Powell-AnarchyInternationalRelations-1994 (Translatix)

Review: Anarchy in International Relations Theory: The Neorealist-Neoliberal Debate
Reviewed Work(s): Neorealism and its Critics. by Robert O. Keohane: Neorealism and
Neoliberalism: The Contemporary Debate. by David A. Baldwin
Review by: Robert Powell
Source: International Organization , Spring, 1994, Vol. 48, No. 2 (Spring, 1994), pp. 313-
344
Published by: The MIT Press
Stable URL: https://www.jstor.org/stable/2706934
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].
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
, The MIT Press , Cambridge University Press and University of Wisconsin Press are
collaborating with JSTOR to digitize, preserve and extend access to International Organization
This content downloaded from

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
All use subject to https://about.jstor.org/terms
Anarchy in international
relations theory: the
neorealist-neoliberal debate
Robert Powell
Robert 0. Keohane, editor. Neorealism and Its Critics. New York: Columbia
University Press, 1986.
David A. Baldwin, editor. Neorealism and Neoliberalism: The Contemporary

Debate. New York: Columbia University Press, 1993.
Two of the most influential contemporary approaches to international relations

theory are neorealism and neoliberalism. The debate between these two
approaches has dominated much of international relations theory for the last
decade. It is now commonplace for an article about some aspect of interna-
tional relations theory to begin by locating itself in terms of this debate. These
two approaches and the debate between them have failed to contribute as
much as they might have to international relations theory. These approaches
suffer from serious internal weaknesses and limitations that the neorealist-
neoliberal debate often has tended to obscure rather than to clarify. Once we
have exposed and clarified these weaknesses and limitations, we will be able to
see several important directions for future theoretical work.
Two books, Neorealism and Its Critics and Neorealism and Neoliberalism: The
Contemporary Debate, make significant contributions to this debate. The former
offered a wide-ranging critique of neorealism when it was published in 1986.
The latter, which has just been published, is more narrowly focused. It takes up
where some of the critiques in Neorealism and Its Critics left off. A review of
these two complementary volumes affords an excellent opportunity to begin to
identify some of the weaknesses and limitations that the neorealist-neoliberal
debate frequently has obscured.
I am grateful to Carol Evans, Jeffry Frieden, Joanne Gowa, Joseph Grieco, Ernst Haas, Peter
Katzenstein, Robert Keohane, David Lake, James Morrow, John Odell, Janice Gross Stein, and
Kenneth Waltz for their thoughtful comments and criticisms of an earlier draft. I also thank Greg
Louden and Michael Sinatra for invaluable research assistance. I gratefully acknowledge the
support of a grant from the National Science Foundation, no. SES-921959.
Intemational Organization 48, 2, Spring 1994, pp. 313-44

?3 1994 by The 10 Foundation and the Massachusetts Institute of Technology

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
314 International Organization
In this review, I discuss four broad avenues of criticism that these volumes
take in evaluating neorealism and specifically Kenneth Waltz's formulation of
it.' The first three avenues are the origins of states' preferences, the agent-
structure problem, and Waltz's specific definition of political structure. These
criticisms generally do not challenge the logical coherence of neorealism. They
focus instead on the limitations of the theory. The first two center on what
neorealism takes for granted, e.g., preferences and intersubjective meanings
and understandings. The third criticism finds Waltz's definition of structure too
confining. The fourth avenue of criticism challenges the internal logic of
neorealism directly. It argues that conclusions claimed to follow from the
assumptions of neorealism actually do not. The neorealist-neoliberal debate
lies along this fourth avenue.
Three issues lie at the center of the neorealist-neoliberal debate. In
reviewing these issues, I try to bring important implicit assumptions to the fore
and show that those assumptions account for many of the important differences
between the two theories. Moreover, many of the differences that have been
thought to be significant, such as the difference between relative and absolute
gains, are not. The first issue at the heart of the debate is the meaning and
implications of anarchy. Although the notion of anarchy has served as a central
organizing concept for much of international relations theory, the emphasis on
anarchy is misplaced. What have often been taken to be the implications of
anarchy do not really follow from the assumption of anarchy. Rather, these
implications result from other implicit and unarticulated assumptions about
the states' strategic environment.
The second central issue is the problem of absolute and relative gains. I
argue that the controversy surrounding this problem generally has mistaken
effects for causes and that this mistake has handicapped analysis of the
problem of international cooperation. More specifically, I try to demonstrate
that the international relations literature generally holds, if at times only
implicitly so, that the extent to which a state is concerned about relative gains
depends on its strategic environment, for example, the offense-defense
balance and the intensity of the security dilemma. But if this is the case, then
the degree to which a state is concerned about relative gains is part of the
outcome to be explained: it is an effect and not a cause. The extent to which a
state is concerned about relative gains, therefore, does not explain the level of
international cooperation. This realization should refocus our attention on
what determines the degree of a state's concern about relative gains.
The third issue is the tension between coordination and distribution. There
are often many ways to realize the joint gains from cooperation, and these
alternatives often lead to different distributions of those gains. Thus, the
potential for joint gains usually creates distributional disputes that tend to
impede cooperation. Although these distributional concerns only recently have
1. Kenneth Waltz, Theory of Intemational Politics (Reading, Mass.: Addison-Wesley, 1979).

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
Neorealism and neoliberalism 315
begun to receive attention in the debate between neorealism and neoliberal-

ism, they hold the promise of clarifying some of the questions that actually do
divide these two approaches.
Neorealism and the structural approach
Much of the neorealist-neoliberal debate can be seen as a reaction to Waltz's

Theory of International Politics and a response to those reactions. A brief
discussion of two of that book's primary objectives is essential to understanding
the debate.2 One objective was to reiterate, reinforce, and refine a line of
argument Waltz began in Man, the State, and War.3 There, he had underscored
the importance of third-image explanations. First-image explanations locate
the causes of international outcomes, say the cause of war, "in the nature and
behavior of men. Wars result from selfishness, from misdirected aggressive
impulses, from stupidity."4 Second-image explanations locate causes in the
internal structure of the state. Imperialism, for example, results from a
particular internal economic structure like capitalism; similarly, international
peace results from a particular form of government like democracy.5 Appealing
to Rousseau's stag hunt and alluding to the then recent development of game
theory, Waltz argued that first- and second-image explanations were insuffi-
cient.6 In a situation entailing strategic interdependence, such as that of the
great powers, an actor's optimal strategy depends on the other actors'
strategies. If, therefore, we want to explain what the actors will do, then, in
addition to looking at the attributes of the actors, we must also look to the
constraints that define the strategic setting in which the actors interact. The
third image locates causes "within the state system."7
A simple example from microeconomic theory illustrates the potential
importance of third-image explanations. The price is higher and the output is
lower in a monopolized market than in a competitive one. But first- and
second-image accounts, which Waltz collectively calls reductive explanations in
Theory of Intemational Politics, do not explain these differences. In both
markets, the attributes of the actors, which are firms in this case, are identical:
every firm tries to maximize its profits and consequently produces the level of
output at which marginal cost equals marginal revenue. What accounts for the
variation in price and output between these markets is not variation in the
attributes of the units but variation in the environments or market structures in
which they act. This is the essence of the third image.
2. For a summary of Waltz's goals, see p. 323 of Kenneth Waltz, "Reflections on Theory of
Intemational Politics, " in Keohane, Neorealism and Its Critics, pp. 322-45.
3. Kenneth Waltz, Man, the State and War (New York: Columbia University Press, 1959).
4. Ibid., p. 16.
5. Ibid., pp. 80-164.
6. Ibid., pp. 172-86 and 201-5.
7. Ibid., p. 12.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
It is important to emphasize two points about the division of explanations

into reductive and systemic accounts. The first is an assumption inherent in this
division: namely, that we can usefully conceive of the actors or units in a system
as separate and distinct from the constraints that define the strategic setting in
which the units interact. The second important point is the kind of conceptual
experiment and explanation that naturally follows from this division. Once a
system has been decomposed into units and constraints, it is natural to ask one
of two questions; or, to put it differently, it is natural to consider two types of
thought experiment. First, how would some aspect of the units' behavior, say
the probability of starting a war, vary if we conceptually change some attributes
of the units while holding the constraints constant? What, for example, would
happen to the probability of war if a state's form of government were
democratic rather than authoritarian? Fixing constraints and varying units'
attributes comprise the essential conceptual experiment underlying reductive
explanations. Second, how would behavior change if the attributes of the units
remained constant and the constraints were changed? What, for example,
would happen to the probability of war if the attributes of the units were
unchanged but the distribution of power changed from bipolarity to multipolar-
ity? Fixing the units' attributes and varying the constraints facing the units comprise
the fundamental conceptual experiment underlying systemic explanations.
After emphasizing the general importance of third-image or systemic
explanations, Waltz turns to a second objective in Theory of Intemational
Politics. He sees structure as a "set of constraining conditions."8 But states may
be constrained by many things-like the distribution of power, the nature of
military technology, or the state's comparative economic advantage. A second
goal for Waltz is to specify a restricted set of constraints that provide a way of
conceiving of a political system and then to demonstrate the power of this
formulation by showing that it tells "us a small number of big and important
things."9 He restricts this set to three elements, defining a political structure in
terms of its ordering principle, the distribution of the units' capabilities, and
the functional differentiation or nondifferentiation of the units.10
Two criteria seem to have guided the selection of these elements and this
definition of political structure. The first is pragmatic. This definition appeared
to lead to interesting insights, which, of course, is the goal of all positive
theories. The second criterion is less general and reflected a trade-off. Waltz
tried to define political structure so that "it would show us a purely positional
picture.""1 The advantage of a positional picture is that many systems can be
seen as similar regardless of the particular substantive context in which the
units interact. "Structure, properly defined, is transposable."12 Thus, firms
8. Waltz, Theory of Intemational Politics, p. 73.

9. Waltz, "Reflections on Theory of Intemational Politics," p. 329.
10. Ibid., pp. 79-101.
11. Kenneth Waltz, "A Response to My Critics," in Keohane, Neorealism and Its Critics, p. 330.
12. Ibid.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
facing a high risk of bankruptcy in an oligopolistic market may be seen to be in

an anarchical, self-help system in much the same way that states facing a high
risk of war in the international system are in an anarchical, self-help system.13
If, therefore, anarchy implies certain behavior, such as the tendency for
balances of power to form, then we would expect to see this behavior obtain
"whether the system is composed of tribes, nations, oligopolistic firms, or street
gangs."'14 The potential advantage of a spare definition of a political structure is
that it may help us see similarities in what initially appeared to be very different
domains. The potential disadvantage of this spare definition is that if the three
dimensions Waltz uses to characterize systems do not sufficiently constrain the
units' interaction, then units in similar systems may not interact in similar ways.
If this is the case, then we shall have to look elsewhere for explanations of these
variations. Recognizing this trade-off, Waltz opts for a spare definition.
Four avenues of criticism
Structural theories decompose a system into units and constraints. This

decomposition makes these theories vulnerable to two broad avenues of
criticism. The first criticism accepts this decomposition but stresses the need
for a theory of preference formation to supplement the structural theory.
Because the units' preferences are exogenously specified in a structural theory,
we need a theory that explains their origins. The second avenue rejects this
decomposition. It emphasizes the agent-structure problem, arguing that agents
and structure are inseparable. In addition to these first two broad avenues of
criticism, any particular structural theory, like Waltz's formulation of neoreal-
ism, is also subject to a third and fourth avenue of criticism. The third focuses
on and questions the specific definition of structure employed in the theory.
The fourth questions whether the conclusions claimed to follow from the
theory do indeed follow.
Preferences are given exogenously
The first avenue of criticism centers on preferences. Structural approaches

take the units' preferences as given. That is, these preferences are exogenously
specified. They become inputs into the analysis rather than the subject of
analysis. This may be an important weakness of the structural approach. As
Robert Jervis cautions, "By taking preferences as given, we beg what may be
the most important question on how they were formed.... Economic theory
treats tastes and preferences as exogenous. Analysis is therefore facilitated, but
13. Waltz, Theory of Intemational Politics, pp. 105 and 111.

14. The quotation is from p. 37 of Kenneth Waltz, "Realist Thought and Neorealist Theory,"
Joumal of IntemationalAffairs 44 (Spring/Summer 1990), pp. 21-37.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
at the cost of drawing attention away from areas that may contain much of the
explanatory 'action' in which we are interested."'15
The first step in assessing the force of the criticism that structural approaches
lack a theory of preferences is to clarify the criticism by distinguishing two types
of preferences. The first type is preferences over outcomes; the second is
preferences over actions or policies. To differentiate these two types, consider a
game in payoff-matrix form. The cells in the matrix correspond to potential
outcomes. The utilities that appear in each cell in the matrix represent the
players' preferences over these potential outcomes. That is, a player's utilities
reflect its preference ranking of the possible outcomes. Given its preferences
over outcomes and its beliefs about what the other players are doing, a player
can rank its potential actions from most to least preferred. In a two-person
game, for example, the row player can rank its actions from best to worst given
its payoffs and its beliefs about what the column player is doing. This induced
ranking defines a player's preferences over actions.'6
Structural theories do not try to explain preferences of one type but do try to
explain preferences of the other type. Structural theories take the units'
preferences over possible outcomes as given and, consequently, lack a theory of
preferences over outcomes. But structural theories try to make predictions
about the units' preferred actions by combining assumptions about the units'
preferences over outcomes with other assumptions about the structural
constraints facing the units. In this sense, structural theories claim to be a
theory of preferences over actions. Game theory, for example, is a theory of
preferences over actions. It attempts to predict the units' optimal actions based
on their preferences over outcomes and the strategic setting in which they
interact. Similarly, Waltz's formulation of neorealism takes the units' prefer-
ences as given. "In a microtheory, whether of international politics or of
economics, the motivation of the actors is assumed rather than realistically
described."'17 In particular, Waltz assumes "that states seek to ensure their
survival" and then attempts to predict the units' actions, albeit in a very general
way, on the basis of this assumption about the units' preferences and other
assumptions about the political structure in which the units interact.18
The two types of preferences are frequently conflated. For example, after
noting that "economic theory takes tastes and preferences as exogenous" and
warning that we may be begging the most important questions by doing so,
Jervis discusses some of the sources of these tastes and preferences over
outcomes. These sources include transnational forces, ideologies, beliefs,
15. Robert Jervis, "Realism, Game Theory, and Cooperation," World Politics 40 (April 1988),
pp. 324-25. For similar warnings, see Joseph Nye, "Neorealism and Neoliberalism," World Politics
50 (January 1988), p. 238.
16. The distinction between preferences over outcomes and over actions is useful, but it should
not be pushed too hard. An outcome in one game may be seen as a policy choice in a larger game.
17. Waltz, Theory of Intemational Politics, p. 91.
18. The quotation is drawn from ibid.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
experience, and knowledge.19 He also sees realism as a source or theory of

preferences over outcomes, saying, "Sometimes we can deduce preferences
from the structure of the system, as Realism suggests. But even a structural
theory of international politics as powerful as Waltz's has trouble producing
precise deductions."20 Jervis confounds the two types of preferences here. He
correctly observes that economic theory takes preferences over outcomes as
given but then treats neorealism, which is a theory of preferences over actions,
as a theory of preferences over outcomes. Robert Keohane similarly conflates
the two types of preferences when he intends to criticize neorealism as a weak
theory of behavior (that is, a weak theory of preferences over actions) but
describes neorealism as a weak theory of preferences over outcomes.2'
Conflating the two types of preferences has at least two negative effects. The
first is to suggest that we cannot use structural or game-theoretic approaches,
which take preferences as given, to study the effects on preferences of changes
in beliefs, experience, or knowledge.22 This suggestion is simply wrong if what
we want to study is how changes in these factors affect preferences over actions
or policies. Indeed, one of the primary uses of incomplete-information games is
to study how interaction affects players' beliefs and, through these beliefs, their
preferred actions. Andrew Kydd, for example, develops an interesting incom-
plete-information model of arms races that he uses to study Jervis's spiral
model of escalation.23 The basic issue in Kydd's game is whether a state will
change from preferring not to arm to preferring to arm because it interprets
another state's arms increase as a sign of hostility rather than of insecurity.
Kydd uses this model to study the circumstances in which two states that have
no hostile intent might arm and eventually go to war because they fear that the
other is hostile. The formal study of dynamic interactions and the learning and
signaling inherent in them is at an early stage in international relations theory.
Many legitimate criticisms can be made of this work.24 But the claim that this
work has nothing to say about learning and changes in preferences (over
19. Jervis, "Realism, Game Theory, and Cooperation," pp. 324-29.

20. Ibid., p. 325.
21. Robert Keohane, "Theory of World Politics," in Keohane, Neorealism and Its Critics, pp.
175-76. One factor contributing to this conflation may be that both Jervis and Keohane focus
primarily on the prisoners' dilemma. There is no strategic interdependence in a one-shot prisoners'
dilemma: a player always does strictly better by playing D rather than C regardless of what the
other player does. In cases in which a player's optimal action is independent of what others do, a
theory of preferences over outcomes also serves as a theory of preferences over actions. The
distinction between the two types of preferences is meaningful only if the game entails a situation
of strategic interdependence in which a player's optimal strategy depends on what it believes others
will do.
22. Jervis, "Realism, Game Theory, and Cooperation," p. 327.
23. Andrew Kydd, "The Security Dilemma, Game Theory, and World War I," paper presented
at the annual meeting of the American Political Science Association, Washington, D.C., 2-5
September 1993. For Jervis's insightful discussion of the spiral model, see his Perception and
Misperception in International Politics (Princeton, N.J.: Princeton University Press, 1977).
24. For an excellent review of some of the limitations of this approach, see David Kreps, Game
Theory and Economic Modelling (New York: Oxford University Press, 1990).

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
actions) because it takes preferences (over outcomes) as given is not one of

these criticisms.
The second negative effect of conflating the two types of preferences is that
doing so confounds two objections to structural approaches that need to be
evaluated separately. The first objection is that these approaches take the units'
preferences as given. The second is that these approaches offer at best very
weak theories of preferences over actions and at worst misleading theories. As
will be seen, the first objection is not very important to the neorealist-
neoliberal debate, while the second objection lies at the heart of it.
The significance of taking the units' preferences as given in a theory or model
depends very much on the theory or model and the purposes for which it has
been constructed. In some models of nuclear crisis bargaining, for example,
there are only three outcomes: a state prevails in the crisis, it backs down but
avoids a nuclear exchange, or the crisis ends in nuclear war.25 Preferences over
these outcomes are exogenously given in these models, but it would seem
bizarre not to assume that a state prefers the first outcome to the second and
the second to the third. Many situations, however, are much more complicated
and what to assume about preferences over outcomes is not obvious. It is not
clear, for example, what to assume about a state's preferences over possible
trade arrangements. Here the work of Jeffry Frieden, Peter Gourevitch, Peter
Katzenstein, David Lake, Helen Milner, Ronald Rogowski, and others in
developing an understanding of the origins of preferences is very important.26
Similarly, a state's preferences over potential national security arrangements,
for example, possible arms control agreements, may not be obvious, and
theories may be needed to explain these preferences.27
That neorealism takes the units' preferences as given is of little consequence
for the neorealist-neoliberal debate. As will be developed more fully below,
this debate largely focuses on the likelihood of cooperation in anarchy and on
the role of institutions in facilitating cooperation. Neorealism maintains that
cooperation will be difficult in an anarchic system composed of units that prefer
survival over extinction. Neoliberalism questions this conclusion but not the
25. See, for instance, the models of nuclear brinkmanship in Robert Powell, Nuclear Deterrence
Theory (Cambridge: Cambridge University Press, 1990).
26. See Jeffry Frieden, "Invested Interests," Intemational Organization 45 (Autumn 1991), pp.
425-51; Peter Gourevitch, Politics in Hard Times (Ithaca, N.Y.: Cornell University Press, 1986);
Peter Katzenstein, ed., Between Power and Plenty (Madison: University of Wisconsin Press, 1978);
David Lake, Power, Protection, and Free Trade (Ithaca, N.Y.: Cornell University Press, 1988); Helen
Milner, Resisting Protectionism (Princeton, N.J.: Princeton University Press, 1988); and Ronald
Rogowski, Commerce and Coalitions (Princeton, N.J.: Princeton University Press, 1989).
27. For example, Adler uses the concept of epistemic communities to explain American
preferences about arms control agreements. See Emanual Adler, "The Emergence of Cooperation
International Organization 46 (Winter 1992), pp. 101-46. For attempts to explain a state's
preferences over military doctrines and the importance of civil-military relations in determinin
those preferences, see Barry Posen, The Origins of Military Doctrine (Ithaca, N.Y.: Cornell
University Press, 1984); Jack Snyder, The Ideology of the Offensive (Ithaca, N.Y.: Cornell University
Press, 1984); and Stephen Van Evera, "The Cult of the Offensive and the Origins of the First
World War," International Security 9 (Summer 1984), pp. 58-107.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
assumption that units are minimally motivated to survive. Indeed, it would

seem bizarre not to assume that units prefer survival over extinction. Thus, the
criticism that these preferences are specified exogenously is unimportant to the
debate about the likelihood of cooperation in anarchy. The potentially
important criticism is that the conclusions claimed to follow from neorealism's
spare assumption about units' preferences and about the political structure in
which these units interact actually do not follow. This is the fourth avenue of
criticism, which will be discussed below.
The inseparability of agents and structure
The structural approach decomposes a system into units and the constraints
facing them. The second avenue of criticism denies the separability of agents
and structure. Drawing on structurationist theories in sociology, Alexander
Wendt argues that agents and structure are "mutually constitutive yet
ontologically distinct entities. Each is in some sense an effect of the other; they
are 'co-determined.' "28
If agents and structure were conceptually inseparable, two consequences
would follow. First, the two conceptual experiments underlying the structural
approach from which this approach derives its explanatory power would
become problematic. We would no longer be able to study the constraining
effects of structure by theoretically holding the units and their preferences
constant while varying the structure in which they interact. If units and
structure are inseparable so that each is at least partly the effect of the other,
then variation in the structure will also change the units.
Second, challenging the separability of units and structure makes the units
an object of inquiry and directs our attention to systemic change and
transformation. If units and structure are mutually constitutive, then it is
natural to ask, How do they evolve, and How do they interact over time?
Thinking of the units as being endogenous shifts our attention away from a
positional model to what David Dessler calls a transformational model. In a
positional model like Waltz's formulation of neorealism, "structure is an
environment in which action takes place. Structure means the 'setting' or
'context' in which action unfolds."29 Structure is, in other words, a set of
constraints. In a transformational theory, "structure is a medium of activity that
in principle can be altered through activity."30 Structure shapes action and is
shaped by action. The goal, therefore, of a transformational theory is to explain
how structure and agent interact. To do this, Robert Cox, Dessler, John
28. See p. 360 of Alexander Wendt, "The Agent-Structure Problem in International Relations
Theory," International Organization 41 (Summer 1987), pp. 335-70.
29. The quotation is from p. 426 of David Dessler, "What's at Stake in the Agent-Structure
Debate," Intemational Organization 43 (Summer 1989), pp. 441-70, emphasis original.
30. Ibid., p. 461.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
Ruggie, Wendt, and others have emphasized identities, interests, rules, roles,
and intersubjective understandings and meanings.3'
As with the first avenue of criticism, the force of the second avenue depends
very much on the particular theory or model being criticized. Cox's distinction
between problem-solving theories and critical theories is helpful here.32 The
former uses the ceteris paribus assumption to restrict the statement of a
specific problem "to a limited number of variables which are amenable to a
relatively close and precise examination."33 Among the many things that
problem-solving theories may exclude by taking them as given and unproblem-
atic are intersubjective understandings and expectations. The ceteris paribus
assumption effectively freezes and thereby assumes away the interaction of
units and structure.
It seems entirely appropriate to assume away this interaction in a problem-
solving theory as long as the applicability or domain of the theory is understood
to be bounded by the ceteris paribus assumption. Structurationists rightly argue
that intersubjective understandings are part of what is being taken as given or
unproblematic in this assumption. If these understandings and meanings differ
significantly from those presumed in the ceteris paribus assumption, then
theories predicated on that assumption may be of little use. Of course, the
ceteris paribus conditions-be they about interests and identities or about the
many other factors left out of a specific theory-are never strictly satisfied. We
do not know a priori whether differences in interests and identities or in the
other excluded factors are important. The best we can do is try to determine the
domain of applicability of problem-solving theories by using them in different
settings. Powerful theories will work in a large domain because the excluded
factors subsumed in the ceteris paribus assumption generally are insignificant.
Weak theories will have a very limited domain. The sociological approach
makes a serious and important criticism and contribution in stressing the
importance of intersubjective meanings and understandings and the interac-
tion between agents and structure.
The sociological approach stresses the inseparability of units and structure.
But it is important not to identify this criticism with this particular approach. A
second line of research is also predicated on the interaction of units and
structure or, more precisely, the interaction of states and the international
structure. The essence of Gourevitch's second-image-reversed argument is that
31. See Robert Cox, "Social Forces, States, and World Orders," in Keohane, Neorealism and Its
Critics, pp. 204-54; Dessler, "What's at Stake in the Agent-Structure Debate?"; John Ruggie,
"Continuity and Transformation in World Polity," in Keohane, Neorealism and Its Critics; John
Gerard Ruggie, "Territoriality and Beyond," International Organization 47 (Winter 1993), pp.
139-74; Wendt, "The Agent-Structure Problem in International Relations Theory"; and Alex-
ander Wendt, "Anarchy is What States Make of It," International Organization 46 (Spring 1992),
pp. 391-425.
32. Cox, "Social Forces, States, and World Orders," p. 208.
33. Ibid., p. 208.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
the international structure shapes domestic institutions and states' prefer-

ences: states and structure interact.
This second form of the criticism that agents and structure are inseparable is
important for two reasons. First, it shows that the agent-structure problem may
arise even in rationalist approaches that take interests and identities as given
and assume that the units act in their own narrow self-interest. Second, the
existing literature illustrates a way of trying to deal with this form of the
agent-structure problem. The potential solution is to redefine the units in the
system. Rather than treating states as unitary actors, states are decomposed
into more basic units. The hope here is that we will be able to separate these
more basic units from the constraints facing them.
To illustrate this approach to dealing with the interaction of states and the
international structure, consider Rogowski's work on the effects of interna-
tional trade on domestic political alignments and states' preferences.34 At the
risk of doing the subtlety of his analysis grave injustice, Rogowski decomposes a
country into three groups or units: landowners, capitalists, and labor. A state's
preferences emerge through competition among these units. Moreover, any-
thing that significantly affects the terms of international trade shifts the
distribution of domestic political power among the units. For example,
technological or political changes, like the advent of railroads and steamships
or the rise of British hegemony, reduce the cost or risk of international trade.
These changes favor and enrich domestic groups that benefit from greater
trade. By assumption, benefited groups become more powerful and the state's
preferences generally become more reflective of the preferences of these
favored groups.35
Rogowski's analysis illustrates a rationalist version of the agent-structure
problem or, more accurately, the state-structure problem. Capital, land, and
labor in Rogowski's argument are acting in their own material self-interest.
Changing intersubjective meanings and understandings are not at issue here.
Nevertheless, we cannot decompose the international system into units and
structure if we treat states as the units. For example, a change in the
international system, like the rise of British hegemony, that reduces the risk
and therefore increases the expected return to international trade will also
34. Rogowski, Commerce and Coalitions.

35. Rogowski readily acknowledges that he is making assumptions about the domestic political
process and does not have a theory of the state. He also emphasizes that although changes in the
terms of trade may make some domestic groups more powerful, they still may lose in the domestic
political struggle (ibid., pp. 4-5). The power of Rogowski's analysis, of course, lies in its ability to
identify the groups that will benefit from greater trade and the domestic cleavages that greater
trade will tend to create. Appealing to the Stolper-Samuelson theorem, Rogowski argues that
greater trade favors the domestic group that controls the relatively abundant factor. So, for
example, land was abundant and capital and labor were scarce in the United States in the latter
part of the nineteenth century, while labor was abundant and capital and land were relatively
scarce in Germany. Accordingly, agriculture in the United States and labor in Germany should
have supported greater openness, while capital and labor in the United States and capital and land
in Germany should have united in support of protectionism (pp. 3-20).

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
tend to change states' preferences. States and structure are interdependent;

each is in part an effect of the other.
This brief illustration also suggests a way of trying to deal with this version of
the agent-structure problem: namely, to decompose the system into different
units that hopefully can be separated from the structure constraining them. In
effect, we enlarge the game by trying to break what we previously took to be a
unitary actor, namely the state, into more basic units.36 Of course, enlarging the
game to include the interaction between domestic and international politics
makes any analysis much more difficult. Needless to say, an approach to dealing
with the inseparability of states and structure is not a theory of their
interaction. Much important work remains to be done on the interaction of
states and structure.37
In sum, theories that take intersubjective meanings and understandings as
given assume away one form of agent-structure interaction. Structural theories
that take the state to be a unitary actor also assume away a rationalist form of
agent-structure interaction. These theoretical simplifications may be appropri-
ate for some questions and not for others. We need to do more to identify those
domains in which this interaction can be disregarded and those in which it
cannot.
Waltz's definition of structure
The first and second avenues of criticism are directed at the structural
approach in general. The third and fourth avenues of criticism apply more
specifically to neorealism and to Waltz's particular formulation of it. The third
criticism focuses on Waltz's spare definition of structure and generally argues
that other elements be included in the description of a system's structure.
Waltz defined a political structure by its ordering principle, the distribution
of capabilities, and the functional differentiation or nondifferentiation of the
units. This definition thus implies that the nuclear revolution in military
technology is a unit-level change and not a structural change.38 Joseph Nye
finds it "particularly odd to see nuclear technology described as a unit
characteristic."39 He and Keohane argue that such factors as "the intensity of
international interdependence or the degree of institutionalization of interna-
tional rules do not vary from one state to another on the basis of their internal
characteristics ... and are therefore not unit-level factors."40 They conclude
36. Clearly this approach does nothing to address the important concerns raised in the
sociological approach to the agent-structure problem.
37. For suggestive discussions of the interaction between states and structure in different
substantive contexts, see Brian Downing, The Military Revolution and Political Change (Princeton,
N.J.: Princeton University Press, 1993); Katzenstein, Between Power and Plenty; and Charles Tilly,
Capital and Coercion (New York: Blackwell, 1990).
38. Waltz, "Reflections on Theory of International Politics, " p. 327.
39. Nye, "Neorealism and Neoliberalism," p. 243.
40. Joseph Nye and Robert Keohane, "Power and Interdependence Revisited," Intemational

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
that "making the unit level the dumping ground for all unexplained variance is
an impediment to the development of theory."41
It is clear why Waltz would not want to include military technology in his
definition of structure. Recall that one of his goals in fashioning his definition
was to give a purely positional picture of a system so the notion of structure
would be transposable from one substantive context to another. One can
readily transpose the idea of the distribution of capabilities from the interna-
tional system where states are the units to, for example, an oligopolistic market
where firms are the actors. But what is the analogue to having a secure,
second-strike force for a firm in an oligopoly? Including military technology in
the definition of structure would seem to make the concept less transposable.
Of course, greater transposability comes at a cost. Waltz's theory cannot
account for variations in outcomes like the probability of war that may be due
to the nuclear revolution. To understand those effects, we have to look to other
theories.
Although it is evident why Waltz would not want to include dimensions like
military technology in his notion of structure given his goal of transposability,
why should the distribution of capabilities across states "be included in the
definition and not other characteristics of states that could be cast in
distributional terms?"42 The answer seems to be a pragmatic one. Waltz
believes that state "behavior varies more with differences of power than with
differences in ideology, in internal structure of property relations, or in
governmental form."43 That is, Waltz believes that a definition of structure
based on the distribution of capabilities rather than on the distribution of
something else seems more likely to have greater explanatory power.44 In
evaluating the theory based on this definition, part of what is being evaluated is
the usefulness of focusing on the distribution of capabilities.
Notwithstanding the prevalence of criticisms of Waltz's spare definition of
structure, there is often a certain hollowness to debates about the proper
definition of structure. Surely the effects of, say, the nuclear revolution on
international politics do not depend on whether we attach the appellation
"4structural" or "unit-level" to this change. Putting a high value on transposabil-
ity, Waltz opted for a definition that made the concept of structure more
readily transposable. Other theorists working on other questions may value
transposability less and may define structure differently. The important issue,
however, is not whether the consequences of the nuclear revolution, different
forms of property relations, varying degrees of institutionalization, or changes
Organization 41 (Autumn 1987), pp. 725-53, and especially p. 746, from which the quotation is
drawn.
41. Ibid.
42. Waltz, "Reflections on Theory of International Politics, " p. 329.
43. Ibid.
44. Buzan, Jones, and Little make a similar point in Barry Buzan, Charles Jones, and Richard
Little, The Logic ofAnarchy (New York: Columbia University Press, 1993), pp. 54-56.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
in other sets of constraints are called "structural" or something else. The issue
is to develop theories that explain these consequences.45 When we debate what
to call these changes rather than develop and test theories about the
consequences of these changes, we appear to believe that the name implies the
consequences.
Neorealism and its implications
The neorealist-neoliberal debate develops primarily along the fourth avenue

of criticism. This criticism questions the conclusions claimed to follow from
Waltz's assumptions and those of neorealism more generally. Neorealism, for
example, claims that international institutions play a minimal role in shaping
international politics and that the prospects for cooperation in anarchy are
bleak.46 Neoliberalism questions these claims in two ways. First, it challenges
the logical coherence of the neorealist argument by trying to show that there is
a mistake in the logic. Second, neoliberalism argues that the explanatory power
of neorealism is weak when compared to neoliberalism. I trace the develop-
ment of this criticism and the neorealist-neoliberal debate in the remainder of
this section. I examine three major disputes in the debate in the next section.
In his contribution to Neorealism and Its Critics, Keohane surveys the
neorealist research program and questions its predictive power.47 He then
describes what a "modified structural research program" would look like. It
would "pay much more attention to the roles of institutions and rules than does
Structural Realism. Indeed, a structural interpretation of the emergence of
international rules and procedures, and of obedience to them by states, is one
of the rewards that could be expected from this modified structural research
program.1148
Keohane challenges neorealism more directly and develops an institutional
approach more fully in his work, After Hegemony.49 The central question is,
"Under what conditions can independent countries cooperate in the world
political economy?"50 Can, for example, states cooperate in the absence of a
hegemon? Keohane begins his analysis of this question "with Realist insights
about the role of power.... [Keohane's] central arguments draw more on the
Institutionalist tradition, arguing that cooperation can under some conditions
develop on the basis of complementary interests and that institutions, broadly
defined, affect the patterns of cooperation that emerge."'51 In short, institutions
45. For a recent effort to do this, see ibid.

46. Joseph Grieco, "Anarchy and the Limits of Cooperation," in Baldwin, Neorealism and
Neoliberalism, pp. 116-42 and pp. 118-19 in particular.
47. Keohane, "Theory of World Politics."
48. Ibid., p. 194.
49. Robert Keohane, After Hegemony (Princeton: Princeton University Press, 1984).
50. Ibid., p. 9.
51. Ibid., p. 9.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
may be a significant factor in promoting international cooperation in ways that

neorealism has failed to appreciate.
Keohane challenged both the logical coherence and the explanatory power
of neorealism. Attacking the logic, Keohane writes, "I propose to show, on the
basis of their own assumptions, that the characteristic pessimism of Realism
does not follow. I seek to demonstrate that Realist assumptions about world
politics are consistent with the formation of institutionalized arrangements,
containing rules and principles, which promote cooperation."52 In sum,
Keohane intends to start with the same set of core assumptions that neorealism
does and then show that cooperation is compatible with these assumptions.
Keohane attempts this demonstration in the context of the repeated
prisoners' dilemma. There are two steps to the demonstration. The first is to
argue that the repeated prisoners' dilemma is a reasonable model for the
international system envisioned in neorealism, that is, that this model is
compatible with realism's central assumptions about the international system.
Although he does not develop this point at length, Keohane claims, "Not all
situations in world politics or international political economy take the form of
Prisoner's Dilemma, but many do."53 As further support for the claim that the
repeated prisoners' dilemma is generally seen to be compatible with realism's
basic assumptions, he might also have referred to Jervis's belief that this game
is an appropriate model for studying the security dilemma.54 The second step in
Keohane's argument is to appeal to the Folk theorem, which shows that the
mutually cooperative outcome can occur in equilibrium in an infinitely
repeated prisoners' dilemma if the actors do not discount the future too
much.55 These two steps taken together imply that cooperation is compatible
with realism.
Writing in 1983, Keohane believed his neoliberal institutional approach
would prove to have greater explanatory power than neorealism. But, a
definitive test of his institutional approach was not yet possible because the
world was "only just entering the posthegemonic era."56 It was too soon to test
the explanatory power of an argument that predicted that international
institutions and cooperation would persist despite the absence of a hegemon.
Instead of a test, Keohane offered a "plausibility probe" of his institutional
approach in the cases of international trade, finance, and petroleum.57
One can envision two general types of response to Keohane's neoliberal
challenge to neorealism. The first addresses Keohane's challenge to the logic of
52. Ibid., p. 67.

53. Ibid., p. 68.
54. Robert Jervis, "Cooperation Under the Security Dilemma," World Politics 30 (January
1978), pp. 167-214 and p. 170 in particular.
55. Drew Fudenberg and Eric Maskin, "The Folk Theorem in Repeated Games with
Discounting or with Incomplete Information," Econometrica 54 (October 1986), pp. 533-54.
56. Keohane, After Hegemony, p. 218.
57. See Robert Keohane, "Institutionalist Theory and the Realist Challenge After the Cold
War," in Baldwin, Neorealism and Neoliberalism, pp. 269-301, and particularly p. 292; and
Keohane, After Hegemony.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
neorealism's analysis of the problem of international cooperation. This

response would show that Keohane had really not based his argument on the
same set of core assumptions that neorealism does. If this were the case, then
Keohane's argument that neorealism's conclusions about the prospects of
international cooperation do not follow from its assumptions would be invalid.
If, more specifically, the repeated prisoners' dilemma is incompatible with
neorealism's core assumptions about the international system, then showing
that cooperation in this game is possible would say nothing about what follows
from neorealism's assumptions.
The second type of response is more empirical. It would say that neorealism
never claimed that international cooperation was logically incompatible with
neorealism's assumptions. So, showing that cooperation is possible given these
assumptions does not contradict neorealism. The real question is how much
international cooperation exists and whether neorealism or neoliberalism does
a better job of accounting for the observed pattern of international coopera-
tion. This response would then go on to compare the relative explanatory
power of these two approaches.
Joseph Grieco developed both types of response to the institutionalist
challenge.58 He argued that Keohane had not started with the same assump-
tions neorealism does. In using the repeated prisoners' dilemma, Keohane
implicitly had assumed that states try to maximize their absolute gains.
According to Grieco, however, neorealism requires a state's utility function to
reflect a concern for relative gains.59 Consequently, Keohane does not "show,
on the basis of their [realists'] own assumptions, that the characteristic
pessimism of Realism does not follow," as he claimed.60 In Cooperation
Among Nations, Grieco tried to assess the relative explanatory power of
neorealism and institutionalism. He considers the case of negotiations over
nontariff barriers during the Tokyo Round of the General Agreement on
Tariffs and Trade, a case that he believes poses a hard test for realism, and
concludes that realism explains this case better than does institutionalism. The
latest round of the debate between realism and liberalism was now fully
engaged.6'
David Baldwin brings a number of previously published contributions to this
debate together in Neorealism and Neoliberalism. 62 Baldwin provides an
58. See the following works of Joseph Grieco: "Anarchy and the Limits of Cooperation";
"Realist Theory and the Problem of International Cooperation," Journal of Politics 50 (Summer
1988), pp. 600-624; and Cooperation Among Nations (Ithaca, N.Y.: Cornell University Press, 1990).
59. Grieco, "Anarchy and the Limits of Cooperation," p. 129. Gowa made the same criticism of
Axelrod's use of the repeated prisoners' dilemma [Robert Axelrod, The Evolution of Cooperation
(New York: Basic Books, 1984)] when he used this game to model international politics. See
Joanne Gowa, "Anarchy, Egoism, and Third Images," International Organization 40 (1986), pp.
167-86 and particularly pp. 172-79.
60. Keohane,After Hegemony, p. 67.
61. See Nye, "Neorealism and Neoliberalism," and the references cited therein for an
introduction to earlier rounds of this debate.
62. These contributions are: Robert Axelrod and Robert Keohane, "Achieving Cooperation

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
overview of the debate, and Grieco and Keohane offer their reflections and
appraisals of the debate in new essays. This volume complements and extends
some of the lines of analysis developed in Neorealism and Its Critics. Neorealism
and Its Critics includes both internal and external critiques of neorealism. The
former share neorealism's problem-solving approach, while the latter adopt a
critical approach.63 The scope of Neorealism and Neoliberalism is narrower,
more focused, and wholly internal. Ali of the contributions exemplify the
problem-solving approach and address various facets of the neorealist-
neoliberal debate. Three issues have dominated this debate, and an assessment
of it requires an examination of each.
At issue
The three issues at the center of neorealist-neoliberal debate are the meaning
and implications of anarchy, the problem of absolute and relative gains, and the
tension between cooperation and distribution. In what follows, I make three
points about these issues. First, although anarchy is often taken to be a
fundamental organizing concept in international relations theory, the emphasis
on anarchy is misplaced. What have often been taken to be the implications of
anarchy do not really follow from that assumption. Rather, these implications
result from other implicit and unarticulated assumptions about states' strategic
environment. Second, the controversy over the problem of absolute and
relative gains generally has mistaken effects for causes in its analysis of the
prospects for international cooperation. Finally, although the debate only
recently has begun to consider distributional concerns, the analysis of these
concerns may help to clarify the differences that do divide neorealism and
institutionalism.
The meaning and implications of anarchy
Much of the neorealist-neoliberal debate centers on the meaning and

implications of anarchy. According to Grieco, neorealism entails five proposi-
Under Anarchy," World Politics 38 (October 1988), pp. 226-54; Grieco, "Anarchy and the Limits of
Cooperation"; Stephen Krasner, "Global Communications and National Power," World Politics 43
(April 1991), pp. 336-66; Charles Lipson, "International Cooperation in Economic and Security
Affairs," World Politics 37 (October 1984), pp. 1-23; Michael Mastanduno, "Do Relative Gains
Matter?" International Security 16 (Summer 1991), pp. 73-113; Helen Milner, "The Assumption of
Anarchy in International Relations Theory," Review of International Studies 17 (January 1991), pp.
67-85; Robert Powell, "Absolute and Relative Gains in International Relations Theory,"American
Political Science Review 85 (December 1991), pp. 1303-20; Duncan Snidal, "Relative Gains and the
Pattern of International Cooperation," American Political Science Review 85 (September 1991), pp.
701-26; and Arthur Stein, "Coordination and Collaboration," International Organization 36 (Spring
1982), pp. 294-324.
63. For an example of the former, see Keohane, "Theory of World Politics"; for one of the
latter, see Richard Ashley, "The Poverty of Neorealism," in Keohane, Neorealism and Its Critics,
pp. 255-300; and Cox, "Social Forces, States, and World Orders."

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
tions. He defines the last three of these as "Third, international anarchy is the
principle force shaping the motives and actions of states. Fourth, states in
anarchy are preoccupied with power and security, are predisposed towards
conflict and competition, and often fail to cooperate even in the face of
common interests. Finally, international institutions affect the prospects for
cooperation only marginally."64 The point of departure for Keohane's analysis
in After Hegemony was to use the prisoners' dilemma to show that anarchy did
not imply a lack of cooperation. Grieco responded by arguing that Keohane's
model was misspecified because he neglected states' concerns for relative gains.
Duncan Snidal then tried to show that anarchy does not imply a lack of
cooperation even if states are concerned with relative gains.65
A review of the neorealist-neoliberal debate about the meaning and
implications of anarchy shows that our continuing emphasis on anarchy is
misplaced. Many of the purported implications of anarchy may be more
usefully traced to other assumptions about the constraints facing the units. This
suggests that we should focus less attention on anarchy and much more
attention on characterizing the strategic settings in which the units interact.
In reviewing the debate about anarchy, it is necessary to begin by distinguish-
ing between two formulations of anarchy. The first is that anarchy means the
"lack of a common government" that can enforce agreements among the states
or more generally among the units.66 Robert Art and Jervis together explain
that "international politics takes place in an arena that has no central
governing body. No agency exists above individual states with authority and
power to make laws and settle disputes. States can make commitments and
treaties, but no sovereign power ensures compliance and punishes deviations.
This-the absence of a supreme power-is what is meant by the anarchic
environment of international politics."67
It is important to emphasize that this formulation of anarchy says nothing
about the means the units have at their disposal as they try to further their ends.
It says only that no higher authority exists that can prevent them from using the
means they have. Thus, for Waltz, firms facing a high risk of bankruptcy may be
in an anarchic self-help system even though the means available to them to
further their interests, like cutting prices or forming alliances to distribute the
64. Grieco, "Anarchy and the Limits of Cooperation," pp. 118-19.

65. Snidal, "Relative Gains and the Pattern of Cooperation." For Grieco's critique of Snidal's
analysis and Snidal's response, see Joseph Grieco, Robert Powell, and Duncan Snidal, "The
Relative Gains Problem for International Cooperation," American Political Science Review 87
(September 1993), pp. 729-43.
66. The quotation is from p. 226 of Axelrod and Keohane, "Achieving Cooperation Under
Anarchy." Also see Kenneth Oye, "Explaining Cooperation Under Anarchy," in Kenneth Oye, ed.,
Cooperation Under Anarchy (Princeton, N.J.: Princeton University Press, 1986), particularly pp.
1-2.
67. Robert Art and Robert Jervis, International Politics, 3d ed. (Boston: Harper Collins), p. 1.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
costs of research and development, have nothing to do with the use of military
force, which is one of the means available to states in the international system.68
One advantage of defining anarchy without reference to the means available
to the units is that it makes the concept of anarchy readily transposable to
different substantive domains. As discussed above, Waltz weighed this advan-
tage heavily in constructing his formulation of structure, so it is hardly
surprising that he would adopt this first definition of anarchy. But he certainly
is not alone, as Milner's survey of different concepts of anarchy shows.69
The second notion of anarchy refers to the means available to the units. In
"Coordination and Collaboration," Arthur Stein begins by observing that many
international relations scholars use anarchy to describe "the classic character-
ization of international politics as relations between sovereign entities dedi-
cated to their own self-preservation, ultimately able to depend only on
themselves, and prepared to use force."70 In effect, this second formulation
adds another dimension to the lack of a central authority: namely, that one of
the means available to the units is the use of force.
The addition of this second dimension has two consequences. First, it makes
the transposability of the concept of anarchy more problematic. What, for
example, is the analogue to using force for a firm facing a high risk of
bankruptcy? If there is no analogue, then a group of firms facing a high risk of
bankruptcy would not form an anarchic system according to this definition. If
we want to argue that there is an analogue, what are the criteria for establishing
that one of the means open to a firm is analogous to a state's ability to resort to
force? Of course, a definition of anarchy that reduces its transposability may
have compensating advantages. Whether these potential advantages outweigh
the disadvantage of a less transposable definition will be discussed below.
Second, adding another dimension raises important questions for interna-
tional relations theory. Do the patterns of behavior generally associated with
anarchic systems, such as the tendencies for balances of power to form and-at
least for neorealists-the limited prospects for international cooperation,
result from the lack of a central authority? Or, are these patterns more heavily
influenced by implicit and unarticulated assumptions about, say, the nature of
military force that are subsumed in the second definition of anarchy?
Two arguments suggest that our emphasis on anarchy has been misplaced if
by anarchy we mean the lack of a central authority. These arguments suggest
that conclusions often claimed to follow from the absence of a central authority
do not. These conclusions require other supporting assumptions. The first
argument is really an empirical observation. Keohane notes in his assessment
of the debate between neorealism and neoliberalism that the modern state
68. Waltz, Theory of International Politics, pp. 105 and 111.

69. Milner, "The Assumption of Anarchy in International Relations Theory."
70. Stein, "Coordination and Collaboration," p. 30.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
system, conventionally dated from 1648, has always been anarchic in the sense
that it lacked a common government.71 Thus, anarchy, while perhaps a
necessary condition, is certainly not sufficient to explain any of the variation in
international politics during the modern era. In particular, anarchy cannot
account for whatever variation in the level of international cooperation and
institutionalization there has been.
The second argument is more theoretical and begins with a recent attempt to
formalize the classic guns-versus-butter problem.72 To summarize the model,
there are two states. In each period a state must decide how much of its output
to consume, how much to allocate to its military sector, and whether or not to
attack the other state. Each state's utility is the discounted sum of its
consumption in each period. As long as neither state attacks, the game
continues. If a state attacks at some time, the game effectively ends in one of
two ways. Either one state or the other will prevail by conquering the other.
The odds that a state will prevail are simply the ratio of its military allocation to
the other state's military allocation. The fact that a state's probability of victory
depends on its military allocation creates a trade-off between current and
expected future consumption. The more a state consumes today, the smaller its
military allocation, and the higher the probability of defeat. Because defeat
means a loss of future consumption, consuming more today reduces expected
future consumption. The formal analysis of the game determines each state's
equilibrium level of consumption and military spending that balances this
trade-off.
The guns-versus-butter model shows that our emphasis on anarchy is
misplaced. Neorealism expects balance-of-power politics to prevail whenever
the system is anarchic and the units want to survive.73 The guns-versus-butter
model indicates that this expectation is too broad. Whether or not the states
balance in the model depends on an assumption about military technology.
Generalizing beyond this model, whether units balance or not depends as much
on other features defining the strategic situation in which they interact as it
does on the presence of anarchy.
To see that balancing depends on underlying assumptions about military
technology, note that the guns-versus-butter game presumes a conventional
military technology in which the probability of victory or defeat depends on the
relative sizes of the opposing military forces. Given this stylized assumption
about military technology, the states balance against each other in the way we
would expect the units to do in an anarchic system.74 Now suppose that
71. Keohane, "Institutionalist Theory and the Realist Challenge After the Cold War."
72. Robert Powell, "Guns, Butter, and Anarchy," American Political Science Review 87 (March
1993), pp. 115-32. The present discussion extends some of the observations made in that essay (see
pp. 126-27).
73. Waltz, Theory of International Politics, p. 121.
74. External balancing through alliances is impossible when there are only two states. Rather,
the states engage in internal balancing. For a discussion of internal and external balancing, see
Waltz, Theory of International Politics, p. 168.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
the states' strategic setting is different. Formalizing and stylizing the nuclear
revolution in military technology, assume that there is, to use Bernard Brodie's
term, an absolute weapon.75 The probability of victory no longer depends on
the relative size of the states' military forces. Rather, once both states have
attained secure second-strike forces, war is certain to take a toll far higher than
any potential gain. If we solve the model based on this assumption about
military technology, the states will spend enough to acquire second-strike
forces. But they will not spend more even if the other state does. There is no
balancing here even though the system remains anarchic and the units still seek
to survive.76 The first notion of anarchy, albeit very transposable, does not imply
balancing.
The guns-versus-butter model, like many models, makes many stark simplifi-
cations and, accordingly, must be used cautiously. On the plus side, models, in
part because of these simplifications, let us vary one factor while holding
everything else constant. Models thereby permit us to isolate the effects of
different factors in ways that historical cases rarely do. When we use the
guns-versus-butter model to isolate the effects of anarchy, we find that
conclusions claimed to follow from the assumption of anarchy depend at least
as much on other unarticulated assumptions about the units' strategic environ-
ment.
The first definition of anarchy is in some sense too transposable, while the
second definition is not transposable enough. As we have seen, if defined as the
absence of a central authority, anarchy encompasses systems in which states do
and do not balance. Conversely, if we define anarchy by adding the notion of
the potential use of force to the lack of a central authority, we find the
transposability of the concept to be greatly limited, even if units generally will
balance in such a system. The disadvantages of this very limited notion of
anarchy are quite high. In particular, this notion does not apply to systems in
which the use of force is for all intents and purposes not at issue. Even if
neorealism's expectations about anarchic systems in which the use of force is a
serious potential concern are correct, the arguments underlying these expecta-
tions cannot be transposed to systems in which the use of force among units is
not at issue. Neorealist expectations about these systems may of course still
prove to be correct, but they lack theoretical foundations.
75. Bernard Brodie, Strategy in the Missile Age (Princeton, N.J.: Princeton University Press,
1959). For other discussions of the effect of the nuclear revolution, see Robert Jervis, The Meaning
of the Nuclear Revolution (Ithaca, N.Y.: Cornell University Press, 1989); Robert Powell, Nuclear
Deterrence Theory (Cambridge: Cambridge University Press, 1990); Thomas Schelling, Arms and
Influence (New Haven, Conn.: Yale University Press, 1966); and Glenn Snyder, Deterrence and
Defense (Princeton, N.J.: Princeton University Press, 1961).
76. Buzan, Jones and Little reach the same conclusion in The Logic of Anarchy. They and
Morrow offer the expansion of the Roman empire as an important example of the failure of
balances to form. See James Morrow, "Social Choice and System Structure," World Politics 41
(October 1988), pp. 75-97.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
The absence of a definition that is less transposable than the first and more
transposable than the second poses an important problem for international
relations theory after the cold war. The problem is evident in some recent
efforts to use neorealism's analysis of anarchy and the problem of absolute and
relative gains to outline the post-cold war contours of international politics.
The neorealist analysis argues that states will start competing and balancing
over economic issues after the cold war much as they competed and balanced
over security issues during the cold war. Samuel Huntington, for example,
bases his assessment of the continued importance of U.S. primacy on a
neorealist analysis.77 Yet, he and others also believe that the prospects of
"military conflict between major states is unlikely."78 The discussion of
transposability shows that neither definition of anarchy provides adequate
theoretical support for the neorealist analysis of international politics if the use
of force is not a relevant concern. The first notion of anarchy can be transposed
readily to a system in which the use of force is not at issue. But as we have seen,
this definition does not support the neorealist claims that anarchy implies
balance-of-power politics.79 The second notion of anarchy, while it may imply
balancing when force is at issue, cannot be transposed to a domain in which
force is presumed not to be at issue.
Huntington, believing that the politics of international economics is more
like a system with conventional military technology, argues for the importance
of international primacy. Jervis, believing that the politics of international
economics is more like a system with an absolute weapon, questions the
importance of international primacy.80 In either case, the neorealist-neoliberal
debate's emphasis on the lack of a central authority is misplaced. As Charles
Lipson puts it in his contribution to the Baldwin volume, "The idea of anarchy
is, in a sense, the Rosetta stone of international relations.... But what was once
a blinding insight-profound and evocative-has ossified and become blinding
in the other sense of the word-limiting and obscuring."8' We need to develop
a more careful specification of the strategic settings in which units interact if we
are to be able to explain the pattern of their interactions. Characterizing this
structure is an important open question for international relations theory.
The problem of absolute and relative gains
The second major issue at the center of the debate between neorealism and
institutionalism is the problem of absolute and relative gains. In what follows, I
77. Samuel Huntington, "Why International Primacy Matters," Intemational Security 17 (Spring
1993), pp. 68-83. See also Robert Jervis, "International Primacy," Intemational Security 17 (Spring
1993), pp. 52-67; and Kenneth Waltz, "The Emerging Structure of International Politics,"
Intemational Security 18 (Fall 1993), pp. 44-79. Jervis uses a neorealist perspective to frame his
discussion, but his conclusions differ from Huntington's.
78. Huntington, "Why International Primacy Matters," p. 93.
79. For a different view, see Waltz, "The Emerging Structure of International Politics,"
especially p. 74.
80. Jervis, "International Primacy," pp. 57-59.
81. Lipson, "International Cooperation in Economic and Security Affairs," p. 80.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
first briefly summarize some aspects of the debate about this problem. Then I
argue that in a narrower methodological sense this debate reflects a basic
misunderstanding of the role of models. More broadly, the debate surrounding
absolute and relative gains generally has mistaken effects for causes and,
therefore, contributed little to the analysis of the problem of international
cooperation. Once we separate causes from effects, we again see the need to
focus our attention on a more elaborate characterization of the strategic
settings confronting states.
To review the debate, neorealism assumes that states are concerned with
relative gains. For Waltz, "states that feel insecure must ask how the gain will
be divided. They are compelled to ask not 'Will both of us gain?' but 'Who will
gain more?' 182 In mounting his institutional challenge in After Hegemony,
Keohane assumes that states are trying to maximize their absolute gains, that is,
the states' preferences "are based on their assessments of their own welfare,
not that of others."83 He then analyzes the problem of cooperation in terms of
the repeated prisoners' dilemma. Grieco in turn criticizes Keohane's assump-
tion that states attempt to maximize their absolute gains. Grieco asserts that
"realism expects a state's utility function to incorporate two distinct terms. It
needs to include the state's individual payoff ... reflecting the realist view that
states are motivated by absolute gains. Yet it must also include a term
integrating both the states' individual payoff ... and the partner's payoff ... in
such a way that gaps favoring the state add to its utility while, more importantly,
gaps favoring the partner detract from it."84 In sum, the debate about absolute
and relative gains became a debate about what to assume about states' utility
functions.
The key to understanding this debate is to distinguish between two
possibilities. The first is that a state's concern or, more precisely, the degree of
its concern for relative gains is the product of the strategic environment in
which the state finds itself. If so, then the degree of concern is likely to vary as
the environment, say the intensity of the security dilemma, varies. In this case,
the strategic setting facing the state induces a concern for relative gains. The
second possibility is that a state's degree of concern does not vary and is the
same regardless of its environment.
Both neorealism and neoliberalism appear to agree that this concern is
induced. Grieco, for example, believes that a state's sensitivity to relative gains
"will be a function of, and will vary in response to, at least six factors."85 These
include the fungibility of power across issues, the length of the shadow of the
future, and whether the relative gains or losses occur over military or economic
matters.86 Neoliberalism also assumes that the degree of concern varies.
Indeed, Keohane emphasizes that both neorealism and neoliberalism presume
82. Waltz, Theory of International Politics, p. 105. See also Waltz, Man, State, and War, p. 198.
84. Grieco, "Anarchy and the Limits of Cooperation," p. 129, emphasis original.
85. Grieco, "Realist Theory and the Problem of International Cooperation," p. 610.
86. Ibid., pp. 610-11.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
that the concern for relative gains is conditional in his appraisal of the
neorealist-neoliberal debate.87
Two important implications follow from the conclusion that the degree of a
state's concern for relative gains is conditional and varies from situation to
situation. The first is that the debate about what to assume about a state's
preferences or utility function is largely irrelevant and reflects a basic
misunderstanding of the role of models. We can formally induce a concern for
relative gains in two ways. First, we can explicitly represent the constraints that
lead to this concern in the model. This is the approach I followed in analyzing
the absolute and relative gains problem.88 I assumed that states were trying to
maximize their absolute gains. But the strategic setting in which they were
attempting to do so induced a concern for relative gains. The second way to
induce a concern for relative gains is to represent this concern in the state's
utility function. When done in this way, the model is in effect a reduced form
for some more complicated and unspecified model in which the strategic
constraints would induce this concern. Grieco's analysis may be seen as an
attempt to work with a reduced form. Rather than specifying a model that
explicitly represents the six factors he believes induce a concern for relative
gains, he abbreviates the influences of these factors through his specification of
the states' utility functions.89
Which approach to modeling a state's concern is better? I do not believe
there is an a priori answer to this question. Models are tools and asking which
approach is better is akin to asking whether a hammer or a saw is better. The
answer depends on whether the task at hand is driving nails or cutting wood.
One advantage of a reduced form is that it is likely to be simpler and easier to
use analytically. A disadvantage is that as long as the more complicated
underlying model remains unspecified, we cannot analyze the purported link
between the constraints that are believed to induce a concern and the
realization of this concern. The link thus remains problematic. Whether the
balance of advantages and disadvantages favors an approach based on a
reduced form or on a more explicit structural form depends on the model as a
whole and on the substantive problem. Thus, debates about what to assume
about preferences cannot be resolved without reference to an overall evalua-
tion of the entire model and the substantive problem being modeled. By
focusing solely on what to assume about preferences and not evaluating this
87. Keohane, "Institutionalist Theory and the Realist Challenge After the Cold War," pp.
418-25.
88. Powell, "Absolute and Relative Gains in International Relations Theory."
89. See Grieco, "Anarchy and the Limits of Cooperation," as well as his "Realist Theory and the
Problem of International Cooperation," and CooperationAmongNations. Although Grieco's model
may be seen as a reduced form, it is not clear that he sees it as such. His assertion that a state's
utility function must incorporate a term reflecting its concern for absolute gains and one reflecting
its concern for relative gains may be true of a particular model, but it does not hold for all models.
His apparent claim that it is true for all models suggests that he does not interpret his model as a
reduced form.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
assumption in the overall context of the entire model, the neorealist-neoliberal

debate about states' preferences seems largely irrelevant. It reflects a basic
misunderstanding of the role of models.
A second important implication follows from the conclusion that the degree
of a state's concern for relative gains depends on, or is a function of, its strategic
environment. This dependency means that the concern for relative gains is part
of the outcome and not part of the explanation. A concern for relative gains is
an effect and not a cause. We cannot explain the presence or absence of
international cooperation because of the presence or absence of significant
concerns for relative gains. Cooperation and concern for relative gains may
co-vary, but one does not cause the other. The causes for both are the
underlying features of the states' strategic environment that jointly induce a
concern for relative gains and thereby make cooperation difficult.
Existing work in international relations theory has to varying degrees
recognized first that relative gains concerns do not explain the level of
cooperation and second the need to look to the underlying strategic environ-
ment. Lipson, for example, tries to relate differences in the strategic environ-
ments inherent in military and economic issues to differences in states' discount
factors and, through the differences in those discount factors, to the likelihood
of international cooperation in military and economic affairs.90 Jervis explicitly
recognizes this need: "The conditions under which states seek to maximize
their relative as opposed to their absolute gains need more exploration."91
A possible explanation of the concern for relative gains might at first seem to
be anarchy: a lack of a central authority leads to balancing and a concern for
relative gains. This answer, however, fails for at least two reasons. Anarchy has
been a constant feature of the modern international system. It cannot therefore
account for variation in the degree of a state's concern for absolute gains. And,
as we have seen above, anarchy does not imply balancing.
As emphasized above, models often must be judged in light of the problem
they are designed to address. If we want to study the problem of international
cooperation and its relation to concerns for relative gains, modeling that
concern in terms of state preferences seems likely to prove a poor approach.
The reduced form would be leaving implicit and unspecified precisely what we
want to know more about, i.e., the link from the states' strategic environment to
their concern for relative gains and the prospects for cooperation. Trying to
make this link more explicit by elaborating a state's strategic setting and the
connection between this setting and the induced concern for relative gains
seems likely to prove a more fruitful approach. As in our discussion of anarchy,
we are led to the need to focus our attention on a more sophisticated
characterization of the strategic situations confronting the units.
90. Lipson, "International Cooperation in Economic and Security Affairs." See also Joanne
Gowa and Edward Mansfield, "Power Politics and International Trade," American Political Science
Review 87 (June 1993), pp. 408-20.
91. Jervis, "Realism, Game Theory, and Cooperation."

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
In sum, the debate surrounding the problem of absolute and relative gains
has betrayed a fundamental methodological misunderstanding of the role of
models. The debate has also mistaken effects for causes. Unfortunately, the
methodological misunderstanding has reinforced the substantive mistake. By
focusing on what to assume about states' preferences, the debate has made it
more difficult to correct the mistake of seeing effects as causes. In a reduced
form in which the concern for relative gains is represented in the states' utility
functions, the degree of this concern is formally an independent variable. Thus,
it is easy to imagine holding everything else constant and asking how changes in
the degree of this concern would affect cooperation. The difficulty is, of course,
that if the degree of concern is really an effect, then one cannot hold everything
else constant while varying this concern. Although formally independent in the
reduced form, the degree of this concern is substantively dependent. The
reduced form thus masks this dependence and makes it more difficult to correct
the mistake of seeing effects as causes. Once we separate effects from causes,
we also appreciate the need for a more careful specification of the units'
strategic setting.
Coordination and distribution
The debate between neorealism and neoliberalism recently has focused on a

third issue. A central contention of the neoliberal approach is that institutions
matter. In particular, they can help states cooperate: "institutions, broadly
defined, affect the patterns of cooperation that emerge."92 In analyzing how
institutions matter, Keohane emphasizes market failures and explains that
institutions can help independent actors overcome these failures by providing
information and reducing transactions costs.93 In short, institutions may make it
possible to realize joint gains and move out toward the Pareto frontier. But
there are often many ways to realize these gains, with some ways giving a larger
share to one state and other ways giving a larger share to another state. "There
are," as Stephen Krasner observes in his contribution to the Baldwin volume,
"many points along the Pareto frontier."94 These multiple ways of achieving the
joint gains from cooperation can create conflicts over how those gains will be
distributed. As Geoffrey Garrett observes of the Single European Act, "the EC
[European Community] members shared the common goal of increasing the
competitiveness of European goods and services in global markets. It is
apparent, however, that there were also substantial differences in national
preferences within this broad rubric."95 Reflecting on the debate, Keohane
92. Keohane,After Hegemony, p. 9.

93. Ibid., p. 246.
94. Krasner, "Global Communications and National Power," p. 235.
95. The quotation is from p. 535 of Geoffrey Garret, "International Cooperation and
Institutional Choice," Intemational Organization 46 (Spring 1992), pp. 533-60. For another
discussion of conflicting interests, see Andrew Moravcsik, "Negotiating the Single European Act,
Intemational Organization 45 (Winter 1991), pp. 19-56.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
U2
A4
A2
XA I
FIGURE 1. Distributional conflicts: U1 = the utility of state 1 (SI); U2 = the

utility of state 2 (S2); Q = status quo; Al through A4 = possible agreements
along the Pareto frontier
now believes that he underemphasized "distributive issues and the complexi-

ties they create for international cooperation" in After Hegemony.96 I will
suggest that a careful analysis of the tension between cooperation and
distribution can illuminate the debate between neorealism and neoliberalism
by clarifying some of the differences that actually do divide these two
approaches.
The distributive problem arises because there are many ways to divide the
cooperative gains. Figure 1 illustrates this problem when two states, SI and S2
are trying to cooperate. SI's utility is measured along the horizontal axis,
S2's utility is measured along the vertical axis. Q is the status quo.Al throughA
are possible agreements that lie along the Pareto frontier. BothA2 andA3 lie on
the Pareto frontier and are Pareto-superior to Q; i.e., both S, and S2 preferA2
to Q and A3 to Q. A2 and A3 are different ways of realizing the joint gains from
cooperation. But, S, prefers A2 to A3 because A2 yields a higher utility.
Similarly, S2 prefers A3. Thus, there is a distributive conflict over A2 and A3.
More generally, S, prefers agreements closer to Al and S2 prefers agreements
closer toA4.
Krasner recently has used these distributional issues to challenge the
neoliberal approach.97 The thrust of Krasner's criticism is that "the nature of
institutional arrangements is better explained by the distribution of national
96. Keohane, "Institutionalist Theory and the Realist Challenge After the Cold War," pp.
446-47.
97. Krasner, "Global Communications and National Power." See also James Morrow, "Model-
ing International Regimes," International Organization, forthcoming.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
power capabilities than by efforts to solve problems of market failure."98 In

terms of Figure 1, the more powerful Sl, the greater will be its share of the joi
gain and the closer the agreement will be to A1.
Viewing the question of whether institutions matter in terms of distributional
issues helps refine and clarify that question. If cooperation can take many
different forms and these alternative forms have distributional consequences,
then the arrangements themselves can become the object of negotiation.
Indeed, given the absence of a supranational authority, the states cannot bind
themselves to any particular initial institutional arrangement and correspond-
ing allocation of cooperative benefits. The institutional structure is always
subject to renegotiation if a state believes it worthwhile.
The perpetual possibility of renegotiation raises an important dynamic
question that must be separated from a more static issue. That is, institutions
might matter in either or both a static and a dynamic way. The static way that
institutions might matter is that they might be a means of overcoming market
failures or, more generally, of realizing joint gains from cooperation. As a
means to an end, the structure of the institution becomes something to be
explained. In his explanation of institutional structure, Keohane emphasizes
monitoring and reducing uncertainty. In emphasizing these factors, he is trying
to explain how institutions can serve as a means to achieving the joint gains
from cooperation. Krasner focuses on another aspect of the explanation of
institutional structure. He argues that the actual institutional arrangement that
will emerge from the set of potential institutional arrangements that fulfill the
functions Keohane describes will tend to reflect the desires of the more
powerful actors. Thus, Keohane's and Krasner's analyses of the static issue
complement each other.
There is also a second, more dynamic way that institutions may matter. If
institutions do matter in this second sense, then they would be part of an
explanation and not part of the outcome to be explained. Figure 1 can be used
to illustrate this second way. Suppose that at some time to two states are at Q.
Both states want to move out to the Pareto frontier. To this end, they create an
institution that reduces transaction costs and uncertainty in the way Keohane
describes. In this way the institution is a means to the end of realizing the joint
gains of cooperation. But there are also distributional conflicts, so both states
also use their political power to shape the institutional arrangements in order
to obtain a larger share of the joint gains. Assume that S1 is more powerful and,
as Krasner argues, the institution through which the states realize the joint
gains will give S, a larger share of the benefits. In particular, suppose th
arrangement moves them from Q to A2. AtA2, S, receives a larger share of the
joint gains, which reflects its greater power. A2 thus reflects Keohane's and
Krasner's complementary analyses of the static dimension of the way that
institutions may matter.
98. Krasner, "Global Communications and National Power," p. 235.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
To examine the dynamic aspect, suppose further that at some later time, say
t1, the balance of power has shifted in favor of S2. Indeed, assume that if the
institution created at time to did not exist and that the states were trying to
create an institution de novo at tl, then S2's greater power would mean that the
institution that would be created would move the states from Q to A3. AtA3, S2
obtains more of the gains, presumably reflecting its greater power.
But the states are not creating a new institution at tl, for they created an
institution at to that moved them from Q to A2. How does the fact that an
institution already exists at tl, when the states must deal with a new distribution
of power, affect the institutional arrangements and distribution of benefits that
will be devised at that time? There are two possibilities.
First, the institutional arrangements existing at to are irrelevant. Institutions
adjust smoothly so that the distribution of benefits always reflects the
underlying distribution of power. In terms of Figure 1, the states will be atA3 at
t1 regardless of the existence of an institution at to. In brief, history does not
matter.
The second possibility is that the institutional arrangements that exist at to
affect those that prevail at tl. To illustrate this possibility, let A in Figure 1
denote the arrangements and associated distribution of benefits that exist at t1
given the arrangements existing at to. Then A will in general differ from A3,
which is what would have prevailed had there been no preexisting institution or
if institutions adjusted smoothly. Intuitively, the fartherA is fromA3, the more
current arrangements are shaped by past arrangements.99 A more concise way
of describing this second way that institutions may matter is that history
matters.100 In terms of Figure 1, the neoliberal claim that institutional history
matters in international relations means thatA will often be very different from
A3. Moreover, the fact that the states originally cooperated means that
cooperation is less likely to collapse and A is more likely to lie on the Pareto
frontier. Cooperation will often continue in the face of a change in the
underlying distribution of power.
The possibility that institutions may not adjust smoothly and that the existing
institutional arrangements and distribution of benefits may not reflect the
underlying distribution of power is a recurrent theme in international politics.
Robert Gilpin, for example, sees this as the cause of hegemonic war. A
hegemon establishes an international order and associated distribution of
benefits that favors the hegemon. Over time, the hegemon's relative power
declines because of uneven economic growth, and the existing order and
distribution of benefits no longer reflect the distribution of power. This sets
99. To simplify matters, I have assumed that institutions are efficient in that they move the states
out to the Pareto frontier. Of course, institutions need not be efficient. For a discussion of
institutions and efficiency, see Douglass North, Institutions, Institutional Change, and Economic
Performance (New York: Cambridge University Press, 1990).
100. North analyzes the problem of institutional change and stability in ibid.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
scene for another hegemonic conflict.'01 Similarly, a disparity between the

distributions of benefits and power is inherent in Krasner's metaphor of
tectonic plates.102 When regimes are first created, they generally reflect the
underlying distribution of power. But the pressure to change the regime builds
over time as the distribution of power changes. In the tectonic plate metaphor,
this pressure does not lead to a smooth adjustment. Rather, the pressure grows
until it is suddenly relieved in an earthquake in which the regime alters in a way
that realigns it with the distribution of power.
In the neorealist-neoliberal debate, Keohane generally emphasizes market
failures, transaction costs, uncertainty, information, and institutions as impor-
tant means of cooperation. But he also argues that international institutional
history matters. Once institutions or regimes are established, actors behave in
ways that, whether deliberately or not, make it costly to change the regime or
build a new one. Thus, even if the original distribution of power underlying the
regime shifts, the now more powerful states will not change the regime unless
the distribution of power has shifted to such an extent that the benefits of a new
regime, which would reflect the new distribution of power, outweigh the cost of
changing the existing regime. The cost of changing or constructing new regimes
thus gives existing regimes some resilience to shifts in the balance of power.
"The high costs of regime-building help existing regimes persist."1103
In contrast, Krasner argues that regimes and institutions do not matter at
least in the case of global communications:
In recent years distributional questions have precipitated conflict over the

allocation of the radio spectrum and over international telecommunica-
tions. The outcome of these disputes has been determined primarily by the
relative bargaining power of the states involved. Whereas previous institu-
tional choices had not imposed much constraint, new interests and power ca-
pabilities conferred by new technologies have led to new institutional ar-
rangements.
This is not to say that institutional arrangements were ever irrelevant:
indeed, they were necessary to resolve coordination problems and to estab-
lish stability. Without regimes all parties would have been worse off. There
are, however, many points along the Pareto frontier: the nature of institu-
tional arrangements is better explained by the distribution of national
power capabilities than by efforts to solve problems of market failure.104
In brief, institutions may serve as a means of achieving the joint gains of

cooperation. But institutional history does not matter. Previous institutional
101. Robert Gilpin, War and Change in World Politics (Cambridge: Cambridge University Press,
1981).
102. Stephen Krasner, "Regimes and the Limits of Realism," in Stephen Krasner, ed.,
International Regimes (Ithaca, N.Y.: Cornell University Press, 1983), pp. 355-68.
104. Krasner, "Global Communications and National Power," p. 235, emphasis added.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
choices do not constrain or significantly affect the future institutional arrange-

ments and the future distribution of benefits.
The neorealist-neoliberal debate leaves us with contrasting claims about the
importance of institutional history. These claims in turn pose two questions for
future research. First, do institutions or regimes actually adjust smoothly to
changes in the distribution of power. Does institutional history matter? Second,
what factors affect the stability or rigidity of a regime or institution and the
rates at which it adjusts? In particular, are there factors in the international
environment that make international regimes and institutions less stable than,
for example, the institution of the medieval law merchant, the Declaration of
Rights and associated institutional changes following the Glorious Revolution
in England, or the current efforts to establish constitutional governments in
Russia and Eastern Europe?105 Douglass North and others offer many
examples in which institutional history seems to be profoundly important.106
We need a better understanding of the conditions under which institutional
history matters and the extent to which the international system satisfies these
conditions. Work on these questions holds the promise of a more unified
understanding of institutions and cooperation.
Before this work can be done, however, two obstacles must be overcome.
First, we need a way of measuring or assessing the constraining effects of
institutions. Figure 1 helps us visualize the issue, but much more than a
visualization is required. Second, we need more powerful theories that make
more specific claims about the extent to which institutions shape future
decisions and actions than neorealism or neoliberalism presently does.
Conclusion
Although the neorealist-neoliberal debate sometimes has obscured as much as

it has clarified, this debate has forced us to examine the foundations of some of
our most influential theories of international politics more carefully. This is an
important contribution. Such examinations deepen our understanding of these
theories by clarifying their strengths and weaknesses. These clarifications in
turn may suggest important directions for future work and ultimately lead to
better theories with greater explanatory power.
As we have seen, both neorealism and neoliberalism see the effects of
anarchy and the degree of concern about relative gains to be conditional. The
105. See Paul Milgrom, Douglass North, and Barry Weingast, "The Role of Institutions in the
Revival of Trade: The Law Merchants, Private Judges, and the Champagne Fairs," Economics and
Politics 2 (March 1990), pp. 1-23; Douglass North and Barry Weingast, "Constitutions and
Commitment," Joumal of Economic History 49 (December 1989), pp. 803-32; and Barry Weingast,
"The Political Foundations of Democracy and the Rule of Law," manuscript, Hoover Institution,
February 1993.
106. North, Institutions, Institutional Change, and Economic Performance.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00
task ahead is to specify these conditions more precisely. We must also explain
more satisfactorily how these conditions lead to particular outcomes like
balancing behavior and a concern for relative gains. Grieco makes a useful start
in this direction by identifying six factors that may affect the degree of a state's
concern for relative gains.107 The next step is to develop a more explicit
characterization of the strategic settings that yield outcomes like balancing and
relative-gains concerns.
When we look beyond the narrowness of the neorealist-neoliberal debate
about anarchy and the relative-gains problem, we see that this debate has
focused our attention on a very broad and important set of issues. These are the
absence of central authority, the potential for joint or cooperative gains, the
distributional conflict these potential gains engender, and the roles of coercion
and institutions in realizing and allocating these joint gains. This nexus of issues
also lies at the heart of the expanding literatures on constitutional design,
governing the commons, and state formation. 108 That a core of common issues
underlies these seemingly disparate substantive concerns makes it possible to
imagine moving beyond what has become a rather sterile debate between
neorealism and neoliberalism in a way that draws on and contributes to these
other literatures.
107. Grieco, "Realist Theory and the Problem of International Cooperation," pp. 611-13. See
also Gowa and Mansfield, "Power Politics and International Trade."
108. See, for example, Barry Weingast, "Constitutions as Governance Structures," Journal of
Institutional and Theoretical Economics 149 (March 1993), pp. 286-311; Elinor Ostrom, Governing
the Commons (New York: Cambridge University Press, 1990); and Tilly, Capital and Coercion.

182.255.0.242 on Sun, 14 Apr 2024 13:34:54 +00:00

Powell-AnarchyInternationalRelations-1994 (Translatix)

Uploaded by

Copyright:

Available Formats

Powell-AnarchyInternationalRelations-1994 (Translatix)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Powell-AnarchyInternationalRelations-1994 (Translatix)

Uploaded by

Copyright:

Available Formats

Review: Anarchy in International Relations Theory: The Neorealist-Neoliberal Debate

Stable URL: https://www.jstor.org/stable/2706934

This content downloaded from

David A. Baldwin, editor. Neorealism and Neoliberalism: The Contemporary

Two of the most influential contemporary approaches to international relations

Intemational Organization 48, 2, Spring 1994, pp. 313-44

This content downloaded from

1. Kenneth Waltz, Theory of Intemational Politics (Reading, Mass.: Addison-Wesley, 1979).

This content downloaded from

begun to receive attention in the debate between neorealism and neoliberal-

Neorealism and the structural approach

Much of the neorealist-neoliberal debate can be seen as a reaction to Waltz's

This content downloaded from

It is important to emphasize two points about the division of explanations

8. Waltz, Theory of Intemational Politics, p. 73.

This content downloaded from

facing a high risk of bankruptcy in an oligopolistic market may be seen to be in

Four avenues of criticism

Structural theories decompose a system into units and constraints. This

Preferences are given exogenously

The first avenue of criticism centers on preferences. Structural approaches

13. Waltz, Theory of Intemational Politics, pp. 105 and 111.

This content downloaded from

This content downloaded from

experience, and knowledge.19 He also sees realism as a source or theory of

19. Jervis, "Realism, Game Theory, and Cooperation," pp. 324-29.

This content downloaded from

actions) because it takes preferences (over outcomes) as given is not one of

This content downloaded from

assumption that units are minimally motivated to survive. Indeed, it would

The inseparability of agents and structure

This content downloaded from

This content downloaded from

the international structure shapes domestic institutions and states' prefer-

34. Rogowski, Commerce and Coalitions.

This content downloaded from

tend to change states' preferences. States and structure are interdependent;

Waltz's definition of structure

This content downloaded from

This content downloaded from

Neorealism and its implications

The neorealist-neoliberal debate develops primarily along the fourth avenue

45. For a recent effort to do this, see ibid.

This content downloaded from

may be a significant factor in promoting international cooperation in ways that

52. Ibid., p. 67.

This content downloaded from

neorealism's analysis of the problem of international cooperation. This

This content downloaded from

The meaning and implications of anarchy

Much of the neorealist-neoliberal debate centers on the meaning and

This content downloaded from

64. Grieco, "Anarchy and the Limits of Cooperation," pp. 118-19.

This content downloaded from

68. Waltz, Theory of International Politics, pp. 105 and 111.

This content downloaded from

This content downloaded from

This content downloaded from

The problem of absolute and relative gains