Political Institutions, Coercive Diplomacy, and the Duration of Economic Sanctions
Author(s): Fiona McGillivray and Allan C. Stam
Source: The Journal of Conflict Resolution, Vol. 48, No. 2 (Apr., 2004), pp. 154-172
Published by: Sage Publications, Inc.
Stable URL: https://www.jstor.org/stable/3176248
Accessed: 13-03-2020 21:06 UTC
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
Sage Publications, Inc. is collaborating with JSTOR to digitize, preserve and extend access to
The Journal of Conflict Resolution
This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:06:19 UTC
All use subject to https://about.jstor.org/terms
Political Institutions, Coercive Diplomacy,
and the Duration of Economic Sanctions
FIONA McGILLIVRAY
Department of Politics
New York University
ALLAN C. STAM
Department of Government
Dartmouth College
A theory of sanction duration that focuses on differences between democratic and nondemocrat
in the structure of leaders' support coalitions is tested, using a hazard model to analyze a data set of 47
tion events with 272 observations. Results show that leadership change strongly affects the duration o
tions only in the case of nondemocratic states. Leadership change in democratic states is unrelated
duration of sanctions; however, leadership change in nondemocratic sender and nondemocratic targ
is strongly related to the ending of economic sanctions.
Keywords: coercive diplomacy; democratic leadership; nondemocratic leadership; econom
tions; sender state; target state; policy change
The effectiveness of international economic sanctions as a tool of coercive d
macy has been debated at length with little resolution to this point (Drezne
Dashti-Gibson, Davis, and Radcliff 1997; Hufbauer, Schott, and Elliot 1990; Ha
O'Sullivan 2000; Hass 1998; Martin 1992; Kaempfer and Lowenberg 1988
1997; Tsebelis 1990). Although our understanding of the coercive success and
of international sanctions is surprisingly limited, we know even less about how
we might expect economic sanctions to last (Drezner 1999; Dorussen and Mo
A cursory glance at the data reveals that in some instances, sanctions remain i
just months before their withdrawal. In other instances, such as the U.S. san
against Cuba, they can remain in place for decades. A better understanding of th
between domestic political institutions and international sanction durations will
something about coercive diplomacy in particular and, more generally, som
about the distinguishing characteristics of the policy choices that emerge f
different political systems.
One reason sanctions end is that they succeed in bringing about a change in
and/or leader in the target state. Leadership change in the target state can trigge
withdrawal of sanctions if it leads to policy changes. Sanctions that fail to bring
AUTHORS' NOTE: Full regression results are available, along with the data, at http://www.yal
unsy/jcr/jcrdata.htm.
JOURNAL OF CONFLICT RESOLUTION, Vol. 48 No. 2, April 2004 154-172
DOI: 10.1177/0022002703262858
? 2004 Sage Publications
154
This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:06:19 UTC
All use subject to https://about.jstor.org/terms
McGillivray, Stam / DURATION OF ECONOMIC SANCTIONS 155
policy concessions in the target state, however, often persist long after it is clear that
they are ineffective foreign policy tools. In some cases, it takes a leadership change in
the sender state to trigger the withdrawal of sanctions. For example, the turning point
in Soviet-Yugoslav sanctions (1949-1954) came in March 1953 with the death of Stalin,
the soviet leader. Nikita Khrushchev took power and, less than a year later, withdrew
many of Stalin's restrictions on trade with Yugoslavia (Hufbauer, Schott, and
Elliot 1990).
Leadership turnover in the sender state, however, does not always lead to a reversal
of sanctions policy. The United States has continued to sanction Cuba throughout nine
U.S. presidential administrations, from Eisenhower to the current Bush administration.
In the U.S./Cuba case, the withdrawal of sanctions is likely to occur following a
leadership change in the target state. If Cuban president Fidel Castro goes, there can be
little doubt that the U.S. president will withdraw sanctions on Cuba.
The timing of coercive diplomacy in general and the duration of sanctions in particular
seem closely related to leadership change in both the target and sender states.
However, the nature of this relationship is unclear. In this study, we try to untangle the
relationship between leadership turnover and sanctions duration. We study how leadership
change affects the choice made by sender states to terminate or continue sanctions
and the choice made by target states to resist or comply with foreign demands.
We argue that domestic political institutions have an important mediating effect on the
influence of leadership change on sanction duration.
We argue that domestic politics in both the sender and target nation drives the application
and withdrawal of sanctions (Smith 1996; Kaempfer and Lowenberg 1988).
The distribution of the economic costs and benefits of sanctions is uneven across
domestic groups and, as a result, so are the political effects of coercive diplomacy. In
the sender state, we argue that the decision to impose sanctions depends on which
domestic groups win representation. The duration of sanctions depends in part on the
persistence of this representation. Because different leaders typically represent different
interests, a leadership change in the sender state is likely to lead to a change in sanctions
policy. Following the same logic, a change in leader in the target state is likely to
trigger the withdrawal of sanctions. If a new leader represents different interests, a
leadership change in the target state could well lead to a change in the policy that initially
aggravated the sender state, causing the withdrawal of sanctions.
Leadership turnover in either the sender or target state can lead to the end of sanctions,
but it does not necessarily mean that it will. It is only when the new leader draws
support from different political groups that leadership change is expected to affect
sanctions policy. As such, domestic political institutions have an important role in
mediating the effect that leadership change has on sanction duration. This is because
the nature of the domestic political institutions determines which political interests
political leaders represent. Perhaps surprisingly, we argue that leadership turnover
affects the duration of sanctions only in the case of nondemocratic states. Our theory
of sanction duration focuses on the size difference between the groups of supporters
that leaders need to win over to stay in power in democratic and nondemocratic states
(Bueno de Mesquita et al. 2002; Bueno de Mesquita et al. 2003).
This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:06:19 UTC
All use subject to https://about.jstor.org/terms
156 JOURNAL OF CONFLICT RESOLUTION
To put together a sufficiently large coalition of supporters to win an ele
ers in democratic systems often need to draw support from many of the s
groups on which their opposition draws. As a result, these groups receiv
tion even when leaders change. Autocratic leaders typically rely on si
smaller groups of supporters. Because of this, newly installed autocra
beholden to the same pivotal groups as their predecessor. Thus, leadersh
nondemocratic states is more likely to alter the interests that are represen
result, the call for policies that initially incurred the sanctions. We argu
leader in a nondemocratic sender state is more likely to withdraw sanction
leader in a democratic sender state, and a new leader in a nondemocratic ta
more likely to trigger the secession of sanctions than a new leader in a dem
Having controlled for leadership change, political institutions are not exp
an independent effect on the duration of sanctions.
The study is structured in three parts. In the first section, we review th
and argue that the standard focus on the success or failure of economic
achieve foreign policy goals has limited our ability to test theories of co
macy. We argue that an examination of sanction duration offers a useful
extending our understanding of this frequently used coercive instrument.
section, we develop our theoretical predictions. We start by examining
war of attrition logic of sanctions between unitary actor nations and id
tions under which sanctions are likely to be observed. We extend this
unpacking the internal structure of nations' political systems. In particular
political institutions moderate the impact of leadership change on sanct
Domestic institutions affect which groups' preferences receive represen
political change takes place and new leaders enter office, political instit
mine the extent to which the preferences receiving representation chan
hypotheses as to how leadership turnover and domestic political institution
duration of sanctions. In the third section, we test these predictions on a s
sanctions episodes using hazard analysis. We conclude with remark
implications of this work for the study of sanctions.
SANCTION SUCCESS AND SANCTION DURATION
Economic sanctions are an example of coercive diplomacy designed to in
get country to change some policy it would not otherwise (Baldwin 1985)
sanctions achieve this by inflicting economic damage on the target state
reduction of customary trade or withdrawal of financial relations by the se
Although most scholars take for granted the usefulness of economic san
the exception of recent work by Dorussen and Mo (2001) and Bolks and
(2000), little scholarly analysis focuses on the duration of sanctions. M
theorists investigate the success of economic sanctions as an instrument of
icy (e.g., Baldwin 1985; Dashti-Gibson, Davis, and Radcliff 1997; R
Lenway 1988; Hass and O'Sullivan 2000; Hart 2000; Hass 1998; Doxey
ponents argue that economic sanctions can be as effective as military fo
This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:06:19 UTC
All use subject to https://about.jstor.org/terms
McGillivray, Stam / DURATION OF ECONOMIC SANCTIONS 157
safer and cheaper (Baldwin 1985). According to its advocates, when costs are effectively
targeted at ruling elites, sanctions can achieve ambitious foreign policy goals
(Hufbauer, Schott, and Elliot 1990). Whether the sanctions have multilateral support
also affects their effectiveness (Martin 1992; but see Drezner 1999).1
More often than not, however, scholars conclude that sanctions simply do not work.
Many argue that sanctions are rarely, if ever, effective tools of coercion (e.g., Pape
1997; Galtung 1967; Morgan and Schwebach 1997; Wallensteen 1968). Klaus Knorr
finds that sanctions clearly failed in 13 of the 17 cases he studies (Knorr 1977). The
trade and business group USA ENGAGE (2004) argues that unilateral sanctions are
largely prone to failure. Although there is a growing consensus that sanctions rarely
succeed, sanctions continue to be used as a foreign policy tool making it likely that this
debate will continue for some time.
One reason for the lack of consensus between policy makers and academics regarding
the efficacy of economic sanctions as a tool of coercive diplomacy revolves around
measurement issues. As Hufbauer, Schott, and Elliot (1990) state, "Judgment plays an
important role in assigning a single number to each element in the success equation."
In their study of sanction episodes, Hufbauer, Schott, and Elliot attempt to minimize
the subjectivity by including evaluations of other scholars. Pape (1997), however,
reevaluates the same data set and comes to a very different set of conclusions about
sanctions' success rate. Scholars are currently in the process of reevaluating Pape's
reevaluation (Hart 2000). This suggests that some rethinking needs to be done about
the reliability of the study of economic sanction outcomes. Because of these measurement
issues, there is little consensus as to which variables affect the success of sanctions.
Among the likely suspects are the power relationships between the two states, the economic
ties binding the two states, the aims of the sender, the domestic political institutions
in the target states, and the time of interaction between the states (Hufbauer,
Schott, and Elliot 1990; Dashti-Gibson, Davis, and Radcliff 1997; Dorussen and Mo
2001; Hart 2000; Wendt, 1999).
Only a small fraction of the large body of research on sanctions has looked at why
some coercion attempts last for decades, but others last only a few months (Dorussen
and Mo 2001; Bolks and Al-Sowayel 2000).2 However, we think this is a useful puzzle
to study for a number of reasons. Although the theoretical focus of the sanctions literature
is firmly set on explaining the success of sanctions and tends to focus on politics of
the target state, we study the duration of sanctions to highlight the importance of politics
in both the target and the sender states.
Another, more practical, reason to study the duration of sanctions is that the
dependent variable is simple to evaluate because the length of economic sanctions is
more clear-cut than either the costs of sanctions or their efficacy. Measurements of
sanction duration are not biased by the normative concerns that plague the measurement
of sanction success. However, there are tricky specification issues in the empirical
analysis of sanctions duration. Sanctions end when either the sender decides to end
1. Others argue that sanctions are often used purely for rhetorical purposes; the failure rate is misleading
because many sanctions are never implemented (Hufbauer, Schott, and Elliot 1990).
2. The literature on war duration, another more violent example of coercive diplomacy, is, in contrast,
much better developed. See, for example, Goemans (2000) for a review.
This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:06:19 UTC
All use subject to https://about.jstor.org/terms
158 JOURNAL OF CONFLICT RESOLUTION
the sanctioning policy or the target complies with demands. In other words
competing risks that sanctions end. When the target complies (i.e., sanctio
cessful), then our ability to observe when the sender would quit is censored. S
if the sender quits, then our ability to observe when the target would acquies
sored. We discuss different ways of dealing with this problem in the empiric
of the study.
Some scholars argue that the problem with the literature on sanction suc
measurement error but rather the weak theoretical underpinnings of the em
erature on sanctions. Empirical work on the success of sanctions suffers from
selection bias that leads to misleading conclusions about the effectivenes
tions. Smith (1996) points out that the success of sanctions and the decision to
are strongly interdependent. When sanctions are likely to succeed and when
political willingness to impose them, no sanctions will actually be impos
their mere threat is sufficient to force the target to change the policies in disput
other extreme, if sanctions are insufficient to force concessions from the tar
try, then from the goal of achieving policy change, they are an ineffective t
sanctioner to continue to use.3 So, we should expect that the mere threat or sh
sition of sanctions produces policy change or that sanctions completely fail
significant policy concessions from the target. Indeed, supporting Smith's c
Dashti-Gibson, Davis, and Radcliff (1997) find empirical evidence consist
this selection effect. They find a "real downward trend" in the likelihood tha
will succeed as time passes (p. 616). The longer sanctions are in place, the l
they are to succeed (see also Hart 2000).
Scholars who study the success of sanctions as foreign policy tools face the
cal problem of sample selection bias. If the sanctions are going to work, th
need to be applied. So why do we observe the hundreds of economic sanc
have been imposed? Why do some states sanction for decades, long after it
sanctions are insufficient to change policy in the target state? We argue that
do not work and are economically costly, they are probably imposed for so
tic political benefit. Next we systematically consider this question, start
Smith's (1996) war-of-attrition approach to studying sanctions.
THE THEORETICAL ARGUMENT
To understand the duration of sanctions, it is necessary to control for th
sanction. Sanctions are boycotts on trading relationships with the target co
literature on sanctions, it is widely assumed that sanctions are used by gov
achieve international political goals, such as influencing policy in the ta
(Lenway 1988). Within this context, sanctions are a battle of wills. They
until the sender decides to give up trying to bring about policy change or un
3. Smith (1996) shows that, except under certain peculiar distributional assumptions, s
ally occur only when sanctions are insufficient to force the target to alter policy and when th
prefers to send sanctions even though it knows they will fail.
This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:06:19 UTC
All use subject to https://about.jstor.org/terms
McGillivray, Stam /DURATION OF ECONOMIC SANCTIONS 159
acquiesces. An appropriate technique for modeling such an interaction is a "war of
attrition," in which both sides pay costs until one side decides to quit and the opponent
wins the prize.
Smith (1996) uses an incomplete information version ofjust such a war-of-attrition
approach. The sender chooses whether to sanction to get the target to give up a prize
that it initially holds-its preferred policy position. The sanctions continue until either
the target gives up the prize by changing policy (a case of successful sanctions) or the
sender decides sanctioning is no longer worthwhile (a case of unsuccessful sanctions).
By characterizing the Bayesian perfect equilibrium of the game, Smith shows that,
absent some particular distributional assumptions, sanctions only occur when they are
insufficient to force compliance by the target but the sender still wishes to send them,
presumably for some domestic political reason.
The logic behind these results is as follows: first, suppose the sanctions are sufficiently
costly that the target would rather comply with the sender's demands than
endure the sanctions indefinitely. In such a setting, because the sender knows the target
will comply, it always sanctions. Knowing the sanctions will persist, the target immediately
complies. Hence, sanctions are either very short or, more likely, their mere
threat is sufficient to bring about the target's compliance. Second, suppose the target
would rather bare the cost of sanctions indefinitely than comply with the sender's
demands. In this setting, the target will never comply, and sanctions are useless for
bringing about policy concessions. Hence, the sender state will never impose the sanction
unless it gains some domestic political benefit that offsets the economic cost of the
sanctions. Given this logic, Smith (1996) predicts the sanctions that we actually see
represent those cases when sanctions are unlikely to work. These sanctions continue
for a long time because both sides prefer their continuance.
Unfortunately, Smith's (1996) model, although useful, assumes nations are monolithic,
unchanging entities. In reality, nations are not atomistic, and their leaders
choose those policies that satisfy the goals of their supporters. This has important
implications for the duration of sanctions. Leadership change in either the target or the
sender state can bring about a shift in sanctions policy. For instance, suppose leadership
change occurs in the sender state. Although the previous leader gained some
domestic benefit from imposing sanctions, the supporters of the new leader might not
gain such benefits and, as such, the leader is likely to terminate sanctions. When political
change brings about a shift in which interests are represented, such changes often
signal the ending of sanctions.
Leadership change in the target state can have similar implications for the ending of
sanctions. Suppose, for instance, that the supporters of the new leader have interests
more in line with the demands of the sender than those of the predecessor's supporters.
Such a shift in leadership change is likely to end sanctions: the new leader shifts policies
to those that the sender desires. Unfortunately, classifying such an event as a success
or failure is fraught with danger because it often conflates success and leadership
turnover. For instance, the recent removal by military force of Iraqi leader Saddam
Hussein led to the end of U.S. sanctions against Iraq. Although the United States
achieved its policy objectives, does this fact constitute sanctions success? In our
empirical section, we shall pay particular attention to this problem.
This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:06:19 UTC
All use subject to https://about.jstor.org/terms
160 JOURNAL OF CONFLICT RESOLUTION
Leadership turnover in either the sender or target state can lead to the
tions, but it does not necessarily mean that it will. The structure of our
straightforward. Leaders adopt those policies most likely to keep them
which are the policies that their supporters desire. As a result, changes
ship's coalition of supporters are liable to lead to a change in sanctions p
the target or in the sender nation. Our predictions as to when leader ch
shift in sanctions policy reduce to an argument about which groups receiv
tion under different political systems. Fortunately, comparative politics
made great inroads along these theoretical lines upon which we draw. We f
ferences in representation between democratic and nondemocratic politic
LEADERSHIP CHANGE AND
TYPE OF POLITICAL SYSTEM
There are several ways to characterize nondemocratic leaders' polit
(Olson 1993). We characterize both nondemocratic and democratic le
ensuring their political survival. They do so by buying rewards for do
whose support they rely on to stay in power. In democracies, election
down of cabinet government often lead to leadership change. In autocrac
coup or the death of the autocrat is a typical mechanism for leadershi
garchies, leadership change is frequently the result of a power struggle
garchy and a shakeup in the distribution of power among political le
All else equal, political leaders typically last longer in nondemocrati
in democratic regimes (Bienen and Van de Walle 1991). Democratic g
have regularly held elections. Although many politicians are reelected,
in power for decades. Indeed, many democracies have term limits on
Nondemocratic regimes do not have regularly held elections, and lead
time in government is highly variable. In some cases, there is rapid le
over. In other cases, the leadership is in power for decades. However, if
leaders are able to avoid a coup in the early years, the risk of turnove
leader acquires political capital and consolidates his or her power.
If leadership change is more common in democracies than nondem
leadership change triggers the withdrawal of sanctions, sanctions wi
duration when either the sender or the target state is democratic. The sh
of all will be between democratic dyads. There is some empirical evid
tions are shorter when the target state is democratic (Bolks and Al
However, we argue next that the higher rate of leadership turnover in d
is not driving this result. Although leadership change is more common
it is less likely to be associated with the withdrawal of sanctions than in
systems. Indeed, in the empirical analysis, we find that having contr
ship change, political institutions by themselves have no significan
duration of sanctions.
Building on Bueno de Mesquita et al.'s (Bueno de Mesquita et al. 2002; Bueno de
Mesquita et al. 2003) approach, we argue that a key difference between democratic
This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:06:19 UTC
All use subject to https://about.jstor.org/terms
McGillivray, Stam / DURATION OF ECONOMIC SANCTIONS 161
and nondemocratic leaders is the size of the coalition of supporters that leaders need to
retain power. In democratic systems, leaders need the support of many voters to survive
in office. Autocrats rely on the support of a much smaller segment of the population.
The size of the coalition of supporters required under different institutional
arrangements affects the probability that leadership change leads to new interests
being represented (Bueno de Mesquita et al. 2002; Bueno de Mesquita et al. 2003).
Leadership change in democratic systems is less likely to change which interests are
represented compared with leadership change in autocratic systems. Because democratic
leaders need to win the support of a much larger coalition of domestic supporters
than do autocratic leaders, in democracies, there is a higher probability that the new
winning coalition will include many of the previous leader's supporters. Leaders in
nondemocratic states stay in power with the support of a smaller coalition of domestic
supporters. The smaller size of their support base means that any overlap between the
supporters of the old and new leaders is likely to be small, and as a result, the general
policy preferences of new coalitions will differ from previous coalitions to a greater
degree in nondemocratic states.4 In this framework, new leaders are more likely to represent
a different set of interests from their predecessors in nondemocratic systems
than in democratic systems.
A democratically elected leader who resists pressure from foreign sanctions does
so with the support of a large coalition of domestic supporters. For these actors, the
benefits of the leader's policy choices outweigh the costs imposed by foreign sanctions.
If this were not the case, the democratic leader would have immediately acquiesced
to pressure from foreign sanctions. Suppose a change in leader occurs, but the
new leader draws some votes from his or her predecessor's coalition of domestic supporters.
If the new leader wants to stay in office, he or she will continue resisting
pressure from foreign sanctions.
In nondemocratic states, a change in leader is more likely to lead to a change in the
interests the government represents. If the supporters of the new leader have interests
more in line with the demands of the sanctions sender than those of the predecessor's
supporters, the leader will likely acquiesce to sanctions pressure. If the new target
leadership is simply more "friendly" toward the sender and hence likely to be receptive
to significant future policy revisions, then the sender might end sanctions to strengthen
the new target leader's domestic position.
A similar logic can be extended to the cases of democratic and nondemocratic
sender states. In democratic states, new leaders follow policies similar to those of their
predecessors. For a large coalition of voters, the benefits of sanctioning outweigh the
domestic costs. Until this changes or the target state acquiesces, sanctions will continue.
In nondemocratic states, a shift in leader is likely to lead to a change in which
interests are represented. Sanctions are imposed because, for these domestic groups,
the benefits associated with continuing the previous leadership's policy are probably
outweighed by the costs of imposing foreign sanctions. The new leader will, as a
result, revoke his or her predecessor's sanctions. We argue that in democracies, new
4. Another way of thinking about nondemocratic systems is that the leader has a clear hand in policy
making as long as he or she rewards the particularistic interests of those on whose support he or she relies.
This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:06:19 UTC
All use subject to https://about.jstor.org/terms
162 JOURNAL OF CONFLICT RESOLUTION
leaders follow similar policies. They continue to sanction in the case of the sen
and continue to resist pressure from sanctions in the target state. However, in
ocratic states, a change in leader is more likely to lead to leaders giving up on
or giving in to sanctions pressure.5
The extent to which new leaders follow new policies does vary within th
democratic and nondemocratic systems. Some scholars argue that democrac
single-party governments experience more radical policy shifts with a ch
leader than multiparty governments. However, not all scholars agree (Dow
Cox 1990; Tsebelis 1999). This is a controversial topic, and there is an extens
ture comparing policy outcomes in different types of democratic systems
Hibbs 1977; Hofferbert and Budge 1993). Far less has been written about lea
turnover across different types of nondemocracies (but see Bunce 1981; B
Mesquita et al. 2002; Bueno de Mesquita et al. 2003; Alesina and Rosent
Olson 1993; North 1981). As in democratic systems, we might expect differ
the type of nondemocratic government (e.g., autocratic or oligarchic) to af
degree to which policy changes with a change in leader. We argue, however,
effect of leadership change on policy shifts is bigger across than within reg
The United States has a unique style of democracy; nonetheless, the effect o
ship change on sanction duration is similar to that in other democratic syst
U.S. sanctions on Cuba are a nice example of this.
In January 1959, Fidel Castro's revolutionary army defeated the Batista's
bent government and seized power in Cuba. By 1960, Castro exhibited a stro
ity with communist states and nationalized all U.S. economic interests in C
U.S. government imposed sanctions that persist today, although the nature of
tions has changed over time.6 Since Eisenhower initiated the sanctions, th
been eight presidential administrations. Some leaders have been Democrat
Republicans; nevertheless, sanctions have remained some of the broadest
stringent ever recorded. In the United States, a change in leader affects which
of interests receives representation. However, because the incumbent leader
lengers seek to win a majority of the vote, many of the same groups of vo
important to all leaders.
The U.S. Constitution developed a presidential style of democracy that
which domestic groups are important to all presidential candidates. The elec
5. In the democratic case, the causes of leadership change are exogenous to the imposition
tions, but in the nondemocratic case, sanctions pressure is potentially a cause of leadership cha
6. In June 1959, Florida Senator George Smathers proposed legislation limiting Cuba's sug
quota to the United States. By February 1960, this legislation was in effect, reducing sugar im
Cuba by 3 million tons, about one-half of Cuba's annual crop, and resulting in more than U.S.$8
lost income (Schreiber 1973). The Eisenhower regime subsequently deepened economic sanc
U.S. exports were banned by October 1960. The Kennedy administration further tightened th
imposed under Eisenhower, calling on the Organization of American States (OAS) to cooperate i
restrictions on Cuba. Although the 1970s saw the relaxing of some restrictions, with the OAS d
ban in 1974 and a softening of sanctions by both Ford and Carter, significant restrictions remained
1973). Castro has made significant policy concessions-particularly in the past few years-nonet
U.S. policy remains largely unchanged. The intensity of U.S. sanctions policy has changed ove
This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:06:19 UTC
All use subject to https://about.jstor.org/terms
McGillivray, Stam / DURATION OF ECONOMIC SANCTIONS 163
lege indirectly elects the U.S. president.7 Presidential candidates need to win a majority
of electoral college votes, not the popular vote. The more heavily populated states
have a higher number of electoral college votes, and in most of these states, the candidate
who carries these states wins all of the electoral votes.8 Although Florida is only 1
of 50 states in the United States, it has about 10% of the votes needed to win. In the
highly competitive state of Florida, winning the votes of the highly organized Cuban
American community has traditionally been key in determining the outcome of the
election (the plurality winner takes all 27 electoral college votes). The intensity of antiCastro
feeling among the large Floridian Cuban American community has meant that
harsh sanctions against Cuba have remained a part of U.S. foreign policy for more than
40 years, regardless of the party or political ideology of the sitting president or major
party candidates (Purcell 1998).
The Indonesia-Malaysia sanctions episode provides a sharp contrast to the U.S.Cuban
case. In the early 1960s, Indonesia's nondemocratic regime was ruled by an oligopoly
headed by President Ahmed Sukarno. In 1963, President Sukaro imposed a
full trade embargo on the newly federated state of Malaysia in protest against the
state's formation (Hufbauer, Schott, and Elliot 1990; Sodhy 1988). The sanctions did
not last long; a reorganization of power within the Indonesian leadership 3 years later
led to the withdrawal of sanctions. In March 1966, General Suharto, commander of the
army, forced Sukarno to transfer much of his executive power to him. Suharto drew
support from the military-a different support base than his predecessor, Sukaro. As
Indonesia's new leader, Suharto reestablished trade connections with Malaysia that
very year.9 In the United States, leaders come and go, but sanctions policy is stable and
resilient. Only in nondemocratic systems, such as Indonesia, should we expect leadership
turnover in the sender state to be key to the termination of sanctions.
We are not the only scholars to explore how domestic political institutions affect
sanction policy. Dorussen and Mo (2001), among the few scholars to explore the duration
question, also examine domestic institutions. They do so, however, from the perspective
that these institutions shape audience costs and the ability of sender nations to
signal their intent (Lenway 1988; Fearon 1997; Schultz 1998; Smith 1996). They
argue that democratic institutions make it costly for leaders to back down. Once a
leader imposes sanctions, it is difficult for her or him to back down without concessions
from the target. Although they may be right that it is hard for democrats to back
down, they fail to explain why successor governments are encumbered with the same
costs to backing down, given they did not impose the sanctions (McGillivray and
Smith 2000; Guisinger and Smith 2002).
7. In the United States, the president typically initiates legislative action on sanctions. Congress,
however, can veto presidential initiatives. Increasingly, the legislature initiates sanctions policy.
8. In theory, in a two-party race, a candidate could carry the electoral college vote with just 25% of
the popular vote by winning a bare majority in the states used to build the electoral college coalition and winning
none of the popular vote in the states outside the electoral college majority. A less extreme example of
this took place in the 2000 election when Bush carried the electoral college vote, whereas Gore won the popular
vote.
9. Barter trade resumed in June 1966, and Indonesia and Singapore officially recognized one another
as states. In August 1966, Malaysia and Indonesia signed a normalization agreement, following which they
reestablished trade and commercial links in March 1967.
This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:06:19 UTC
All use subject to https://about.jstor.org/terms
164 JOURNAL OF CONFLICT RESOLUTION
Bolks and Al-Sowayel (2000) argue that sanctions are of shorter duration if th
get is a democratic state because democratic leaders are more vulnerable than
cratic leaders to pressure from the domestic groups suffering from the e
effects of sanctions. They find empirical evidence to this effect. However, ju
more representative does not mean that democratic states are more likely to acq
Whether democratically elected leaders acquiesce to sanctions pressure depe
whether key domestic supporters believe that the costs of sanctions outweigh th
fits of retaining existing policy. Only if the costs outweigh the benefits to d
supporters will democratic leaders give in, and then they are likely to do so
However, if the benefits of keeping the existing policy are greater than the
sanctions to the majority of voters, leaders will resist pressure from foreign sa
Moreover, we argue that leadership change will be unrelated to the duration
tions, and the eventual withdrawal of sanctions will be brought about for an ex
reason-perhaps leadership change in the sender state. Indeed, we find that al
the duration of sanctions is shorter with democratic targets than autocratic
leadership change is not the mechanism that drives this.
In summary, we hypothesize the following:
Hypothesis 1: Leadership change in the sender nation increases the probability tha
tions end.
Hypothesis 2: Leadership change in the target nation increases the probability tha
tions end, although such events can often be conflated with sanctions success.
Hypothesis 3: The impact of leadership change on the duration of sanctions is gre
nondemocratic rather than democratic states.
Next, we test these hypotheses using a Weibull hazard model fit to a data set of 56
sanction episodes with 363 observations.
EMPIRICAL TESTS
RESEARCH DESIGN: CASE SELECTION, VARIABLES, AND DATA
The data on sanctions come from Hufbauer, Schott, and Elliot (1990); we generated
the control variables using EUGene (Bennett and Stam 2000); and the leadership data
come from Bueno de Mesquita et al. (2003).'0 We discuss the sanctions data below in
more detail. The dependent variable is the length of sanctions between sender nation A
and target nation B. Hazard analysis estimates the probability that sanctions end at any
point in time conditional on sanctions not having already ended." We use both the
parametric Weibull model and the semiparametric technique of Cox's proportionate
hazard model. The results in Table 1 present the results of the Weibull models. The dif-
10. Many thanks to Randy Siverson (University of California, Davis) for sending us these data.
11. Hazard models are also referred to as survival models, duration models, or event history models. A
basic introduction to such analysis is in Allison (1984). Comprehensive sources for duration analysis include
Greene (1993) and Lancaster (1990). Recent applications of hazard analysis to political science include
Warwick (1994), Bueno de Mesquita and Siverson (1994), Bennett and Stam (1996), Bennett (1998), and
Goemans (2000).
This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:06:19 UTC
All use subject to https://about.jstor.org/terms
TABLE 1
Weibull Estimates of International Sanctions Duration: Parameter Estimates Pr
Model 1: Model 2: Mode
Censored Termination of All Sanction Censore
Successful Sanctions ("Sender's and Target's Successful
("Sender's Decision to Quit") Aggregate Decision to Quit") (
Leader change A
Leader change A* polity A
Polity A
Leader change B
Leader change B* polity B
Polity B
Ratio of power capabilities
Similarity of interests
Dyadic age
Distance
United States
Success
Duration dependence parameter, p
Observations
Cases, n
Log likelihood
38.9** (2.04)
0.014** (2.18)
2.05 (0.57)
6.31*** (2.36)
0.215 (1.21)
0.786 (0.28)
0.210 (1.25)
1.01 (0.27)
1.50* (1.24)
2.59** (2.33)
1.43*** (6.04)
262
47
-34.5
20.5** (1.78)
0.035** (1.86)
1.88 (0.53)
4.94*** (2.39)
0.220* (1.44)
2.02 (1.11)
1.06 (0.06)
0.962 (0.77)
1.49* (1.55)
1.65** (1.74)
1.14*** (7.25)
272
47
-50.13
12.2*
0.074
1.11
4.67*
0.156
4.19*
6.02
1.01
2.11*
3.79*
0.124*
1.52*
1.36*
262
47
-11
*p <. 10. **p < .05. ***p < .01. All one-tailed tests. We present our results as hazard ratios rather than t
creased risk of failure and hence shorter durations. Z scores appear in parentheses next to the paramete
http://www.yale.edu/unsy/jcr/jcrdata.htm/.
This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:06:19 UTC
All use subject to https://about.jstor.org/terms
166 JOURNAL OF CONFLICT RESOLUTION
ference between these results and those obtained using the Cox estimator
markable and omitted to save space. The Weibull setup models the hazard rate
the probability that sanctions terminate conditional on not already having
h(t) = p t p-l, where X represents the regressors, k = exp(xj p) and p is an
parameter that determines whether the hazard rate increases or decreases o
Using such hazard approaches, we can assess whether factors such as le
change or regime type increase or decrease the probability that sanctions end
nately, some care is needed. Sanctions end when either the sender decides
sanctioning policy or the target complies with demands. There are compet
that sanctions end.12 Suppose we focus on when the sender quits. The recent e
sanctions against Iraq provides a useful illustration. In 2003, the United Stat
sanctioning Iraq. However, it did not do so because it chose to "quit"; rather, t
vation for the sanctions-the removal of Saddam Hussein's regime-was no
there. Although U.S.-Iraq sanctions ended in 2003, it is clear the U.S. willin
maintain sanctions persists should they still be required. In assessing when
quits, the U.S.-Iraq sanctions case shows that the United States was prepare
tion until at least 2003. We do not know whether the United States would hav
tioned in 2004 because this became a moot point following the Second Gu
Cases such as these are best thought of as censored.
In general, when the target complies (i.e., when sanctions are successful)
ability to observe when the sender would quit is censored. Similarly, if
quits, then our ability to observe when the target would acquiesce is censor
tunately, as the contentious debate over sanctions success indicates, it is extre
ficult in many cases to ascertain which side quits. If we could make such a dis
then a standard competing hazards approach could be used (Gordon 2002; D
and Stevenson 2000).13 Here we adopt a compromise between lumping toget
effects into a single hazard equation and a competing hazards approach.
Schott, and Elliot (1990) provide a 16-point scale of sanctions success. Alt
already indicated, this scale is controversial, we take success ratings greater t
be instances of clear success; hence, from the perspective of the sender's
quit, such events are censored. In addition, we estimate the hazard mode
assuming such censoring-in effect, lumping both sides' decisions to quit in
equation.
We corrected and dropped observations of sanctions following Pape's (1997) arguments.
Hufbauer, Schott, and Elliot (1990) identified 115 cases of sanctions. Pape
argues that 40 of these should not be considered cases of sanctions (e.g., trade wars), or
the sanctions ended for reasons unrelated to political factors at either the initiator or
target state (e.g., the target state ceased to exist). Because we are focusing on the duration
of economic sanctions as a function of political factors in the sending state, we
dropped sanctions that ended due to military defeat and coups or by military force in
the target state. There is also a growing body of evidence to suggest that international
12. See Gordon (2002) for a discussion of competing hazard models in political science.
13. Dependencies between competing hazards can be modeled using frailty models. Unfortunately,
given the small sample size, such techniques are impractical here (Gordon 2002).
This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:06:19 UTC
All use subject to https://about.jstor.org/terms
McGillivray, Stam / DURATION OF ECONOMIC SANCTIONS 167
relations changed fundamentally after World War II (Gowa 1999). As a result, we also
dropped cases including and before World War II. Our data therefore cover the years
from 1948 to 1990. We begin with 60 cases; we drop 13 more due to missing data. We
analyze 47 cases of international economic sanctions with 363 yearly observations.
Each of the remaining incidents of sanctions has multiple observations, with a range of
2 to 31.'4 An observation reflects one of two occurrences: the ending of the sanctions
or the end of a calendar year. Following Hufbauer, Schott, and Elliot (1990), as well as
Pape (1997), we code sanctions as bilateral, not scoring joiners but only the initiator
and target states.
The variable "leader change A" is a dummy variable that codes whether any leadership
change took place in a particular year in the sender nation, A. There is a corresponding
variable for leadership changes in the target state, B. Using the number of
instances of leadership change, rather than simply presence or absence of leadership
change, yields much the same results.
Domestic political institutions are coded on the Polity III scale. Specifically, polity
A is the Polity III democracy score minus the polity autocracy score for state A. Both
these scores range between 0 and 10. We normalize the difference between these indices
by adding 10 and dividing by 20. In our hazard models, polity A ranges between 0
and 1. Democratic states score high on this measure. Polity B is the corresponding
measure for the target state.
In addition to the political variables, we include the following control variables:
1. Ratio ofpower capabilities: power A/(power A + power B), where power A represents
the Correlates of War (COW) composite power index for nation A.
2. Similarity of interests: We measure the similarity of the two states' interests with Bueno
de Mesquita's (1981) global tau-b measure. This is a-1 to 1 correlation of the two states'
alliance portfolios.15
3. Dyadic age: We measure the amount of time in years that the two states have existed simultaneously
in the international system. If one state drops from the system and then reappears,
the age clock reverts to zero and starts anew. We use the logarithm of dyad age
plus 1 year.
4. Distance: This variable measures the logarithm of the geographic distance between the
two states' capitals in kilometers.
5. United States: A dummy variable coding whether the United States is the sender state.
6. Success: This variable measures the potency of the sanctions on the target state in terms
of proportional cost relative to gross national product (GNP). It is a 1 to 16 scale.16
RESULTS
The estimates of the hazard models are shown in Table 1. Two different app
are presented: estimates of the sender's decision to quit and the aggregate effect
14. The mean number of observations is 6.
15. We had missing data for roughly 5% of the cases on this variable. For these cases, we substituted
the mean value for the rest of the series.
16. For details on the success scale, a product of two 1 to 4 scales, see Hufbauer, Schott, and Elliot
(1990).
This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:06:19 UTC
All use subject to https://about.jstor.org/terms
168 JOURNAL OF CONFLICT RESOLUTION
the sender's and target's decisions to quit. As discussed above, sanction
because either the sender quits or the target complies, although in man
impossible or controversial to distinguish between these outcomes. In a
estimate the sender's decision to quit, it is not appropriate to consider even
target clearly complied as instances of quitting. In such a scenario, we d
when the sender would have quit, only that she or he did not quit prior to
compliance. Hence, when sanctions were definitely successful (> 12 on t
scale), the sender's decision to quit is censored. We refer to this case as
decision to quit." We also present analyses in which we do not (or mor
cannot) distinguish who quit. We refer to this case as the "sender's an
aggregate decision to quit."
Before moving to the prominent questions of political and leadership
briefly describe the impact of the control variables. The relative strengt
parties, as measured by the power ratio variable, has no significant impact
tion of sanctions. The similarity of interests measured by Bueno de Me
measure also has no significant impact on the length of sanctions. The a
that the longer a dyad has interacted, the shorter that sanctions are
Although this variable is statistically significant in some models, it is n
Furthermore, the substantive impact of this variable is small. Across all
the greater the distance between sender and target, the more likely sanctio
The ancillary parameter, p, assesses the extent to which the risk of sanc
increases or decreases over time. The coefficient greater than 1, reported i
models in Table 1, indicates that the longer that sanctions last, the more li
to end.
Models 3 and 4 include controls for whether the United States is the sen
level of eventual success of the sanction. Both variables have a powerfu
dicting the length of sanctions. All else equal, the United States is only
likely to end sanctions compared to other nations. As the experience
against Cuba shows, the United States is prepared to impose long-term
The success variable strongly influences the ending of sanctions. How
not surprising. The success of sanctions is an ex post evaluation and, in m
equivalent to the end of sanctions. In models 1 and 3, we censor those
events when sanctions ended with clear success. Although the inclusion
variable improves the model fit, it does not substantively alter the impact
cal and leadership variables. It is these variables that are of most interes
POLITICAL INSTITUTIONS
AND LEADERSHIP CHANGE
The variable "leader change A" is coded 1 if leadership turnover occu
sender state in a particular yearly observation and 0 otherwise. The statistic
cant hazard ratio of 38 reported in model 1 means that for an autocratic send
likelihood of quitting sanctions is 38 times greater if leader change occurs
This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:06:19 UTC
All use subject to https://about.jstor.org/terms
McGillivray, Stam /DURATION OF ECONOMIC SANCTIONS 169
change appears to have an enormous impact on the duration of sanctions. However,
this interpretation applies only to autocratic states. The variable "leader change A*
polity A" is the product of the leadership change dummy and the institutional arrangements
in nation A. When the sender state is fully autocratic, and hence polity A = 0,
then this variable has no impact on the duration of sanctions. For such a state, only the
"leader change A" variable is relevant in assessing the impact of leadership change.
However, when the sender state is democratic, then the impact of leadership change
depends on both terms. The coefficient on the interaction term in model 1 is 0.014. For
a fully democratic state (polity A = 1), the impact of leadership change on the decision
to quit sanctioning is negligible. This pattern is repeated throughout the other models
in Table 1, although the magnitude of the effect differs somewhat. Consistent with our
expectations, leadership turnover reduces the duration of sanctions in autocratic, but
not democratic, senders.
Leadership change in target states also drastically affects the probability of sanctions
ending. The variable "leader change B" is coded 1 if leadership change occurs in
the target state and 0 otherwise. Across the four models presented in Table 1, the proportionate
hazard on this variable is around 4, meaning that leadership change in the
target state makes sanctions four times more likely to end when the target is autocratic.
This result should not surprise us. In many cases, the implicit goal of sanctions is
regime change.
When the target is democratic, the impact of leadership change is negligible. The
variable "leader change B* polity B" is the product of any leadership change and the
target's regime type. When the target is democratic (polity B = 1), then the net effect of
leadership change is the product of the proportionate hazards (equivalently, the sum of
the coefficients). The net effect of leadership change in democracies is 0. As our theoretical
arguments predict, to put together a sufficiently large coalition of supporters,
leaders in democratic systems often need to draw support from many of the same
domestic groups. Therefore, the interests of these groups receive representation even
when leaders change. Autocratic leaders rely on smaller groups of supporters. As
such, they are less beholden to the same pivotal groups as their predecessor, and so
leadership change alters the interests that are represented and hence the calls for
policies that incur sanctions.
Having controlled for leadership change, political institutions by themselves have
no significant effect on the duration of sanctions. The variables polity A and polity B
code the regime type of sender. Neither of these variables alters the length of sanctions
directly. As argued earlier, there are competing influences determining how institutions
affect sanctions. Although the impact of leadership change is greater in autocratic
systems, leadership turnover is more common in democracies. The results suggest
that neither effect dominates the other, so taken together, regime type does not
alter the length of sanctions.
The results also show that leadership turnover often ends sanctions in autocratic
nations, with the effect being much less in democracies. Leadership change is relevant
in both senders and targets. It is worth returning to the distinction between models 1
and 3 and models 2 and 4 to consider the relevance of these results. In models 1 and 3,
This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:06:19 UTC
All use subject to https://about.jstor.org/terms
170 JOURNAL OF CONFLICT RESOLUTION
we focus on instances when the sender quits sanctioning by censoring thos
end in success (e.g., when the target complied). Despite that the focus of th
is on the sender, leadership change in the target has a significant impact o
decision to quit sanctions. Several factors contribute to this result. First, a
not explicit, reason for sanctions is regime change. Hence, although t
might not be a coercive success, once a major reason for the sanctions ha
the sender quits. Second, if the new target leadership is friendlier towa
and hence likely to be receptive to significant future policy revisions, the
might end sanctions to strengthen the new target leader's domestic pos
the sender's decision, leadership change in the target has significant im
decision to maintain sanctions.
Models 2 and 4 make no attempt to distinguish how sanctions end. As such, they
represent the aggregate effect of both the sender's and the target's decision to quit. The
analyses produce very similar results to those in models 1 and 3. Leadership change in
autocratic states often leads to the termination of sanctions because new autocratic
leaders often shift their support base relative to their predecessors, which leads to different
interests receiving representation. In democratic systems, the requirement to
form a large coalition of supporters often forces both incumbents and challengers to
seek the support of the same domestic groups. Because satisfying the goals of these
key domestic groups is essential for political success, leadership change has less
impact on sanctions policy. Whether from the perspective of the sender's decision to
quit or the aggregate of both the sender's and target's decisions, the impact of leadership
change in both the sender and target on the length of sanctions depends on the
institutional context in which leaders serve. In democracies, leadership change has little
impact of the cessation of sanctions. In contrast, leadership change in autocratic
states is likely to bring about the end of sanctions.
CONCLUSIONS
In this study, we developed a model linking leadership change and do
cal institutions to the duration of economic sanctions. Domestic politic
play an important role in mediating the effect that leadership change has
because institutions determine which political interests political leaders
focus on differences between democratic and nondemocratic states in the st
leaders' support coalitions. We argue that leadership change strongly aff
tion of sanctions only in the case of nondemocratic systems. A change
cratic leadership is more likely to trigger the withdrawal of sanctions th
democratic leadership-in either the sender or the target state. We fit a h
a data set of 47 sanction events with 272 observations. Our findings str
our predictions. Leadership change in democratic sender states and dem
states is unrelated to the duration in sanctions. However, leadership ch
nondemocratic target and sender states is a strong predictor of sanctio
This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:06:19 UTC
All use subject to https://about.jstor.org/terms
McGillivray, Stam /DURATION OF ECONOMIC SANCTIONS 171
REFERENCES
Alesina, A., and H. Rosenthal. 1995. Partisan politics, divided government, and the econom
UK: Cambridge University Press.
Allison, P. 1984. Event history analysis. Beverly Hills, CA: Sage.
Baldwin, D. A. 1985. Economic statecraft. Princeton, NJ: Princeton University Press.
Bennett, D. S. 1999. Parametric models, duration dependence, and time-varying data revisi
Journal of Political Science 43:256-70.
Bennett, D. S., and A. C. Stam. 1996. The duration of interstate Wars. American Political Sc
90:239-57.
.2000. EUGene: A conceptual manual. International Interactions 26:179-204.
Bienen, H., and N. Van de Walle. 1991. Of time andpower: Leadership duration in the moder world. Stanford,
CA: Stanford University Press.
Bolks, S. M., and D. Al-Sowayel. 2000. How long do economic sanctions last? Examining the sanctioning
process through duration. Political Research Quarterly 53:241-65.
Bueno de Mesquita, B. 1981. The war trap. New Haven, CT: Yale University Press.
Bueno de Mesquita, B., J. D. Morrow, R. Siverson, and A. Smith. 2002. Political institutions, policy choice
and the survival of leaders. British Journal of Political Science 32 (4): 559-90.
Bueno de Mesquita, B., and R. Siverson. 1995. War and the survival of political leaders: A comparative study
of regime types and political accountability. American Political Science Review 89:841-55.
Bueno de Mesquita, B., A. Smith, R. M. Siverson, and J. D. Morrow. 2003. The logic of political survival.
Cambridge, MA: MIT Press.
Bunce, V. 1981. Do new leaders make a difference ? Executive succession and public policy undercapitalism
and socialism. Princeton, NJ: Princeton University Press.
Cox, G. 1990. Centripetal and centrifugal incentives in electoral systems. American Journal of Political Science
34:903-35.
Dashti-Gibson J., P. Davis, and B. Radcliff. 1997. On the determinants of success of economic sanctions: An
empirical analysis. American Journal of Political Science 41:608-18.
Diermeier, D., and R. Stevenson. 2000. Cabinet terminations and critical events. American Political Science
Review 94:627-40.
Dorussen, H., and J. Mo. 2001. Ending economic sanctions: Audience costs and rent-seeking as commitment
strategies. Journal of Conflict Resolution 45:395-427.
Downs, A. 1957. An economic theory of democracy. New York: Harper & Row.
Doxey, M. 1987. A framework for analysis with special reference to the UN and Southern Africa. International
Organization 26:527-50.
Drezner, D. 1999. The sanctions paradox. Cambridge, MA: Cambridge University Press.
.2000. Bargaining, enforcement and multilateral economic sanctions. International Organization
54:73-102.
Fearon, James D. 1997. Signaling foreign policy interests: tying hands versus sinking costs. Journal of Co
flict Resolution 41:68-90.
Galtung, J. 1967. On the effects of economic sanctions. World Politics 19:378-416.
Goemans, H. E. 2000. Fighting for survival: The fate of leaders and the duration of war. Journal of Confl
Resolution 44:555-80.
Gordon, S. C. 2002. Stochastic dependence in competing risks. American Journal of Political Science
46:200-17.
Gowa, Joanne. 1999. Ballots and bullets: The elusive democratic peace. Princeton, NJ: Princeton Univer
Press.
Greene, William H. 1993. Econometric analysis. 3rd ed. New York: Macmillan.
Guisinger, A., and A. Smith. 2002. Honest threats: The interaction of reputation and political institutio
international crises. Journal of Conflict Resolution 46:172-200.
Hart, R. A., Jr. 2000. Democracy and the successful use of economic sanctions. Political Research Quar
53:267-84.
This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:06:19 UTC
All use subject to https://about.jstor.org/terms
172 JOURNAL OF CONFLICT RESOLUTION
Hass, R. N., ed. 1998. Economic sanctions and American diplomacy. Washington, DC: Broo
sity Press.
Hass, R. N., and M. L. O'Sullivan, eds. 2000. Honey and vinegar Washington, DC: Brook
Press.
Hibbs, D. A., Jr. 1977. Political parties and macroeconomic policy. American Political
71:1467-87.
Hofferbert, R., and I. Budge. 1993. Controversies: Party platforms and government spending. Americ
Political Science Review 87:744-59.
Hufbauer, G. C., J. J. Schott, and K. E. Elliot. 1990. Economic sanctions reconsidered. Washington, DC:
Institution for International Economics.
Kaempfer, W., and A. D. Lowenberg. 1988. The theory of international economic sanctions: A public choice
approach. American Economic Review 4:786-93.
Knorr, K. 1977. International economic leverage and its uses. In Economic issues and national security,
edited by Klaus Knorr and Frank N. Trager, 99-126. Lawrence, KS: Regents.
Lancaster, T. 1990. The econometric analysis of transition data. Cambridge, UK: Cambridge University
Press.
Lenway, S. A. 1988. Between war and commerce: Economic sanctions as a tool of statecraft. International
Organization 42:397-426.
Martin, L. 1992. Coercive cooperation: Explaining multilateral economic sanctions. Princeton, NJ: Princeton
University Press.
McGillivray, F., and A. Smith. 2000. Trust and cooperation through agent-specific punishments. International
Organization 54:809-24.
Morgan, C., and V. Schwebach. 1997. Fools suffer gladly: The use of economic sanctions in international crises.
International Interactions 21:27-50.
North, D. C. 1981. Structure and change in economic history. New York: Norton.
Olson, M., Jr. 1993. Dictatorship, democracy and development. American Political Science Review 87:567-
76.
Pape, R. 1997. Why economic sanctions do not work. International Security 2:90-136.
Purcell, S. 1998. Cuba. In Economic sanctions andAmerican diplomacy, edited by R. N. Hass. Washington,
DC: Brookings University Press.
Rogers, E. S. 1996. Using economic sanctions to control regional conflicts. Security Studies 5:43-72.
Schreiber, A. P. 1973. Economic coercion as an instrument of foreign policy: U.S. measures against Cuba
and the Dominican Republic. World Politics 25:387-413.
Schultz, K. A. 1998. Domestic opposition and signaling in international crises. American Political Science
Review 92:829-44.
Smith, A. 1996. The success and use of economic sanctions. International Interactions 21:229-45.
Sodhy, P. 1988. Malaysian-American relations during Indonesia's confrontation against Malaysia, 1963-6
Journal of Southeast Asian Studies 19:111-36.
Tsebelis, G. 1990. Are sanctions effective? A game theoretic analysis. Journal of ConflictResolution 34:2-28.
1999. Veto players and law production in parliamentary democracies: An empirical analysis. Amer
ican Political Science Review 93:591-608.
USA ENGAGE. 2004. Statement of purpose. http://www.usaengage.org/archives/background/purp
html/.
Wallensteen, P. 1968. Characteristics of economic sanctions. Journal of Peace Research 3:248-67.
Warwick, P. 1994. Government survival in parliamentary democracies. New York: Cambridge University
Press.
Wendt, A. 1999. A social theory of international politics. Cambridge, UK: Cambridge University Press.
This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:06:19 UTC
All use subject to https://about.jstor.org/terms