also guard against exploitationof such cooperation.Whether conciliationor continued (or greater)firmnessis
the betterpolicy for any state will depend on its adversary'spreferences-which the state
does not know, although presumablyit knows its own. The adversarydilemma in Table
I simplyassumes theextremes-the opponentis eitherstatus-quo-oriented
or expansionisteven thoughhe may have mixed motivesor be expansionistin different
degree in different
situations.Formally speaking, adversariesplay a series of sub-games within the general
contextof supergame "DD." Some of these may be prisoner'sdilemmas,but some may be
othergames thatare eithermoreor less conflictualthantheprisoner'sdilemma.The dilemma
for the state,then,arises fromits uncertaintyas to what game is being played, which in
turnstemsfromitsuncertainty
about theadversary'spreferences.
Choosing the beststrategy
thusrequiresan estimateof the opponent'spreferencerankingsforpossibleoutcomes.Space
forbidsa discussionof the various possiblesub-games:in Table i, theyare implicitly"Stag
Hunt" (when the opponent prefersthe statusquo) and a variantof "Chicken" (when the
opponent is expansionistbut preferspeace over expansion by war). For a comprehensive
treatment,see Glenn H. Snyder and Paul Diesing, ConflictAmong Nations (Princeton:
PrincetonUniversityPress, I977), chap. 2.
The secondaryadversarydilemma is similar to the dichotomybetween the axioms of
deterrencetheoryand thoseof the "spiral model" as describedby RobertJervisin Perception
and Misperception
in International
Politics(Princeton:PrincetonUniversityPress, I976), chap.
3. Jervismakes the point that the choice between deterrenceand spiral-modelaxiomsbetweenfirmnessand accommodation-depends essentiallyon one's estimateof the adversary'sultimateaims. Jervisdoes not apply the term"securitydilemma" to thisbroad choice,
but uses it in its traditionalsense as an explanationof why the search for securityamong
status-quostatesmay be self-defeating-i.e.,as the structuralbasis of the spiral model.
ALLIANCE
SECURITY
DILEMMA
471
A "C" strategyof conciliatingthe adversarywill have the desirable
side effectof restrainingthe ally, thus reducing the risk of entrapment.
The ally, observing one's improving relationswith the opponent, will
have less confidencethat one will stand four-squarebehind him in a
crisis; consequently,he will be more cautious in his own dealings with
the opponent. He may even interpretthe improved relationswith the
adversary as a sign that one is consideringrealignment;he may then
become more amenable in order to discourage one's defection.Conciliating the opponent also keeps open one's option of realignmentwith
him, which is desirable forits own sake as well as forthe enhancement
of one's bargaining leverage over the presentpartner.Accommodating
an adversarymay also weaken his alliance, as his partnersbegin to doubt
his loyaltyand seek alternativepartners.
The most undesirable side effectof conciliatingthe adversaryis that
it entails the risk of abandonment by the ally. His fear that one is
contemplatingrealignmentmay induce him, not to try to discourage
thisby becomingmore accommodative,as suggestedabove, but to realign
preemptivelyor at least move closer to the opponent. This risk is an
importantconstrainton conciliationbetween adversaries,possibly offsettingthe benefitsfromgreatercooperationin the adversarygame.7
THE
DETERMINANTS
OF
CHOICE
What determineschoices in the alliance securitydilemma? Thus far,
we have suggested only a superficialanswer to this question: choice
involves estimatesof, and trade-offsamong, the various benefits,costs,
and riskslisted in Table I. But what determinesthe magnitude of these
values and consequentlythe severityof the dilemma?
Probably the most importantdeterminantis the relativedependence
of the partnerson the alliance-how much theyneed each other'saidand their perceptionsof each other's dependence. Thus, the more dependent a state is, and/orthe less dependent the ally appears to be, the
7 There are some interesting
analogies betweenthe dynamicsof the two games. A strategy
of conciliationin the adversarygame may produce a "fallingdomino" effect:the opponent
interpretsone's overtureas weakness and pushes harderon both presentand futureissues.
The alliance analogue is the entrapmenteffect.As an opponentmay be emboldened if he
is appeased, the ally may become more intransigentand aggressiveif he is supported.The
alliance analogue to deterrenceof the opponentis restraintof the ally. Deterrenceinvolves
a threatof forceagainst an adversary;restraintmay be accomplishedby threateningnot to
use forcein supportof the ally. In the alliance game, strengthening
one's commitmentto
the ally tends to forecloseone's alternativealliance options; in the adversarygame, firming
up one's commitmentagainst the opponentmay foreclosecompromisesettlementoptions.
Bargainingpower over the ally is enhanced by a weak, ambiguous commitment;leverage
over the adversaryis strengthenedby a firm,explicitone.
472
WORLD POLITICS
more likely it is that the costs and risksof abandonment will outweigh
the costs and risks of entrapment.Dependence is compounded of (i) a
state's need for assistance in war as a functionof the extent to which
its militarycapability falls short of its potential adversary'scapability;
(2) its partner'scapacityto supplythe assistance(the greaterthe partner's
strength,the more one is dependent on him, up to the point where the
combined strengthprovides sufficientsecurity);(3) the state's degree of
conflictand tension with the adversary (the greater the conflictand
tension,the more likely one will have to call on the partnerfor help);8
and (4) the state's realignment alternatives(the more numerous the
alternatives,and the more satisfactory
theyare, the less the dependence
on the presentpartner).These factorswill of course change over time;
some may change in opposite directions. For example, by I9I4 the
increased militarystrengthof France and Russia (especially the latter)
had reduced theirdependence on England, but this was partiallyoffset
by theirincreased conflictwith Germany and Austria. Also, the factors
are not entirelyindependent.Thus, increasedconflictwith the adversary
will reduce realignmentalternativesby tending to close offthe option
of realigningwith him.
Another determinantis the degree of strategicinterestthat the parties
have in defendingeach other.Strategicinterestis an interestin keeping
the ally's power resourcesout of the opponent's hands. Analytically,this
is differentfrom dependence as just defined,since it refersnot to the
need foraid in case one is attacked,but to the need to block an increase
in the adversary'spower. In practice,the two are closely linked since
the ally's independent existence is a prerequisite to receiving his aid.
Strategicinterestmight be termed"indirectdependence," and need for
the ally's assistancemightbe labeled "directdependence." Most alliances
in a multipolarworld involveboth kinds of dependence,but theyshould
be kept separate because allies may be dependent in differentdegree on
each dimension. The most importantcauses of such asymmetriesare
geographical factorsand disparityof power betweenallies. The English
Channel, for instance, minimized Britain's direct dependence on her
continentalallies, but she was indirectlydependent on them because
domination of the Continentby a single power would have neutralized
the protection of the Channel. In the pre-I9I4 decade, France was
directlydependent on British aid, but had little operational strategic
8 "Conflict"
ofinterest;
and "tension"
areanalytically
Conflict
is incompatibility
separable.
willproducewar in
tensionmaybe definedcrudelyas thefeltlikelihoodthattheconflict
theimmediate
future.
A highdegreeof conflict
maybe accompanied
bylow tension;the
oppositeis alsopossible,
thoughlesslikely.The twoarelinkedherebecauseoftheirsimilar
on alliancedependence.
effect
ALLIANCE
SECURITY
DILEMMA
473
interestin defending England because England could defend herself
against direct attack by Germany. The effectsof power disparitymay
be seen in the German-Austrianalliance of I879. Because of her relative
weakness, Austria was much more dependent in a directsense on German assistance against Russia than vice versa, but Germany was indirectlydependent because of her strategicinterestin preventingAustria
frombeing absorbed by Russia.9
Asymmetriesin indirectdependence chieflyaffectthe partners'relative fearsof abandonment.Thus, when one statehas a strongerstrategic
interestin its partner than vice versa, the firstwill worrymore about
abandonment than the second, although this differentialmay be offset
if the second state is more dependent in the directsense. Differencesin
strategicinteresthelp to explain why the most powerful state in an
alliance often has little leverage over its partners: when the stronger
state's strategicinterestis well known, it cannot crediblythreatendefectionor realignment.
in the alliance agreeA thirddeterminantis the degree of explicitness
ment. A vague or ambiguous agreement tends to maximize fears of
abandonment; an explicit one minimizes such worries,but it does not
eliminate them. Conversely,entrapmentmay be a less worrisomepossibilitywith a vague agreement since the partnerscan assert that they
are not committed.The flip side of this, however, is that while states
explicitlyallied may be entrapped over the partner'sintereststhat are
covered in the agreement, they may find it quite easy to dissociate
themselvesfrom the ally in contingenciesnot mentioned because they
can be fairlyconfidentof each other'sloyaltyin theultimatecontingency.
But when the agreement is ambiguous, they may find it necessary to
stand by the ally in all situationsto prove their loyalty.Thus, France
and Russia, bound by a quite specificagreementlimited to the contingencies of German or Austrian militaryattack,were each able to withhold support in several crises in the pre-I9I4 decade where only the
partner's interestswere at stake, with little damage to the alliance.
England, however,tied onlylooselyto France and Russia, feltcompelled
9 Austria also had a strategicinterestin Germany's not being absorbed by Russia, but
thisinterestwas abstract-i.e., non-operational-sinceGermanywas quite capable ofholding
her own against Russia, provided Russia had not previouslyconquered Austria. Thus, the
situationwas analogous to the Franco-Britishrelationship.The general point is that the
strongerally will be less directlydependentthan its partnerbut more indirectlydependent.
Asymmetryof this kind is especiallypronounced in a bipolar world because of the wide
disparityof power between the superpowersand theirallies. Thus the European members
of NATO are highlydependenton the U.S. in the directsense,but the U.S. is not similarly
dependenton them; on the otherhand, the U.S. is indirectlydependenton its allies because
of its strategicinterestin containingSoviet power.
WORLD POLITICS
474
lesttheydoubtherfidelity.io
to supportthemin nearlyall confrontations
factoris onlya modifierof the
At bottom,however,theexplicitness
effectsof the more basic dependenceand interestfactors.Vagueness
tendsto inhibittheireffects.
allows thesefactorsfullplay;explicitness
Thus,asymmetrical
dependencebyitselfwillcausethemoredependent
butthisanxietywill be reducedbya formal,
allyto fearabandonment,
sucha contract
willmakeit harderforthe
explicitcontract.
Conversely,
at leaston issuesspecifiedin
less dependentally to escapeentrapment,
theagreement.
A fourthdeterminant,
affecting
boththe riskof abandonmentand
is the degreeto whichthe allies' interests
that
the riskof entrapment,
are similar
are in conflict
withtheadversary
are shared.If theseinterests
and valuedwithaboutequal intensity,
bothriskswillbe minimizedfor
bothparties,sincepresumably
(ceteris
paribus)theywillbe aboutequally
are quite
readyto fightoverthem.On theotherhand,iftheirinterests
each partnerwillworryaboutbeingtrappedinto"pullingthe
different,
out of thefire,"but each will also fearthattheother
other'schestnuts
For instance,in the
are threatened.
maystandaside ifhisown interests
alliancebeforeWorldWar I, Francehad littleintrinsic
Franco-Russian
withAustriain the Balkans,while Russia
interestin Russia'sconflicts
had littleinterestin France's conflictswith Germany.When France
hercommitment
toRussiain I9I2, sheencouragedRussian
strengthened
thusincreasingherown dangerof beingentrapped.But
aggressiveness,
thiswasmorethanoffset
bythereduceddangerofRussia'sabandonment
of thealliancein a Franco-German
war.The Frenchbythistimewere
convincedthata war was virtuallyinevitable:it was desirablethatit
fracasin the Balkans,therebyensuring
beginwitha Russian-Austrian
Russian participation.
forcurrentstrategy
and disincentives
options
Finally,the incentives
will be affectedby one's own and others'behaviorin the recentpast.
The determinants
justdescribedare situationalratherthanbehavioral.
of someitemsin thecalculus,
Althoughtheyare fairlygood indicators
or entrapment,
suchas thecostsofabandonment
theyyieldonlygeneral
and uncertainjudgmentsabout theirlikelihood
and about such other
itemsas one's own or the ally'sloyaltyreputation.
Expectationsabout
--The best treatmentof this series of crises,beginningwith Morocco (I905) is still that
of Luigi Albertini,The Originsofthe Warof 1914 (London: Oxford UniversityPress, I952),
Vol. I.
* The term"abandonment"is used here in the senseof renegingon one's specificalliance
obligation,not as realignmentor de-alignment.
of the French
On Franco-Russianrelationsbetween I9I2 and I9I4 and the strengthening
commitment,see G. P. Gooch, Beforethe War: Studiesin Diplomacy (London: Longmans,
Green, i938), chap. 2.
475
allies' future behavior and their probable expectationsof one's own
behavior cannot be very preciselyarrived at from guesses about the
others' "dependence," "interest,"and so on. Behavioral evidence supplements the situational elements to yield more specificand confident
expectations.Thus, Britainand France were generallyconcernedabout
Russian defection throughout the pre-I9I4 decade because of their
knowledge of Russia's ratherlow dependence on the Triple Entente
mostlybecause her option of realigningwith Germany was open until
late in theperiod. They became more specificallyand intenselyconcerned
when Russia negotiateda settlementof certaindisputes with Germany
in i910. Of course, the ally's recentbehaviormay be in responseto one s
own priorbehavior.The pointis thatthecontinuingsequence of strategic
choicesbyall actorsyieldsa streamof behavioralevidencewhich interacts
with estimatesof the general situationalfactorswhen the parties assess
the probable consequences of currentstrategyoptions.
These factors direct and indirectdependence, explicitnessof commitment,disparityof interestsin conflictwith the opponent, and the
behavioral record are the principal determinantsof the values and
likelihoods that the parties impute to the various possible consequences
of strategyoptions as listed in Table i. These values and likelihoods,
in turn,are the proximatedeterminantsof strategychoice itself.
Thus, ifa statefeelshighlydependenton itsally,directlyor indirectly,
if it perceives the ally as less dependent, if the alliance commitmentis
vague, and if the ally's recent behavior suggests doubtful loyalty,the
state will fear abandonment more than entrapment.It will therefore
tend to reassure the ally of its commitment,support him in specific
confrontationswith the opponent,and avoid conciliatingthe opponent.
The reverseconditions will tend to induce opposite strategies.
These propositionsmay be illustrated,though imperfectly,
by certain
differencesbetween Britishand French behaviortoward each otherand
toward Germany from approximately I905 to 1909. Britain was not
dependent on French aid in case of a direct German attack the navy
could handle that. She was indirectlydependent on the Entente,however, because of her strategicinterestin a continentalbalance of power
and her derivativeinterestin preventinga German conquest of France
or a German-French alliance, eitherof which would make the British
Isles vulnerable. France was perceived as less dependent because she
already had an alliance with Russia in addition to an apparent option
of realignmentor rapprochementwith Germany while England had no
alliance alternatives.The vague language of the Entente Cordiale of
I904
placed little constrainton French realignment.The British did
476
WORLD POLITICS
in a Franco-German
war overspeworrysomewhataboutentrapment
cificFrenchinterests,
but thisconcernwas less importantthan their
anxietiesaboutFrenchdefection.
The resultof theseconsiderations
was a policyofreassurance
toward
Franceand firmness
towardGermany.This reducedthe riskof abandonmentat thecostof someincreasedriskof entrapmentin case the
of
towardGermanyin theirconfidence
Frenchbecametoo intransigent
Britishsupport.However,the lattercost was small: Britainhad little
leverageforrestraining
Francein anycasebecauseofFrenchknowledge
of Britishstrategicinterests.
The firmstancetowardGermanywould
deterGermanyifshewereaggressively
inclined,butitriskedprovoking
an insecurity
spiral if she were not. Sir Edward Grey,the Foreign
as a prudent
Minister,
entertained
bothpossibilities,
butchosefirmness
with
in the adversarygame; it was also consistent
worst-casestrategy
his alliancegame strategy
of reassuring
France.
AlthoughFrance valued the Entente,she was less dependenton it
thanEngland not becauseshe would not need help in a war against
Germany,butbecauseshe did notbelieveEnglandcouldprovidemuch
perceivedthestrongBritishstrategic
helpin a land war.Francecorrectly
a Franco-Germanreinterestin defendingFrance and in preventing
alignment.She was also reassuredbyunqualifiedBritishsupportin the
Moroccocrisisof I905. Therefore,France had littlereasonto doubt
Britishloyaltyand feltlittleneed to reassurethe Britishof her own
loyaltyin orderto avoid abandonment.Instead,she exploitedBritish
anxietiesaboutFrenchloyaltyin orderto get a firmerBritishcommittalksand an implicit
ment principally
in theformofjointmilitary
staff
forceto theContinentin case
Britishpledgeto send an expeditionary
of war. France had some fearof entrapment
in a war resultingfrom
the Anglo-Germannaval race; partlyfor this reason,she adopted a
forconof Germany.Otherincentives
policyof moderateconciliation
ciliation
camefromtheFranco-German
adversary
game.Powerful
groups
in Francewanteda rapprochement
withGermanyat thistime,bothto
compensateforFrenchand Russianmilitaryweaknessand to secure
commercialand colonialbenefitsby deals withGermanyin the Near
withGermany
East and Africa.These incentives
producedan agreement
over Moroccoin i909, which,of course,further
increasedBritishconcernsabouta possibleFrenchrealignment.12
12On
Anglo-French-Germanrelationsduring this period, see K. A. Hamilton, "Great
and D. W. Sweet, "Great Britain and Germany, I905Britain and France, 1905-1911,"
I9II,"
in F. H. Hinsley,ed., BritishForeignPolicy UnderSir Edward Grey(Cambridge,:At
the UniversityPress, 1977), chaps. 5 and ii.
ALLIANCE
SECURITY
DILEMMA
477
Thus, actual and perceived asymmetriesof direct and indirect dependlence,differencesin particularinterestsvis-at-visthe opponent, the
vagueness and fragilityof the Entente,and the behavioral record combined to produce differentstrategiesby France and Britain in both the
alliance and adversarygames.
ON
SPIRALS
In the adversarysecuritydilemma, "D" strategiesof firmness,played
between adversarieswho believe each other to be potentiallyaggressive
but who really are not, produce an "insecurityspiral" i.e., a spiral of
power/securitycompetition that feeds on each party's fears that the
An analogous spiral
other'sdefensivemoves are aggressivelymotivated.&3
in the alliance securitydilemma is an "integrativespiral,"in which allies
move progressivelycloser out of theirmutual fearof abandonment.The
integrativespiral interlockswith the insecurityspiral in the adversary
game i.e., the two spirals are mutually reinforcing.Thus, allies play
reciprocal"C" strategiesout of fear that the partnermay defectif they
do not show support; theyfurtherattemptto reduce the partner'sanxieties by standing firm against the opponent; the opponent then feels
threatenedand encircled, and his hostile response increases the allies'
incentive to close ranks; when they do, the adversary becomes more
fearfuland hostile,and, feelingmore dependenton his own allies,moves
closer to them; in reaction,the firstalliance furthersolidifies,and so on.
The Anglo-French Entente of I904 and the German reaction to it
set in motion an alliance integrativespiraland an interlockingadversary
insecurityspiral that continued until the outbreak of war in I9I4. The
Entente had not initiallybeen intended as a quasi-alliance against Germany,but only as a colonial settlementand as a hedge against involvement in the impendingwar betweenthe parties'respectiveallies Japan
and Russia. But Germany did see it as a threatand challenged it in the
Morocco crisisof I905 in an attemptto break it up by exposing British
infidelity.However, Britain stood firmlybehind France; the result of
the Germans' coercion was a self-confirmation
of theirown hypothesis:
the Entente was transformedinto a quasi-alliance against the newly
revealed German threat.
The tighteningof the Entente set in motion an insecurityspiral in
3 What I call the insecurityspiral is sometimesreferredto as the securitydilemma. I
believe this is a mislabeling,however,since the spiral is an outcomeof both adversaries'
choosing one of the options in the securitydilemma, not theproblemof choice itself.The
of choosingbetween two options,each of
definingfeatureof any dilemma is the difficulty
which will have more or less unsatisfactory
consequences.
478
WORLD POLITICS
theadversary
game,chiefly
betweenGermanyand England.The latter,
feelingmore dependenton France as evidenceof German hostility
carefully
mounted(especiallyvia theAnglo-German
navalcompetition)
movestowardGermanythat
avoided,or limited,any accommodative
mightraise doubts in French minds about Britishloyaltyand thus
possiblytrigger
a Frenchrealignment
withGermany.'4
Shealsoincreased
For
hercommitment
to Francebyinitiating
military
staffconversations.
of Ententehostility
now seemedto be even
Germany,the hypothesis
morestrongly
confirmed
by theBritishreserveand thespectacleof the
partners
movingclosertogether.
Germanysteppedup hership-building
and drewcloserto Austria.Thus,theincreasedsolidarity
in one alliance,
interacting
withan insecurity
spiralin the adversarygame,increased
thecohesionof theotheralliance.
These spiralscontinuedwiththeAnglo-Russian
Ententeof I907 and
the Bosnian crisisof i908-i909.
The Anglo-Russian agreement,like the
earlierAnglo-French
ofvariouscolonial
one,was on itsfacea settlement
conflicts,
but forBritainit was also (unlikethe FrenchEntente)conthebalanceof poweragainstGermany.
sciouslyintendedto strengthen
and felt
Germanyperceivedit as anotherstep in her encirclement,
increasingly
dependenton Austria.During the Bosniancrisis,fearing
and, by issuing
Austriandefection,
she supportedher ally completely
an ultimatum,
she forcedRussiato back down. She also expandedthe
to Austria.Whereastheallianceforscopeof heralliancecommitment
mallycalledformutualsupportagainstRussiaonlyin a defensivewar,
war.
Germanynow assuredAustriaof fullsupportevenin an offensive
not
to
thepoint
Britainalso supportedRussiaduringthecrisis,although
of offering
military
assistance.The resultof thecrisiswas thusto conto extendthescopeof theGerman-Aussolidatebothalliancesystems,
trianone, and to increasethelevelof tensionbetweenthetwo.'5
The integrative
processcontinuedon the Ententeside withBritish
supportof France duringthe second Morocco crisis(I9 II),i6 and a
betweenFranceand England,
firmingand extensionof commitments
connection
and betweenFranceand Russia,in I9I2. The French-British
was tightenedby an agreementon a naval divisionof labor in which
and the Britishfleet
the Frenchfleetwas to guard the Mediterranean
was to guard the Atlanticcoastsof bothcountries.In returnforthis
'4George Monger, The End of Isolation(London: Nelson, I963),
chap. I 2.
BernadotteSchmitt,The Annexationof Bosnia (Cambridge: At the UniversityPress,
I937).
i6
Ima C. Barlow, The Agadir Crisis(Chapel Hill: Universityof North Carolina Press,
5
I 940).
479
arrangement,France successfullydemanded an exchange of notes that
strengthenedthe Britishdiplomatic commitment.'7
The bolsteringof theFrench-Russianalliance occurredmainlybecause
of France's increasinganxietyabout possible Russian defection.Russian
dependence on France was declining because of Russia's steadily increasing militarystrength.Moreover, she had demonstratedsome independence by failingto supportFrance during the I9II Morocco crisis
and by settlingsome minor conflictswith Germany in the Near East.
France became worried that Russia mightbe induced to remain neutral
in a Franco-German war. She dealt with the problem by extendingthe
scope of her commitmentto Russia, implicitlypromisingaid not just in
case of a German (or German-Austrian)attack on Russia (as per the
termsof the treaty)but also in case Russia was forcedto attack Austria
because of an Austrian militaryinitiative in the Balkans. This was
analogous to the extension-ofthe German commitmentto Austria after
the Bosnian crisis.I8
The integrative/insecurity
spiral was partially-but only temporarwhen England
ily interruptedduring the Balkan Wars (1912-I9I3),
and Germany managed to fashion a detente based on collaborative
restraintof theirallies.
THE
RESTRAINING
THE
BALKAN
ALLIES
WARS
DILEMMA:
INTERLUDE
When members of opposite alliances become involved in a crisis
confrontation,theirallies may find themselvesin a special formof the
composite securitydilemma in which the alliance and adversarycomponents are closely linked. The dilemma is not just whetherto support
or restrainthe ally, but whether to support the ally or to collaborate
with the noninvolved state on the opposite side in restrainingboth
protagonists.As in the classic prisoner'sdilemma, the opposite restraining stateshave a good deal to gain by cooperating-in keeping the peace
and avoiding entrapment-but theyare also motivatedagainstcooperation by the desire to avoid alienating their allies. The best outcome
'7Samuel R. Williamson,Jr.,The Politicsof Grand Strategy(Cambridge: Harvard UniversityPress, i969), providesan excellentaccount of these French-Britishnegotiations;see
chaps. i i and I 2.
x8 It is interesting
that France tightenedher alliance with Russia by increasingher own
commitment,but strengthenedher Britishconnectionby extractinga firmercommitment
from England. The explanation for the differenceis that the "balance of dependence"
betweenFrance and Russia favoredRussia,whereasbetweenFrance and England, it favored
France. In other words, France enjoyed more bargainingpower over England than over
Russia.
480
WORLD POLITICS
one's own ally while the admay be obtainedby secretlysupporting
versaryrestrainshis; but if bothpartiespursuethisstrategy
(perhaps
theresultis thetypical
out of suspicionthattheotheris doublecrossing)
dilemma in thiscase,mutualentrap"DD" outcomeof theprisoner's
ment.
The BalkanWars of I9I2-I9I3
engaged Britainand Germany in this
sortof dilemma. Their allies, Russia and Austria respectively,
supported
opposite sides among the Balkan states who firststripped Turkey of
most of her European territoryand then foughtanother war among
themselvesover thedivisionof the spoils.England and Germanyreached
an informalagreement:England was to restrainRussia fromintervening,
Germany to restrainAustria. The collaborationworked initially.Acting
througha conferenceof all the great powers, the two peacemakers did
restrain their allies and enforced compromise settlements.Formally,
England and Germany both played "C" in the adversarygame and "D"
in the alliance game, risking the loss of their allies in order to avoid
entrapmentin a conflictover issues in which neitherhad much intrinsic
interest.Apparently,the integrative/insecurity
spiral of the preceding
years had been stopped.
This cooperative achievement,however, was due to a special combinationof factorsand, ironically,its success generatedforcesthatmade
its repetitionunlikely.For Germany,the payofffor collaborationwith
England was illusorilymagnifiedby the vain hope that it would lead
to an Anglo-German neutralitypact, which Germany had been seeking
for years. This hope outweighed the risk of alienating Austria, a risk
thatwas low because the governmentin Vienna was itselfdivided about
riskinga confrontationwith Russia. As forEngland, she was now more
willing than before to take a chance on Russian defectionbecause her
dependence on the Russian and French ententeshad declined. England's
dependence had always been indirect,based on her strategicinterestin
a continentalequilibrium.In I9I2-I9I3,
such an equilibriumwas at least
being approached, by virtueof a substantialincreasein Russian military
strengthand the growing solidarityof the Russian-Frenchalliance. A
British commitmentwas thereforeless necessary to deter the central
powers, and Russia and France were less likely to realign if theybegan
to doubt Britishloyalty.In the adversarygame, the Britishbalance of
incentiveshad shiftedsomewhat fromdeterrenceto conciliation,and in
the alliance game her balance of worries had shifted from fears of
abandonmentto fearsof entrapment.The Britishpayoffin collaborating
with Germany was furtherincreased by Sir Edward Grey's desire to
dampen the insecurityspiral by easing German fears of encirclement,
481
and by the desireto gain more leverageover Russia in the persistent
squabbleswith her over Persia. Finally,Britain'scost in restraining
Russiawas minimizedby thefactthattheRussiangovernment
was as
dividedand vacillatingabout riskingwar as the Austriangovernment
was.'9
THE
SIR
STRADDLE
EDWARD
GREY
STRATEGY:
IN
JULY
I9I4
The Anglo-German cooperation of the Balkan Wars did not carry
over into the crisis of JulyI9I4, chieflybecause Germany's balance of
incentiveshad changed sharplyby then.The growthof Russian military
power, the initiation of Anglo-Russian naval conversations,and the
increasingtensionbetween the continentalalliances had increased Germany's sense of dependence on Austria. This, combined with Austria's
reproaches for her weak support during the Balkan Wars, increased
German fears of Austrian defection: Austria must not be restrained
again. Germany's feelingthatwar was inevitableled to the thoughtthat
it had betterbe now than later, when Russia would be even stronger.
Her incentiveto cooperate with England had declined with the evaporation of the dream of a neutralitypact, although German leaders
continued to believe that England might well stand aside in a war in
which Russia could be made to appear the aggressor.By Julyof I9I4,
Germany was no longerin a security"dilemma," strictlyspeaking,since
she had resolvedit on the side of definitesupportof her ally and firmness
toward her opponents.The same was trueof France and Russia: Russia,
like Austria, was determinednot to back down again, and France repeatedly avowed her unflinchingloyalty.
In England, however, Sir Edward Grey's dilemma was as severe as
ever: his incentivesfor eitheroption in both the alliance and adversary
games were about evenlybalanced. In the alliance game, thiswas largely
the resultof the apparent equilibriumof power on the continent,which
lessened Britain's need to stand by her allies. If they could hold their
own against the central powers, they were less likely to defect out of
fear of British defection.In the adversarygame, Grey hoped to keep
alive the detentewith Germanythathad been nourishedby theamicable
settlementof certaincolonial disputesin the springand summerof I9I4.
Thus Grey's worries about the intransigenceof Britain'sallies were at
19The classic studyof the Balkan Wars is E. C. Helmreich,The Diplomacyof theBalkan
emWars (Cambridge: Harvard UniversityPress, I938). For a more recentinterpretation
phasizingAnglo-Germanrelations,see R. J.Crampton,TheHollow Detente(London: George
Prior Publishers,I979).
482
WORLD POLITICS
least as strongas his fearsof theirdesertion,and he was as eager to
avoidprovoking
Germanyas todeterher.The twohornsofthedilemma
appearedaboutequallyattractive
(orunattractive).
Greytherefore
adopted
a straddle a mixed "C-D" strategy
in boththe allianceand the adversarygames. He gave vague reassurances
to France and Russia,rejectingtheirpleas foran unambiguousdeclarationof armed support.
He issueda fewmild warningsto Austriaand Germany,but concentratedon revivingthe joint Anglo-Germanmediationof the Balkan
Wars.The lattereffort
failedbecauseGreydid notappreciatetheextent
towhichGermanpayoffs
had changedsinceI9I2-I9I3.
Germanyagreed
to collaborate in joint restraintof Austria and Russia, but defected by
goading Austria on. In the overall straddle, it was Grey's hope that
Russia would be restrained without being alienated while Germany
would be deterredwithoutbeing provoked. This strategyfailed because
France-Russia and Austria-Germany,all thinkingwishfully,drew oppositeconclusionsfromit. France and Russia counted on Britishsupport
while Germany and Austria expected Britishneutralityuntil the very
last moment. Thus, in strivingfor the best of all worlds, Grey got the
worst. If, early in the crisis,he had declared his firmsupportof Russia
and France and warned Germany unequivocally,or if he had clearly
declared to France and Russia that Britain would not fight,the war
might have been averted.20This is not to argue that Grey's ambiguity
was a centralcause of the outbreak of World War I. But England was
the only major power that still had both polar options available; in
failingto exerciseeitherone, she gave up a chance to preventthe war.2'
We should not conclude fromthissinglecase, however,thata straddle
strategyin the composite securitydilemma cannot work, thatone must
firmlygrasp one horn or the other.Special circumstancesin I9I4 made
it virtuallyinevitablethatthe continentalpowers would interpretBritish
intentionswishfully.These circumstanceswere,first,thatthecontinental
powers were firmlylocked into a collision course: Austria and Russia
via theirdeep conflictof interestin the Balkans, Germany and France
because of theirhigh dependence upon and firmcommitmentsto their
Historiansare divided as to whetherGreycould have avertedthe war by takinga clear
stance one way or the other.Albertini'sopinion (fn. 10, Vol. II, 514), is that he could have.
Masteryin Europe (London: Oxford
For a contraryview, see A.J.P. Taylor, The Strugglefor
UniversityPress, I954), 525. For a discussionof Grey's dilemma, see Michael J. Eckstein,
"Great Britain and the Triple Entente on the Eve of the Sarajevo Crisis," in Hinsley (fn.
I 2), chap. i 8.
considerationsfrom
21It is true thatGrey feltconstrainedby domesticand constitutional
issuing an unequivocal commitment.The factthateventuallyhe did issue a clear warning
to Germany that England could not remain neutral-although it came too late to affect
German decision making-indicates that these constraintswere not absolute.
483
spiral within and between the contiallies. The integrative/insecurity
so
tightly
wound by the summer of I9I4
nental alliances had become
that,in order to have any chance of reversingit, Grey would have had
to choose one of his extreme options. Second, British power was not
considered so decisive for the outcome of a war that the other states
were unwilling to gamble with the uncertaintyof its being used. Third,
both sides on the Continent had grounds for optimisticexpectations
about England, generated by recentinteractionswith her.
With differentbackgroundconditions,a straddlemightbe successful.
Success means instillingcautious (and, hence, opposite) expectationsin
the ally and the adversary,ratherthan wishful(and opposite) ones, and
doing so without unduly antagonizing eitherparty.One wants the ally
to think one probably will not fight,and the opponent to think one
probablywill. The ambiguityof a straddleis mostlikelyto be interpreted
cautiously in circumstancesmore or less opposite to those of 1914: the
conflictof interestbetween ally and adversaryis relativelymild, they
are not firmlycommitted against each other, and one's own military
power is expected to determinethe outcome of a war. Then the other
parties' costs of being restrainedor deterred,or of acceptinga compromise settlement,would be low compared to the expected cost of a war
(and defeat)resultingfroma mistakenpredictionofone's own intentions.
A mixed past record containing instances of both non-supportof the
ally and strongresistanceto the opponent would also be helpful. Such
conditionswere approximatedin relationsbetween Russia, Austria,and
Germany in the i870s and i88os, when Bismarck played the straddle
option successfullyon several occasions.22
THE
ALLIANCE
SECURITY
DILEMMA
IN A BIPOLAR
SYSTEM
The alliance securitydilemma is sharplytruncatedin today's bipolar
system because one of the central "bads" that of abandonment is
22These remarkscan no more than suggestthe complexityof the straddleproblem.One
aspect of this complexityis that one is tryingto optimize among four objectives: restrain
the ally,but avoid alienatinghim; deterthe opponentbut avoid provokinghim. Justwhat
mixtureof communicationsis likely to optimize will be difficultto estimate,even if one
could assume thattheywill be interpretedas desired by both other parties.Anotheraspect
is the necessaryambiguityof one's messages, which makes it unlikelythat they will be
interpretedexactlyas desired.The possibleporosityof diplomaticcommunicationchannels
limitsthe extentto which signals sent to the ally and adversarycan be inconsistentwith
each other-e.g., signalsto the ally thatlean toward restraintand messagesto the opponent
thatemphasize firmness.The "pure" and consistentstrategiesof restrainingthe ally while
conciliatingthe adversary,or supportingthe ally while warningthe opponent,are less likely
to be misread,and eitherone may well keep the peace, but theyrisk alienatingthe ally or
theselatterrisks,a mixedstrategy
theopponent's
promByminimizing
hostility.
increasing
outcomethaneitherofthepureones,butalso risksa worseoutcome,
especially
isesa better
conditions
whenthebackground
approachthosein I9I4.
484
WORLD
POLITICS
highlyunlikely.At leastthatis so fortheEuropeancentralarenaof the
The superto whichthefollowingdiscussionwill be limited.23
system,
powersare solidlycommittedby theirown intereststo defendtheir
allies, hence theirde-alignmentis irrational.Realignmentwith each
otheris logicallyimpossible,
simplybecausethereis no otherstatepowerfulenough to providea motive.The European allies theoretically
could shiftsidesifleftto theirown devices.However,thesuperpowers
havepowerfulincentives
to preventtheirrealignment byforceifnecessary as the Sovietshave demonstrated.
The West Europeanstates
have no motiveto realign,sincetheSovietUnionis theprincipalthreat
Finally,
totheirsecurity
and theUnitedStatesis theirnaturalprotector.24
simplede-alignment
by the smallerstatesis ultimately
illusory,since
theirprotector
will defendthemno matterwhatpoliticalposturethey
assume.
is possible,however.Conceivably,
boththesuperpowers
Entrapment
and theiralliescould be pulledunwillingly
in Europe by
intoa conflict
one partner'sinitiative forinstance,involvement
of theWestGerman
The smalleralliesmustworryin
armyin an East Germanrevolution.
additionabout extraregional
entrapment
(the spilloverintoEurope of
a superpowerconflictoriginatingin otherareas),and about nuclear
nuclearwarfarein Europe in the
entrapment
(a superpowerinitiating
is a more
hope thatit mightbe confinedthere).In general,entrapment
seriousconcernforthe lesserallies thanforthe superpowersbecause
because the
theyshare only a portionof the latter'sglobal interests,
havea muchgreatercapacityfortakinginitiatives
superpowers
(notably
nuclearinitiatives),
and becausethe allies' capacityto restrainthe superpowersis muchsmallerthanvice versa.
Since the alliance dilemma is mostlya functionof tensionbetween
the risk of abandonmentand the risk of entrapment-reducing
one
tendsto increasetheother-the dilemmais weak in a bipolaralliance,
becauseonlyone of theserisksis significantly
present.That risk,enfromtheally'spolicy,
can be dealtwithsimplybydissociation
trapment,
or by variousmeans of restraining
the ally,withoutconcernthatthe
ally may defectin consequence.The superpowermay exerteconomic
pressureor simplywithholdsupportfromtheally'sadventures(as the
23 Jbelieve the presentsystemshould still be classifiedas bipolar,even though there has
been some movementtoward multipolarity.Although some of my theoreticalstatements
in this section apply to bipolarityin Europe in general, they are much more relevantto
NATO than to the Warsaw Pact, as my examplesclearlyindicate.Bipolar alliances outside
Europe have somewhatdifferentdynamics,which cannot be explored here.
24 Some mightargue that the West European countriesare allied with the United States
withthe Soviet Union), not beqause
out of culturaland ideologicalaffinities
(and disaffinities
did not exist,alignmentwith the
of structuralcompulsion.However, even if theseaffinities
U.S. would be dictatedby theirsecurityinterests.
485
United States did during the Suez crisisof I956). The weaker allies may
also fail to support, or even hinder, the superpower's actions (as the
Europeans did during the Yom Kippur War and on the recent gas
pipeline issue); theymay also be able to exercisesome restrainingleverage
by appealing to consensusnormsor by exploitingthe superpower'sneed
for collective legitimization.In so doing, they do not risk losing the
superpower's protection.Neither the superpower nor its protegesneed
feel any compulsion to supporteach other (and therebyaccept increased
risksof entrapment)solelyto insureagainst the alliance's collapse. When
support is provided against the adversary,it is chieflybecause the allies
share each other's interestsand perceptionsin the adversarygame, not
because they fear the partnermight otherwise leave the fold. Finally,
neitherthe greaternor the smaller allies are inhibitedfromconciliating
the adversaryby the worry-typical of multipolaralliances-that such
conciliationmight precipitatethe ally's realignment.
Because of the weakness of the alliance securitydilemma, the adversary dilemma dominates. That dilemma can be dealt with by each
partneraccording to its own preferences,with littleconcernforalliance
"side effects."The superpowersand theirallies may disagreeabout policy
toward the adversary, and, as in NATO at present, they may even
pursue contradictorypolicies. But they need have little fear that their
own association is therebyendangered. Indeed, it is largelybecause the
alliance itselfis fundamentallystable that such policy divergencesmay
develop and persist. The alliance is stable because it is essentiallya
product of the structureof the system and of the common security
interestsgenerated thereby.So long as thatstructureand those interests
persist,the allies are free to disagree.
By contrast,in a multipolarsystemwhere the alliance and adversary
dilemmas are roughly co-equal, allies are much more constrained in
their play of the adversary game by the possible consequences to the
alliance. In particular,theyare inhibitedfromconciliatingthe adversary
forfear the ally will take it as a prelude to realignment,and will realign
preemptively.Since multipolar alignmentsare not determinedby the
structureof the system,but are formedby choice among severaloptions,
theyare unstable and vulnerable to policy disagreement.Worries about
alliance disintegrationthus tend to induce similarityof policy toward
the opponent. In a mulitpolaralliance there is a close link between the
degree of policy solidarityand the stabilityof the alliance; in a bipolar
alliance that link is tenuous at best.25
25 For a similarargument,
Politics(Reading,
see KennethN. Waltz, TheoryofInternational
Mass.: Addison Wesley, I979), esp. i69-70. This essay has been considerablyinfluencedby
Waltz's impressivework.
486
WORLD
POLITICS
Although the alliance securitydilemma is weak in a bipolar system,
it is not entirelyabsent. There are partialsurrogatesforabandonment
various degrees of movementaway fromtheally or toward the opponent
that give rise to truncatedformsof the dilemma. Differenttypes can
be discerned at the political and militarylevels.
At the political level, the most extreme conceivable approaches to
abandonment are "condominium" for the superpowers and "finlandization" or "neutralization"forthe lesser states.Condominium means a
high degree of collaboration by the superpowers to maintain order in
the system,and subordination of their competitiveinterestsand the
interestsof theirallies to thisgoal. Logically,condominiumis more akin
to multipolar"concert"than to realignment,and is not inconsistentwith
a continued intentionto defend allies against actual attack. However,
the allies might still have reason to worry about not being supported
over less-than-ultimateissues,and about losing whateverpoliticalinfluence theycurrentlyenjoy as a consequence of superpower rivalry.
Finlandization or neutralizationare also just barely conceivable. In
an ultimate sense, both are illusions. Although it is possible that some
European NATO members might driftin these directionsout of fear
of being abandoned by the United States or, conversely,of being entrappedin a conflictbetweenthe superpowers,such stanceswould amount
only to "pseudo-abandonment" so long as the U.S. retainedits intention
to defend them. NATO, like any bipolar alliance, is essentiallya guarantee of the lesser states by the superpower; as long as the guarantee
holds, the alliance holds-whatever postures the smaller states may
strike.This does not mean, however,thatthe degree of West European
accommodation with the Soviets is unimportant.If West Germany,say,
were to come under significantSoviet influence,importantcosts forthe
U.S. and the Western alliance would be entailed, even if the alliance
per se remained intact.
Within these extremes-that is, within the realm of the possibleare various degrees of "detente"with the adversary,each involvingsome
degree of cooperative benefitsand some dependence on the adversary
for continuance of the benefits,and hence, some influenceby the adversary.Specific moves toward detente, by either the superpowers or
theirproteges,may generatefearsthatfurthermoves thatwould weaken
alliance obligations are impending. Typically, the fears exceed what is
likely to happen, much as fearsof realignmentin the multipolarworld
are quite disproportionateto the real possibilities-states are notoriously
paranoid -but, of course, it is the subjective fears ratherthan the objective possibilitiesthat influencetheirbehavior. Thus, in the truncated
487
alliance game of bipolarity,simply the reduction of tension between
adversaries,or its possibility,is functionallyequivalent to potentialdefectionor realignmentas a source of abandonment anxiety.
An obvious militarysurrogatefor abandonment would be a partial
U.S. troop withdrawal fromEurope. (A completewithdrawal is hardly
conceivable.) Although this would not seriouslyimpair the U.S. commitmentto Europe, it would certainlystimulateever-lurkingEuropean
worries about a U.S. retreatto the American "fortress."
The alliance dilemma forthe U.S. at the politicallevel is a by-product
of the adversarydilemma-of the choice between conciliatingor confrontingthe adversary.Unlike themultipolarversion,it does not involve
a trade-offbetween risksof abandonmentand risksof entrapment,but
rathera balancing between two kinds of incentivesfor partial "abandonment" by the smaller states.As Henry Kissingerhas noted,the West
Europeans tend to oscillate between fearsof collusion and fearsof collision betweentheUnited Statesand theSoviet Union, and theirresponse
to bothfearsis to conciliatethe Soviets.26Their motivesforthisresponse
to superpower collusion presumablyare (i) to hedge against a possible
weakening of the American will to protecttheirinterests;(2) to preserve
some bargainingpower vis-a-visthe United States;or (3) simplyto share
in the benefitsof betterrelationswith the Soviet Union. When theyfear
collision (entrapment),their incentivesto conciliate the Soviet Union
may be (i) to restrainU.S. intransigence;(2) to insulate theirown good
relationswith the Soviets fromthe superpowerconfrontation;and perhaps (3) to offsetU.S. belligerence,thus moderatingthe overall alliance
stance and reducing the risk of provoking the Soviets.
Thus, forthe United States,thealliance dilemma presentsthe problem
of findingthe optimum blend of firmnessand accommodation toward
the Soviets-that which generates the least incentivefor the allies to
"neutralize" or "finlandize." Since the U.S. will have its independent
preferencesin the adversary game, which might point to a different
optimum stance in that game, the overall problem is one of optimizing
between two optimums.
Henry Kissingerstatedtheproblemclearlywhen he warned President
Nixon in I973:
We had to remainsoberin our dealingswiththe SovietUnion. If we
ofa U.S.
becametooimpetuoustheEuropeannationswouldgrowfearful
...
Sovietdeal. This would cause themto multiplytheirown initiatives
withtheU.S.S.R.
to protect
themselves,
bymakingtheirown arrangements
the same would happenif the U.S. remainedin the
But paradoxically,
26
Kissinger,The WhiteHouse Years(Boston: Little,Brown, 1979), 382.
488
WORLD POLITICS
trenches
oftheCold War. In thatcaseEuropeanleaderswouldbe tempted
to appearbeforetheirpublicsas "mediators"betweenbellicosesuperpowers. The U.S. had to conducta carefulpolicytowardstheSovietUnion:
sufficiently
strongto maintainthe interestin the commondefense;sufflexibleto preventour alliesfromracingto Moscow.27
ficiently
This strategyis appropriate when a superpower is most concerned
about the effectsof its own strategyupon that of the allies. When the
concern is that the ally is "defecting"on its own initiative,however, a
betterstrategymightbe to preempt.For example, Nixon and Kissinger
worried in the early seventiesthat the West German Ostpolitik
might
go too far; this stimulated them to move ahead with theirown negotiations lest Germany be left alone in the Soviet embrace. Kissinger
reportsthat he dreaded the moment "when no German chancellorcan
affordthe hostilityof the Soviet Union . . .":
This uneasinessabout the driftof Brandt'spolicieswas sharedby his
principalpartnersin the Westernalliance.... To forestall[independent
GermanmaneuveringbetweenEast and West],or perhapsoutflankit,
each of Brandt'scolleagues-includingNixon-sought to pre-emptGermanyby conductingan activedetentepolicyof its own. In thissense,
to a race
farbeyondthoseintended.It contributed
Ostpolitikhad effects
to Moscowand overtimeheightened
mutualsuspicionamongtheallies.28
When the U.S.-Soviet detente negotiationswere well underway,the
bogeyof "condominium" was raised in Europe- notablyaftertheagreement on Prevention of Nuclear War in I973, which seemed to imply
jointSoviet-U.S. managementof crisesand disputesbetweenotherstates.
At that point, the European states felt impelled to deepen their own
detentein order to hedge against the possibilitythatWashington's commitment to its allies was weakening. Willy Brandt noted that one of
his motives for Ostpolitik
was his belief that only a part of U.S. forces
would still be in Europe at the end of the seventies.29
However, the most importantmotives for detente on both sides of
the Atlanticwere autonomous ones stemmingfromthe adversarygame,
and the reduced cohesion of NATO was more the resultof the reduced
Soviet threat than of mutual fears of defectionamong the allies. Nor
did fearsof the partners'defectionseriouslyinhibitthe detente policies
of either the U.S. or Europe-despite Kissinger's warning.
There is no bipolar parallel to the integrativespiral that occurred in
Ibid., 382.
Kissinger,Yearsof Upheaval(Boston: Little,Brown, I980), I45-46.
29 Willy Brandt,Begegnungen
und Einsichten.Die Jahreig60-i975 (Hamburg: Hoffmann
& Campe, I976), 348, cited in Karl Kaiser and Hans-Peter Schwarz, Americaand Western
Europe: Problemsand Prospects(Lexington,Mass.: Lexington Books, I977), 229.
27
28
ALLIANCE
SECURITY
DILEMMA
489
the pre-I9I4 alliances. Such a spiral may occur in a multipolar system
because alliance commitmentsare always somewhat in doubt; henceespecially during periods of rising tension-there are tendencies to
strengthenand expand commitmentsin order to guard against abandonment,tendenciesthat at the same time generate and are reinforced
by an insecurityspiral between adversaries. In a bipolar alliance such
as NATO, the basic alliance commitmentis not in doubt since it is
structurallyordained; there is little scope or incentiveto strengthenit
by acts of policy,and thereis hence no integrativespiral. An insecurity
spiral may develop independentlyin the adversarygame, and NATO's
policy cohesion may increase in response to common perceptionsof a
risingexternalthreat,but these processesare not additionallydriven by
fearsof desertionby allies.30
THE CURRENT NATO
CRISIS AND THE SECURITY DILEMMA
The presentconflictbetween the United Statesand itsEuropean allies
is chieflya disagreementabout how to deal with the adversarysecurity
dilemma. Each side of the alliance prefersa differentstrategytoward
the Soviet Union, and each either advocates or independentlypursues
its own preferredstrategy.The United States prefersa "D" strategyof
toughnessand confrontation;the Europeans generallya moderate "C"
strategyof conciliation. Their strategiesin the alliance game are consistentwith, though more or less incidental to, theirfavored strategies
in the adversarygame. The United States perceivesitselfas increasing
its support for its allies by building up its capabilityfordeterrenceand
defense, in Europe as well as in peripheral areas where it perceives
European intereststo be at least as great as its own. The Europeans are
divided, but mostly skeptical about U.S. nuclear policies. They resist
U.S. pressuresto increase theirown conventionalforces;theyalso resist
U.S. tendenciesto make detentea hostageto Soviet behaviorin the third
world,and theyobstructWashington'sattemptsto apply economic sanctions against the Soviets.
Some of the reasons forthis divergenceare extraneousto the present
3 Conversely,it ought to be easier for the bipolar superpowers than for multipolar
adversariesto dampen or avoid insecurityspirals by conciliatingeach other, since their
conciliationis so much less constrainedby alliance concerns.In the pre-I9I4 decade, Sir
Edward GreyempathizedwithGermany'sinsecuritiesand fearsof encirclement;he realized
that the steadytighteningof the Entente was provokingher, but he could do littleabout
it because he fearedabandonmentby Russia and France if he tried.The U.S. and the Soviet
Union do not have a similar problem.This is not to say that it is easy to escape fromor
weaken insecurityspiralsin a bipolarsystem-it is merelyeasierbecause theattemptinvolves
significantrisks in the adversarygame only.
490
WORLD POLITICS
theme-notably,the rise in Europe's economicstrength(which has
generateda new spiritof assertiveness
againsttheU.S. in the security
as well as theeconomicfields),and a decliningfaithin America'sleadershipcapacities.A roughassessmentof the weightsthatthe parties
attachto the"goods"and "bads" ofthecompositesecurity
dilemmaand
of whytheseweightshave changed,mayprovidea further
explanation.
The reasonsfallintofoursomewhatoverlapping
categories:
(i) divergent
images of the motivesand intentionsof the adversary;(2) different
valuationsofdetente;(3) Europeanperceptions
ofa decliningcredibility
oftheU.S. deterrent
inlightoftheSovietUnion'sachievement
ofnuclear
adventures
to thethirdworld,
parity;and (4) a shiftofSovietaggressive
which has precipitateda conflictbetweenthe global interestsof the
of theEuropeanallies.
UnitedStatesand themorelimitedperspectives
boththealliancedilemmaand theadversary
These changeshaveaffected
dilemma,butmostlythelatter.
and value loadingson theadversarydilemmahave
The perceptions
Mostimportantly,
changedsharplyforEuropesincetheearlyseventies.31
of course,thepoliticaland economicgainsfromdetentehave substantiallyincreasedthe incentiveto conciliatethe Soviets.The European
becausetheyperstatessee no "resolvereputation"
costsin conciliation
ceiveSovietintentions
as essentially
defensive,
possiblybecausetheSovietshave not been aggressivein Europe formorethan two decades.
from
This imagemaybe partlya wishfulone,supportedbythebenefits
detente.
towardtheSovietUnion
The "goods"and "bads" ofa toughstrategy
havealso changedwitha netincreaseforthe"bads."For theEuropeans,
the deterrentvalue of firmnesshas declinedbecause of theirpacific
imageof theSoviets,whilethedangerofprovokingan insecurity
spiral
has increased.Europe is disposedto invoke
throughexcessivefirmness
the "I9I4 analogy":the greatestdangerarisesnot fromthe powerful
whichmightproduce
butfromthedynamicsofconflict
adversary,
itself,
a "war that nobody wanted."32Finally, too tough a stance might un-
of detente.
dermineprospectsforarmscontrol,thecenterpiece
Americanperceptionsand values in the adversarydilemma have
changedin theoppositedirection.
ViewingSovietbehaviorand detente
3 The terms"European" and "American"are intendedto referto theapparentlydominant
views among European and Americanelites.Obviously,thereis a wide spectrumof opinion
on both sides of the Atlantic.
32 The I9I4 analogyis poorlytaken,sincethatspiralfedon factorspeculiarto a multipolar
system-notably high and equal interdependenceamong allies. Also, recenthistoriography
has suggestedthat World War I was a war that"somebody" (Germany) did want. See, for
example, Fritz Fischer,Germany's
Aimsin theFirstWorldWar (New York: W. W. Ndrton,
i967).
491
froma global perspective,
the UnitedStatesbegan to devaluedetente
in'thelateseventies
afterSovietadvancesin thethirdworld,and virtually
wroteit offafterthe invasionof Afghanistan.
The Reagan AdministrationperceivestheSovietUnionas an inherently
aggressivestatethat
will seize everyopportunity
to expandby force,and will interpret
any
as weakness.Accordingly,
conciliation
and firmness
deterrence
are valued morehighly,while the riskof stimulating
Sovietinsecurity
and a
spiralof risingtensionhas been downgraded,if not ignored.On the
economicside,a strategy
of economicdeprivation,
designedto weaken
theSovietUnion'smilitary
potentialand economy,has beensubstituted
forthe formerstrategy
of inducingSovietaccommodation
by weaving
theU.S.S.R. intoa web of economicinterdependence.
In thealliancegame,theEuropeansworryprincipally
aboutentrapment.They fearthatexcessiveU.S. bellicosity
mightset offa severe
insecurity
spiralwiththeSovietUnion,whichcould explodeintocrisis
or violence.A more specificconcernis thatof extraregional
entrapconflict
causedbyan American
ment-being caughtup in a superpower
over-reaction
toSovietadvancesin thethirdworld.Talk byU.S. military
plannersof "horizontalescalation"has no doubt fueledthis anxiety.
However,thefearprobablyis notso muchone of beingpulledintoan
area as it is thattensionsgenerated
actualwar begunin someperipheral
in such areas mightbecomegeneraland destroydetentein Europe.
of the Soviet hold on the East
Anothercost would be a tightening
Europeancountries,
endingthe dreamof at leasta partialintegration
is a
of all of Europe. (Europeanconcernabout "nuclearentrapment"
specialcase whichwill be discussedbelow.)
The alliance dilemmaforthe European allies is how to escape or
minimizetheserisksofentrapment
withoutseriously
riskingsomeform
of partialU.S. abandonment.The lattermightconsistof troopwithofEuropeandefensein
of thepriority
drawals,Americandowngrading
favorof otherareas such as the PersianGulf,or a furtherdriftto
unilateralism.
Althoughthe European alliancedilemmais somewhat
sharperthan the U.S. dilemmadiscussedearlier,it is stillnot severe
becauseany such partialabandonmentby theUnitedStateswould fall
or itsnuclear
well shortof totalwithdrawalof thelatter'scommitment
deterrent.
THE NUCLEAR DIMENSION
I have chosento treattheissueof nuclearstrategy
separately
because
it poses the risksof abandonmentand entrapment
(and consequently
thealliancedilemmaitself)in different
forms.Simplyput,nuclearaban-
492
WORLD
POLITICS
donment means the loss of U.S. will to use its strategicweapons in
defense of Western Europe; more precisely,that the credibilityof U.S.
deterrencedrops below the level required to deter the Soviets. Europe
has had reason to worry about this since the Soviet Union achieved a
second-strikecapabilityin thelate fiftiesor earlysixties:therecentadvent
of full parityapparentlyoperates as a kind of perceptualthresholdthat
has greatlyheightenedEuropean anxiety.Of course,European concerns
about such "decoupling" are not so much fears of actual U.S. nuclear
abandonment as theyare worries that U.S. strategicnuclear forcesare
no longer effectivein deterringthe Soviets. American proposals for a
policy of no firstuse do raise fearsof deliberatenuclear abandonment.
Thus, to Europe nuclear abandonmentmeans eithera withdrawalof
the U.S. nuclear deterrentor an evaporation of its credibility.Nuclear
entrapmentmeans the actual use of nuclear weapons in case deterrence
fails,especiallyin a way thatmakes Europe the principalbattleground.
Thus, the tension between nuclear abandonment and entrapmentis
virtuallyequivalent to the familiartension between deterrenceand deor a capabilityand doctrinefor
fense,the latterdefinedas war-fighting,
fightinga nuclear war "rationally."Europeans have always favoredpure
deterrenceover war-fightingpostures. Some, if not most, probably (if
secretly)would prefera strategyof pure nuclear bluff:deterrenceby a
threat of massive retaliation,but no retaliationat all if the threat is
ineffective.Nuclear war-fightingand limited nuclear war strategiesfavored by the United States are resisted for at least three reasons: (i)
they imply a greater likelihood, and greater degree, of devastation in
Europe than a strategyof assured destruction,which seems to hold out
at least some possibilityof a nuclear war being conducted intercontinentally,"over Europe's head"; (2) they might encourage the U.S. to
initiatenuclear war too casually in the event of conventionalattack or
to take excessive risks in a crisis; and (3) they are provocative to the
Soviet Union, both politicallyand militarily.
The nuclear issue presentsthe alliance dilemma to the Europeans in
its severestform,and it is the issue about which Europe is most ambivalentor divided. By itself,the dilemma is whetherto supportAmerican effortsto strengthenextended deterrence,including intermediate
nuclear forces (INF), at the risk of greater destructionin Europe if
deterrencefails, or to resistthese efforts,therebysacrificingsome deterrencein the hope of limitingdamage to Europe. As on the political
side of the dilemma, thereis a trade-offbetween the risks of abandonment and entrapment.The nuclear dilemma links up to the broader
securitydilemma throughthe risk of alienating the United States if its
ALLIANCE
SECURITY
DILEMMA
493
preferencesare opposed, and of damaging the Europe-Soviet detenteif
theyare accepted. A furtherlink is providedby theoptionof conciliating
the Soviets: if deterrenceis weakened by parity,perhaps the U.S.S.R.'s
incentivesto attack can be weakened by increasingits stake in economic
intercoursewith Western Europe-even if this means increasing Europe's vulnerabilityto Soviet influence.
European governmentshave tentativelyresolvedthe nuclear dilemma
by a contingentacceptance of INF, intended both to recouple the U.S.
strategicdeterrentto Europe and to balance offtheSoviet SS-20. Damage
to the European detentewill be minimized,theyhope, with only American fingerson the triggers.They tacitly,if reluctantly,accept the U.S.
war-fighting
doctrine.The package is roundedout bypoliticalassurances
to the Soviets-or "reinsurance,"as formerchancellor Schmidt called
it-presumably against eitherlower deterrencecredibilityif INF is not
deployed or against Soviet ire if it is. Deployment was made contingent
on a prioreffortat arms control;thesetalks,at thiswriting,have broken
down and deploymentis underway, but whetherit will be completed
as planned is uncertain.
This is not the place to rehearse all the arguments,paradoxes, and
political currentsgenerated by the INF issue. It is of interestfor our
presenttheme in that it presentsEurope's nuclear alliance dilemma in
quintessentialform:it is seen simultaneously(or by differentgroups) as
insurance against American nuclear abandonment and as a source of
European nuclear entrapment.33
Probably INF would shore up deterrence marginally. In contemplatinga conventionalattack, the Soviets
probablywould perceive a greaterlikelihood of U.S. retaliationagainst
their homeland with nuclear weapons based in Europe than with intercontinentalones, since the U.S. could at least hope the Soviets would
limit theircounter-retaliationto Europe. (Indeed, Moscow would have
a strongincentiveto so limit it to act as if the physicallocation of the
weapons rather than the nationalityof the triggerfingerdictated the
directionof theirresponse in the hope of avoiding settingoffthe whole
U.S. strategicarsenal.) Alternatively,the Soviet Union might feel that
an attack on Europe would have to startwith a preemptiveassault on
INF, and would shrink from that for fear that so provocativean act
would triggerfull-scale U.S. retaliation.But both of these deterrent
arguments lead straightto the point made by the European Left that
INF means greaterdamage to Europe in case deterrencefails,and the
firstargument supports the view that it makes the United States more
33 A similarpoint is made by RobertJ. Art in "Fixing AtlanticBridges,"ForeignPolicy,
No. 46 (Spring 1982), 77.
494
WORLD POLITICS
willing to initiate nuclear war. The Left in the European political spectrumtendsto focuson the danger of nuclear entrapmentthatINF poses,
whereas the Right emphasizes its reductionof the risk of nuclear abandonment. No doubt the moderates feel the dilemma most intensely.
CONCLUSION
The securitydilemma occurs in relations between allies as well as
between adversaries.In a multipolarsystem,the alliance and adversary
dilemmas are of roughlyequal importanceand are closelyintertwined.
Choices in each dilemma are constrainednot onlyby predictedor feared
effectsinternal to itself,but also by side effectsin the other dilemma.
In particular,strategychoices in the adversarygame-conciliation or
firmness-are constrainedby fears of abandonment or entrapmentby
allies.
The alliance dilemma is more severe in a multipolarthan in a bipolar
systembecause high mutual dependence coexistswith plausible realignment options. An ally's defectionis a calamity,yet distinctlypossible.
Conciliating the adversary,or weakening one's support of the ally to
guard againstentrapment,are bothconstrainedby fearsof abandonment.
But attemptsto insure against abandonment by supportingthe ally and
avoiding accommodation with the opponent increase the risk of entrapment;hence the dilemma. On the whole, however, abandonment
worries outweigh entrapmentfears. The mutual fear of abandonment
tends to promote convergenceof policy-typically convergenceon mutual support and firmnesstoward the adversary.
The alliance securitydilemma is less severe in a bipolar systembecause-although fears of entrapmentexist-the risks of abandonment
are low. The adversary dilemma dominates; strategychoices in that
dilemma are not much constrainedby side effectsin the alliance game.
The allies may adopt independent,indeed contradictory,
policies toward
the opponent with littlefearthatthe partnerwill defectin consequence.
Thus, the tendency is toward divergence rather than convergence of
policy.Although minor policydifferencesmay be resolvedby consensus
norms,major differencestend to persistbecause thereis littlestructural
pressuretoward theirresolution.And theirpersistencedoes not seriously
threatenthe existenceof the alliance.34
It follows that those who see NATO's currentcrisisas heralding its
collapse tend to confuse cause and effect.Although the disagreements
34 These remarksare, in effect,
an elaborationof Kenneth Waltz's axiom that flexibility
of alignmentin multipolarityproduces rigidityof strategy,while rigidityof alignmentin
of strategy.Waltz (fn. 25).
bipolaritypromotesflexibility
495
have arisen froma varietyof proximatecauses,theypersistlargely
because the alliance cannot break up. Since NATO is a product of the
of the system,it cannotcollapseor changebasically
bipolarstructure
untilthatstructure
guaranteeagainstdisintechanges.This structural
and inhibitscompromise.
grationencouragesunilateralism
Policyconflictsmay not be resolvedbecausethe costof not resolvingthemdoes
notincludea riskto theallianceitself.In contrast,
thestructural
instabilityof multipolaralliances-the factthattheycouldcollapse- tended
to promoteconflict
resolution
and policyconsistency
amongtheirmembers.
The onlywayin whichintra-NATOconflict
mightconceivably
bring
about the alliance'sdemise is througha radical revisionof the U.S.
conceptionof its interests
such thatit is no longersubjectto structural
compulsion.It is conceivablethattheU.S., disillusioned
withEuropean
obstinacy,
mightrelapseinto hemispheric
isolation.This would mean
as RobertTuckerand othershaveargued,35
recognizing,
thatAmerica's
in thenuclearage is notdependent
physicalsecurity
on keepingWestern
Europe out of Soviethands;it would also indicatea substantial
devaluationofthecultural,economic,and othernonsecurity
valuesthathave
It is also conceivablethatWestern
motivatedthe U.S. commitment.
foritsown defenseand
Europemightuniteand takefullresponsibility
of
nucleardeterrence.
This would amountto a changein thestructure
orunnecessary.
theinternational
theallianceirrelevant
system,
rendering
The likelihoodof eitherof theseeventsin the foreseeablefutureis
will
low. The alliancewill survive,whileitsinternalconflicts
extremely
continueto producemuchargumentbutlittleresolution
untiltheSoviet
Union once again appearsthreatening
to
to Europe or nonthreatening
U.S. interests
elsewhere,or untilpoliticalchangesin eithertheUnited
Statesor Europe (or in both)generatemore harmoniouspolicypreferencestowardtheadversary.
35RobertW. Tucker and Linda Wrigley,eds., The AtlanticAllianceand Its Critics(New
York: Praeger, I983), chap. 6.

The Security Dilemma in Alliance Politics

Transcription

Similar documents

Speech of the Prime Minister

Alliance Aluminum Extrusions

A Global Guide to Business Etiquette

Moscow 2015

Corporate Responsibility - Alliance Financial Group Berhad

Stony Plain Alliance Church Women`s Retreat Camp Nakamun

Star(Jones) - Tuberous Sclerosis Alliance

A Techno-Organic Cybertron, and Maximal Scientists attrap and

media kit - Alliance Wakeboard

May 2011 - Village of Alliance