Review 2

153

S-
m
,r

at
li-
,t-
~e
:es
n-
et-
,m
rs-

y,”

,le
ue
m.
ire

ral
p-
h-
ns
k-

th

he
.5,
in
ce

w

al
id
:),

in
rt ,-
z,
)-

k:
is
i-

ROBERT O. KEOHANE / Theory of World Politics

tical Science does not contain a section devoted to
the problem of political change. Nor does the entry
“political change” appear in its cumulative index. See
Fred I. Greenstein and Nelson W. Polsby, eds., Hand-
book of Political Science (Reading, Mass.: Addison-
Wesley, 1975).
11. Joseph Schumpeter, History of Economic Analysis
(New York: Oxford University Press, 1954), p. 964.
12. K. J.Holsti, “Retreat from Utopia,” Canadian Jour-
nal of Political Science, Vol. 4 (1971 ), pp. 165-177.
13. Ibid., p. 167.
14. Ibid., p. 171.
15. Several important books have recently indicated

revival of interest in general theory. See Choucri and
North, Nations in Conflict, Hedley Bull, The Anarchical
Society (New York: Columbia University Press, 1977};
Keohane and Nye, Power and Interdependence; Stanley
Hoffmann, Primacy or World Order (New York:
McGraw-Hill, 1978); Ralph Pettman, State and Class
(London: Croom Helm, 1979); Kenneth N. Waltz, The-
ory of International Politics (Reading, Mass.: Addison-
Wesley, 1979). Marxist scholars, of course, never lost
interest in “grand theory.”
16. Three recent exceptions are Evan Luard, Types of

International Society (New York: Free Press, 1976);
Robert G. Wesson, The Imperial Order (Berkdey: Uni-
versity of California Press, 1967); Martin Wight, Systems

of States, ed. Hedley Bull (Leicester, England; Leicester
University Press, 1977).
17. Martin Wight, “Why Is There No International

Theory?” in Herbert Butterfield and Martin Wight
(eds.), Diplomatic Investigations (London: George Allen
and Unwin, 1966), pp. 17-34.
18. Albert 0. Hirschman, “The Search for Paradigms

as a Hindrance to Understanding,” World Politics, Vol.
22 (1970), pp. 329-43.
19. John Burton, International Relations: A General
Theory (Cambridge, MA: Cambridge University Press,
1965), pp. 71-72.
20. E. H. Carr, The Twenty Years’ Crisis (London:
Macmillan, 1951).
21. John Harsanyi, “Rational-Choice Models of Politi-
cal Behavior vs. Functionalist and Conformist Theories,”
World Polirics, Vol. 21 (1969), pp. 513-38.
22. R. L. Curry fr. and L. L. Wade, A Theory of Political
Exchange (Englewood Cliffs, N.J.: Prentice Hall, 1968),
p. 49; and Lance E. Davis and Douglass C. North, Insti-
tutional Change and American Economic Growth (Cam-
bridge England: Cambridge University Press, 1971 ), p. 40.
23. Trout Rader, The Economics of Feudalism (New
York: Gordon and Breach, 1971), p. 50.
24. Henry A. Kissinger, A World Restored (Boston:
Houghton Mifflin, 1967), pp. 1-2.
25. Carr, Twenty Years’ Crisis, p. 132.

Theory of World Politics: Structural
Realism and Beyond

ROBERT 0. KEOHANE

The author surveys and critically evaluates the work of realists and “structural
realists.” Keohane argues that structural realism helps us understand world pvl-
itics by focusing on its systemic aspects that establish the context for state action.
On the other hand, to explain state behavior and outcomes, Keohane looks not
just to international political structure but also to its relation to domestic politics
and decision making 1 internal-external interactions, and the workings of inter-
national institutions.

From Political Science: The State of the Discipline, ed. Ada W. Finifter (Washington, DC: American Political
Science Association, 1983 ). Reprinted by permission of the APSA.

154 ROBERT O. KEOHANE / Theory of World Politics

For over 2000 years, what Hans J. Morgenthau
dubbed “Political Realism” has constituted the
principal tradition for the analysis of interna-
tional relations in Europe and its offshoots in the
New World (Morgenthau, 1966). Writers of the
Italian Renaissance, balance of power theorists,
and later adherents of the school of Machtpolitik
all fit under a loose version of the Realist rubric.
Periodic attacks on Realism have taken place; yet
the very focus of these critiques seems only to
reconfirm the centrality of Realist thinking in the
international political thought of the West.1

Realism has been criticized frequently dur-
ing the last few years, and demands for a “new
paradigm” have been made. Joseph S. Nye and I
called for a “world politics paradigm” a decade
ago, and Richard Mansbach and John A. Vas-
quez have recently proposed a “new paradigm
for global politics.” In both these works, the new
paradigm that was envisaged entailed adopting
additional concepts-for instance, “transna-
tional relations,” or “issue phases” (Keohane &
Nye, 1972, esp. pp. 379-386; Mansbach & Vas-
quez, 1981, Chapter 4). Yet for these concepts to
be useful as part of a satisfactory general theory
of world politics, a theory of state action-
which is what Realism purports to provide-
is necessary. Understanding the general princi-
ples of state action and the practices of govern-
ments is a necessary basis for attempts to refine
theory or to extend the analysis to non-state
actors. Approaches using new concepts may be
able to supplement, enrich, or extend a basic
theory of state action, but they cannot substitute
for it.2

The fixation of critics and reformers on the
Realist theory of state action reflects the impor-
tance of this research tradition. In my view,
there is good reason for this. Realism is a neces-
sary component in a coherent analysis of world
politics because its focus on power, interests,
and rationality is crucial to any understanding
of the subject. Thus any approach to interna-
tional relations has to incorporate, or at least
come to grips with, key elements of Realist

thinking. Even writers who are concerned prin-
cipally with international institutions and rules,
or analysts in the Marxist tradition, make use of
some Realist premises. Since Realism builds on
fundamental insights about world politics and
state action, progress in the study of interna-
tional relations requires that we seek to build on
this core.

Yet as we shall see, Realism does not pro-
vide a satisfactory theory of world politics, if we
require of an adequate theory that it provide a
set of plausible and testable answers to questions
about state behavior under specified conditions.
Realism is particularly weak in accounting for
change, especially where the sources of that
change lie in the world political economy or in
the domestic structures of states. Realism, viewed
dogmatically as a set of answers, would be worse
than useless. As a sophisticated framework of
questions and initial hypotheses, however, it is
extremely valuable.3

Since Realism constitutes the central tradi-
tion in the study of world politics, an analysis,
like this one, of the current state of the field
must evaluate the viability of Realism in the
penultimate decade of the twentieth century.
Doing this requires constructing a rather elabo-
rate argument of my own, precluding a compre-
hensive review of the whole literature of
international relations. I have therefore selected
for discussion a relatively small number of
works that fit my theme, ignoring entire areas of
research) much of it innovative. 4 Within the
sphere of work dealing with Realism and its lim-
itations, I have focused attention on several
especially interesting and valuable contribu-
tions. My intention is to point out promising
lines of research rather than to engage in what
Stanley Hoffmann once called a “wrecking oper-
ation” (Hoffmann, 1960, p. 171).

Since I have written on the subject of Real-
ism in the past) I owe the reader an explanation
of where I think my views have changed, and
where I am only restating, in different ways,
opinions that I have expressed before. This

chapter
sympath
vious w1
consiste1
dence.In
theory a
internati
1977, Pr
models,
Realist o
internati,
as adapt
beyond i1

Adm
dependen
live ideal
which ex1
Realist a
types wet
under wt
change w
“Realist,”
which st,
chies of i
instrume1
23-29). T
Realist tr.
and it see1
our view
Chapter 2
draws on
account fc

Top,
theoretica
Structural
conceptio1
explicated
Imre Lak,
concept as
of scientifi
regarded a
lion in Th,
digm.6 n
programs.
assumptio1

155 ROBERT O. KEOHANE I TheoryofWorldPolitics

·chapter deals more systematically and more
sympathetically with Realism than does my pre-
vious work. Yet its fundamental argument is
consistent with that of Power and Interdepen-
dence. In that book Nye and I relied on Realist
theory as a basis for our structural models of
international regime change (Keohane & Nye,
1977, pp. 42-46). We viewed our structural
models as attempts to improve the ability of
Realist or neo-Realist analysis to account for
international regime change: we saw ourselves
as adapting Realism, and attempting to go
beyond it, rather than rejecting it.

Admittedly, Chapter 2 of Power and Inter-
dependence characterized Realism as a descrip-
tive ideal type rather than a research program in
which explanatory theories could be embedded.
Realist and Complex Interdependence ideal
types were used to help specify the conditions
under which overall structural explanations of
change would or would not be valid; the term,
“Realist,”was used to refer to conditions under
which states are the dominant actors, hierar-
chies of issues exist, and force is usable as an
instrument of policy (Keohane & Nye, 1977, pp.
23-29). Taken as a full characterization of the
Realist tradition this would have been unfair,
and it seems to have led readers concerned with
our view of Realism to focus excessively on
Chapter 2 and too little on the attempt, which
draws on what I here call structural realism, to
account for regime change (Chapters 3-o). 5

To provide criteria for the evaluation of
theoretical work in international politics-
Structural Realism, in particular-I employ the
conception of a “scientific research programme”
explicated in 1970 by the philosopher of science,
lmre Lakatos (1970). Lakatos developed this
concept as a tool for the comparative evaluation
of scientific theories, and in response to what he
regarded as the absence of standards for evalua-
tion in Thomas Kuhn’s (1962) notion of a para-
digm.’ Theories are embedded in research
programs. These programs contain inviolable
assumptions (the “hard core”) and initial condi-

tions, defining their scope. For Lakatos, they
also include two other very important elements:
auxiliary, or observational, hypotheses, and a
“positive heuristic,” which tells the scientist
what sorts of additional hypotheses to entertain
and how to go about conducting research. In
short, a research program is a set of method-
ological rules telling us what paths of research to
avoid and what parts to follow.

Consider a research program, with a set of
observational hypotheses, a “hard core” of
irrefutable assumptions, and a set of scope con-
ditions. In the course of research, anomalies are
bound to appear sooner or later: predictions of
the theory will seem to be falsified. For Lakatos,
the reaction of scientists developing the research
program is to protect the hard core by con-
structing auxiliary hypotheses that will explain
the anomalies. Yet any research program, good
or bad, can invent such auxiliary hypotheses on
an ad hoc basis. The key test for Lakatos of the
value of a research program is whether these
auxiliary hypotheses are “progressive,” that is,
whether their invention leads to the discovery of
new facts (other than the anomalous facts that
they were designed to explain). Progressive re-
search programs display “continuous growth”:
their auxiliary hypotheses increase our capacity
to understand reality (Lakatos, 1970, pp. 116-
122, 132-138, 173-180).

Lakatos developed this conception to assess
developments in the natural sciences, particu-
larly physics. If we took literally the require-
ments that he laid down for “progressive”
research programs, all actual theories of interna-
tional politics-:-and perhaps all conceivable the-
ories-would fail the test. Indeed, it has been
argued that much of economics, including oli-
gopoly theory (heavily relied upon by Structural
Realists), fails to meet this standard (Latsis,
1976). Nevertheless, Lakatos’ conception has the
great merit of providing clear and sensible crite-
ria for the evaluation of scientific traditions, and
of asking penetrating questions that may help us
to see Realism in a revealing light. Lakatos’ ques-

_..,,,

156 ROBERT O. KEOHANE I Theory of World Politics

tions are relevant, even if applying them without
modification could lead to premature rejection
not only of Realism, but of our whole field, or
even the entire discipline of political science.7

The stringency of Lakatos’ standards sug-
gests that we should supplement this test with a
“softer,” more interpretive one. That is, how
much insight does Realism provide into con-
temporary world politics?

For this line of evaluation we can draw
inspiration from Clifford Geertz’s discussion of
the role of theory in anthropology. Geertz argues
that culture «is not a power, something to which
social events, behaviors, institutions, or processes
can be causally attributed; it is a conte>…i:-some-
thing within which they can be intelligibly-that
is, thickly-described” (1973, p. 14). The role of
theory, he claims, is “not to codify abstract regu-
larities but to make thick description possible,
not to generalize across cases but to generalize
within them” (ibid., p. 26). This conception is
the virtual antithesis of the standards erected by
Lakatos, and could alltoo easily serve as a ratio-
nalization for the proliferation of atheoretical
case studies. Nevertheless, culture as discussed by
Geertz has something in common with the inter-.
national system as discussed by students of world
politics. It is difficult to generalize across systems.
We are continually bedeviled by the paucity of
comparable cases, particularly when making sys-
temic statements-for example, about the oper-
ation of balances of power. Much of what
students of world politics do, and what Classical
Realism in particular aspires to, is to make the
actions of states understandable ( despite obfus-
catory statements by their spokesmen): that is,
in Geertz’ s words, to provide “a context within
which they can be intelligibly described.” For
example, Morgenthau’s discussion of the con-
cept of interest defined in terms of power,
quoted at length below, reflects this objective
more than the goal of arriving at testable gener-
alizations.

This essay is divided into four major sec-
tions. The first of these seeks to establish the

basis for a dual evaluation of Realism: as a
source of interpretive insights into the operation
of world politics, and as a scientific research
program that enables the investigator to dis-
cover new facts. I examine the arguments of
Thucydides and Morgenthau to extract the key
assumptions of Classical Realism. Then I discuss
recent work by Kenneth N. Waltz, whom I
regard as the most systematic spokesman for
contemporary Structural Realism.

Section II addresses the question of inter-
pretation and puzzle-solving within the Realist
tradition. How successful are Realist thinkers in
making new contributions to our understanding
of world politics? In Section III, I consider the
shortcomings of Realism when judged by the
standards that Lakatos establishes, or even when
evaluated by less rigorous criteria, and begin to
ask whether a modified version of Structural
Realism could correct some of these faults. Sec-
tion IV carries this theme further by attempting
to outline how a multi-dimensional research
program, including a modified structural the-
ory, might be devised; what its limitations would
be; and how it could be relevant, in particular, to
problems of peaceful change.

The conclusion emphasizes the issue of
peaceful change as both a theoretical and a prac-
tical problem. Realism raises the question of
how peaceful change could be achieved, but
does not resolve it. Understanding the condi-
tions under which peaceful change would be
facilitated remains, in my view, the most urgent
task facing students of world politics.

I. Structural Realism as Research Program

To explain the research program of Realism, I be-
gin with two classic works, one ancient, the other
modem: The Peloponnesian War, by Thucydides,
and Politics Among Nations, by Morgenthau. 8 The
three most fundamental Realist assumptions are
evident in these books: that the most important
actors in world politics are territorially organized
entities (city-states or modern states); that state

behavior ,
states seek
terms of pi
national SJ

The F
attempt to
the Fifth C
by Athem
Thucydid,
pose, he rn
city-states
Morgenth,
ence of int
states. Real

Both a
world polit
ining then
authoritati
they woul,
encountere
both, there
reconstructi
not have ti
given durin

It was ir
wordfo
been to
opinion
sions, of
the gen
(Thucyc
Modern

Morger
stand foreig

We put c
who mm
icy undei
selves w:
whicha f
these rat:
man, act
to choof
hypothes

157

a
n
h ,_
,f
,y
5S

,r

r-
st
n
1g
ie

,e
n
0
al

,g
h

.d
😮

,f

,f
1!
1-

>e
1!

,-
·s,
1e
re
11
,d
te

RO BERT O. KEOHANE I Theory of World Politics

behavior can be explained rationally; and that
states seek power and calculate their interests in
terms of power, relative to the nature of the inter-
national system that they face.

The Peloponnesian War was written in an
attempt to explain the causes of the great war of
the Fifth Century B.C. between the coalition Jed
by Athens and its adversaries, led by Sparta.
Thucydides assumes that to achieve this pur-
pose, he must explain the behavior of the major
city-states involved in the conflict. Likewise,
Morgenthau assumes that the subject of a sci-
ence of international politics is the behavior of
states. Realism is “state-centric.” 9

Both authors also believed that observers of
world politics could understand events by imag-
ining themselves, as rational individuals, in
authoritative positions, and reflecting on what
they would do if faced with the problems
encountered by the actual decision-makers. They
both, therefore, employ the method of rational
reconstruction.Thucydides admits that he does
not have transcripts of all the major speeches
given during the war, but he is undaunted:

It was in allcasesdifficult to carry [ the speeches]
word for word in one’s memory, so my habit has
been to make the speakers say what was in my
opinion demanded of them by the various occa-
sions, of course adhering as closely as possible to
the general sense of what they really said.
(Thucydides, Book I, paragraph 23 [ Chapter I,
Modern Library edition, p. 14I).

Morgenthau argues that in trying to under-
stand foreign policy,

We put ourselves in the position of a statesman
who must meet a certain problem of foreign pol-
icy under certain circumstances, and we ask our-
selves what the rational alternatives are from
which a statesman may choose … and which of
these rational alternatives this particular states-
man, acting under these circumstances, is likely
to choose. It is the testing of this rational
hypothesis against the actual facts and their con-

sequences that gives meaning to the facts of
international politics and makes a theory of pol-
itics possible. (Morgenthau, 1966, p. 5)

In reconstructing state calculation, Thu-
cydides and Morgenthau both assume that
states will act to protect their power positions,
perhaps even to the point of seeking to maxi-
mize their power. Thucydides seeks to go
beneath the surface of events to the power reali-
ties that are fundamental to state action:

The real cause [ of the war] I consider to be the
one which was formally most kept out of sight.
The growth in the power of Athens, and the
alarm which this inspired in Lacedemon, made
war inevitable. (Thucydides, Book I, paragraph
23, Chapter I, Modern Library Edition, p. 15)10

Morgenthau is even more blunt: “Inter-
national politics, like all politics, is a struggle
for power” (I 966, p. 25, see also Morgenthau,
1946). Political Realism, he argues, understands
international politics through the concept of
”interest defined as power”:

We assume that statesmen thinkand act in terms
of interest defined as power, and the evidence of
history bears that assumption out. That assump-
tion allows us to retrace and anticipate, as it
were, the steps a statesman-past, present, or
future-has taken or will take on the political
scene. We look over his shoulder when he writes
his dispatches; we listen in on his conversation
with other statesmen: we read and anticipate his
very thoughts. (1966, p. 5)

The three assumptions just reviewed define
the hard core of the Classical Realist research
program:

1. The state-centric assumption: states are the
most important actors in world politics;
2. The rationality assumption: world politics can
be analyzed as if states were unitary rational
actors, carefully calculating costs of alternative
courses of action and seeking to maximize their

158 ROBERT O. KEOHANE / Theory of World Politics

expected utility, although doing so under condi-
tions of uncertainty and without necessarily
having sufficient information about alternatives
or resources (time or otherwise) to conduct a
full review of all possible courses of action;’ 1
3. The power assumption: states seek power
(both the ability to influence others and
resources that can be used to exercise influence);
and they calculate their interests in terms of
power, whether as end or as necessary means to
a variety of other ends.

More recently, Kenneth N. Waltz (1959)
has attempted to reformulate and systematize
Realism on the basis of what he called, in Man,
the State and War, a “third image” perspective.
This form of Realism does not rest on the pre-
sumed iniquity of the human race-original sin
in one form or another-but on the nature of
world politics as an anarchic realm:

Each state pursues its own interests, however
defined, in ways it judges best. Force is a means
of achieving the external ends of states because
there exists no consistent, reliable process of rec-
onciling the conflicts of interests that inevitably
arise among similar units in a condition of anar-
chy. (p. 238)12

Even well-intentioned statesmen find that
they must use or threaten force to attain their
objectives.

Since the actions of states are conceived of
as resulting from the nature of international
politics, the paramount theoretical task for Real-
ism is to create a systemicexplanation of inter-
national politics. In a systemic theory, as Waltz
explains it, the propositions of the theory specify
relationships between certain aspects of the
system and actor behavior (1979, pp. 67-73).
Waltz’s third-image Realism, for instance, draws
connections between the distribution of power
in a system and the actions of states: small coun-
tries will behave differently from large ones, and
in a balance of power system, alliances can be
expected to shift in response to changes in

power relationships. Any theory will, of course,
take into account the attributes of actors, as well
as features of the system itself. But the key dis-
tinguishing characteristic of a systemic theory is
that the internal attributes of actors are given by
assumption rather than treated as variables.
Changes in actor behavior, and system out-
comes, are explained not on the basis of varia-
tions in these actor characteristics, but on the
basis of changes in the attributes of the system
itself. A good example of such a systemic theory
is microeconomic theory in its standard form. It
posits the existence of business firms, with given
utility functions (such as profit maximization),
and attempts to explain their behavior on the
basis of environmental factors such as the com-
petitiveness of markets. It is systemic because its
propositions about variations in behavior de-
pend on variations in characteristics of the sys-
tem, not of the units (Waltz, 1979, pp. 89-91,
93-95, 98).

To develop a systemic analysis, abstraction is
necessary: one has to avoid being distracted by
the details and vagaries of domestic politics and
other variables at the level of the acting unit. To
reconstruct a systemic research program, there-
fore, Structural Realists must devise a way to
explain state behavior on the basis of systemic
characteristics, and to account for outcomes in
the same manner. This needs to be a coherent
explanation, although it need not tell us every-
thing we would like to know about world politics.

Waltz’s formulation of Structural Realism
as a systemic theory seeks to do this by develop-
ing a concept not explicitly used by Morgenthau
or Thucydides: the structure of the international
system. Two elements of international structure
are constants: (!) the international system is
anarchic rather than hierarchic, and ( 2) it is
characterized by interaction among units with
similar functions. These are such enduring
background characteristics that they are consti-
tutive of what we mean by “international poli-
tics.”13 The third element of structure, the
distribution of capabilities across the states in

the syster
over time
the distrib
importan< capabiliti, actors. Str actors tha maJor one Accor pal deterr. level: struc discouragf tended cor obtain the power of o ForW aninternat terns of st, their intere culations al tern. The 1 actor beha assumption diet that le, and constra Taking rati, attribute va tions in cha tern. Othen- accounted f ability of sta of Structurai tory power) assumption- balance of pc oretical clain The mo, tural theory system has a conceptualiz; geneous and achieve resul1 out significar tics becomes l respects, pow 159 urse, well dis- •rylS n by bles. out- 1na- the .tern eory n.It iven on), the ,m- e its de- sys- -91, Ill IS l by and To ere- ' to nuc sin ·ent ,ry- ics. ism op- .1au nal ure 1 is t is ,jth ing sti- ,!i- the m ROBERT O. KEOHANE I TheoryofWor!dPolitics the system, varies from system to system, and over time. Since it is a variable, this element- the distribution of "power" -takes on particular importance in the theory. The most significant capabilities are those of the most powerful actors. Structures "are defined not by all of the actors that flourish within them but by the major ones" (Waltz, 1979, p. 93). According to Waltz, structure is the princi- pal determinant of outcomes at the systems level: structure encourages certain actions and discourages others. It may also lead to unin- tended consequences, as the ability of states to obtain their objectives is constrained by the power ofothers (1979, pp. 104-111). For Waltz, understanding the structure of an international system allows us to explain pat- terns of state behavior, since states determine their interests and strategies on the basis of cal- culations about their own positions in the sys- tem. The link between system structure and actor behavior is forged by the rationality assumption, which enables the theorist to pre- dict that leaders will respond to the incentives and constraints imposed by their environments. Taking rationality as a constant permits one to attribute variations in state behavior to varia- tions in characteristics of the international sys- tem. Otherwise, state behavior might have to be accounted for by variations in the calculating ability of states; in that case, the systemic focus of Structural Realism (and much of its explana- tory power) would be lost. Thus the rationality assumption-as we will see in examining Waltz's balance of power theory-is essential to the the- oretical claims of Structural Realism. 14 The most parsimonious version of a struc- tural theory would hold that any international system has a single structure of power. In such a conceptualization, power resources are homo- geneous and fungible: they can be used to achieve results on any of a variety of issues with- out significant loss of efficacy. 15Power in poli- tics becomes like money in economics: "in many respects, power and influence play the same role in international politics as money does in a mar- ket economy" (Wolfers, 1962, p. 105). In its strong form, the Structural Realist research program is similar to that of micro-eco- nomics. Both use the rationality assumption to permit inferences about actor behavior to be made from system structure. The Realist defini- tion of interests in terms of power and position is like the economist's assumption that firms seek to maximize profits: it provides the utility func- tion of the actor. Through these assumptions, actor characteristics become constant rather than variable, and systemic theory becomes pos- sible.16 The additional assumption of power fun- gibility simplifies the theory further: on the basis of a singlecharacteristic of the international sys- tem (overall power …