Lake AnarchyHierarchyVariety 1996

Anarchy, Hierarchy, and the Variety of International Relations
Author(s): David A. Lake

Source: International Organization , Winter, 1996, Vol. 50, No. 1 (Winter, 1996), pp. 1-33
Published by: The MIT Press
Stable URL: https://www.jstor.org/stable/2706997
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

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
All use subject to https://about.jstor.org/terms
Anarchy, hierarchy, and the variety
of international relations
David A. Lake
Following World War II, the United States entered into an alliance with the
countries of Western Europe. At the same time, the Soviet Union created an
informal empire in Eastern Europe. America's alliances were anarchic in
character. While each partner agreed explicitly and implicitly to defend the
others, all retained substantial discretion over the implementation of this
agreement. The Soviet Union's informal empire was more hierarchic. Not only
were the parties committed to mutual defense but the Soviet Union intruded
deeply into the internal political, economic, and social affairs of its partners in
order to enforce their allegiance.
Realism cannot explain this difference in relations. Two countries occupying
similar positions within the international system should adopt similar strategies
for building power or security. The nature of the political regimes in the two
superpowers provides at best only a partial explanation. Regime type cannot
explain Soviet restraint in areas it might have dominated, such as Finland or
Yugoslavia, or American informal imperialism in the Caribbean and Central
America.' Nor are ideological or cultural approaches-whether primordial,
instrumental, or constructivist in nature-particularly promising alternatives.
Earlier versions of the essay were presented at the annual meeting of the American Political
Science Association, Chicago, 3-6 September 1992, and in seminars at Indiana, Princeton, and
Columbia Universities; the University of California, Berkeley, Los Angeles, Santa Barbara, and
San Diego; and the University of Chicago. Numerous colleagues have provided many helpful
suggestions and criticisms. I would like to acknowledge the especially important guidance of Jeff
Frieden, Joanne Gowa, Robert Keohane, Wendy Lake, John Odell, Robert Pahre, Paul
Papayoanou, Robert Powell, Beth Yarbrough, and several anonymous reviewers. Ellen Comisso,
Matthew Evangelista, Peter Gourevitch, Jack Snyder, and Celeste Wallander were kind enough to
comment on the penultimate draft on short notice. Scott Bruckner, Risa Brooks, Kathleen
Hancock, and Adam Stulberg provided research assistance. This research has been supported by
the Academic Senates of the University of California at Los Angeles (UCLA) and at San Diego
(UCSD), the Center for International Relations and the Office of International Studies and
Overseas Programs at UCLA, and the Institute on Global Conflict and Cooperation at UCSD.
1. On the essential similarity between relations in the Soviet and U.S. spheres of influence, see
Triska 1986.
Intemational Organization 50, 1, Winter 1996, pp. 1-33

? 1996 by The IO Foundation and the Massachusetts Institute of Technology

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
2 International Organization
Marxism-Leninism is ambiguous on this score, both promising an anarchic

withering away of the state and excusing a hierarchic dictatorship of the
proletariat at home and, by extension, abroad. In the United States, the
prevailing foreign policy ideology, rooted in George Washington's Farewell
Address, explicitly eschewed "entangling alliances," yet these became the
instrument of choice after 1945. The difference in paths chosen by the two
superpowers remains a genuine analytic puzzle.
This essay draws upon theories of relational contracting, first developed in
economics but now finding increasing application in political science, to
examine the alternative relations available to all states.2 The theory presented
below is informed by a central metaphor, specifically, that the state is a firm
producing security.3 Whenever a state chooses to manufacture security in
association with another, it must choose a relationship-ranging from an
anarchic alliance to a hierarchic empire-to govern interactions with its
partner. In this metaphor, an alliance is analogous to an arm's-length contract
between separate firms, while an empire is akin to integration within a single
firm. The choice between these alternatives, I posit, is a function of two main
variables: the expected costs of opportunism, which decline with relational
hierarchy, and governance costs, which rise with relational hierarchy. After a
brief review to locate my approach within the broader international relations
literature, I define the dependent variable and develop the theory at some
length. The theory is self-consciously general and designed to apply in a variety
of circumstances. To explicate the central logic and assess the initial plausibility
of the theory, however, I close by examining briefly the alternative paths
selected by the Soviet Union and the United States and unraveling the analytic
puzzle posed by their contrasting choices.
The logic of security
Grand strategy has been the topic of considerable scholarly debate. Defined by
Barry Posen as "a state's theory about how it can best 'cause' security for
itself," the subject is, appropriately, nearly synonymous with the study and
practice of international relations.4 As a field, we have made substantial
progress in understanding the sources and variations of grand strategy.5
2. For an example from economics, see Williamson 1985. For a review of the economic
literature, see Eggertsson 1990. For an application to American politics, see Weingast and
Marshall 1988. Robert Keohane was the first to employ this approach in international relations; see
Keohane 1983 and 1984. For applications, see Frieden 1994 and Simmons 1993. For a view from
two economists who speak directly to the concerns of international relations, see Yarbrough and
Yarbourgh 1992.
3. This metaphor has been employed with great effect in economic history. See Lane 1979 and
North 1981.
4. The quotation is from Posen 1984, 13.
5. For reviews, see Walt 1991; and Nye and Lynn-Jones 1988.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
Anarchy 3
Nonetheless, two related problems remain. While these problems are natural
consequences of the magnitude of the subject and will not be remedied fully
here, they continue to limit further progress.
First, while scholars have given much attention to the many techniques for
causing security, they typically fail to consider adequately the alternatives
available to states. In other words, the range of possible variation in strategy-
the dependent variable-commonly is abridged.6 For example, scholars focus
on alliances or empires but fail to ask how, in what ways, and to what extent
these relationships are substitutes for one another.7 When states choose to seek
empire, for instance, they are simultaneously choosing not to form an alliance.
The net benefits of imperialism by themselves do not explain action; rather, it is
the net benefits of the alternatives relative to one another that drive state
choice. Any full explanation of grand strategy must therefore compare all
alternatives.8
Second, analysts continue to work within what Benjamin Most and Harvey
Starr describe as separate "islands" of theory and fail to appreciate or build
upon common political problems and independent variables.9 This is true even
in the comparatively well-developed literatures on alliance formation and
imperialism. The standard international relations texts, for instance, nearly
always discuss alliances and empires in separate chapters and often in very
different theoretical contexts: the first tend to be covered under the headings of
realism, systems-level theories, or the balance of power; the second under
Marxism, unit-level theories, or North-South relations. Yet in practice, all
states are concerned with problems of aggregating and pooling resources-
building power-and coping with opportunism by their partners-restraining
self-seeking actions. These practical problems are also central independent
variables in our theories of international politics, even if they are often implicit.
Without seeking to minimize differences in the research agendas that have
grown up around these and other topics, I want to suggest that a theoretical
core does exist-a core that informs the theory discussed in the remainder of
this article.
The current literature on alliance formation is dominated by a "capability
aggregation" model. As Bruce Russett and Harvey Starr write in their widely
used text, "Throughout history the main reason states have entered into
6. Among the best recent efforts to elucidate the full range of strategies are Friedberg 1988;
Snyder 1991; and Morrow 1993. Emerson Niou and Peter Ordeshook ask similar questions, but
take a different approach than discussed here. See Niou and Ordeshook 1994.
7. See Walt 1987; and Doyle 1986.
8. David Baldwin suggests that international political economists may not be sufficiently
attentive to this fundamental tenet of rational choice theory either. See Baldwin 1985, 8-18 and
29-40. Much of the work on trade policy, for instance, has proceeded without sufficient reference
to exchange rate manipulation as an alternative to tariff protection. For a typical example of this
failing, see Lake 1988.
9. Most and Starr 1984. See also Siverson and Starr 1991.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
alliances has been the desire for the aggregation of power. "10 In this model,
states form alliances primarily to counter common threats larger than each
individually.
Within this model, the costs of alliances-and the reasons why states
sometimes eschew foreign entanglements-are understood to be reduced
autonomy and freedom of action. Glenn Snyder has explicated these costs
more fully. In the "alliance dilemma," he writes, states may be "abandoned,"
defined broadly to include both free riding on the efforts of others and shirking
on agreements, and "entrapped," or embroiled in conflicts they might
otherwise avoid. More recently, Thomas Christensen and Jack Snyder have
termed these respective costs "buck passing" and "chain ganging."'1
The literature on imperialism, on the other hand, has long been rent by three
competing perspectives: metrocentric theory focuses on dispositional features
of imperial states, systemic theory highlights competition between the great
powers, and pericentric theory emphasizes conditions in peripheral states and
territories. In several recent studies, scholars have succeeded in integrating
these approaches into a consistent and powerful explanation of imperialism.12
With obvious differences, metrocentric and systemic theories of imperialism
share with theories of alliance formation an emphasis on capability aggrega-
tion. At the most basic level, both hypothesize that states form empires to
capture important resources-raw materials, manpower, markets, strategic
locations-otherwise unavailable to groups at home or the polity in general.
Although many theories of imperialism emphasize the domestic implications of
expansion, explanations of alliance formation and imperialism share a common
focus on capability aggregation.
More than the other approaches, pericentric theories attempt to explain the
form of imperialism and in turn focus on opportunism. Following the
pioneering work of John Gallagher and Ronald Robinson, such theories
hypothesize that dominant states prefer informal empire when possible but
create formal empires when local elites are either unwilling or unable to
provide acceptable domestic political orders; in other words, formal empires
are created when local rulers possess different policy preferences or lack the
ability to carry out the interests of dominant states.13 Whether emphasizing
intent or ability, the failure to comply with or carry out the desires of the core
10. Russett and Starr 1989, 91 (emphasis original). For reviews of the voluminous literature on
alliances, see the essays collected in Friedman, Bladen, and Rosen 1970; and in Holsti, Hopmann,
and Sullivan 1973. Another explanation of alliance formation focuses on national attributes, such
as ideology or regime type. For that view, see Barnett and Levy 1991; David 1991; and Siverson and
Emmons 1991. The historical literature also points to the desire to control partners as a motivation
for alliances, but this empirical insight has not been incorporated into the theoretical literature; see
Schroeder 1976. For an alternative approach, see Morrow 1991.
11. See Snyder 1984 and 1990; and Christensen and Snyder 1990.
12. See Doyle 1986; Smith 1981; and Cain and Hopkins 1993. The labels for the three theories
are from Doyle 1986, 22-30.
13. Gallagher and Robinson 1953. On the ensuing debate, see Louis 1976.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
Anarchy 5
state in pericentric theories is functionally equivalent to abandonment in

alliance theories. Within this parallel, however, lies an apparent contradiction,
resolved in the theory below: whereas the potential for opportunism thwarts
alliances, it stimulates formal empires.
In both alliances and empires, then, our dominant theories understand states
as being motivated by the desire to aggregate capabilities. Both literatures also
view problems of opportunism as determinants of behavior. This should not be
surprising, given the need to build power and manage relations between
partners in the conduct of international relations. This common theoretical
core is central to the approach developed below. In this way, the theory builds
upon existing approaches in the field of international relations. Yet, because
scholars fail to appreciate this common practical and theoretical core, they
have not made a sufficient effort to explicate the continuum underlying
alternative relations or identify how opportunism varies across these options.
While clearly building on and indebted to the existing literature, the theory
developed here seeks to explain better these variations in relations across
partners and over time.
This common core also ignores the important role of governance costs, which
have made a significant contribution to the theory of international regimes and
form a necessary part of the analysis below.14 A second goal of the theory is thus
to incorporate into the core a focus on issues of international governance-a
topic often believed to be empirically or theoretically unimportant to security
studies.15 As seen below, governance, broadly conceived as the management of
relations between actors, is carried out through a range of alternative policies.
As a generic problem of politics, it has a wide variety of possible solutions.
While international organizations and regimes may be of marginal importance
for security issues-itself a debatable proposition-governance costs are
necessary to any explanation of grand strategy.
The variety of international relations
Security, the starting point for any discussion of grand strategy, is too often
undefined, and when it is defined, it is usually tailored either to the specifics of
time and place or the idiosyncratic preferences of the author.16 I define security
here as the ability to consume, invest, or otherwise use national wealth as a
polity sees fit. In other words, a country is secure to the extent that its wealth
cannot be coerced or otherwise extorted from it. It follows from this definition
that a polity will be concerned with two essential national freedoms: the
freedom to possess wealth, traditionally associated with the territorial integrity
14. On regimes and transactions costs, see Keohane 1984.

15. See Jervis 1983; and Mearsheimer 1994-95. For alternative views, see Wallander and
Keohane 1995; and Lipson 1994 and 1995.
16. On the concept of security, see Buzan 1991.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
of the nation-state, and the freedom to choose its own form of rule, embodied
in constitutional independence and a defining characteristic of sovereignty.'7
Central to the latter freedom is the country's ability to decide autonomously
how to accumulate and use its wealth, which requires that the collective choice
mechanism be free from foreign control. So defined, security is aspired to
rather than realized; it is a continuous rather than a dichotomous variable. A
country is more secure the lower the probability that other states can seize or
constrain effectively the disposition of its wealth.
Security is a complex phenomenon, dependent upon a state's assessment of
the likely behavior of another. The behavior of the threatening state is in turn a
function of its intentions and military capabilities, which for simplicity are
assumed to be exogenous throughout this essay (see below), and the first state's
own defense efforts and security relations, which it can alter in response to
changing preferences or circumstances. A state's defense effort is influenced by
many factors, including threats from others, the aggregated preferences and
risk propensities of its constituents, and the costs of producing defense relative
to other goods in the economy. These factors are captured in the traditional
guns or butter trade-off, which reflects the willingness of the polity to trade
other valued goods for increased defense effort and, in turn, security.
When confronted by a common threat, a state and its partner may choose to
pool their resources, abilities, and efforts with one another in what I call here a
security relation.18 Security relations can take a variety of forms and can vary by
dyad. The dyadic nature of these relations is particularly important. Each state
has many potential partners, and there are many degrees of hierarchy within
each possible relationship (see below). Accordingly, relations can differ across dyads,
taking one form with one partner and another with its neighbor, depending on
circumstances. In the early postwar period, for example, the United States entered
into an alliance with Australia and New Zealand but, in the same general region,
expanded its imperial outpost in Guam to include all of Micronesia.
Security relations vary along a continuum defined by the degree of hierarchy
between the two parties (see Figure 1).19 The degree of hierarchy is in turn
17. James 1986. Today, as factors of production become more mobile internationally, the
traditional concern with territorial integrity may be less central; see Rosecrance 1986.
18. States also possess the option of unilateralism, in which they choose not to pool resources
and efforts in the joint production of security. Prior to 1945, this was the prevailing policy of the
United States, and it is the principal alternative to alliances today. In addition to the expected costs
of opportunism and governance costs, the choice between unilateralism and the "relational"
alternatives discussed in this article is also influenced by scale economies in producing security.
More specifically, the analysis must expand to include the division of labor between the parties, the
technology of production, and positive externalities in the unilateral and relational alternatives.
For reasons of space, I do not develop this additional dimension of policy or set of causal variables
in this article. As scale economies are important only for the choice between unilateralism and the
optimal relationship between parties defined below, the analysis presented here stands on its own.
For a full treatment, however, see Lake forthcoming. I also do not address alternatives to
balancing, such as appeasement or detente with threatening states.
19. Although relationships may be hierarchic, the international system remains anarchic. See
Waltz 1979.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
Anarchy 7
Anarchy Hierarchy
XWgce O~~cwtet ,geaOe

aX~~~~~~~
FIGURE 1. A continuum of security relations
defined by the locus of rights of residual control.20 All relationships, whether

entered into voluntarily or as a result of coercion, can be considered as based
upon some "contract" between the two parties specifying explicitly or implicitly
the terms under which they will pool their defense efforts and the residual
rights of control retained by each. Contracts are, simply, self-enforcing
agreements that define the terms of the relationship between two parties.
Parties can completely or incompletely specify their contracts. Completely
specified contracts detail numerous contingencies and set forth appropriate
responses by the parties; incompletely specified ones contain holes that are
filled in by the parties as necessary. In other words, the parties may either
"contemplate all conceivable bridge crossings in advance, which is a very
ambitious undertaking," as Oliver Williamson notes, or address actual bridge
crossings as events unfold.21 With costly information (see below), no contract
can address all contingencies in all possible states of the world. In practice, all
contracts are specified imperfectly and thus possess a varying "residual" of
unspecified rights, obligations, and actions. Which party has the ability de jure
or de facto to make decisions in this residual defines the rights of control. The
term "right" does not necessarily imply a de jure and formal recognition by
both parties of their authority to exercise control over the residual. The term
can simply reflect a de facto and informal ability of one state to control the
behavior of the other over some areas. Rights differ from mere influence,
however, by constituting an enduring pattern of control within an ongoing
relationship.
In anarchy, each party to the relationship possesses full residual rights of
control; while constrained by its environment, each state is master of its own
fate in that area of rights not previously ceded in the contract. In hierarchy, one
party-the dominant member-possesses the right to make residual decisions,
while the other party-the subordinate member-lacks this right. Thus, the
dominant state possesses control over all resources and assets of the subordi-
nate actor that have not been specifically reserved to the latter in the contract.
Whether the parties have entered into the relationship voluntarily or through
the pain of battle is irrelevant to the definition of hierarchy; the process of its
creation does not define a relationship's degree of relational hierarchy.
20. Surprisingly, economists typically fail to define hierarchy in a formal way. The definition
used here follows most closely from Grossman and Hart 1986.
21. Williamson 1985, 20.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
As Figure 1 illustrates, the range of relations is continuous in principle.

When relations move from anarchy to hierarchy, the rights of residual control
possessed by the subordinate party decline. Nonetheless, certain salient
relational forms have been observed historically. The placement of these forms
along the continuum is only approximate.22
In an alliance, which lies at the anarchic end of the continuum, states pool
resources in pursuit of some common objective while retaining complete rights
of residual control. Most modern defense alliances, including the Rio and
Australia-New Zealand-United States pacts formed after World War II, take
this form. The North Atlantic Treaty Organization (NATO) is also essentially
anarchic except for the position of the Supreme Allied Commander, Europe
and the special provisions governing Germany, both of which were added later
as the organization matured and reflect some transfer of residual rights to the
organization itself or between states within the alliance.
Empire lies at the hierarchic end of the continuum and occurs when one
partner cedes substantial rights of residual control directly to the other. Yet,
imperial amalgamations are seldom complete and vary considerably in form
and extent; the locus of residual control need not reside completely in the
dominant unit for us to regard the new entity as hierarchic.23 As implied above,
imperialism can be freely negotiated-as in the recent merger of the German
Democratic Republic into the Federal Republic of Germany-or coerced, as
witnessed in the late nineteenth-century scramble for Africa and in Germany's
quest in the first half of this century for a continental empire; under both
circumstances, it is the reduction in the rights of residual control that defines
the hierarchic relationship.
Between these extremes lie at least two intermediate relationships. In a
protectorate, one state cedes control to another over important areas of
national policy, most notably foreign affairs. Although the terms vary, such
grants of control are typically broad, made for extended periods of time, and
are not revocable. Such delegations of authority transfer residual rights of
control in the designated areas from the "protected" state to the "protector"
and severely constrain the former's ability to influence the policy choices the
latter makes for it. Britain extended a de jure protectorate over the so-called
native states of India and a de facto protectorate over what is now the United
Arab Emirates during the nineteenth and early twentieth centuries. The
United States has transformed parts of its Trust Territory of the Pacific Islands
established as U.S. empire under United Nations auspices in 1947, into
protectorates; under "free association" agreements adopted in 1986, the
United States now recognizes the Federated States of Micronesia and the
Republic of the Marshall Islands as sovereign states but continues to accept
22. One of the best, if dated, sources on relational forms is Willoughby and Fenwick [1919] 1974.
23. Compare the discussion of the United States in Waltz 1967 with that of hierarchy in Waltz
1979.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
Anarchy 9
responsibility for their defense. Relations between Russia and severa

Soviet successor states also appear to be evolving into protectorates.24
Informal empire exists when one state controls indirectly substantial residual
rights in the other; this typically occurs through a nominally sovereign but
functionally dependent and therefore controllable agent in the subordinate
state. Informal empire differs from formal empire, according to Michael Doyle,
principally in the mechanism of control, "which informal imperialism achieves
through the collaboration of a legally independent (but actually subordinate)
government in the periphery." This creates the reality of empire without the
form.25 In this sense, informal empire is rule by proxy or, more meaningfully, a
form of delegated authority. To the extent that the local government is
dependent upon the dominant state, it must obey the latter's commands and
anticipate its wishes. This dependence effectively, if not legally, transfers broad
residual rights of control to the dominant state. The greater the dependence of
the local government upon the dominant state and the less costly it is for the
latter to replace the former, the greater the transfer of rights.
An informal empire is difficult to recognize, though, since the exercise of
imperial power is evident only in "out-of-equilibrium" behavior. When the
dependence of the client and costs to the patron of changing local rulers are
well understood by both parties, no resistance occurs, no overt coercion is
necessary, and the local government complies with the wishes of the dominant
state as if in an alliance. Only if the client tests its dependence or the patron's
patience will the imperial controls become manifest. As a result, and as Doyle
suggests, informal empire can be distinguished from other relations only by
observing interactions over some extended period, thereby increasing the
chances for encountering anomalous behaviors.26
The most striking modern example of an informal empire was the Soviet bloc
established in Eastern Europe following World War II and subsequently
embodied in the Warsaw Pact. As events attested, the local communist
governments were fundamentally dependent upon Soviet backing; once Presi-
dent Mikhail Gorbachev withdrew that support, the regimes quickly toppled.
Moreover, the subordinate peoples consistently tested the limits of informal
Soviet rule, sparking suppressions of local dissent in East Germany in 1953,
Hungary in 1956, Czechoslovakia in 1968, and (indirectly) Poland in 1981. In
this example, recurring out-of-equilibrium behavior clearly affirmed the exis-
tence of an informal empire. As recent out-of-equilibrium events in Haiti,
Nicaragua, and Panama also suggest, many states in the Caribbean and Central
America are similarly caught in an informal U.S. empire.
While the lines between these historical forms may be difficult to discern in
the absence of attempts by subordinate actors to assert their independence, the
24. Roeder 1995.

25. Doyle 1986, 38 and 42.
26. Doyle 1986, 39 and 43.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
distinctions nonetheless capture important differences in relations. Not all

relevant relations are anarchic and, despite their nominal sovereignty, not all
countries have the same freedom of residual choice. "Pure" cases of anarchy or
hierarchy seldom exist. Even in what are mostly consultative relations between
allies some degree of control may be exercised over residual choices. And in
empire, local officials often are given some independence in responding to local
conditions. In the following, the terms "alliance" and "empire" are used as
shorthand for policies that lie closer to the anarchic or hierarchic end of the
continuum, respectively.
Contracting for security
To explain the choice of relations by states, I build upon theories of relational

contracting. Analysts first used these theories to explain the institution of the
firm, but they constitute an approach to understanding social organization
more generally. The key insights of the relational contracting approach are,
first, that the transaction is the unit of analysis-in the case here, the pooling of
efforts to produce security-and, second, that transactions can be carried out in
a variety of governance structures, or what I have called relations above. Stated
another way, the structures in which particular transactions occur are variable
and endogenous. The approach is highly suggestive for the kinds of compara-
tive analyses of grand strategy now lacking in international relations.
Transactions in turn are embodied in contracts between the relevant parties.
As noted above, contracts vary in both their specificity and the rights of residual
control possessed by each member; indeed, the latter is the defining attribute of
relational hierarchy. Contracts also contain safeguards created by the parties to
ensure mutual compliance. As Williamson suggests, such safeguards typically
include: (1) actions designed to modify opportunity costs (sanctions for failing
to perform the required actions, the exchange of hostages, side-payments), (2)
"specialized governance structure(s) to which to refer and resolve disputes,"
and (3) "regularities that support and signal continuity intentions."27 More
generally, the degree of relational hierarchy and coercion can be considered to
be safeguards. In constructing contracts, states are defining the terms of their
transaction; the potential for cheating, defection, and other forms of opportun-
ism; and the means for controlling one another's behavior. In this approach,
contracts are instruments through which to control the behavior of others.
The central hypothesis of the contracting approach is that parties choose
relations to economize on resources-in other words, to maximize benefits
from exchange and minimize transaction costs. In short, actors choose the
relationship that is most efficient for conducting the transaction. By doing so,
the parties maximize the resources that can be used for other valued purposes;
27. Williamson 1985, 34.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
Anarchy 11
for firms, this implies profit maximization and for states (under assumptions to
be discussed shortly), cost minimization. At a general level, this implication-
while clearly artificial-is nonetheless plausible. When decision makers are
perfect agents for their principals, whether these be shareholders in a firm or
voters in a democracy, they have no incentive to conduct relations in ways that
intentionally waste resources. Yet at this level, the implication is also a virtual
truism, as it is relatively easy to concoct post hoc stories about why any observed
relationship is efficient for the parties involved; indeed, this is perhaps the most
common criticism of the approach. The true test of a theory of relational
contracting comes from specifying more concretely the range of alternatives
and the determinants of efficiency-issues scholars continue to debate, thus
yielding multiple theories united by a common approach.
Below, I develop a theory of relational contracting tailored to international
security affairs. As we shall see, the optimal relationship is principally a
function of the expected costs of opportunism, which decline with relational
hierarchy, and the governance costs of creating and maintaining the relation-
ship, which increase with hierarchy.
Assumptions and limitations
In the following theory, I make several assumptions, which I will note here
but not defend at length. First, actors are rational, but information is costly and
therefore limited. The first part of this assumption is common but frequently
criticized in theories of international politics. By rationality, I mean simply that
actors possess transitive preferences and act purposively. While states may
know their own defense effort with certainty, they can observe features of their
environment-including their partner's defense effort-only at some cost. Due
to diminishing marginal returns, states never acquire complete or perfect
information about their partners. Rather, they estimate probabilities of certain
behaviors and update their beliefs as they receive new information. In short,
states are Bayesian decision makers and choose relations on the basis of
expected utility calculations.28
Second, security is a single dimension of policy; the relevant "selectorate"
(whether voters or Politburo members) possess single-peaked preferences; and
the government is a perfect agent for its selectorate. These are the familiar
conditions behind the median voter theorem, applied here more generally to a
wider variety of domestic choice mechanisms.29 This assumption produces a
highly stylized view of domestic politics in which policy always reflects the
position of the median selector and allows us to treat states as unitary actors. In
28. For an introduction to Bayesian decision making in a fully strategic setting, see Morrow
1994, 161-87.
29. For a good introduction to public choice models, see Schwartz 1987. For applications of the
concept of selectorate to nondemocratic polities, see Roeder 1993, 24-27; and Shirk 1993, 71-72.
For a related approach, see Achen 1988.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
making this assumption, I do not intend to imply that domestic politics are
unimportant. Rather, I want to focus attention on relations between rather
than within states. I have addressed some consequences of relaxing this
assumption elsewhere.30
Third, states do not possess preferences over particular relations; that is, the
selectorate does not innately prefer alliances or imperialism but values these
alternatives only as instruments toward some other, unspecified but valued
ends. States may be expansionary or status quo powers, power-seeking or
wealth-seeking; this assumption merely restricts states from having preferences
for or against particular instruments of policy. Like the second, this assumption
is often violated in reality, but its effect is to concentrate our attention on the
environment of states rather than their dispositions. Whether environmental
theories, such as this, or dispositional theories prove more useful is an
empirical question not addressed here.
Together, these first three assumptions imply that states choose the
relationship that minimizes their costs of producing a desired level of security.
By so choosing, states and their selectorates maximize their utility by conserv-
ing resources that can then be used to obtain other valued ends.
Fourth, the threats faced by dyads are exogenous. Positing exogenous threats
is an analytic convenience that produces a point of entry into the ongoing cycle
of action and reaction in international politics. Phrased differently, the theory
begins with the existence of a security threat from a third party and seeks to
explain how the members of a dyad choose a particular response. As noted
above, the threat from others is a complex phenomenon, driven by the third
party's goals and risk propensities, the state's own goals, the security dilemma,
and many other factors. Without explaining where threats originate, it is
nonetheless appropriate to ask how others respond.
Finally, when members of a dyad would otherwise choose different relations,
the stronger of the two parties prevails by coercing the weaker party into
accepting its preferred alternative. The current theory is decision theoretic
rather than game theoretic in construction. In essence, states are understood as
playing against "nature" when in fact they are interacting with other strategic
actors. This construction is problematic when coupled with the assumption of
costly information, as there is then no guarantee that actors will share identical
estimates of the probability distributions of the variables in the model-
although over time we may expect these estimates to converge. As a result,
states in the same dyad may choose different relations, a problem that arises in
the real world as what I have above called out-of-equilibrium behavior. This
assumption is, admittedly, rather severe-especially in dyads where the actors
are of relatively equal strength and, if locked in a coercive struggle, might
logically become caught in an infinite spiral. It is less problematic in the
unequal dyads, such as those between the superpowers and their respective
30. Lake 1992.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
Anarchy 13
partners, that motivate this article; in very practical ways, the superpowers were
relationship "makers" and others relationship "takers." Recognizing that this
is an important limitation, I nonetheless offer the current theory for consider-
ation and, in the spirit of ongoing debate, invite readers to challenge, extend,
and possibly refine its logic.
Opportunism
As discussed above, opportunism is commonly understood to be ubiquitous

in international relations-and, in the relational contracting approach, in other
areas of social life as well. It is defined by Williamson as self-seeking behavior
"with guile" and is assumed to be a trait common to all actors.31 Opportunistic
actors do not honor contracts out of a sense of obligation but press for
individual advantages whenever possible. Opportunism by partners-whether
intentional or not-is always a costly possibility. It is also a major determinant
of the choice of relations.
States face three kinds of opportunistic behavior by their partners. As Glenn
Snyder has posited, they may be abandoned. More generally, partners may
shirk by formally or informally abrogating agreements. In ways large and small,
partners may fail to fulfill their commitments. States may be entrapped by their
partners into undesired conflicts and wars.32 As such, entrapment is often a
problem of moral hazard where commitments, once issued, cause the partner
to act in a more risky, negligent, or aggressive fashion than before. Finally,
states may be exploited-a closely related form of opportunism not addressed
by Snyder. Having settled on an initial division of the benefits and costs of the
relationship, partners can subsequently seek to alter the terms of agreements
and obtain more favorable distributions of the joint gains.
When it occurs, opportunism is costly to the state. Its partner either
contributes less to the joint enterprise than promised, forces the state to divert
its own resources toward undesired ends, or seizes a greater share of the joint
gain than anticipated. Abandonment, for instance, not only reduces the
contribution of the partner but it can also reduce the efficacy of the state's own
defense efforts; if the state has specialized in a land-based army and its partner
has agreed to provide the complementary naval defense, for instance, opportun-
ism by the latter may leave the former more vulnerable than if it had produced
both an army and navy of its own.
The cost to the state of such behavior is determined by the state's
opportunity cost. When it possesses alternatives that it values almost as much
as the one in question, a state's opportunity cost is relatively low; it can shift
from relationship to relationship easily and without a significant loss in welfare.
When it has no attractive alternatives, the state's opportunity cost is high, and
31. Williamson 1985,30.

32. On abandonment and entrapment, see Snyder 1984.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
changing relations is a more consequential decision with more deleterious

welfare implications. The greater the state's opportunity costs, the greater is
the harm-or actual cost-the partner can inflict upon it.
Opportunity costs are determined by the degree to which assets are
relationally specific, or possess more value in one use and relationship than in
others.33 Assets can be highly flexible and easily transferred from one
application to another or highly specialized and difficult to redirect. The more
relationally specific the asset, the greater are the opportunity costs incurred by
the state. In security affairs, relationally specific assets range from military
forces requiring large investments in dedicated infrastructure or hardware
(e.g., heavily armored divisions designed for a land war against the former
Soviet Union or U.S. intermediate-range missiles developed for overseas
deployment) to ports uniquely located in strategic areas (e.g., Guam). Nearly
all divisions of labor entail some asset specificity; moving from specialized and
integrated forces to an autonomous military necessarily requires retooling,
retraining, redeployment, and time. In sum, the more specific the asset, the
greater the state's opportunity costs and thus the greater the costs inflicted by
the partner's opportunistic behavior.
The probability that the partner will act opportunistically, in turn, is a
function of the governance structure that the actors choose to construct. If
actors have safeguarded the provisions of the contract adequately (that is, if the
provisions are self-enforcing), their ability to act opportunistically is deter-
mined by their rights of residual control. The greater the residual rights, the
greater the discretion of the actor to behave in ways that may-intentionally or
unintentionally-undermine the joint defense effort. Thus, by implication, the
probability that the partner will engage in opportunistic behavior decreases
with relational hierarchy.
In an alliance, for instance, the partner retains complete residual rights and
thus wide discretion. Even though it may agree to declare war if a third state
attacks the first, the partner nonetheless retains the right to decide who
attacked and how many resources it will actually commit to the conflict. In
anarchic relations, ceteris paribus, the probability that the partner will behave
opportunistically is comparatively high. In an empire, at the other extreme,
states merge their formerly autonomous decision-making processes and trans-
fer rights of residual control to the dominant member. The dominant state now
decides-to continue with the same example-who is the victim of any attack
(presumably itself), who is the aggressor, and how many resources its partner
must mobilize. In practice, however, even in imperial relationships, localities
may possess some decision-making authority and, thus, some potential for
opportunism. Nonetheless, the tighter the imperial relationship and the more
33. This is also known as the asset's quasi-rent; see Klein, Crawford, and Alchian 1978. The
more specialized the asset, the greater are its quasi-rents. Asset specificity can also be understood
to affect the transactions costs of changing relations.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
Anarchy 15
integrated the amalgamated territories, the smaller the probability that the
subordinate partner will act opportunistically.
The expected cost of opportunism is a function of its actual cost, if it occurs,
and its probability. Expected costs can be absorbed directly by the state; here,
the state simply accepts some opportunism by its partner as a necessary evil.
Conversely, the state can deploy some of its own resources as a hedge against
possible opportunism. Even when a partner has agreed to help protect it, for
instance, the state may maintain redundant forces for fear that the partner will
not produce the promised benefits. The state may also forgo the most efficient
technology-or continue to use a less efficient technology in some portion of its
forces-for fear that its partner will renege on necessary basing rights or other
matters. Or, the state may waive some degree of specialization for fear that the
other will not live up to the terms of the agreement. Hedging is a form of
insurance against opportunism. Like actual opportunism, it is costly to the
state.
In principle, the expected costs of opportunism can be measured by the
resources necessary to hedge completely against this possibility, given existing
information. This conversion provides a useful metric and allows us to map the
resources required to produce a constant level of security onto the degree of
relational hierarchy, creating an "isosecurity" contour that reflects the change
in costs as a function of the probability of opportunism. This is illustrated in the
O contour in Figure 2. The shape of the contour reflects the probability that the
partner will act opportunistically. Following from the arguments above, it is
negatively sloped throughout. The more effectively hierarchical governance
structures control opportunism, the steeper the contour will be. The height of
the contour in turn is positively related to the state's opportunity costs; O'
represents either an exogenous increase in the state's opportunity cost or,
comparatively, another dyad with higher costs of opportunism.
All other things considered, states should prefer to bind their partners in
more hierarchic relations. The fewer the residual rights of control retained by
the partner, the lower the potential for opportunism and thus the expected
costs. The specific relationship chosen, however, also depends upon gover-
nance costs.
Governance costs
States incur governance costs in creating and maintaining relations. These

costs take three distinct forms: distorted incentives in the subordinate partner,
safeguards on the dominant state, and coercion. Despite the prominence
attached to coercion in international relations-defined broadly to include all
threats or actions designed to reduce the welfare of the partner unless certain
actions are performed (compellence) or not performed (deterrence)-its role
and importance are more easily understood once the logic of voluntary
contracting is explicated. Where opportunism declines with the degree of

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
Costs
G'
E3 El E2
Anarchy Hierarchy
(alliance) (empire)
FIGURE 2. Optimal se
opportunism (0) and govemance costs (G)
relational hierarchy, governance costs increase with relational hierarchy. As a

result, governance costs deter states from pursuing more hierarchic relations.
Distorted incentives in the subordinate partner. The shift of residual

control from one partner to the other distorts incentives for the efficient use of
resources in the subordinate member of the dyad. In a purely voluntary
relationship, this creates for the dominant state both additional costs of
monitoring and control and the need for greater side-payments.
With complete residual control, as in anarchy, parties to an agreement are
motivated to produce security in the most efficient manner possible, as the
resources thereby saved from less productive alternatives can be used for other
valued purposes. Just as individuals are best motivated when they are claimants
on the profits of a firm, so are states best motivated by the freedom to use their
resources in any way they choose. As the subordinate party's residual control
shrinks, its incentives are altered; the resources released by efficient production

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
Anarchy 17
are subject to increasing control by the dominant state and, thus, are less
valuable to the subordinate. As the subordinate party's residual control
recedes, incentives in other areas of economic and political life are also
distorted. By definition, as residual rights decline, the dominant state exercises
control over a greater range of behavior and, by implication, directs resources
to uses the subordinate party would not choose on its own. As the residual
rights of control shift from one party to the other, distortions in the subordinate
member multiply.34
These distortions-both in the production of security and other areas-can,
in part, be corrected by additional contractual provisions that mandate certain
specified actions by the subordinate party. When states adopt this contractual
solution, however, the dominant state must employ additional resources to
monitor and safeguard the new provisions. As the distortions increase with
relational hierarchy, the resources employed by and therefore the costs to the
dominant state must increase as well.
Contracting cannot remove distortions entirely. Monitoring and safeguard-
ing additional areas of behavior are subject to diminishing marginal returns. It
is also efficient to shift the locus of residual control from one party to the other
precisely when it is difficult to specify future contingencies; as a result, it is very
costly to safeguard against all possible distortions. Because of its inherent and
uncorrectable distortions, increasing hierarchy-ceteris paribus-reduces the
welfare of the subordinate party. To gain the subordinate party's willing
consent to a hierarchic relationship, the welfare losses created by these
distortions must be compensated by some transfer or side-payment from the
dominant state-increasing costs to the latter. As the subordinate partner's
residual control declines and the distortions increase, so must the compensa-
tion package offered by the dominant state.
Safeguards on the dominant state. Relational hierarchy shifts the locus of

opportunism from the subordinate partner to the dominant state. As the
partner's rights of residual control decline, the rights of the dominant state over
it expand-thereby increasing the potential that the latter will act opportunisti-
cally toward the former.35 Especially important here is the enhanced ability of
the dominant state to exploit the subordinate party; by using its enhanced
rights of residual control, the state can alter the terms of the initial contract to
its advantage. In the absence of coercion, the subordinate partner will not
submit to this vulnerable position unless the behavior of the dominant state is
adequately safeguarded in the agreement.36
34. This argument has a direct analog in the case of private firms; see Grossman and Hart 1986.
35. On opportunism by hierarchical authorities, see Dow 1987.
36. Safeguards render the dominant state's commitment not to exploit the subordinate partner
credible. On the problem of credible commitments in international relations, see Powell 1990;
Bueno de Mesquita and Lalman 1992; Fearon 1992; Papayoanou 1992; and Cowhey 1993. James
Fearon focuses on precisely the problem here-that after consolidation the weaker party will

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
Central to any contract are safeguards employed by both parties to protect

their interests, ensure compliance, and thereby render the agreement enforce-
able. Safeguards can apply both to the items agreed to in the contract and the
residual rights of control. They impose costs on states directly by committing
them to particular undertakings designed to modify opportunity costs and
signal continuity intentions-such as deploying troops in exposed "tripwire"
positions to assuage allied fears of abandonment. Safeguards also produce
costs through the creation and maintenance of specialized governance struc-
tures, ranging from, say, NATO today to Britain's imperial apparatus at home
and abroad in the nineteenth century. The more extensive and severe the
safeguards, the more costly they are; the greater the safeguards, in other words,
the greater the resources that are employed to alter opportunity costs, signal
continuity intentions, and create and maintain specialized institutions.37
Any partner will insist upon adequate safeguards against future opportunism
prior to entering voluntarily into a relationship. Given the shift in the locus of
residual control, subordinate partners will be increasingly attentive to this need
as the degree of relational hierarchy increases. As a result, dominant states
must undertake costly actions to bind themselves to the terms of the contract
and, especially, must commit to exercise their expanded residual control in a
nonopportunistic manner. Again, the latter requirement is particularly difficult
to meet, as it is most efficient to shift rights of residual control from one state to
another under conditions of uncertainty; if future contingencies cannot be
specified in the contract, neither can those circumstances under which
safeguards would be employed. By their very nature, safeguards on the
enhanced ability of the dominant state to exploit its subordinate partner will
tend not to be credible-suggesting why coercion, discussed immediately
below, is so prevalent in relational hierarchies. Nonetheless, the relationship
between increasing hierarchy and increasing safeguards on the dominant party
is continuous; the greater the relational hierarchy, the greater the safeguards
the dominant state must accept in order to gain the voluntary compliance of the
subordinate partner.
This insistence on increasing safeguards can be demonstrated amply by a
comparison of alliances and federal states. Alliance treaties, at most, obligate
states to come to one another's aid under some specified circumstances and, at
a minimum, oblige them only to consult with one another. Such treaties
typically contain few safeguards other than the right of withdrawal. Alliance
treaties contrast with, say, the Constitution of the United States, which
embodies one of the great voluntary amalgamations of previously or at least
potentially sovereign states. While it contains some specific provisions, the
framers of the American Constitution made no attempt to mandate actual
become even weaker-and finds that under a wide range of circumstances the weaker side will
choose to fight rather than accept an incredible agreement. See Fearon 1993.
37. Safeguards may also create maladaption costs, which arise when circumstances change and
render existing contracts less appropriate. See Williamson 1985, 21; and North 1990.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
Anarchy 19
behaviors; rather, the agreement focuses on the design of institutions-the

separation of powers, the composition of the Senate, the creation of a
nonpartisan Supreme Court-to safeguard the important interests of the
constituent states within the new hierarchy and, especially, the rights of the
smaller members who might otherwise be overwhelmed by their larger and
more powerful neighbors. As relational hierarchy increases, the safeguards on
the dominant state and the governance costs it incurs must increase as well.
Coercion. Although it fits awkwardly within the neoclassical economic

approach that informs relational contracting theories, coercion is a fact of life
in international relations (and elsewhere). The governance costs incurred in
correcting and compensating distortions in the subordinate party and safeguard-
ing against the potential for exploitation by the dominant party both are rooted
in problems of contracting under costly information. Coercion is a substitute
for contracting. It can be used by the dominant state to correct distortions by
eliciting appropriate behaviors from the subordinate party in difficult-to-
specify contingencies. Likewise, it can be used to secure compliance with a
hierarchical relationship without granting compensation for the resulting
distortions. It also can be used in lieu of effective safeguards on the dominant
state. Rather than binding itself to costly actions that alter opportunity costs,
demonstrate commitment, or create specialized governance structures, the
dominant state can force its partner to accept a contract that lacks the
safeguards that would be necessary in a purely voluntary relationship. In short,
coercion is an alternative to the concessions and compromises that would
normally be required in any voluntarily negotiated contract between two
parties.
Coercion also can enable states to overcome differences in their preferred
relations by forcing the weaker party to acquiesce to the wishes of the stronger
(this underlies the fifth assumption above). Just as muggers force their victims
to choose between their money and their lives, so powerful states can force less
powerful ones to choose between a subordinate position in a hierarchic
relationship and enormous harm. The asymmetrical ability to use coercion
often leaves the weaker party with only the illusion of choice-and sometimes
not even that. Nonetheless, the alternative relations and determinants de-
scribed in the theory remain salient, defining both the advantages to the weaker
party of not conceding to the other and the amount of force the stronger must
employ. Given the difficulty of safeguarding against the potential for future
exploitation of the subordinate party by the dominant state, coercion may be
the only means of effectively establishing informal and formal empires.
Throughout history, the terms of the peace have been written by the victors, not
the losers-the powerful, not the weak. The same holds for hierarchical
contracts.
Like contracting, coercion requires the use of scarce resources and, there-
fore, is costly to countries that use or threaten to use it. This is clearly true for

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
direct military action, but it holds equally for other sanctions, such as trade
embargoes (which, if effective, reduce the sanctioning country's terms of
trade). States will choose between contracting and coercion according to their
relative costs. Countries that possess a comparative advantage in coercion-
whether from sheer size, military prowess, resource endowments, or whatever-
will tend to rely upon this instrument. Coercion does appear to be the more
cost-effective and, at least for hierarchic relations, the more frequently used
instrument. The process of consolidating political authority into larger territo-
rial units, the state-building process that produced the modern nation-states in
today's developed world, frequently was violent.38 Likewise, the threat of
coercion by technologically superior European states was key to the process of
empire building in the periphery of the global system.
As with the other governance costs, the costs of creating and maintaining a
relationship through coercion typically increase with hierarchy. The more
hierarchic the imposed relationship, the smaller the subordinate actor's rights
of residual control, and therefore the more it can be expected to resist its loss of
freedom both at the outset and throughout the course of the relationship. The
greater the resistance, the greater the coercion necessary to support a given
relationship.
Thus, the governance costs of acquiring control over others and maintaining
a relationship either by contract or coercion increase as relations move from
anarchy to hierarchy. As Figure 2 illustrates, defined in terms of the resources
used in correcting and compensating distortions, safeguarding against the
potential for exploitation by the dominant state, and coercing the subordinate
party, these costs aggregate into a positively sloped governance schedule (G). I
assume here that such costs rise at an increasing rate and that each move
toward hierarchy becomes progressively more expensive to the dominant state.
The absolute quantity of resources devoted to governance determines the
height of the G schedule. Its shape is derived from the relative costs of
alternative governance structures.
Optimal relations
The expected costs of opportunism vary over time and across countries
depending upon the opportunity costs of the state and its partners. They also
vary across alternative relations, declining with hierarchy. Governance costs
vary across time and space as well, depending upon the ease of safeguarding
against and coercing partners. These costs rise with the degree of relational
hierarchy, deterring states from imperialism.
Together, the expected costs of opportunism and governance determine the
optimal relationship between a state and its partner.39 For any dyad, this
38. On state building, see Tilly 1990.

39. Optimal relations are also influenced by feedback effects, which occur when relations in one

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
Anarchy 21
relationship is defined by the intersection in Figure 2 of the isosecurity contour

and the governance cost schedule. At the point of intersection, one factor of
production devoted to increasing hierarchy releases one factor from hedging
against the risk of opportunism. To the left of this point, an additional factor
devoted to governance releases more than one unit from hedging. To the right,
an additional factor spent on hierarchy releases less than one unit from
hedging. Exogenous increases in the expected costs of opportunism raise the
isosecurity contour and shift the equilibrium toward greater hierarchy (E2).
Likewise, exogenous increases in governance costs shift the schedule upward
and move the equilibrium toward greater anarchy (E3). Changes in the shapes
of the curves produce analogous effects (see Figure 3 below for examples).
States possess unique sets of curves for each partner, suggesting that optimal
relationships will differ across dyads.
In equilibrium, we should observe anarchic relations most commonly when
the expected costs of opportunism are low or governance costs are high, and
hierarchic relations when the expected costs of opportunism are high or
governance costs are low. These equilibrium conditions reconcile the apparent
contradiction revealed in the discussion of opportunism in the literatures on
alliance formation and imperialism above.
The common core in international relations, focusing on capability aggrega-
tion and opportunism, is central to the relational contracting theory developed
in this article. In producing security, states aggregate resources and worry
about opportunism by their partners. The expected costs of opportunism play
an important role in determining the choice of relations, leading states to
prefer more rather than fewer hierarchic alternatives. In a way not sufficiently
appreciated in the common core, however, the risk of opportunism is partly
endogenous. The expected costs of opportunism are influenced by circum-
stances, but they are also a function of the relations states choose, declining
with the degree of hierarchy. States mold their relations to control this risk,
pursuing greater hierarchy when it increases and less hierarchy when it
decreases.
Missing from the common core identified above are governance costs, which
rise with relational hierarchy and constrain states from pursuing more
imperialist alternatives. The importance of these costs is evident from the
counterfactual: in their absence or if instead of rising they declined with
relational hierarchy, we would observe an inevitable trend toward empire in
most dyads-a trend clearly at variance with reality. Somewhat counterintu-
itively, relational hierarchy is costly to dominant states. They must safeguard
distortions or offer side-payments to their subordinate partners, expend
resources to limit their own ability to exploit their partners in the future, or
dyad allow third parties to update their prior beliefs about the expected costs of opportunism or
governance of their relations with the members of that dyad, and scale economies in governance
costs, which may promote multilateral relationships. For reasons of space, these extensions of the
basic model are not developed here. See Lake forthcoming.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
coerce their partners. In the end, states weigh the efficacy of hierarchy agai
increasing governance costs when choosing their security relations.
Superpower strategies
Assessing the theory developed above faces three major hurdles. First, the
independent variables are not easy to operationalize. The expected costs of
opportunism and governance costs both are difficult to measure precisely; in
turn, we lack systematic data. This same problem also has bedeviled tests of
relational contracting theories in economics. Direct tests of the theory based on
good proxies for these variables will be nearly impossible for the foreseeable
future.
Second, and relatedly, both variables actually are defined as probability
distributions: states do not know the expected costs of opportunism or
governance costs with certainty but base their decisions on some set of prior
beliefs that they update as they receive new information. The probabilistic
nature of the variables makes definition and measurement difficult. Case
selection, then, is extremely consequential. For example, focusing on promi-
nent historical events-especially noteworthy policy failures where the strate-
gies employed performed far under expectations-risks limiting analysis to
cases with extreme values for one or both variables. Deterrence failures
resulting from buck passing or entrapments from chain ganging, often
described as pathologies resulting from dysfunctional political or cognitive
biases, may simply be the result of unfortunate events drawn from the
"unlucky" tails of the probability distributions.40 Fortunately, the dyadic nature
of the present theory multiplies the number of possible cases.
Third, and most important, we can observe the relationship chosen within a
dyad, but we cannot observe the relationships not chosen and the costs and
benefits associated with those counterfactuals.41 In other words, we can
observe, presumably, the intersection of the isosecurity and governance cost
contours, but we cannot observe directly the height and shape of the curves
themselves. Yet these unobservable relationships are central to the theory and,
as argued above, structure the choice of relations by states.
All three of these problems are quite common in the social sciences but
typically are glossed over; they certainly are not unique to the current theory.
Given these problems, one appropriate research strategy is to use historical
episodes in which there were relatively clear differences in the independent
variables to compare the static predictions of the theory. Through such cases,
we can observe the direction of the independent variables and predict the
40. Christensen and Snyder 1990, 138 and 141.

41. This is a form of the fundamental problem of causal inference. See King, Keohane, and
Verba 1994, 79.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
Anarchy 23
range of possible variation in relations even if we cannot assign precise values

or fully explicate the counterfactuals. Predictions can then be assessed against
the record. In advocating this strategy, I am not suggesting that we should
abandon scientific criteria or eschew systematic large-n research designs.
Rather, I suggest that carefully selected historical cases are a necessary first
step in assessing a theory with difficult-to-measure variables.
In the remainder of this essay, I undertake a preliminary and highly tentative
assessment of the theory, contrasting U.S. and Soviet relations with their
partners in Western and Eastern Europe, respectively, in the years immedi-
ately after World War 11.42 This assessment is necessarily abridged, but it
suggests that the theory can usefully illuminate empirical puzzles. In turn, the
thumbnail sketches offered do not appear to disconfirm the theory. For brevity,
I use Western and Eastern Europe as the units below; this is acceptable given
the broad similarities in circumstances and policies within each region. For
reasons of space, I cannot develop the interesting within-region variations.
Nonetheless, it is important to remember that-despite this shorthand-this
assessment is based on a series of dyadic observations, increasing the number
of cases and consequently our confidence in the initial fit between theory and
history.
As Figure 3 illustrates in brief, the United States faced a comparatively low
risk of opportunism by its partners after World War II, even in anarchy (Ous in
Figure 3), and rapidly escalating governance costs (Gus), implying that a
relatively anarchic relationship would be adopted between the United States
and Western Europe (such as that at US). By comparison, the Soviet Union
confronted a higher risk of opportunism (OussR) and lower governance costs
(GussR), suggesting that a more hierarchic relationship with Eastern Europe
would be selected (USSR). Though speculative, the isosecurity contours and
governance cost schedules in Figure 3 illustrate how the variables combine to
predict broad patterns of relations. I begin with an overview of the relations
chosen and then briefly discuss each variable.
Anarchy versus hierarchy
In 1948, the United States began negotiations with Britain (and Canada)
over what would eventually become NATO. The treaty was signed in April
1949 and ratified three months later. Although the Korean War heightened
America's commitment to Europe, deepened its involvement, and institutional-
ized the alliance, the basic direction of U.S. foreign policy already had been
determined prior to June 1950. Article 5, the core of the treaty, simply states
that "the Parties agree that an armed attack against one or more of them in
42. For an extended treatment of the U.S. case and more complete citations, see Lake
forthcoming, chapt. 4. For the Soviet case, the single best source remains Brzezinski 1967. Other
noteworthy studies include Holden 1989; Jones 1990; Mastny 1979; and Ulam 1974.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
Costs Gus
GUSSR
OUSSR
------- ~ ?us
US USSR
Anarchy Hierarchy
(alliance) (empire)
FIGURE 3. Optimal sec

during the early post
G = govemance costs
Europe or North America shall be considered an attack against them all" and
calls upon each member to "assist the Party or Parties so attacked by taking
forthwith ... such action as it deems necessary, including the use of armed
force, to restore and maintain the security of the North Atlantic area." Article
3 specifies that the member states should engage in "continuous and effective
self-help and mutual aid."43 While based on mutual defense and assistance,
each state is free under the treaty to interpret circumstances and choose its
response according to its own designs. Each remains in control of its own policy
not only in all areas outside the alliance but in virtually all areas covered by the
agreement as well.
The Soviet Union's informal empire in Eastern Europe was more opaque.
While each state remained nominally sovereign, the Soviet Union exerted
control over significant areas of political, economic, and social activity in
Eastern Europe through two principal means. First, the Soviet Union imposed
43. The treaty is reprinted in Kaplan 1984, 227-28.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
Anarchy 25
ideologically compatible states on the local populations, acting under the

general rule that noncommunist regimes meant anti-Soviet governments.44
More important than ideology, however, these regimes typically lacked popular
support. As a consequence, the new communist leaders in Eastern Europe
were dependent upon Moscow for continued rule. Rather than serving as
agents of their local populations, these leaders served more or less at the behest
of Moscow and as agents of their Soviet principals. The structural weakness of
the East European regimes gave the Soviet Union substantial de facto residual
rights of control over those areas in which it chose to exercise those rights.45
Second, the Soviet Union partially integrated the economies of Eastern Europe
into its own centralized command economy. Over time, the political depen-
dence of the local communist rulers was reinforced by the economic depen-
dence of the state and society. The command system and economic depen-
dence further solidified and, indeed, institutionalized the transfer of residual
rights to Moscow.46
In Western Europe, as many revisionist historians have argued, the United
States also was concerned about the composition of local governments and
patterns of economic transactions.47 Yet, the U.S. role in Europe tended to be
proscriptive rather than prescriptive. The United States was involved in the
internal politics of several European states after the war. It worked against
communism in France, Greece, and Italy, but it did not typically support
domestic elites who lacked substantial indigenous support. It also sought to
prevent economic dependence on the Soviet Union, to destroy existing imperial
economic systems, and to strengthen the international market economy in
which it was the dominant competitor. But, in all cases, the United States
cultivated a more diversified and diffuse form of economic dependence that
stopped short of dictating patterns and levels of exchange. As the entire theory
above suggests, residual control is a question of degree rather than kind.
Without denying the importance of the United States in European politics
after 1945, it is nonetheless apparent that the Soviet Union exercised far
greater residual control over Eastern Europe.
Opportunism
The risk of opportunism by America's partners in Western Europe was

relatively low. In the security arena, the Europeans were challenged by the
Soviet Union, dependent upon the United States, and limited in their
44. Campbell 1984.

45. Others emphasize direct control through the secret police, Soviet ambassadors, and other
forms of monitoring and control. See Brzezinski 1967, 116. While important, the effectiveness of
these instruments depended upon the structural weakness of the East European regimes.
46. Brzezinski 1967, 99-100.
47. The classic statement of the revisionist position is Williams 1972. For a more nuanced
revisionist account, see LaFeber 1993. The revisionist perspective has generated a large
counterliterature, including the voluminous history by Leffler 1992.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
alternatives. The United States also promised to be the postwar center of

economic dynamism, and Europe's recovery was closely tied to access to the
American market. With high political and economic opportunity costs, the
Europeans were fixed in the American orbit, and the United States rested
secure that it was unlikely to be abandoned by its partners-even in a relatively
anarchic alliance. The United States was concerned with free riding by its
smaller partners and possible entrapment by its more adventurous or re-
vanchist allies. Such forms of opportunism, however, could be effectively
safeguarded at minimum cost through creative institutional solutions within
NATO (see below).
The potential for opportunism in the Soviet sphere in Eastern Europe, on
the other hand, was quite high. As a great power, the Soviet Union had long
been a regional threat and it had a demonstrated record of aggrandizement
against its neighbors. Indeed, before and during the war many in Eastern
Europe perceived Germany as the lesser of two evils and sought to use the
German reich as a counterweight to an even more distasteful Soviet imperial-
ism.48 The Soviet Union's regional dominance only increased with the defeat of
Germany. Rather than emerging as a "good neighbor," the Soviet Union exited
the war as a revisionist regional power against whom the East Europeans would
normally balance rather than bandwagon.
Similarly, the East Europeans, and particularly the northern tier states of
Czechoslovakia, East Germany, Hungary, and Poland, had been previously
integrated into the European market economy. Economically, as well as
politically and socially, these states were part of "Europe," and the economic
turn toward the Soviet Union was, to put it simply, unnatural. The attraction of
the West, moreover, was reinforced by America's postwar economic vitality.
Czechoslovakia and Poland's initial interest in participating in the Marshall
Plan provided continuing evidence to the Soviet Union of the pull of the
Western economic and political systems. Russia was justifiably concerned that
the states of Eastern Europe would abandon it at the first opportunity-as
indeed they did some forty years later.
Governance costs
The conduct of the war left the Soviet Union and United States with
dramatically different governance costs in their respective spheres in Europe.
The United States fought alongside its Western allies during the war and, with
the exception of its occupation zone in Germany, did not claim exclusive areas
of jurisdiction upon victory. To assert control over the areas it liberated would
have forced the United States either to forfeit the considerable gains it
acquired from pooling resources with its partners during the conflict or to tu
its war machine against its allies. The Soviet Union, conversely, fought by itself
48. Dallin 1942, 242.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
Anarchy 27
on the eastern front. While they cooperated with local resistance groups, often
communist, Soviet troops liberated most areas from Axis control, directly
occupied them after the war, and-in the ensuing flux-ensured that commu-
nists and other pro-Soviet groups formed important parts of every postwar
government. As a result, the costs of establishing control over and coercing
their postwar partners were absorbed in the defeat of Germany. From the
perspective of postwar relations these costs were "sunk." In other words,
finding themselves in a dominant position in Eastern Europe at the close of the
war, past costs of coercing partners were not factors in the choice of future
relations. The costs that theoretically-and in the West, practically-impeded
the pursuit of hierarchy dropped out of the Soviet calculus. All the Soviets
needed to consider were the future costs of governing territories they already
effectively controlled.
The governance costs of maintaining hierarchic relations also differed
between the Soviet Union and the United States. For the United States, the
costs of an alliance were small but rose rapidly with relational hierarchy. In this
regard, the U.S. occupation of Germany provides a useful case through which
to probe the relevant counterfactual. Even though the defeat of Germany left
U.S. troops dominant in their zone and thus in a position similar to that of their
Soviet counterparts, the ongoing costs of the occupation were prohibitive,
suggesting that greater hierarchy elsewhere also would have been extremely
costly. The Germans were expected to pay all costs associated with the
occupation, and they bore a substantial burden during a period when many
were living below subsistence levels: between 1946 and 1950, for instance,
between 42 and 26 percent of all tax revenues collected in Germany were used
to pay for the occupation.49 Yet, the United States maintained that it was
spending far more than it received from Germany in reimbursements-and
may have been spending four times more for the occupation than the Germans
themselves. Even in 1946, during the initial, punitive peace phase when little
sympathy was shown the defeated Germans and before Marshall Plan and
other forms of aid began to flow to the war-devastated state, the occupation of
the then-consolidated Anglo-American zone was costing taxpayers in Britain
and the United States over $600 million per year.50
Closer to its occupied territories, less attentive to local needs, and more
willing to use coercion as a substitute for voluntary compliance, the costs to the
Soviet Union of its informal empire were relatively low in the early postwar
years-taking the form mostly of low paid occupation troops.5' Moreover, these
49. Grosser 1955, 90. See also Davidson 1961, 261, 265, and 297.
50. Botting 1985, 216 and 110.
51. For suggestive evidence that leaders within the Soviet Union considered the maintenance
costs of the informal empire and decided nonetheless to continue the relationship, see the growing
literature on the so-called Beria Affair, especially Richter 1993, in which the secret police chief
appears to have proposed loosening Soviet control over East Germany-possibly to the point of
allowing reunification.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
direct governance costs were offset by substantial resource extractions from

Eastern Europe. In the period 1946-56, reparations paid to the Soviet Union
totaled $14-20 billion. Substantial implicit trade subsidies added to this
amount. By one estimate, the Soviet Union extracted on net $1 billion per year
from Eastern Europe until 1956, when the rebellion in Hungary forced a
dramatic reversal in policy.52 Even if it did not produce a profit in the early
postwar years, the informal empire clearly did not constitute a significant drain
on Soviet coffers.
Finally, U.S. fears of entrapment and exploitation could be safeguarded at
relatively low cost within the alliance. The United States avoided any automatic
commitment to Europe in NATO, retaining instead its freedom to respond at
the time and in the manner of its own choosing even to direct attacks on the
territories of its allies. And as its original role deepened with the outbreak of
the Korean War, the United States also secured a commitment from its allies
that the expanded and newly integrated forces in NATO would be led by a U.S.
general directly responsible to the U.S. President. The ability of the Europeans
to entrap the United States was thereby limited. The United States also
emphasized the importance of mutual aid within NATO. Embodying this
emphasis, Article 3 of the treaty was used with some effect to pry greater
(in-kind) contributions out of Europe than might have occurred without the
treaty-including base rights and, despite France's opposition, the eventual
rearmament of Germany. This emphasis on mutual aid also signaled to the
Europeans that they would be expected to carry a greater share of the burden
once they had recovered from the war. In short, the institutionalization of the
alliance within NATO and the safeguards thereby created served to reduce the
risks of opportunism by the Europeans at little cost to the United States,
increasing the attractiveness of more anarchic relations. Safeguards against
free riding and entrapment were also available at similar costs to the Soviet
Union-and in some cases were taken up, as when Soviet officers served in
military command and advisory positions in Eastern Europe-but they were
largely inadequate for solving the more thorny problem of potential abandon-
ment.53
I have painted this initial assessment in broad brush strokes, losing subtlety
and masking local variations in the process. Nonetheless, the evidence is
broadly consistent with the theory. The theory, in turn, helps unravel the
empirical puzzle that began this article. Facing relatively low expected costs of
opportunism and escalating governance costs, the United States opted for
anarchic alliance-based relations. With comparatively high expected costs of
opportunism and lower governance costs, especially as the costs of coercion
52. The net extraction estimate is from Brzezinski 1967, 285-86. The debate over implicit price
subsidies is large; see Marrese and Vanous 1983; Marer 1984a; 1984b; van Brabant 1984; Crane
1986; and Poznanski 1988. On the politics of implicit trade subsidies, see Stone forthcoming. The
history of Soviet-East Europe economic relations is surveyed in Bunce 1985.
53. Brzezinski 1967, 122, 173, 459.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
Anarchy 29
were sunk, the Soviet Union chose more hierarchic relations in Eastern
Europe. While the evidence is supportive, it provides only a preliminary and
tentative assessment of the theory. I have neither explicated fully the variables,
probed within-region variations, nor controlled for plausible rival hypotheses-
such as regime type. Even so, the broad fit provides a source of optimism for the
likely success of more extensive tests.
Conclusion
Grand strategy must be understood as a choice across alternative relations, not

as the choice of a single policy. Relations vary along a continuum from anarchy
to hierarchy, or alliance to empire. The choice across alternative relations, I
have argued above, is a function of the expected costs of opportunism and
governance.
The theory developed above and the difference in relations it explains are
not simply of academic or historical interest. The concerns central to the
variation in superpower relations are central to every state's foreign policy-
especially for those seeking their way in the post-cold war world. Whenever
they seek to cooperate, states must invest in governance structures designed to
hedge against the expected costs of opportunism by their partners, with
important consequences for themselves and others.
Although the theory is explicated here only for international security issues,
the approach and the kinds of trade-offs it identifies are more general. States
worry about opportunism, create safeguards, and coerce others in all of their
interactions-including trade and financial relations, international resource
and environmental negotiations, and human rights disputes. Indeed, interna-
tional relations in general can be considered as a network of transactions
embodied in implicit and explicit contracts and possibly modeled in ways
similar to those here. By emphasizing the contractual nature of international
relations, the theory calls to our attention the ways in which states shape their
environments, choose whether or not to invest in greater safeguards, and
decide how much "cheating" they are willing to accept. Where current theories
of international cooperation tend to emphasize exogenously given circum-
stances, the approach developed here highlights the endogenous nature of
institutions and relations. And where the present literature sees institutions as
facilitating cooperation, they are also seen here as instruments of control.54
In recent years, the discipline of international relations has focused too much
on the fact of systemic anarchy and has been insufficiently attentive to
variations in hierarchy among polities. The systemic condition has blinded
scholars to variations in dyadic forms.55 Once central, theories of imperialism
54. See Stein 1990; Martin 1992; and Keohane 1984.

55. This is true even for critics of neorealism who challenge Waltz's view of structure. See
Ruggie 1983 and Wendt 1992.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
have all but disappeared from the field. With the current movement toward
political integration in Europe, and the ongoing dissolution of the Russian
empire, it is time that we reclaimed hierarchy as an interesting and variable
characteristic of international relations.
More generally, focusing on transactions rather than actors may even allow
international relationists to begin probing the rapidly changing patterns of
political authority within the present global system. The breakup of the postwar
communist empire, the formation of the European Union, and the rise of
global markets and firms are momentous events that are defining, for better or
worse, our new world order. Changing forms of relational contracting may well
be creating, in John Ruggie's words, a postmodern, multiperspectival world.56
But the concerns of international relations properly understood endure and are
shaping the actors now emerging on the world stage.
References
Achen, Christopher H. 1988. A state with bureaucratic politics is representable as a unitary

rational actor. Paper presented at the annual meeting of the American Political Science
Association, Washington, D.C., 31 August-3 September.
Baldwin, David A. 1985. Economic statecraft. Princeton, N.J.: Princeton University Press.
Barnett, Michael N., and Jack S. Levy. 1991. Domestic sources of alliances and alignments: The
case of Egypt, 1962-1973. Intemational Organization 45:369-95.
Botting, Douglas. 1985. In the ruins of the reich. Boston: Allen and Unwin.
Brzezinski, Zbigniew K. 1967. The Soviet bloc: Unity and conflict. Rev. ed. Cambridge, Mass.:
Harvard University Press.
Bueno de Mesquita, Bruce, and David Lalman. 1992. War and reason: Domestic and intemational
imperatives. New Haven, Conn.: Yale University Press.
Bunce, Valerie. 1985. The empire strikes back: The transformation of the Eastern bloc from a
Soviet asset to a Soviet liability. Intemational Organization 39:1-46.
Buzan, Barry. 1991. People, states, and fear: An agenda for intemational security studies in the
post-cold war era. 2d ed. Boulder, Colo.: Lynne Rienner.
Cain, P. J., and A. G. Hopkins. 1993. British imperialism. 2 vols. New York: Longman.
Campbell, John C. 1984. Soviet policy in Eastern Europe: An overview. In Soviet policy in Eastem
Europe, edited by Sarah Meiklejohn Terry. New Haven, Conn.: Yale University Press.
Christensen, Thomas J., and Jack Snyder. 1990. Chain gangs and passed bucks: Predicting alliance
patterns in multipolarity. Intemational Organization 44:137-68.
Cowhey, Peter F. 1993. Domestic institutions and the credibility of international commitments:
Japan and the United States. Intemational Organization 47:299-326.
Crane, Keith. 1986. The Soviet economic dilemma of Eastern Europe. Document R-3368-AF,
Rand Corporation, Santa Monica, Calif.
Dallin, David J. 1942. Soviet Russia's foreign policy, 1939-1942. New Haven, Conn.: Yale Universit
Press.
David, Steven R. 1991. Explaining Third World alignment. World Politics 43:233-56.
Davidson, Eugene. 1961. The death and life of Germany: An account of the American occupation.
New York: Knopf.
Dow, Gregory K. 1987. The function of authority in transaction cost economics. Joumal of
Economic Behavior and Organization 8:13-38.
Doyle, Michael W. 1986. Empires. Ithaca, N.Y.: Cornell University Press.
56. Ruggie 1993.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
Anarchy 31
Eggertsson, Thrainn. 1990. Economic behavior and institutions. New York: Cambridge University
Press.
Fearon, James D. 1992. Threats to use force: Costly signals and bargaining in international crises.
Ph.D. diss., University of California, Berkeley.
. 1993. Ethnic war as a commitment problem, University of Chicago. Photocopy.
Friedberg, Aaron L. 1988. The weary titan: Britain and the experience of relative decline, 1895-1905.
Princeton, N.J.: Princeton University Press.
Frieden, Jeffry A. 1994. International investment and colonial control: A new interpretation.
Intemational Organization 48:559-93.
Friedman, Julian R., Christopher Bladen, and Steven Rosen, eds. 1970. Alliance in intemational
politics. Boston: Allyn and Bacon.
Gallagher, John, and Ronald Robinson. 1953. The imperialism of free trade. The Economic History
Review 2d ser., 6:1-15.
Grosser, Alfred. 1955. The colossus again: Westem Germany from defeat to rearmament. New York:
Frederick A. Praeger.
Grossman, Sanford J., and Oliver D. Hart. 1986. The costs and benefits of ownership: A theory of
vertical and lateral integration. Joumal of Political Economy 94:691-719.
Holden, Gerard. 1989. The Warsaw pact: Soviet security and bloc politics. New York: Basil Blackwell.
Holsti, Ole R., P. Terrence Hopmann, and John D. Sullivan. 1973. Unity and disintegration in
intemational alliances: Comparative studies. New York: Wiley.
James, Alan. 1986. Sovereign statehood: The basis of intemational society. Boston: Unwin and
Hyman.
Jervis, Robert. 1983. Security regimes. In Intemational Regimes, edited by Stephen D. Krasner.
Ithaca, N.Y.: Cornell University Press.
Jones, Robert A. 1990. The Soviet concept of "limited sovereignty" from Lenin to Gorbachev: The
Brezhnev doctrine. London: Macmillan.
Kaplan, Lawrence S. 1984. The United States and NATO: The formative years. Lexington: University
Press of Kentucky.
Keohane, Robert 0. 1983. The demand for international regimes. In International Regimes, edited
by Stephen D. Krasner. Ithaca, N.Y.: Cornell University Press.
. 1984. After hegemony: Cooperation and discord in the world political economy. Princeton,
N.J.: Princeton University Press.
King, Gary, Robert 0. Keohane, and Sidney Verba. 1994. Designing social inquiry: Scientific
inference in qualitative research. Princeton, N.J.: Princeton University Press.
Klein, Benjamin, Robert G. Crawford, and Armen A. Alchian. 1978. Vertical integration,
appropriable rents, and the competitive contracting process. Joumal of Law and Economics
21: 297-326.
LaFeber, Walter. 1993. America, Russia, and the cold war, 1945-1992. 7th ed. New York:
McGraw-Hill.
Lake, David A. 1988. Power, protection, and free trade: Intemational sources of U.S. commercial
strategy, 1887-1939. Ithaca, N.Y.: Cornell University Press.
. 1992. Powerful pacifists: Democratic states and war. American Political Science Review
86: 24-37.
. Forthcoming. Entangling relations: American foreign policy in its century.
Lane, Frederic C. 1979. Profits from power: Readings in protection rent and violence-controlling
enterprises. Albany: State University of New York Press.
Leffler, Melvyn P. 1992. A preponderance ofpower: National security, the Truman administration, and
the cold war. Stanford, Calif.: Stanford University Press.
Lipson, Charles. 1994. Understanding international contracts: Building institutions and sustaining
agreements in an anarchic world. Paper presented at the annual meeting of the American
Political Science Association, 1-4 September, New York.
. 1995. Techniques and problems of international contracting. Paper presented at the
annual meeting of the International Studies Association, 21-25 February, Chicago.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
Louis, Wm. Roger, ed. 1976. Imperialism: The Robinson and Gallagher controversy. New York: New
Viewpoints.
Marer, Paul. 1984a. Intrabloc economic relations and prospects. In The Warsaw pact: Alliance in
transition, edited by David Holloway and Jane M. 0. Sharp. Ithaca, N.Y.: Cornell University
Press.
. 1984b. The political economy of Soviet relations with Eastern Europe. In Soviet policy in
Eastem Europe, edited by Sarah Meiklejohn Terry. New Haven, Conn.: Yale University Press.
Marrese, Michael, and Jan Vanous. 1983. Soviet subsidization of trade with Eastern Europe: A
Soviet perspective. Research series no. 52, Institute of International Studies, University of
California, Berkeley.
Martin, Lisa L. 1992. Coercive cooperation: Explaining multilateral economic sanctions. Princeton,
N.J.: Princeton University Press.
Mastny, Vojtech. 1979. Russia's road to the cold war: Diplomacy, warfare, and the politics of
communism, 1941-1945. New York: Columbia University Press.
Mearsheimer, John J. 1994-95. The false promise of international institutions. Intemational
Security 19:5-49.
Morrow, James D. 1991. Alliances and asymmetry: An alternative to the capability aggregation
model of alliances. American Joumal of Political Science 35:904-33.
. 1993. Arms versus allies: Trade-offs in the search for security. Intemational Organization
47: 207-33.
. 1994. Game theory for political scientists. Princeton, N.J.: Princeton University Press.
Most, Benjamin A., and Harvey Starr. 1984. International relations theory, foreign policy
substitutability, and "nice" laws. World Politics 36:383-406.
Niou, Emerson M. S., and Peter C. Ordeshook. 1994. Alliances versus federations: An analysis with
military and economic capabilities distinguished. Durham, N.C.: Duke University, and Pasa-
dena: California Institute of Technology, respectively. Photocopy.
North, Douglass C. 1981. Structure and change in economic history. New York: W. W. Norton.
. 1990. Institutions, institutional change, and economic performance. New York: Cambridge
University Press.
Nye, Joseph S., Jr., and Sean M. Lynn-Jones. 1988. International security studies: A report of a
conference on the state of the field. Intemational Security 12:5-27.
Papayoanou, Paul A. 1992. Economic interdependence and the balance of power: The strategy of
commitment and great power politics. Ph.D. diss., University of California, Los Angeles.
Posen, Barry R. 1984. The sources of military doctrine: France, Britain, and Germany between the
world wars. Ithaca, N.Y.: Cornell University Press.
Powell, Robert. 1990. Nuclear deterrence theory: The search for credibility. New York: Cambridge
University Press.
Poznanski, Kazimiera Z. 1988. Opportunity cost in Soviet trade with Eastern Europe: Discussion of
methodology and new evidence. Soviet Studies 40:290-307.
Richter, James. 1993. Re-examining Soviet policy towards Germany in 1953. Europe-Asia Studies
45:671-91.
Roeder, Philip G. 1993. Red sunset: The failure of Soviet politics. Princeton, N.J.: Princeton
University Press.
. 1995. From hierarchy to hegemony: Patterns of security among the Soviet successor states.
In Regional orders: Building security in a new world, edited by David A. Lake and Patrick M.
Morgan, Institute on Global Conflict and Cooperation, University of California, San Diego.
Typescript.
Rosecrance, Richard. 1986. The rise of the trading state: Commerce and conquest in the modem world.
New York: Basic Books.
Ruggie, John Gerard. 1983. Continuity and transformation in the world polity: Toward a neorealist
synthesis. World Politics 35:261-85.
. 1993. Territoriality and beyond: Problematizing modernity in international relations.
Intemational Organization 47:139-74.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00
Anarchy 33
Russett, Bruce, and Harvey Starr. 1989. World politics: The menu for choice. 3d ed. New York: W. H.
Freeman.
Schroeder, Paul W. 1976. Alliances, 1815-1945: Weapons of power and tools of management. In
Historical dimensions of national security problems, edited by Klaus Knorr. Lawrence: University
Press of Kansas.
Schwartz, Thomas. 1987. Votes, strategies, and institutions: An introduction to the theory of
collective choice. In Congress: Structure and policy, edited by Mathew D. McCubbins and Terry
Sullivan. New York: Cambridge University Press.
Shirk, Susan L. 1993. The political logic of economic reform in China. Berkeley: University of
California Press.
Simmons, Beth A. 1993. Why innovate? Founding the Bank for International Settlements. World
Politics 45:361-405.
Siverson, Randolph M., and Juliann Emmons. 1991. Birds of a feather: Democratic political
systems and alliance choices in the twentieth century. Joumal of Conflict Resolution 35:2
Siverson, Randolph M., and Harvey Starr. 1991. The diffusion of war: A study of opportunity and
willingness. Ann Arbor: University of Michigan Press.
Smith, Tony. 1981. The pattem of imperialism: The United States, Great Britain, and the late-
industrializing world since 1815. New York: Cambridge University Press.
Snyder, Glenn H. 1984. The security dilemma in alliance politics. World Politics 36:461-95.
1990. Alliance theory: A neorealist first cut. Joumal of IntemationalAffairs 44:103-22.
Snyder, Jack. 1991. Myths of empire: Domestic politics and intemational ambition. Ithaca, N.Y.:
Cornell University Press.
Stein, Arthur A. 1990. Why nations cooperate: Circumstance and choice in intemational relations.
Ithaca, N.Y.: Cornell University Press.
Stone, Randall, forthcoming. Satellites and commissars: Strategy and conflict in the politics of
Soviet-bloc trade. Princeton, N.J.: Princeton University Press.
Tilly, Charles. 1990. Coercion, capital, and European states, AD 990-1990. Cambridge, Mass.: Basil
Blackwell.
Triska, Jan F., ed. 1986. Dominant powers and subordinate states: The United States in Latin America
and the Soviet Union in Eastem Europe. Durham, N.C.: Duke University Press.
Ulam, Adam B. 1974. Expansion and coexistence: Soviet foreign policy, 1917-1973. 2d ed. New York:
Praeger.
van Brabant, Jozef M. 1984. The USSR and socialist economic integration: A comment. Soviet
Studies 36:127-38.
Wallander, Celeste A., and Robert 0. Keohane. 1995. An institutional approach to alliance theory.
Working paper 95-2, Center for International Affairs, Harvard University, Cambridge, Mass.
Walt, Stephen M. 1987. The origins of alliances. Ithaca, N.Y.: Cornell University Press.
. 1991. The renaissance of security studies. Intemational Studies Quarterly 35:211-39.
Waltz, Kenneth. 1967. Foreign policy and domestic politics: The American and British experience.
Berkeley: University of California Institute of Governmental Studies Press.
. 1979. Theory of intemationalpolitics. Reading, Mass.: Addison Wesley.
Weingast, Barry R., and William J. Marshall. 1988. The industrial organization of Congress: Or,
why legislatures, like firms, are not organized as markets. Joumal of Political Economy
96: 132-63.
Wendt, Alexander. 1992. Anarchy is what states make of it: The social construction of power
politics. International Organization 46:391-425.
Williams, William Appleman. 1972. The tragedy ofAmerican diplomacy. 2d rev. ed. New York: Dell.
Williamson, Oliver E. 1985. The economic institutions of capitalism: Firms, markets, relational
contracting. New York: Free Press.
Willoughby, W. W., and C. G. Fenwick. [1919] 1974. The inquiry handbooks, vol. 16. Wilmington,
Del. Reprint. Originally published as Types of restricted sovereignty and of colonial autonomy.
Washington, D.C.: U.S. Government Printing Office.
Yarbrough, Beth V., and Robert M. Yarbrough. 1992. Cooperation and governance in intemational
trade: The strategic organizational approach. Princeton, N.J.: Princeton University Press.

142.157.199.173 on Fri, 01 Sep 2023 13:41:10 +00:00

Lake AnarchyHierarchyVariety 1996

Uploaded by

Copyright:

Available Formats

Lake AnarchyHierarchyVariety 1996

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

Lake AnarchyHierarchyVariety 1996

Uploaded by

Copyright:

Available Formats

Anarchy, Hierarchy, and the Variety of International Relations

Author(s): David A. Lake

Stable URL: https://www.jstor.org/stable/2706997

This content downloaded from

Intemational Organization 50, 1, Winter 1996, pp. 1-33

This content downloaded from

Marxism-Leninism is ambiguous on this score, both promising an anarchic

The logic of security

This content downloaded from

This content downloaded from

This content downloaded from

state in pericentric theories is functionally equivalent to abandonment in

The variety of international relations

14. On regimes and transactions costs, see Keohane 1984.

This content downloaded from

This content downloaded from

XWgce O~~cwtet ,geaOe

FIGURE 1. A continuum of security relations

defined by the locus of rights of residual control.20 All relationships, whether

This content downloaded from

As Figure 1 illustrates, the range of relations is continuous in principle.

This content downloaded from

responsibility for their defense. Relations between Russia and severa

24. Roeder 1995.

This content downloaded from

distinctions nonetheless capture important differences in relations. Not all

Contracting for security

To explain the choice of relations by states, I build upon theories of relational

27. Williamson 1985, 34.

This content downloaded from

Assumptions and limitations

This content downloaded from

30. Lake 1992.

This content downloaded from

As discussed above, opportunism is commonly understood to be ubiquitous

31. Williamson 1985,30.

This content downloaded from

changing relations is a more consequential decision with more deleterious

This content downloaded from

States incur governance costs in creating and maintaining relations. These

This content downloaded from

relational hierarchy, governance costs increase with relational hierarchy. As a

Distorted incentives in the subordinate partner. The shift of residual

This content downloaded from

Safeguards on the dominant state. Relational hierarchy shifts the locus of

This content downloaded from

Central to any contract are safeguards employed by both parties to protect

This content downloaded from

behaviors; rather, the agreement focuses on the design of institutions-the

Coercion. Although it fits awkwardly within the neoclassical economic

This content downloaded from

38. On state building, see Tilly 1990.

This content downloaded from

relationship is defined by the intersection in Figure 2 of the isosecurity contour

This content downloaded from

40. Christensen and Snyder 1990, 138 and 141.

This content downloaded from

range of possible variation in relations even if we cannot assign precise values

Anarchy versus hierarchy