Rand Rr4293

C O R P O R AT I O N
BRYAN FREDERICK, JENNIFER KAVANAGH, STEPHANIE PEZARD,

ALEXANDRA STARK, NATHAN CHANDLER, JAMES HOOBLER,
JOOEUN KIM
Assessing Trade-Offs in
U.S. Military Intervention
Decisions
Whether, When, and with What Size Force
to Intervene
For more information on this publication, visit www.rand.org/t/RR4293.
About RAND
The RAND Corporation is a research organization that develops solutions to public policy
challenges to help make communities throughout the world safer and more secure, healthier
and more prosperous. RAND is nonprofit, nonpartisan, and committed to the public interest. To
learn more about RAND, visit www.rand.org.
Research Integrity
Our mission to help improve policy and decisionmaking through research and analysis is
enabled through our core values of quality and objectivity and our unwavering commitment
to the highest level of integrity and ethical behavior. To help ensure our research and analysis
are rigorous, objective, and nonpartisan, we subject our research publications to a robust and
exacting quality-assurance process; avoid both the appearance and reality of financial and
other conflicts of interest through staff training, project screening, and a policy of mandatory
disclosure; and pursue transparency in our research engagements through our commitment to
the open publication of our research findings and recommendations, disclosure of the source
of funding of published research, and policies to ensure intellectual independence. For more
information, visit www.rand.org/about/principles.
RAND’s publications do not necessarily reflect the opinions of its research clients and sponsors.
Published by the RAND Corporation, Santa Monica, Calif.
© 2021 RAND Corporation
is a registered trademark.
Library of Congress Cataloging-in-Publication Data is available for this publication.
ISBN: 978-1-9774-0506-7
Cover photo by Sean Taylor, U.S. Army.
Limited Print and Electronic Distribution Rights
This document and trademark(s) contained herein are protected by law. This representation
of RAND intellectual property is provided for noncommercial use only. Unauthorized posting
of this publication online is prohibited. Permission is given to duplicate this document for
personal use only, as long as it is unaltered and complete. Permission is required from RAND
to reproduce, or reuse in another form, any of its research documents for commercial use. For
information on reprint and linking permissions, please visit www.rand.org/pubs/permissions.
Preface
This report documents research and analysis conducted as part of a

project entitled Costs of Not Intervening, sponsored by the Office of the
Deputy Chief of Staff, G-3/5/7, U.S. Army. The purpose of this proj-
ect was to develop a framework to assess the trade-offs between delay-
ing or forgoing interventions and intervening in the early stages of an
armed conflict in order to inform U.S. Army and U.S. Department of
Defense (DoD) planning and decisions regarding whether and when to
initiate U.S. military interventions.
This research was conducted within RAND Arroyo Center’s
Strategy, Doctrine, and Resources Program. RAND Arroyo Center,
part of the RAND Corporation, is a federally funded research and
development center (FFRDC) sponsored by the United States Army.
RAND operates under a “Federal-Wide Assurance”
(FWA00003425) and complies with the Code of Federal Regulations for
the Protection of Human Subjects Under United States Law (45 CFR 46),
also known as “the Common Rule,” as well as with the implementa-
tion guidance set forth in DoD Instruction 3216.02. As applicable, this
compliance includes reviews and approvals by RAND’s Institutional
Review Board (the Human Subjects Protection Committee) and by the
U.S. Army. The views of sources utilized in this study are solely their
own and do not represent the official policy or position of DoD or the
U.S. Government.
iii
Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxvii
Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxix
CHAPTER ONE
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Theoretical Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Organization of This Report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
CHAPTER TWO
What Do We Know? Reviewing Existing Literature . . . . . . . . . . . . . . . . . . . . . . 13
The Intervention Decision: When Do Interventions Improve
Outcomes? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
The Timing Decision: Is It Better to Intervene Early or Late? . . . . . . . . . . . . . 24
The Size Decision: Should the Intervention Be Small or Large? . . . . . . . . . . . 30
Synthesis and Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
CHAPTER THREE
Quantitative Analysis: Studying Trends and Patterns . . . . . . . . . . . . . . . . . . . . 39
Data Set of Intervention and Nonintervention Cases . . . . . . . . . . . . . . . . . . . . . . . . 39
Exploring Outcomes by Intervention Timing and Size . . . . . . . . . . . . . . . . . . . . . 48
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
v
vi Assessing Trade-Offs in U.S. Military Intervention Decisions
CHAPTER FOUR
Lessons from Focused Case Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Assessment and Insights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
Implications and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
CHAPTER FIVE
Counterfactuals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
Summary of the Counterfactual Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
CHAPTER SIX
Conclusion and Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
The Effect of Interventions on Outcomes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
APPENDIX
A. List of Cases of Potential U.S. Intervention . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
ONLINE APPENDIX
Available for download at www.rand.org/t/RR4293
B. Counterfactual Case Studies
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
Figures
S.1. The Intervention Decision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii

1.1. The Intervention Decision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
vii
Tables
1.1. Intervention Outcomes of Interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.1. Summary of Prior Research on Military Interventions and
Outcome Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.2. Summary of Prior Research on Intervention Timing and
Outcome Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.3. Summary of Prior Research on Intervention Size and
Outcome Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.1. Intervention Categories and Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.2. War and Crisis Interventions, by Timing and Size . . . . . . . . . . . . . 47
3.3. Conflict Duration by Intervention Type for War Cases . . . . . . . . 53
3.4. Conflict Duration by Intervention Type for Crisis Cases . . . . . 54
3.5. Change in Average Annual Battle Deaths, by Intervention
Category . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.6. Conflict Recurrence, by Intervention Category. . . . . . . . . . . . . . . . . . 57
3.7. Percentage of Objectives Successfully Achieved, by
Intervention Category . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.8. Change in Economic Assistance to Host Nation, by
3.9. Change in Military Assistance to Host Nation, by
3.10. Changes in Polity Score, by Intervention Category . . . . . . . . . . . . 64
3.11. Changes in GDP per Capita, by Intervention Category . . . . . . . 66
3.12. Changes in Infant Mortality Rate, by Intervention
Category . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.13. Summary of Quantitative Investigations into Relationship
Between U.S. Interventions and Outcomes . . . . . . . . . . . . . . . . . . . . . . 69
4.1. List of Cases Selected . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
ix
x Assessing Trade-Offs in U.S. Military Intervention Decisions
5.1. List of Selected Counterfactual Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

5.2. Summary of Effects on Outcomes Illustrated by
Counterfactual Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
5.3. Summary of Effects of Intervention for Counterfactual
Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
5.4. Summary of Effects of Timing in Bosnia Counterfactual
Case. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
5.5. Summary of Implications for Intervention Decisions
Illustrated by Counterfactual Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
6.1. Summary of Findings for Outcome Measures . . . . . . . . . . . . . . . . . 138
A.1. List of Cases of Potential U.S. Intervention . . . . . . . . . . . . . . . . . . . . 154
Summary
The research reported here was completed in November 2019, followed by

security review by the sponsor and the Office of the Chief of Public Affairs,
with final sign-off in November 2021.
When faced with the outbreak of a war or the onset of an interna-

tional crisis that affects important U.S. interests, U.S. policymakers
must consider how best to respond. They have a variety of options and
tools available to them: doing nothing, engaging diplomatically, pro-
viding military or economic assistance to one or more of the involved
parties, or committing U.S. forces either in a unilateral intervention or
in a multilateral coalition. Even after an initial choice is made, policy-
makers still have several decisions to make. For instance, if the decision
is made to use military assistance, how much aid should be provided,
and should that aid consist of weapons, supplies, or money? In the case
of military interventions, policymakers must decide the size and com-
position of the forces to be employed. Moreover, even if an initial deci-
sion is made to not become involved militarily, that decision can always
be revisited, and a later intervention can be undertaken.
As policymakers consider these options, the costs and benefits
of not intervening militarily might be difficult to assess in compari-
son with the costs and benefits of undertaking a military intervention.
There are cases in which it is possible that a quick U.S. intervention
to reestablish stability and force a settlement might have prevented a
longer, more costly conflict that might then prompt a later but more
challenging U.S. intervention. Failure to intervene might have strate-
gic costs as well, such as an increase in regional or global instability or
the loss of access to key strategic territory, bases, or ports. However, a
decision to forgo intervention could have notable benefits. Depending
on the situation, these might include not only economic savings or sav-
xi
xii Assessing Trade-Offs in U.S. Military Intervention Decisions
ings in terms of lives lost, but also the enhancement of U.S. reputation
and diplomatic influence, the strengthening of other regional actors
that are then empowered to handle future crises, or even the potential
to prompt U.S. adversary intervention into what becomes, for them,
a quagmire. An improved understanding of the potential costs and
benefits of intervention and nonintervention, and the factors on which
they depend, could improve the information available to policymakers
and could lead to substantial benefits for the United States, given the
often highly consequential nature of military intervention decisions.
Objective
To this end, the objective of this report is to create a framework that

can be used to rigorously consider the trade-offs between interven-
ing militarily immediately following the outbreak of a war or crisis,
intervening later, and not intervening at all, as well as the trade-offs
involved with decisions regarding the size of the potential intervention
force to be employed. This framework can be used to provide a better
understanding of the relationships between intervention timing, inter-
vention size, and intervention outcomes to inform future debates about
whether and when to undertake a military intervention.
Although policymakers might decide to undertake a military
intervention at any point during a war or crisis, we conceptualize the
decision to intervene as occurring in two distinct stages. First, policy-
makers decide whether to intervene in the early stages of a crisis or war.
This “early” decision period is intended to reflect the first U.S. consid-
eration of a potential intervention in a crisis or war, notwithstanding
any limited potential delays imposed because of the complexity of the
U.S. policy process or the gathering of information on the events in
question. If a decision is made to intervene in this initial period, poli-
cymakers would also need to decide on the size and scope of the U.S.
intervention. Indeed, these decisions are likely to be made together
because the set of feasible intervention options will likely affect the
advisability of the intervention itself. Policymakers might have a wide
variety of options for the forces they commit to an intervention, but we
Summary xiii
stylize these choices down to three: the commitment of small, medium,

or large forces. As these sizing decisions are considered, policymakers
might view the problem in many different ways, such as deciding to
“throw everything” at a particularly acute problem, conducting a very
limited intervention designed only to monitor the situation or signal
interest, or opting for something in between.
If policymakers decide not to intervene in this first decision
period, they could (if the war or crisis continues) face a second decision
period regarding whether to intervene later in the crisis or war and, if
so, what size force to commit at that time. Taken together, this model
suggests seven possible choices for a state considering whether to inter-
vene. Figure S.1 illustrates a decision tree that can be used to concep-
tualize the decision points and available options.
In this report, we are interested in understanding how the deci-
sions about whether to intervene, when to intervene, and with how
many forces to intervene are associated with intervention and con-
flict outcomes. In other words, we aim to understand how outcomes
differ as a result of the policymaker decision to choose one of the seven
options shown in Figure S.1. We identify three categories of outcomes
as being of particular interest: conflict outcomes, outcomes for the
Figure S.1
The Intervention Decision
Decision1
2 3 4
Early, large Early, medium Early, small
1 Decision2
No intervention
5 6 7
Late, large Late, medium Late, small

xiv Assessing Trade-Offs in U.S. Military Intervention Decisions
intervening state (the United States), and outcomes for the host nation.
Throughout the report, we explore the relationships between interven-
tion decisions and these outcome measures, with the goal of identifying
patterns and insights that can be used by planners and decisionmakers.
Research Approach
We assessed that it would be unlikely that a single approach would

adequately allow us to assess the trade-offs inherent in the decisions
regarding whether, when, and with what size forces to intervene, given
the complexity of these decisions. Moreover, different approaches
might have different strengths and weaknesses for addressing different
aspects of these issues. We therefore investigated these questions using
four distinct methods, then combined the results to develop a more
comprehensive assessment of the trade-offs involved in the set of deci-
sions described in the previous section.
First, we conducted an extensive review of academic and policy
literature on relevant topics, including (1) the relationship between
third-party interventions and conflict duration and intensity, (2) the
costs and benefits of an intervention for the intervening state, and
(3) the effects of third-party interventions on the host nation. We used
the literature review to determine what previous research has already
suggested about the relationship between interventions, intervention
timing, and intervention size and conflict outcomes, outcomes for the
intervener, and outcomes for the host. This literature helps to bound
our expectations for what we might find using other methods as well
as highlights aspects of our research question that have (or have not)
previously been carefully investigated.
Second, we conducted a quantitative analysis using a database of
286 crises and wars since 1945 that we identified as involving U.S.
interests to a degree that made a U.S. military intervention plausible.
This database includes instances in which the United States intervened
and those in which it did not. We categorized these cases according
to three criteria: (1) whether there was an intervention, (2) whether an
intervention (if it occurred) was early or late, and (3) whether an inter-
Summary xv
vention (if it occurred) was large, medium, or small in size. We then

compared these categories of cases along several dimensions that reflect
the three main categories of outcomes on which we focus: effects on
the conflict, effects on U.S. interests, and effects on the host nation.
Third, we conducted a set of 45 short, focused case studies. These
case studies include interventions and noninterventions, each address-
ing a consistent set of questions. Topics covered include an assessment
of why an intervention did or did not occur, why it occurred when it
did (early or late), and what the consequences of the intervention or
nonintervention choice were for the United States and the host nation
(or potential host, in the case of noninterventions). In particular, these
case studies helped us to isolate the effects of intervention size and
timing decisions on U.S. interests from selection effects that might
drive more-general correlations or apparent relationships.
Finally, we conducted four in-depth counterfactual illustrations
focused on the 1992–1995 Bosnian War, the 1950–1953 Korean War,
the 2011 Libyan Civil War, and the 1979 Iranian Revolution. In these
counterfactuals, we first describe what happened historically and why,
then we explore what might have happened if the United States had
made a different intervention decision, such as choosing not to inter-
vene in the Korean War. We highlight the role of key actors and their
interests and capabilities and explore how they likely would have acted
in the alternative scenario. We explore the costs and benefits for the
United States of this alternate decision and identify which factors
would have been central to determining the counterfactual outcomes.
We use the insights from these cases to illustrate broader observations
about the trade-offs between different intervention decisions and out-
come measures.
Results
The findings from these four research approaches provided insights

into how intervention decisions affect all three categories of outcome
measures that we assessed: effects on the conflict, effects on U.S. inter-
ests, and effects on the host nation. Next, we summarize our results of
xvi Assessing Trade-Offs in U.S. Military Intervention Decisions
greatest interest to policymakers (the effects of intervention decisions

on U.S. interests) through the lens of the three key intervention deci-
sions our research was designed to inform. A full discussion of our
results related to other outcome measures is presented in Chapter Six.
The Intervention Decision: Should the United States Intervene?

Although military interventions can be very effective at advancing U.S.
interests in some contexts and situations, they could have the opposite
effect in other contexts, creating long-term entanglements and increas-
ing economic and strategic costs without realizing U.S. objectives. Our
analysis suggested that the United States achieves its political objec-
tives in military interventions about 60 percent of the time. More-
over, U.S. intervention is not a requirement for protecting U.S. inter-
ests in all cases. Many nonintervention cases resolve in ways that do
not have significant implications for U.S. interests; historically, more
nonintervention cases resulted in improved outcomes for U.S. interests
than resulted in deteriorated outcomes. Rather than presenting a bias
toward intervening to solve problems, then, this study highlights for
decisionmakers the importance of carefully considering local condi-
tions in order to assess whether conducting a military intervention is
likely to advance U.S. goals.
The factors that distinguish between intervention opportuni-
ties that advance U.S. interests and those that have the opposite effect
appear to be contextual. In particular, our focused case studies, coun-
terfactual case studies, and review of the literature all highlight the
centrality of understanding how an outside intervention might shift
the local balance of power for determining how the intervention could
affect the course of the conflict and the leverage available to the United
States to achieve its interests. If the U.S. intervention would serve to
shift the power of the favored side in a conflict from either a stalemated
or a disadvantaged position to a clearly advantaged position, then we
would expect the intervention to help end the conflict and increase
the likelihood that the United States is able to advance its other policy
goals.
Historical examples of this dynamic can be seen in the 1991 Gulf
War or the 1995 Bosnian intervention. Other dynamics, however,
Summary xvii
could force policymakers to make explicit trade-offs across different

outcomes. In the 1950 intervention into the Korean War, for exam-
ple, U.S. intervention helped avoid the defeat of the Republic of Korea
(ROK), but it also served to substantially prolong the conflict, likely
increasing the near-term destruction of the Korean peninsula, partic-
ularly once Chinese intervention pushed the conflict into a lengthy
stalemate. Although the end result helped to advance U.S. strategic
interests and placed the ROK on a positive developmental trajectory,
these achievements came at the cost of involvement in a lengthy con-
flict with a substantial human toll. In other cases, illustrated by our
counterfactual study of the 1979 Iranian Revolution and by our analy-
sis of the 2007 war in Yemen, any plausibly sized U.S. intervention
likely would have been unable to dramatically alter the local balance
of power in the desired direction, and intervention was therefore likely
best avoided. Finally, a U.S. intervention could be largely superfluous
if it occurs in a context in which the supported side already has the
upper hand. This was partly the case in the 1965 Dominican Republic
intervention, where a combination of local and regional actors likely
could have achieved the same, generally positive outcome without U.S.
intervention, or at least with a much smaller U.S. force. In the 1948
Costa Rican civil war, the U.S. decision not to undertake a military
intervention at all helped to facilitate a regional response that settled
the immediate conflict and put the country on a sustainable, positive
trajectory for decades.
Our analysis also highlights that, despite the importance of deter-
mining the local balance of power when anticipating intervention out-
comes, such a determination can be difficult in practice, particularly if
the motivations and capabilities of third-party spoilers are not properly
understood. The Chinese intervention to reverse U.S. advances in the
Korean War, for example, was not anticipated, although it proved deci-
sive for shifting the outcome of the war. The role Pakistan has played
in the post-2001 conflict in Afghanistan provides another example.
The Timing Decision: Is It Better to Intervene Early?

Our analysis suggests that early interventions are associated with both
the highest rewards and increased risks. In our quantitative analysis,
xviii Assessing Trade-Offs in U.S. Military Intervention Decisions
early interventions were somewhat more likely to achieve U.S. political

objectives than later ones were. The short, focused case studies suggest
more-equal rates of success between early and late interventions in terms
of achieving U.S. objectives, but we also find many more instances
where the early timing of an intervention was vitally important to those
successes, and we find only one potential case in which the lateness of
the intervention might have been beneficial. For example, quick U.S.
responses to the 1991 Kurdish humanitarian crisis in northern Iraq
and the rapid mobilizations and deployments undertaken in response
to the 1962 Cuban Missile Crisis were successful in good part because
of the fact that they occurred early in the crises. Although some later
interventions, such as the 2014 intervention into Iraq to combat the
Islamic State of Iraq and the Levant (ISIL), were also largely successful,
our research suggests this degree of success was, for the most part, not
because they began late in these conflicts.1 One potential exception to
this trend was the U.S. intervention in Bosnia. Although our counter-
factual analysis suggests that an earlier U.S. intervention could have
led to a similar outcome (while decreasing the humanitarian toll of the
conflict), it would only have done so if the United States had been will-
ing to pay a substantially higher price, including increased risk to its
ground forces. The later intervention that occurred historically in 1995
required fewer trade-offs for the United States because the local bal-
ance of power was already moving away from the Bosnian Serbs before
robust U.S. intervention began. Waiting to undertake an intervention
later in a conflict could therefore provide a local balance of power that
is more conducive to advancing U.S. interests, thereby limiting U.S.
costs or increasing the feasibility of intervention—though there is, of
course, no guarantee that such changes will occur.
More broadly, however, it is important to emphasize that the deci-
sion for policymakers about timing is really an iterative one that is
closely intertwined with the decision regarding whether to intervene at
1 To clarify, we assessed that a U.S. intervention prior to the fall of Mosul could have been
more effective (and at lower cost) than the historical intervention that took place afterward.
Once Mosul had fallen, however, U.S. patience in rebuilding Iraqi security forces before
mounting a sustained counteroffensive appears to have paid dividends.
Summary xix
all. Early interventions, from this perspective, simply reflect instances

when policymakers decided to intervene at their first opportunity to do
so, while later interventions occur when policymakers decide to inter-
vene at subsequent points in the conflict. But policymakers do not
typically make a decision at one point in time to intervene at some
later point in time, so there is no actual decision to “intervene late.”
The decision that policymakers face is therefore similar regardless of
when in the conflict it occurs. The analysis in the prior section regard-
ing whether the United States should intervene at all therefore remains
relevant regardless of when in the course of the conflict an intervention
is being contemplated.
The Size Decision: Is It Better to Intervene with Larger Forces?

There is a tendency to assume that “more is always better” when decid-
ing how many forces to commit to an intervention. Certainly, robust
interventions appear both warranted and highly likely in response to
direct threats posed to U.S. national security, such as the invasion
of Afghanistan following the September 11, 2001, attacks and the
response to the 1962 Cuban Missile Crisis.
We also have evidence suggesting that larger interventions could
be associated with better outcomes more generally. In our focused cases,
for example, we found that large and medium interventions more often
achieved U.S. political objectives. Further, we observed that there are
more instances in which the size of a larger intervention is central to its
success than is the case for smaller interventions. Large interventions,
such as the 1991 Gulf War, highlight the value that larger forces can
provide, particularly in cases of traditional combat and when facing
large, powerful adversaries.
That said, we also find reasons for caution in the employment
of larger forces. To begin with, a blanket application of large forces
in interventions could often prove to be unnecessary, adding addi-
tional costs without realizing additional benefits, as was the case in the
large U.S. intervention in the Dominican Republic in 1965. But the
potential concerns with large forces extend beyond a lack of efficiency.
Our analysis indicates at least three circumstances in which forces of
a more-limited size might be preferable. First, a large U.S. force might
xx Assessing Trade-Offs in U.S. Military Intervention Decisions
produce a substantial backlash among either the local population or

important regional actors, undermining the intervention’s ability to
achieve its objectives and potentially strengthening adversary forces.
Second, a large U.S. force might seem highly threatening to an adver-
sary and prompt unwanted escalation in the conflict or crisis. Third,
a large force, particularly a large ground force, might take substantial
time to mobilize and deploy, such that policymakers might prefer to
send a smaller force that can be ready more quickly.
Even beyond these conditions that might suggest a preference for
a force of a more-limited size, the introduction of large forces effec-
tively increases the “bet” the United States is placing on the outcome
of the intervention. As noted above, large forces can help to increase
success, but in other instances, such as the post-2001 intervention in
Afghanistan, even greatly increased forces might prove unable to fun-
damentally alter the dynamics of the conflict, increasing the costs of
the intervention without substantially increasing its benefits. In our
focused case studies, it is notable that large interventions encompass
some of the most notable successes and the most notable failures for
U.S. interventions. Small interventions, by contrast, tend to have less
effect overall on U.S. interests and are, for the most part, associated
with more-limited gains; they are also, in other circumstances, associ-
ated with more-circumscribed losses.
What determines when larger forces are a good bet to make in
an intervention? Although this determination is partly affected by the
U.S. interests involved, how large forces would be expected to affect
the local balance of power and the degree of leverage the United States
can exercise in the host nation appear to be central. When large forces
allow the United States to tip the balance of power clearly in favor
of the side it supports, it can generally expect these larger forces to
help lead to better outcomes. The United States has tremendous mili-
tary capabilities, so this dynamic is often relevant. There are, however,
exceptions in which the introduction of large forces could prompt local
backlash, unwanted escalation, inadvisable delays, or third-party inter-
vention to counterbalance U.S. forces—such as the Chinese interven-
tion during the Korean War—or could simply be insufficient to deci-
sively alter the local balance of power. The decision to intervene with
Summary xxi
large forces therefore represents a higher-risk, higher-reward strategy,

the advisability of which should be carefully assessed according to the
local context and the criteria noted above.
Recommendations
Using the key insights of our analysis, we make three recommenda-

tions to inform U.S. and Army planning and decisionmaking regard-
ing future interventions.
Select Interventions Judiciously: Context Is Likely to Be Decisive

Our analysis clearly shows that military interventions can advance U.S.
national interests in certain circumstances. But we also find suggestive
evidence that, since 1945, the United States could be over-intervening
in aggregate. Instances of U.S. interventions that failed to achieve their
objectives and where U.S. interests might have been better served by
not undertaking the intervention in the first place, such as the post-
2003 intervention in Iraq, are often well-known. Equally interesting,
however, is that we found few cases in which the United States did not
intervene but likely would have been better off if it had, such as the
1978 Nicaraguan Civil War. More numerous were cases in which the
United States did not intervene and saw its interests advanced partly
because of its restraint, such as the 2006 Iranian nuclear crisis and the
1948 civil wars in Costa Rica and Colombia. In general, the costs of not
intervening may have been overemphasized in analyses of U.S. inter-
vention policy in comparison with the potential benefits of restraint.
However, this general observation is of more-limited value for
policymakers charged with identifying which specific interventions
are most essential to undertake and which can be safely avoided. Our
analysis identifies four key criteria that can be assessed when deciding
whether a military intervention is likely to advance U.S. interests or
prove counterproductive or wasteful:
• The local balance of power: In crises and wars where the balance of
power already tilts toward the side favored by the United States,
xxii Assessing Trade-Offs in U.S. Military Intervention Decisions
intervention might be unnecessary. However, where the United

States’ interests are under threat, intervention could make more
sense, provided the U.S. intervention would be capable of turning
the tide.
• The role of spoilers: If external actors might respond to a U.S.
intervention by increasing their involvement in a crisis or war
sufficiently to counteract the introduction of U.S. forces, then
intervention could prove ineffective. U.S. adversaries might also
respond to the intervention by taking actions in other theaters
that undermine U.S. interests in a manner that a narrow analysis
of the host nation might not account for. During the Cold War,
for example, the United States had to carefully consider how the
Soviet Union would react to its overseas activities in order to pre-
vent unintentional escalation.
• The potential risks of escalation: If a U.S. intervention has the poten-
tial to directly threaten, or to be perceived as directly threatening,
the security of a near-peer or other highly capable U.S. adversary,
then the intervention has the potential to draw the United States
into a wider or more-dangerous confrontation. Although such cir-
cumstances were limited in the cases we assessed, their potential
consequences highlight the importance of considering this factor
when making intervention decisions.
• The degree of legitimacy: Interventions taken without international
legitimacy can affect U.S. relations with allies and the U.S. repu-
tation more broadly. Legitimacy can also be vital for avoiding a
backlash in the host nation; such a backlash could undermine
the ability of the intervention to achieve its goals. This concern is
most acute for interventions with large numbers of U.S. forces. In
other cases (most notably, those involving humanitarian concerns
for which a U.S. or multilateral intervention might be increas-
ingly expected), it could be that a failure to intervene to prevent
civilian suffering could sap the legitimacy of the United States as
a global leader.
Summary xxiii
Earlier, Larger Interventions Are Higher Risk, Higher Reward

Our analyses suggest that earlier and larger interventions are overall
more likely to result in positive outcomes for the United States. More-
over, our case studies illustrate clearly how both the timing and size
of these interventions can be essential factors in their success in this
regard. However, earlier and larger interventions are also associated
with some of the most negative outcomes that we observe and, even
when successful, involve substantial economic costs for the United
States. These interventions therefore constitute “betting big.” They
have a clear potential to achieve U.S. objectives, and in some instances,
they might be the only way to do so. But there are several circum-
stances, most notably including the presence of highly capable third-
party spoilers, where such interventions might be apt to fail nonethe-
less. When such interventions fail to achieve their objectives or when
they generate additional, unforeseen complications, the strategic, repu-
tational, and economic costs they can impose on the United States are
enormous. Earlier, larger interventions should be undertaken only after
a careful analysis of their likelihood of success, informed by an assess-
ment of the factors noted above and an assessment of potential alterna-
tives to military intervention.
Prioritize Collection and Analysis of Key Intelligence to Inform

Intervention Decisions
The above two recommendations emphasize the importance of under-
standing the local context in order to inform intervention decisions.
These factors, as much as anything the United States does in the actual
conduct of the intervention, can determine whether an intervention
advances or harms U.S. interests. This is an area in which focused
intelligence and data gathering and subsequent careful analysis can
have a significant impact. Intelligence collection to inform interven-
tion decisions should focus on the following questions:
• Warring parties and the balance of power: Who are the warring
parties? What is the balance of power between relevant parties?
Would the balance be changed in a way favorable to U.S. interests
with an intervention?
xxiv Assessing Trade-Offs in U.S. Military Intervention Decisions
• Motivations of the parties: Is the war or crisis one that seems ame-
nable to settlement? How long is each side willing to fight? Are
the issues at stake likely to remain intractable, or could they be
resolved through negotiation?
• External spoilers and influencers: Which external actors could
become involved in ways that counter U.S. interests? What are
their goals and capabilities? Are they likely to intervene with mili-
tary forces or to exert other types of influence?
• Local populations: Will the local population of the host nation be
supportive of a U.S. intervention? If not, what type of backlash
is likely, and how severe will it be? What are the relationships
between the other parties to the conflict and the local population?
• Adversary responses: How will adversaries respond to a U.S. inter-
vention? Are they likely to feel threatened by it? Are there impor-
tant escalation risks to consider? Do adversaries have the ability
and motivation to undermine U.S. interests elsewhere in response
to the intervention?
• International community: What will be the response of the inter-
national community to the intervention? Will the United States
face backlash from allies for intervening or not intervening?
Ensuring high-quality and up-to-date information on these ques-

tions can support better decisions about whether and when to intervene
and with what size force. Of course, the U.S. military, and the U.S.
government more broadly, might already be collecting much of this
information for some locations. We make this recommendation partly
to emphasize the importance of filling any gaps that might exist with
regard to countries where an intervention is being contemplated but
also to emphasize the information that decisionmakers will need in
order to apply the framework developed in this report to the interven-
tion decision questions we explore.
The theme running throughout this report is that decisions about
whether, when, and with what size force to intervene can have impor-
tant implications for U.S. interests. These decisions can be optimized
to maximize benefits to the United States but only if contextual factors
are carefully weighed and analyzed. Simplistic advice for these com-
Summary xxv
plex intervention decisions is not possible because there is no single

approach that will work every time. Instead, each intervention decision
must be made on its own terms. Our aim with this report is to reduce
the complexity of this decision for senior leaders by identifying the key
factors that should be weighed when making this decision, describ-
ing the mechanisms that link these factors to intervention outcomes,
providing detailed historical examples, and assessing systematically the
likely implications of specific intervention decisions executed in par-
ticular contexts.
Acknowledgments
Many people helped with the research and completion of this report.
The authors thank MG Christopher McPadden for his sponsorship
and support throughout the project. We also thank MG (ret) William
Hix for asking the question that led to this project and for support-
ing the effort. Tony Vanderbeek and Mark Calvo provided valuable
guidance and feedback throughout the research and writing of this
report. Stephen Biddle of Columbia University, Karin von Hippel of
the Royal United Services Institute, and Karl Mueller of RAND pro-
vided insightful reviews that led to substantial improvements in this
document. The authors are also grateful to Bruce Bennett, Ambas-
sador (ret) James Dobbins, Edward Geist, Benjamin Fishman, Jeffrey
Martini, Laurel Miller, and Ariane Tabatabai for their insights and
expertise on specific cases included in this report. Lynn Davis provided
essential guidance and insights for our research throughout. Finally,
Sally Sleeper and Stephen Watts offered valuable comments on earlier
drafts. All mistakes are the authors’ own.
xxvii
Abbreviations
DPRK Democratic People’s Republic of Korea

GDP gross domestic product
ICB International Crisis Behavior
ISIL Islamic State of Iraq and the Levant
JCPOA Joint Comprehensive Plan of Action
NATO North Atlantic Treaty Organization
PLO Palestinian Liberation Movement
ROK Republic of Korea
RUMID RAND Unclassified Military Intervention
Dataset
UCDP Uppsala Conflict Data Project
UN United Nations
UNPROFOR United Nations Provisional Force
xxix
CHAPTER ONE
Introduction
When faced with the outbreak of a war or the onset of an interna-

tional crisis that affects important U.S. interests, U.S. policymakers
must consider how best to respond. They have a variety of options and
tools available to them: doing nothing, engaging diplomatically, pro-
viding military or economic assistance to one or more of the involved
parties, or committing U.S. forces either in a unilateral intervention or
in a multilateral coalition. Even after an initial choice is made, policy-
makers still have several decisions to make. For instance, if the decision
is made to use military assistance, how much aid should be provided,
and should that aid consist of weapons, supplies, or money? In the case
of military interventions, policymakers must decide the size and com-
position of the forces to be employed. Moreover, even if an initial deci-
sion is made to not become involved militarily, that decision can always
be revisited, and a later intervention can be undertaken.
The decision about which course of action to take will depend
on a variety of factors, including the U.S. interests at stake and assess-
ments of the likely course of the crisis or war absent U.S. involvement.
In the case of military interventions—the focus of this report—the
decision will also be influenced by several other considerations, such
as the available resources (forces and money), public support, and an
assessment of the likelihood that a military intervention will allow the
United States to achieve its objectives and protect its national interests.
Further, ongoing shifts in the international distribution of power and
influence, notably including the rise of China, are likely to affect U.S.
1
2 Assessing Trade-Offs in U.S. Military Intervention Decisions
calculations in this regard to an increasing degree in the future, in ways

that could differ from more-recent U.S. experiences.
However, although the use of military force is often considered
in response to a war or crisis, military interventions are less common.
Only a fraction of all international conflicts and crises provoke a third-
party military intervention, and only a small fraction of those are inter-
ventions undertaken by the United States. In the majority of cases,
then, policymakers choose not to intervene militarily, opting instead to
limit their engagement to diplomatic pressure or providing some other
sort of assistance or support (or opting to do nothing).
There are many cases where a decision not to intervene or to
“wait and see” (effectively delaying an intervention decision) reflects an
assessment that an intervention would be too costly, that it would tie
up U.S. forces in a long-lasting engagement, that it would be politically
unpopular, or that an intervention would be unlikely to advance U.S.
national interests. For instance, the U.S. decision not to initially inter-
vene militarily in Rwanda in response to the 1994 genocide was, accord-
ing to notes from the National Security Council at the time, (1) the
result of concerns over the costs of such an intervention, (2) uncer-
tainty about whether an intervention force could arrive in time to sub-
stantially alter the course of the conflict and save lives, (3) the lack of
appetite (among the general public and policymakers) for another U.S.
intervention in sub-Saharan Africa in the wake of the perceived failure
of the intervention in Somalia, and (4) the fact that Rwanda was seen
as peripheral to U.S. interests.1 The costs were therefore perceived as
high, and the benefits were seen as uncertain or limited. In the recent
Syrian civil war, involving a country arguably more directly relevant
to U.S. strategic interests, the United States showed great hesitancy to
intervene, waiting almost four years after the outbreak of the war in
1 Marc Grossman, Executive Secretary, U.S. Department of State, “Discussion Paper for
Deputies Committee on Peacekeeping Options in Rwanda,” memorandum to William H.
Itoh, Executive Secretary, National Security Council, Washington, D.C., May 16, 1994;
Susan E. Rice, National Security Council, “Updated Talking Points on Rwanda,” memoran-
dum to W. Anthony K. Lake et al., Washington, D.C., May 7, 1994; Susan E. Rice and Nick
Rasmussen, “Deputies Committee Meeting on African Peacekeeping Issues,” memorandum
to Samuel R. Berger, Washington, D.C., September 20, 1993.
Introduction 3
2011 before initiating a military intervention in the country and, even

then, limiting the intervention in size and scope. In this instance, the
decision to delay reflected a fear of being dragged into a long-running
intervention, concerns about a potential confrontation with Russia
and Iran (which were also active in the conflict), fatigue following the
long conflicts in Iraq and Afghanistan, and a sense that a military
intervention was unlikely to meaningfully advance U.S. objectives and
interests.2 The anticipated costs and uncertain benefits of undertaking
interventions can often deter or delay their initiation.
For policymakers, what can be more difficult to assess in these
discussions about whether and when to intervene, however, is the extent
to which the lack of U.S. intervention could also have important costs
or benefits. There are cases (Syria potentially included) in which it is
possible that a quick U.S. intervention to reestablish stability and force
a settlement might have prevented a longer, more costly conflict that
might, in turn, prompt a later but more challenging U.S. intervention.
Similarly, there could be reputational and strategic costs from a deci-
sion not to intervene. The limited U.S. response to the 1994 genocide
in Rwanda presented a challenge to U.S. reputation and credibility as
an advocate of human rights.3 Failure to intervene might have strategic
costs as well, such as an increase in regional or global instability or the
loss of access to key strategic territory, bases, or ports. One of the key
reasons that the United States has been so active in the Middle East in
recent decades, for instance, has been to ensure that key sea passages
remain open and that oil markets remain stable.4 However, a decision
to forgo intervention could have notable benefits as well. Depending
on the situation, these might include not only economic savings or
savings in terms of lives lost, but also the enhancement of U.S. repu-
2 Dominic Tierney, “Syria and the Cycle of American Intervention,” The Atlantic, Octo-
ber 24, 2016.
3 Samantha Power, “Bystanders to Genocide,” The Atlantic, September 2001.
4 Richard N. Haass and Martin Indyk, “Beyond Iraq: A New U.S. Strategy for the
Middle East,” Foreign Affairs, Vol. 88, No. 1, 2009; Stacie L. Pettyjohn and Jennifer Kava-
nagh, Access Granted: Political Challenges to the U.S. Overseas Military Presence, 1945–2014,
Santa Monica, Calif.: RAND Corporation, RR-1339-AF, 2016.
tation and diplomatic influence; the strengthening of other regional

actors, which are then empowered to handle future crises; or even the
potential to prompt U.S. adversary intervention into what becomes, for
them, a quagmire.
The costs and benefits of not intervening might be difficult to
assess in comparison with the costs and benefits of undertaking an
intervention and could, therefore, tend to be discounted by policy-
makers when making their intervention decision. However, an
improved understanding of these costs and benefits could improve the
information available to policymakers and lead to substantial benefits
for the United States, given the often highly consequential nature of
military intervention decisions.
Objective
To this end, our objective with this report is to create a framework that
can be used to rigorously consider the trade-offs between interven-
ing militarily immediately following the outbreak of a war or crisis,
intervening later, and not intervening at all, as well as the trade-offs
involved with decisions regarding the size of the potential intervention
force to be employed. This framework can be used to provide a better
understanding of the relationships between intervention timing, inter-
vention size, and intervention outcomes to inform future debates about
whether and when to intervene.
To construct this framework and explore these trade-offs, we will
need to consider not only the historical U.S. experience with inter-
ventions of different timings and sizes, but also the historical record
of cases in which the United States did not intervene. In addition to
investigating the historical relationships between intervention deci-
sions and relevant outcomes, we will explore the possible effects of
alternative decisions. In cases where the historical decision was made
to intervene, we will endeavor to estimate the possible costs and bene-
fits of the alternative choice: a decision not to intervene. In cases where
the historical decision was made not to intervene, we will explore the
potential costs and benefits to the United States had the decision been
Introduction 5
made to send military forces. Counterfactual comparisons are inher-

ently challenging, especially given the heterogeneity of U.S. interven-
tion and nonintervention cases. A full accounting of relevant costs and
benefits will include not only economic costs and benefits, but also
the strategic and reputational costs and benefits of each intervention
and nonintervention decision; such costs and benefits are challenging
to estimate, even for historical cases. In the remainder of this chapter,
we discuss our theoretical framework and methodological approach to
addressing these challenging questions in a rigorous way.
Theoretical Framework
Although policymakers might decide to undertake a military interven-

tion at any point during a war or crisis, we conceptualize the decision
to intervene as occurring in two distinct stages. First, policymakers
decide whether to intervene in the early stages of a crisis or war. This
“early” decision period is intended to reflect the first U.S. consider-
ation of a potential intervention in a crisis or war, notwithstanding any
limited potential delays imposed because of the complexity of the U.S.
policy process or the gathering of information on the events in ques-
tion. If a decision is made to intervene in this initial period, policymak-
ers would also need to decide on the size and scope of the U.S. inter-
vention. Indeed, these decisions are likely to be made together because
the set of feasible intervention options will likely affect the advisability
of the intervention itself. Policymakers might have a wide variety of
options for the forces they commit to an intervention, but we styl-
ize these choices down to three: the commitment of small, medium,
or large forces. As these sizing decisions are considered, policymakers
might view the problem in many different ways, such as deciding to
“throw everything” at a particularly acute problem, conducting a very
limited intervention designed only to monitor the situation or signal
interest, or opting for something in between. For many situations—
and depending on the issues involved—very large and very small inter-
ventions might not be plausible options for policymakers. The inclu-
sion of a middle category that is more plausible in a wider range of

circumstances is therefore an important aspect of this framework.
If policymakers decide not to intervene in this first decision
period, they could (if the war or crisis continues) face a second deci-
sion period regarding whether to intervene later in the crisis or war
and, if so, what size force to commit at that time. It is important to
note that although we have reduced these decision periods to just two
for analytical purposes, policymakers have the opportunity to revisit
these decisions at any time; also of note is that the continuous nature
of determining when and how to respond to these issues represents a
challenge. Policymakers could face situations that become incremen-
tally more concerning but without dramatic events that force a recon-
sideration of U.S. policy. Our reduction of the decision process to two
distinct decision points could, therefore, make the process of deciding
whether to intervene appear to be simpler than it would be in real-
ity. In addition, it is important to clarify that because policymakers
cannot make binding decisions never to intervene in a crisis or war—
changed circumstances could always force a reconsideration—the “No
Intervention” option therefore reflects circumstances for which policy-
makers have decided not to intervene at least at one point in time (and
possibly more), and the crisis or war ended before this decision to forgo
intervention was altered. Taking these issues into account, it is clear
that this model of U.S. intervention decisions is highly stylized, but
we use it in this report because it captures the main trade-offs facing
policymakers and will enable us to group relatively more-similar deci-
sions for analysis.
Altogether, this model suggests seven possible choices for a state
considering whether to intervene. Figure 1.1 illustrates a decision tree
that can be used to conceptualize the decision points and available
options.
In this report, we are interested in understanding how the deci-
sions about whether to intervene, when to intervene, and with how
many forces to intervene are associated with intervention and conflict
outcomes. In other words, we aim to understand how outcomes differ as
a result of the policymaker decision to choose one of the seven options
Introduction 7
Figure 1.1
The Intervention Decision
Decision1
2 3 4
Early, large Early, medium Early, small
1 Decision2
No intervention
5 6 7
Late, large Late, medium Late, small
shown in Figure 1.1.5 We identify three categories of outcomes as being

of particular interest: conflict outcomes, outcomes for the intervening
state (the United States), and outcomes for the host nation. Within
each of these categories are several more-specific outcomes of interest,
identified in Table 1.1. For example, conflict outcomes might include
the conflict’s duration, its intensity, and the likelihood that the con-
5 There are, of course, numerous other factors that could affect intervention outcomes
beyond these high-level decisions made at the outset of an intervention. Perhaps most notable
among these, from the perspective of policymakers, are part of a series of more-operational
decisions regarding how to actually execute the intervention, including decisions regarding
rules of engagement, targeting, and a host of other issues. Although these decisions poten-
tially are important to determining outcomes, we largely placed consideration of the effects
of such decisions outside the scope of this report to focus more directly on the three essen-
tial high-level decisions discussed above. That said, how interventions are executed has the
potential to confound our analysis if patterns we attribute to timing and size decisions turn
out to be attributable to execution decisions. Because of this possibility, when we considered
in our focused case studies and counterfactual cases (summarized in Chapters Four and Five)
whether size and timing decisions made independent contributions to the advancement of
U.S. interests, we assessed whether the execution of mission decisions was instead responsible
for the observed patterns. We did not have a similar ability to account for these issues in our
quantitative chapter (Chapter Three); this should be understood as an additional limitation
of that analysis.
Table 1.1
Intervention Outcomes of Interest
Conflict Outcomes Intervening State Interests Host-Nation Outcomes
Duration • Economic costs and • Economic trajectory

benefits
Intensity • Casualties • Political trajectory
Recurrence • Reputational costs and • Costs imposed by the

benefits conflict
• Strategic costs and
benefits
• Domestic political costs
and benefits
NOTE: This report does not focus on either domestic political costs or military
casualties, although both are discussed in Chapter Two.
flict will recur. Host-nation outcomes might focus on that country’s

political and economic trajectory and the costs of the conflict borne by
that country (e.g., economic costs or physical costs, such as destruction
of property and loss of lives). Finally, the outcomes for the intervener
might include the economic, strategic, reputational, and political costs
and benefits of the intervention,6 the possible loss of U.S. military or
civilian personnel, and whether the intervention or nonintervention
advances state interests. Throughout the report, we will explore the
relationships between intervention decisions and these outcome mea-
sures, with the goal of identifying patterns and insights that can be
used by planners and decisionmakers.
6 There is a significant literature that considers the political costs and benefits of conduct-
ing military interventions at various points in the political and election cycles. We consider
this literature in Chapter Two. We do not, however, directly consider political costs in the
remainder of the analysis. The focus of this report is on how decisions about when and with
how many forces to intervene shape the outcomes of those interventions and their ability to
advance U.S. national interests. Concerns about the political costs of interventions to spe-
cific actors or parties is therefore outside the scope of the report.
Introduction 9
Approach
We assessed that it would be unlikely that a single approach would

adequately allow us to assess the trade-offs inherent in the decisions
regarding whether, when, and with what size forces to intervene, given
the complexity of these decisions. Moreover, different approaches
might have different strengths and weaknesses for addressing differ-
ent aspects of these issues. We therefore investigated these questions
using four distinct methods, then combined the results to develop a
more comprehensive assessment of the trade-offs involved in the set of
decisions described in the previous section. Although we include a full
discussion of the relevant methods in each of the main chapters of this
report, we provide a brief overview here.
First, we conducted an extensive review of academic and policy
literature on relevant topics including (1) the relationship between
third-party interventions and conflict duration and intensity, (2) the
costs and benefits of an intervention for the intervening state, and
(3) the effects of third-party interventions on the host nation. We used
the literature review to determine what previous research has already
suggested about the relationship between interventions, intervention
timing, and intervention size and conflict outcomes, outcomes for the
intervener, and outcomes for the host. This literature helps to bound
our expectations for what we might find using other methods as well
as highlights aspects of our research question that have (or have not)
previously been carefully investigated.
Second, we conducted a quantitative analysis using a database of
286 crises and wars since 1945 that we identified as involving U.S. inter-
ests to a degree that made a U.S. military intervention plausible. This
database includes instances in which the United States has intervened
and those in which it did not. We categorized these cases according
to three criteria: (1) whether there was an intervention, (2) whether an
intervention (if it occurred) was early or late, and (3) whether an inter-
vention (if it occurred) was large, medium, or small in size. We then
compared these categories of cases along several dimensions, includ-
ing conflict or crisis duration and recurrence, likelihood of achieving
U.S. political objectives, military and economic assistance to the host
nation (or location of the conflict or crisis), and political and economic
outcomes for the host nation (or location of the conflict or crisis). We
use these comparisons as another perspective on how decisions about
whether to intervene, decisions about intervention timing, and choices
about intervention size might be related to different outcome measures.
Third, we conducted a set of 45 short, focused case studies. These
case studies include interventions and noninterventions, each address-
ing a consistent set of questions. Topics covered include an assessment
of why an intervention did or did not occur, why it occurred when it
did (early or late), and what the consequences of the intervention or
nonintervention choice were for the United States and the host nation
(or potential host, in the case of noninterventions). In particular, these
case studies helped us to isolate the effects of intervention size and
timing decisions on U.S. interests from selection effects that might
drive more-general correlations or apparent relationships. The case
studies address many of the same questions as the first two method-
ological approaches described, though they do not focus on the effects
on conflict intensity because these appeared to be well-considered else-
where. However, they allow for the inclusion of context and nuance as
well as cross-case comparisons and observations.
Finally, we conducted four in-depth counterfactual illustrations
focused on the 1992–1995 Bosnian War, the 1950–1953 Korean War,
the 2011 Libyan Civil War, and the 1979 Iranian Revolution. In these
counterfactuals, we first trace out what happened historically and why,
then we explore what might have happened if the United States had
made a different intervention decision, such as choosing not to inter-
vene in the Korean War. We highlight the role of key actors and their
interests and capabilities and explore how they most likely would have
acted in the alternative scenario. We explore the costs and benefits for
the United States of this alternate decision and identify which factors
would have been central to determining the counterfactual outcomes.
We use the insights from these cases to illustrate broader observations
about the trade-offs between different intervention decisions and out-
come measures.
Introduction 11
Limitations
Although our use of four methods will give us advantages over the
use of any single method, there remain limitations to the conclusions
we will be able to draw. The dynamics of war and crisis situations can
be immensely complex, and isolating the effects of U.S. interventions
from the context in which they occur is quite difficult. Our analysis
will not provide a single or simple answer. Instead, it will show how
the relative costs and benefits of different intervention timing and size
decisions depend very much on the context and that the relationship
between intervention decisions and outcomes is highly conditional. We
aim to provide decisionmakers with a framework for navigating the set
of choices surrounding whether, when, and how to intervene, includ-
ing insights into when early or late interventions are more likely to lead
to favorable outcomes and when interventions tend to have costs that
are not offset by their benefits. Finally, the purpose of this report is not
to directly evaluate any notable previous or anticipated U.S. interven-
tion decision. Such assessments certainly could be useful to learn les-
sons or inform specific decisions, but because each intervention deci-
sion and each context is unique, they might not provide, in isolation,
more-general guidance to policymakers facing these choices in unan-
ticipated contexts in the future. Instead, our objective is to consider the
broader U.S. historical experience and to identify patterns, trends, and
trade-offs to develop a generalizable framework that can help guide
U.S. intervention decisions going forward.
Organization of This Report
We approach the question of how intervention decisions are associ-

ated with outcomes from a different perspective in each of the next
four chapters. In Chapter Two, we cover the review of existing litera-
ture. In Chapter Three, we provide our quantitative analysis of the
case universe. In Chapter Four, we discuss the key insights from our 45
short, focused case studies, including the patterns observed in the cases
and what these patterns mean for the relationships between interven-
tion timing, size, and outcomes. In Chapter Five, we discuss the key
takeaways from our four longer counterfactual case studies, including

what we can learn about the circumstances in which different interven-
tion decisions might be most or least effective in working toward key
U.S. objectives at minimum cost. Finally, in Chapter Six, we synthesize
these different methods to more comprehensively describe the trade-
offs between different intervention decisions.
We also include two appendixes. Appendix A provides additional
detail on our list of potential U.S. intervention cases. Appendix B,
available online, provides the full text of our counterfactual case stud-
ies and can be downloaded at www.rand.org/t/RR4293.
CHAPTER TWO
What Do We Know? Reviewing Existing

Literature
As noted in Chapter One, states must make a series of decisions when

faced with a new conflict or crisis regarding whether to intervene mili-
tarily, at what time, and with what sized force. At each decision point,
decisionmakers should consider three types of outcomes: the effect of
the intervention on the outcome of the conflict (e.g., whether it esca-
lates, ends, recurs), on the intervener’s own national interests, and on
the trajectory of the host nation.
For our first of four methodologies, we consider in this chap-
ter what existing research can tell us about each of these questions
and outcomes. Specifically, we explore key insights from previous work
about the relationships between the presence, timing, and size of third-
party interventions and three types of outcomes: conflict outcomes,
outcomes for the intervening state, and outcomes for the host nation.
We then identify the remaining gaps in this literature, with a focus
on which gaps could be addressable using the other methodologies
employed in this report. Significantly, the least has been written about
how timing affects outcomes. Our report therefore has the potential to
fill an important gap in prior analyses.
13
The Intervention Decision: When Do Interventions

Improve Outcomes?
Conflict Outcomes
Even with an extensive body of previous research, there is no defini-
tive answer to the question of whether third-party interventions into
conflicts or crises tend to improve or worsen the outcomes of those
conflicts and crises. Instead, it seems that the effect of the interven-
tion depends heavily on the context: the nature of the conflict and its
participants, the intensity of the conflict, and the relative strength of
the participants.
Duration and Intensity
There are several reasons why interventions might lengthen the dura-
tion of a conflict or crisis. Watts, Johnston, et al., 2017, finds that
“[i]nterventions—especially those of the United States—are associated
with much lengthier periods of fighting. More specifically, interven-
tions are associated with wars lasting some 50 to 60 percent longer than
those not experiencing intervention.”1 The two key variables mediating
the relationship between interventions and conflict duration appear to
be the balance of power between warring parties and the side on which
the intervener chooses to intercede. Specifically, when the intervener
chooses to enter the conflict on the side of the weaker party, effectively
lowering the cost of conflict for that party or preventing it from being
destroyed, the conflict is likely to last longer than it would have with-
out such an intervention. By allowing weaker parties to continue fight-
ing and reducing their cost of war, third-party interventions can delay
the parties’ willingness to reach a settlement or delay the point at which
they accept defeat.2 Third-party interventions could also lengthen the
1 Steven Watts, Patrick B. Johnston, Jennifer Kavanagh, Sean M. Zeigler, Bryan Freder-
ick, Trevor Johnston, Karl P. Mueller, Astrid Stuth Cevallos, Nathan Chandler, Meagan L.
Smith, Alexander Stephenson, and Julia A. Thompson, Limited Intervention: Evaluating
the Effectiveness of Limited Stabilization, Limited Strike, and Containment Operations, Santa
Monica, Calif.: RAND Corporation, RR-2037-A, 2017, p. 125.
2 Donald Wittman, “How a War Ends: A Rational Model Approach,” Journal of Conflict
Resolution, Vol. 23, No. 4, December 1979; Ibrahim A. Elbadawi and Nicholas Sambanis,
“External Interventions and the Duration of Civil Wars,” Washington, D.C.: World Bank,
What Do We Know? Reviewing Existing Literature 15
duration of conflict if they make the conflict more difficult to resolve

by involving additional actors, triggering internal backlash, or other-
wise changing conflict dynamics.3 Not all interventions extend the
duration of conflict, however. An intervention that provides support to
the side that already has the upper hand or military advantages could
bring a quick resolution of conflict.4 Interventions might also speed
the resolution of conflict when the intervener can serve as a guarantor
of peace.5
Recurrence
Third-party interventions are also one of several factors that can affect
conflict recurrence and settlement stability, but once again, this effect is
conditional, mediated by several other factors.6 Past research finds that
previous foreign backing on the side of the rebels increases the likeli-
hood of renewed conflict because rebel actors might anticipate future
aid and thus be more willing to contest settlement terms. Foreign assis-
tance to governments was not found to affect the chances of conflict
relapse.7 The type of intervention—that is, whether it is explicitly a
peacekeeping, combat, or humanitarian mission—could also matter.
Some research has found that multilateral United Nations (UN) peace
September 2000; Patricia L. Sullivan and Johannes Karreth, “The Conditional Impact of
Military Intervention on Internal Armed Conflict Outcomes,” Conflict Management and
Peace Science, Vol. 32, No. 3, 2015.
3 David Kilcullen, The Accidental Guerrilla: Fighting Small Wars in the Midst of a Big One,
New York: Oxford University Press, 2009; David M. Edelstein, “Occupational Hazards:
Why Military Occupations Succeed or Fail,” International Security, Vol. 29, No. 1, Summer
2004, pp. 50–51.
4 Richard K. Betts, American Force: Dangers, Delusions, and Dilemmas in National Security,
New York: Columbia University Press, 2012, p. 31.
5 Barbara F. Walter, “The Critical Barrier to Civil War Settlement,” International Organi-
zation, Vol. 51, No. 3, Summer 1997, p. 363.
6 This research has mixed findings, in part, because of differing definitions, focal points,
and time frames.
7 Niklas Karlén, “The Legacy of Foreign Patrons: External State Support and Conflict
Recurrence,” Journal of Peace Research, Vol. 54, No. 4, 2017.
operations can prevent recurrence, perhaps because such operations

provide a mechanism of third-party enforcement.8
However, to fully understand the effect of interventions on con-
flict recurrence, it is worth considering some of the many other factors
that might matter. Most importantly, research shows that some coun-
tries (especially those with low income) are more likely to get stuck in
a cycle of poverty, low development, and conflict, with these factors
fueling one another—termed a conflict trap.9 In other words, there are
a wide variety of factors that might affect conflict recurrence. There is
evidence that the presence of a third-party intervention is one of those
factors, but the characteristics of the host nation matter to that out-
come in key ways.
Intervening State Interests

When deciding whether to intervene in a conflict or crisis, states must
weigh the likelihood that a military solution will advance their inter-
ests alongside the potential economic, strategic, and geopolitical costs
of the intervention. This assessment involves considering the nature
of the conflict or crisis, the actors involved, and the balance of power
between those actors. States might be unwilling to intervene in cases
where they expect that their intervention will not shift the balance of
power sufficiently to change the outcome. For example, research has
found that states are less likely to intervene in support of a very weak
opposition group that has little chance of defeating the state.10 If states
in the international system had perfect information about the bal-
ance of power in a given conflict (along with good information about
8 Michael Doyle and Nicholas Sambanis, “International Peacebuilding: A Theoretical and

Quantitative Analysis,” American Political Science Review, Vol. 94, No. 4, December 2000;
Caroline Hartzell, Matthew Hoddie, and Donald Rothchild, “Stabilizing the Peace After
Civil War: An Investigation of Some Key Variables,” International Organization, Vol. 55,
No. 1, Winter 2001; Virginia Page Fortna, “Does Peacekeeping Keep Peace? International
Intervention and the Duration of Peace After Civil War,” International Studies Quarterly,
Vol. 48, No. 2, June 2004.
9 Barbara F. Walter, “Does Conflict Beget Conflict? Explaining Recurring Civil War,”
Journal of Peace Research, Vol. 41, No. 3, May 2004.
10 Watts, Johnston, et al., 2017.
other relevant factors), they would only intervene in conflicts and crises
that they could influence to their desired outcome. In reality, how-
ever, states have much more trouble accurately assessing the balance of
power and face obstacles in the form of private and asymmetric infor-
mation. There are many cases in which U.S. policymakers might have
made a different choice about an intervention if they had full informa-
tion at the initial decision point. Because miscalculation does occur,
interventions only sometimes advance the intervener’s interests and
could bring unexpected economic and other costs.11 In this section, we
consider what previous research has suggested about the implications
of military interventions for the intervening party.
National Interests
Our understanding of which factors influence the relative success of
military interventions at achieving political objectives (used here as a
proxy for national interests) is based largely on the U.S. experience,
because few other states have a sufficient number of interventions for a
rigorous analysis. Previous RAND Arroyo Center work suggests that
the type of objectives defined for an intervention directly influences
the likelihood the intervention is able to achieve those objectives. U.S.
military interventions are more likely to achieve their political objec-
tives when those objectives are narrowly defined and are less success-
ful when objectives are ambitious or sweeping in scope (e.g., build
democracy, rebuild institutions, prevent the spread of communism).
In practice, this relationship has meant that the United States has been
less successful in interventions since the Cold War than was the case
before and during it because intervention goals have become increas-
ingly broad in scope.12 Success can vary by specific objective as well.
For instance, the U.S. record on democratization-focused interven-
tions is (as would be expected) mixed, with some research finding that
the military interventions have been successful in increasing levels of
11 Stephen Biddle, Military Power: Explaining Victory and Defeat in Modern Battle, Princ-
eton, N.J.: Princeton University Press, 2004.
12 Jennifer Kavanagh, Bryan Frederick, Alexandra Stark, Nathan Chandler, Meagan L.
Smith, Matthew Povlock, Lynn E. Davis, and Edward Geist, Characteristics of Successful U.S.
Military Interventions, Santa Monica, Calif.: RAND Corporation, RR-3062-A, 2019.
democracy in the host nation13 and some research finding little or no

relationship.14
Other characteristics of the intervention and intervention context
are likely to matter as well. First, the intervening state’s military capa-
bilities, both in absolute terms and compared with adversaries; strategy
and tactics, and use of nonmilitary resources or tactics (economic or
diplomatic) could also have direct implications for the likelihood of
success (at achieving the desired end state) because these factors shape
the course of the intervention, the ease with which the intervener is
able to operate, and the resources at the intervener’s disposal.15 Second,
factors related to the host nation, such as whether the host supports the
intervention and the economic and political development of the host
nation,16 could be relevant to whether the intervention is able to suc-
cessfully achieve political objectives or advance the intervening state’s
interests. For instance, weak political institutions in the host nation
13 James Meernik, “United States Military Intervention and the Promotion of Democ-
racy,” Journal of Peace Research, Vol. 33, No. 4, 1996, p. 400; Margaret G. Hermann and
Charles W. Kegley, Jr., “The U.S. Use of Military Intervention to Promote Democracy:
Evaluating the Record,” International Interactions, Vol. 24, No. 2, 1998, p. 108.
14 Jeffrey Pickering and Mark Peceny, “Forging Democracy at Gunpoint,” International
Studies Quarterly, Vol. 50, No. 3, September 2006, p. 555; Bruce Bueno de Mesquita and
George W. Downs, “Intervention and Democracy,” International Organization, Vol. 60,
No. 3, 2006, p. 647.
15 The United States has typically been more successful at achieving its political objectives
with combat interventions than during stabilization or deterrence missions. This success
could be because combat missions tend to be more heavily resourced, allowing for the appli-
cation of overwhelming force supported by technological superiority. Kavanagh, Frederick,
Stark, et al., 2019. Also see Patrick M. Regan, “Conditions of Successful Third-Party Inter-
vention in Intrastate Conflicts,” Journal of Conflict Resolution, Vol. 40, No. 2, June 1996; and
Patrick M. Regan and Aysegul Aydin, “Diplomacy and Other Forms of Intervention in Civil
Wars,” Journal of Conflict Resolution, Vol. 50, No. 5, October 2006.
16 Host-nation support will be a key consideration, since interventions could face many
more obstacles when the host nation is opposed. Edelstein, 2009, argues that host nations are
most likely to support a large intervening force when they share a common threat perception
or desired outcome with the intervener. An example in the U.S. context would be the high
willingness of South Korea to host U.S. forces since the end of the Korean War. David M.
Edelstein, “Foreign Militaries, Sustainable Institutions, and Postwar Statebuilding,” in
Roland Paris and Timothy D. Sisk, eds., The Dilemmas of Statebuilding: Confronting the
Contradictions of Postwar Peace Operations, New York: Routledge, 2009.
can severely complicate any stabilization mission.17 The presence of

third-party spoilers, either from within the host nation or outside it,
might also be detrimental to the intervener’s efforts to achieve national
objectives through the intervention.18
Economic Costs and Benefits
Interventions also have significant financial costs for the intervener
and can quickly tax the intervening nation’s resources, both financial
and military. There is the risk of overextension for intervening coun-
tries, which could jeopardize their security if a sudden crisis requires
resources that are already mobilized elsewhere.19 More broadly, what
appears initially as a “simple” military intervention could quickly take
on broader aims that perpetuate the intervention and increase its costs,
in what some have described as mission creep.20 The so-called Pottery
Barn principle (“you break it, you own it”) illustrates that interventions
can quickly morph into a constellation of complex nation-building
tasks.21
Another way to consider the costs of the intervention is to explore
the opportunity cost—that is, the loss of other alternatives (in this case,
other forms of spending) that is the result of the decision to spend
money on the military intervention. Most obviously, money spent on a
military intervention is not available for domestic issues, such as edu-
cation and healthcare, and the taxes from which military spending is
17 Kavanagh, Frederick, Stark, et al., 2019; Matthew Adam Kocher, Adria Lawrence, and
Nuno P. Monteiro, The Rabbit in the Hat: Nationalism and Resistance to Foreign Occupa-
tion, Chicago, Ill.: American Political Science Association 2013 Annual Meeting, August 20,
2013.
18 Kavanagh, Frederick, Stark, et al., 2019.
19 Richard N. Haass, “Military Force: A User’s Guide,” Foreign Policy, No. 96, 1994, p. 21.
20 See Michael Mandelbaum, “The Reluctance to Intervene,” Foreign Policy, No. 95,
Summer 1994, pp. 6–9; and Watts, Johnston, et al., 2017, p. 108.
21 Walter, 1997, p. 361; Bob Woodward, Plan of Attack, New York: Simon and Schuster,
2004, p. 150.
drawn might otherwise have contributed to economic growth through

consumer spending, investment, and commercial expansion.22
Reputational and Strategic Costs and Benefits
Interventions can have reputational and strategic consequences, both
positive and negative, for the intervening state. Unilateral interventions
could create tensions between the intervener and other states in the
international system that disapprove of or oppose the original inter-
vention. Such tension can have many implications, including spark-
ing a competing intervention or increasing the challenges associated
with basing access.23 Interventions could complicate relationships with
other major powers, including allies that disapprove or adversaries that
feel threatened or provoked.24 Military interventions that lack inter-
national support could also incur more-diffuse costs in the long term,
for instance by spurring resentment against international norms, law,
and institutions and contributing to their erosion.25 The presence of an
international coalition could advance the legitimacy of the intervention
or provide a necessary aggregation of capabilities (especially important
for small states). However, having more states involved creates coordi-
nation challenges.26
Reputational costs could also be relevant. Research has explored
whether deliberations over an intervention might have implications for
22 Among many others, see, for example, James H. Lebovic and Ashfaq Ishaq, “Military
Burden, Security Needs, and Economic Growth in the Middle East,” Journal of Conflict Res-
olution, Vol. 31, No. 1, March 1987; and Jülide Yildirim and Nadir Öcal, “Military Expen-
ditures, Economic Growth and Spatial Spillovers,” Defence and Peace Economics, Vol. 27,
No. 1, 2016, p. 89.
23 Pettyjohn and Kavanagh, 2016.
24 Eelco van der Maat, “Sleeping Hegemons: Third-Party Intervention Following Territo-
rial Integrity Transgressions,” Journal of Peace Research, Vol. 48, No. 2, 2011.
25 Barry R. Posen, Restraint: A New Foundation for U.S. Grand Strategy, Ithaca, N.Y.: Cor-
nell University Press, 2014.
26 Sarah E. Kreps, “Multilateral Military Interventions: Theory and Practice,” Political Sci-
ence Quarterly, Vol. 123, No. 4, Winter 2008/2009.
a state’s credibility and reputation in future international disputes.27

Indeed, U.S. policymakers have frequently argued for the central
importance of credibility in U.S. foreign policy and national securi-
ty.28 Although there is some work that supports this view, the major-
ity of recent research finds that operational context plays a significant
role in mitigating any lasting reputational costs from past intervention
decisions.29
Domestic Political Costs and Benefits
An additional consideration for intervening states that is frequently
discussed in the literature (although it is not a focus of this report) is
the domestic political costs of military interventions. Here, too, find-
ings have been mixed and ambiguous. The “diversionary hypothesis”
has enjoyed significant popular adherence and some scholarly support,
suggesting that leaders facing unpleasant headlines or flagging polls
could be tempted to engage in foreign adventures to distract from the
unpopularity of their domestic policies.30 Related is work on the “rally
effect,” which finds that political leaders can receive a surge in popular
support following the start of a new military intervention or war—
27 Thomas C. Schelling, The Strategy of Conflict, Cambridge, Mass.: Harvard University

Press, 1960; Christopher J. Fettweis, “Credibility and the War on Terror,” Political Science
Quarterly, Vol. 122, No. 4, 2007/2008; Stephen M. Walt, “The Credibility Addiction,” For-
eign Policy, January 6, 2015.
28 See, for example, John McCain, “Obama Has Made America Look Weak,” New York
Times, March 14, 2014; and Schelling, 1960.
29 John Mitton, “Selling Schelling Short: Reputations and American Coercive Diplomacy
After Syria,” Contemporary Security Policy, Vol. 36, No. 3, 2015.
30 Russett, Bruce, ‘‘Economic Decline, Electoral Pressure, and the Initiation of Interstate
Conflict,’’ in Charles S. Gochman and Alan Ned Sabrosky, eds., Prisoners of War? Nation-
States in the Modern Era, Lexington, Mass.: Lexington Books, 1990.
although this bump could be short-lived.31 Different types of interven-

tions might also generate different levels of public support.32
Although interventions might provide some domestic political
benefits, they almost always also have some political costs.33 The type
and magnitude of domestic costs can vary according to any number of
factors. War represents a political hazard for leaders, which can be mit-
igated or aggravated by various factors—such as whether the leader is
democratic or authoritarian or whether the war is incurring high costs
or even potentially resulting in defeat.34 Military casualties also affect
the level of public support for an intervention (especially under certain
conditions), but this relationship is complex.35 In general, it seems that
it might not be absolute numbers but instead the marginal increase
in casualties that affects public opinion most. However, under cer-
tain conditions—especially when the stakes of the intervention seem
high—the public could have reasonably high (but not unlimited) toler-
31 John E. Mueller, “Presidential Popularity from Truman to Johnson,” American Political

Science Review, Vol. 64, No. 1, March 1970; Scott Sigmund Gartner and Gary M. Segura,
“War, Casualties, and Public Opinion,” Journal of Conflict Resolution, Vol. 42, No. 3, June
1998, p. 281; Patrick James and Jean-Sébastien Rioux, “International Crises and Linkage
Politics: The Experiences of the United States, 1953–1994,” Political Research Quarterly,
Vol. 51, No. 3, September 1998, p. 781; John R. Oneal and Anna Lillian Bryan, “The Rally
‘‘Round the Flag Effect in U.S Foreign Policy Crises, 1950–1985,” Political Behavior, Vol. 17,
No. 4, December 1995.
32 Bruce W. Jentleson, “The Pretty Prudent Public: Post Post-Vietnam American Opin-
ion on the Use of Military Force,” International Studies Quarterly, Vol. 36, No. 1, 1992;
John R. Oneal, Brad Lian, and James H. Joyner, Jr., “Are the American People ‘Pretty Pru-
dent’? Public Responses to U.S. Uses of Force, 1950–1988,” International Studies Quarterly,
Vol. 40, No. 2, June 1996, p. 273.
33 A potentially important political cost of military intervention that is not discussed exten-
sively in existing literature is the demand such deployments place on senior policymakers’
time and attention and the opportunity cost of not having those policymakers focused on
key domestic, foreign policy, and national security issues.
34 Bruce Bueno de Mesquita and Randolph M. Siverson, “War and the Survival of Politi-
cal Leaders: A Comparative Study of Regime Types and Political Accountability,” Ameri-
can Political Science Review, Vol. 89, No. 4, December 1995; Charles W. Ostrom, Jr., and
Dennis M. Simon, “Promise and Performance: A Dynamic Model of Presidential Popular-
ity,” American Political Science Review, Vol. 79, No. 2, June 1985, p. 355.
35 John E. Mueller, War, Presidents, and Public Opinion, New York: Wiley, 1973.
ance for casualties.36 In the U.S. context, support for additional mili-
tary interventions has tended to decline for a time after conflicts that
involve large numbers of casualties, especially after military defeats.37
Host-Nation Trajectory
Previous work has also explored how external interventions affect the
governance, economies, and societies of host countries. We focus on
political and economic outcomes.
Political Outcomes
Work focused on political outcomes has emphasized how interventions
affect the host nation’s level of democracy and its record on politi-
cal and civil rights. The results are mixed. Several studies have found
that, in general, the existence of an intervention does not make the
host nation more likely to democratize.38 Other studies have found that
there does seem to be such a relationship but that it depends heavily
on context. The strategy of the intervener (both military and nation-
building) and host-nation political institutions could affect the out-
come.39 However, the literature is more pessimistic about the duration
36 Gartner and Segura, 1998; Eric V. Larson, Casualties and Consensus: The Historical Role
of Casualties in Domestic Support for U.S. Military Operations, Santa Monica, Calif.: RAND
Corporation, MR-726-RC, 1996, pp. 99–101; James Burk, “Public Support for Peacekeep-
ing in Lebanon and Somalia: Assessing the Casualties Hypothesis,” Political Science Quar-
terly, Vol. 114, No. 1, 1999.
37 Jennifer Kavanagh, Bryan Frederick, Matthew Povlock, Stacie L. Pettyjohn, Angela
O’Mahony, Stephen Watts, Nathan Chandler, John Speed Meyers, and Eugeniu Han, The
Past, Present, and Future of U.S. Ground Interventions: Identifying Trends, Characteristics, and
Signposts, Santa Monica, Calif.: RAND Corporation, RR-1831-A, 2017.
38 Bueno de Mesquita and Downs, 2006; Meernik, 1996; Alexander B. Downes and Jona-
than Monten, “Forced to Be Free? Why Foreign-Imposed Regime Change Rarely Leads to
Democratization,” International Security, Vol. 37, No. 4, 2013; William Easterly, Shanker
Satyanath, and Daniel Berger, Superpower Interventions and Their Consequences for Democ-
racy: An Empirical Inquiry, Cambridge, Mass.: National Bureau of Economic Research,
Working Paper No. 13992, May 2008; Jeffrey Pickering and Emizet F. Kisangani, “Politi-
cal, Economic, and Social Consequences of Foreign Military Intervention,” Political Research
Quarterly, Vol. 59, No. 3, September 2006; Mark Peceny, “Forcing Them to Be Free,” Politi-
cal Research Quarterly, Vol. 52, No. 3, September 1999.
39 Meernik, 1996; Hermann and Kegley, 1998; Downes and Monten, 2013; Peceny, 1999.
of observed changes to democracy, finding progress toward democracy

is typically temporary.40 There are comparatively few studies regarding
the effect of intervention on a broader set of political outcomes, such
as civil liberties, political rights, or human rights, and results on these
outcomes are mixed (and sometimes, things get worse before they get
better, especially in the area of human rights).41
Economic Outcomes
Research on the effect of interventions on host-nation economies is
limited, and results are mixed. Overall, third-party interventions
do not appear to have any discernable long-term effect on economic
growth in the host nation, regardless of the regime type of the host or
intervener.42
Table 2.1 summarizes what we do and do not know about the
relationship between military interventions and our three types of out-
comes of interest.
The Timing Decision: Is It Better to Intervene Early or

Late?
An intervening state can decide to intervene early (while the conflict or

crisis is in its beginning stages), or it can opt not to do so but revisit the
decision at a later time and decide to intervene then. Research provides
some insight into the implications of the timing decision for conflict
40 Nils Petter Gleditsch, Lene Siljeholm Christiansen, and Håvard Hegre, Democratic
Jihad? Military Intervention and Democracy, Washington, D.C.: World Bank, Working Paper
No. WPS4242, June 2007.
41 Frederic S. Pearson, Scott Walker, and Stephanie Stern, “Military Intervention and Pros-
pects for Democratization,” International Journal of Peace Studies, Vol. 11, No. 2, 2006;
Taylor B. Seybolt, Humanitarian Military Intervention: The Conditions for Success and Fail-
ure, Oxford, United Kingdom: Stockholm International Peace Research Institute and
Oxford University Press, 2007, p. 270; Dursun Peksen, “Does Foreign Military Intervention
Help Human Rights?” Political Research Quarterly, Vol. 65, No. 3, 2012.
42 Marie Olson Lounsbery, Frederic Pearson, and Andrea Kathryn Talentino, “Unilateral
and Multilateral Military Intervention: Effects on Stability and Security,” Democracy and
Security, Vol. 7, No. 3, 2011; Pickering and Kisangani, 2006.
Table 2.1
Summary of Prior Research on Military Interventions and Outcome
Measures
Effect of Military Interventions Found by

Category Outcome Prior Research
Conflict Duration and intensity Mixed. Depends on balance of capabilities.
Recurrence Mixed. Depends on domestic context.
Intervening National interests Mixed. Depends on type of intervention,

state (general) international context and coalition, relative
military capabilities.
Economic costs and Few economic benefits. Costs include real

benefits financial costs and opportunity costs.
Strategic and Mixed. States must balance credibility gain

reputational costs and from an intervention with possible negative
benefits reaction from international community.
Domestic political costs Interventions can trigger short term bump

and benefits in presidential approval, but leaders could
lose power over failed interventions,
especially if gains seem low and military
fatalities are high.
Host nation Political and economic Few long-term, lasting improvements.

trajectory
outcomes and for the interests of the intervener and host nations, but
the body of evidence surrounding the timing question is limited.
Conflict Outcomes
Research on the timing of military and nonmilitary interventions sug-
gests that there are two points during which an intervention (military
or nonmilitary) might be able to shorten the duration of a conflict or
crisis (or prevent a crisis from escalating). The first occurs immediately
after the start of the conflict or crisis, before states have invested time
and money in the conflict and before battle lines are fully drawn.43
43 I. William Zartman and Saadia Touval, “International Mediation: Conflict Resolution

and Power Politics,” Journal of Social Issues, Vol. 41, No. 2, 1985; Patrick M. Regan and
Allan C. Stam, “In the Nick of Time: Conflict Management, Mediation Timing, and the
In the case of a crisis, most interventions will occur quickly (given

the event’s shorter duration) and will either successfully resolve the
crisis and prevent escalation or fail, in which case the resulting con-
flict could escalate. The second point during which an intervention
could be successful (and which likely applies only to conflicts) occurs
after the point of ripeness: the point at which the parties to a conflict
might be ready to come to the table and negotiate.44 Interventions that
occur after this point of ripeness can be effective at resolving conflicts
through a few different mechanisms. First, after an extended period of
fighting, both states might grow tired of conflict, run out of resources,
or revise their expectations of success.45 Second, it is possible that, at
some point in the conflict, the stakes will be so high and the risk of
crisis so large that warring parties are willing to come to the table to
avoid that catastrophic outcome. Third, as the conflict wears on, new
opportunities for solutions might emerge that bring parties back to the
table.46 Once any one of these points is reached, the intervention could
be more effective at providing the final impetus that brings parties to
the negotiating table.
Recurrence
There is also limited research that explicitly considers whether the
timing of an intervention is associated with the likelihood of conflict
recurrence. As already noted, the likelihood of conflict recurrence is
largely determined by the characteristics of the host nation, rather than
the intervention itself. Regardless of whether the intervention occurs
early or late, preventing recurrence means dealing with the factors—
Duration of Interstate Disputes,” International Studies Quarterly, Vol. 44, No. 2, June 2000;
Christopher R. Mitchell, Cutting Losses: Reflections on Appropriate Timing, Fairfax, Va.: Insti-
tute for Conflict Analysis and Resolution, Working Paper No. 9, 1995; T. David Mason,
Joseph P. Weingarten, Jr., and Patrick J. Fett, “Win, Lose, or Draw: Predicting the Outcome
of Civil Wars,” Political Research Quarterly, Vol. 52, No. 2, June 1999, p. 264.
44 Zartman and Touval, 1985.
45 Stephen John Stedman, Peacemaking in Civil War: International Mediation in Zimbabwe,
1974–1980, Boulder, Colo.: Lynne Reinner, 1991.
46 Zartman and Touval, 1985; Stedman, 1991; Christopher R. Mitchell, The Structure of
International Conflict, London: Macmillan, 1981, Ch. 8.
such as continued grievances and supplies of weapons—that can

increase the risk of future conflict in the intervention or subsequent
postconflict agreement.47

States make the decision to intervene by weighing costs against the
likelihood that they will achieve their political objectives and advance
national interests by intervening. Timing can be an important consid-
eration that has implications for the likelihood of success and expected
costs.
National Interests
Research suggests that the relationship between intervention timing
and the likelihood that the intervention advances national interests is
shaped by the type of intervention and the type of objective. Differ-
ent intervention timings might be appropriate for different types of
interventions. For example, although it could be worth waiting to see
whether a political crisis (for instance, the noncompliance of a state with
some international rules or a weapons inspection program) or a limited
territorial dispute can resolve itself through diplomatic means early
interventions could be essential during certain humanitarian crises to
forestall mass atrocities, which often occur early on (e.g., Bosnia and
Rwanda).48 The strength of the adversary might also vary with the
timing of the intervention, but it is not clear in what direction. In some
cases, an earlier intervention might confront a weaker adversary before
it reaches full strength; in others, the adversary might be weaker if the
intervention is late, after it has been exhausted by fighting.49
47 Stephen L. Quackenbush and Jerome F. Venteicher, “Settlements, Outcomes, and the

Recurrence of Conflict,” Journal of Peace Research, Vol. 45, No. 6, November 2008.
48 Jon Western and Joshua S. Goldstein, “Humanitarian Intervention Comes of Age: Les-
sons from Somalia to Libya,” Foreign Affairs, Vol. 90, No. 6, November/December 2011,
p. 57.
49 Stephen Biddle, comments to the authors, January 2019.

There is little research that directly considers the link between inter-
vention timing and economic costs for the intervener. We can draw
some inferences, however. An early intervention that is able to quickly
end a conflict might ultimately be less costly than a late intervention,
which might face a more complex conflict and challenge. On the other
hand, a late intervention that is well timed could be cost-effective if
that larger size allows the intervener to more quickly achieve interven-
tion goals. An early intervention that does not successfully end a con-
flict quickly becomes resource intensive.
Strategic and Reputation Costs and Benefits
The timing of interventions could, however, have a larger effect on the
strategic or reputational costs faced by the intervening power. The per-
ceived legitimacy of the intervention and the stance of the international
community are important for understanding the effects of interven-
tion timing on the intervening state’s strategic interests and reputation.
Participating in interventions once they are sanctioned by the interna-
tional community rarely results in significant reputational costs. But
intervening early when international sanction is lacking or delaying
participation (e.g., Bosnia) in interventions that are sanctioned and
that aim at humanitarian goals could damage the state’s reputation
and could have strategic costs.50
Finally, there could be domestic political costs associated with the
timing of interventions for democratic and autocratic leaders. The rel-
evant mechanism is based on public opinion and approval. Domes-
50 Kreps, 2008/2009. For example, interventions that occur before any diplomatic efforts
toward resolution could face questions of legitimacy. The United States has been faulted in
the case of Iraq for intervening without giving the processes of diplomacy sufficient oppor-
tunity to take their course. As Kreps argues, intervening as part of a coalition could help
to reduce some of the strategic and reputational costs that might come with an early inter-
vention. Also see Jon Western, “Sources of Humanitarian Intervention: Beliefs, Informa-
tion, and Advocacy in U.S. Decisions on Somalia and Bosnia,” in James M. McCormick,
ed., The Domestic Sources of American Foreign Policy: Insights and Evidence, 6th ed., New
York: Rowman and Littlefield Publishers, 2012, p. 399; and Seybolt, 2007.
tic political costs might be higher when the domestic public does not
support the intervention. For this reason, election timing could play a
role in decisions about when to launch a military intervention.51 Early
interventions might generally be harder to sell to the public, especially
in the case of an emerging crisis or conflict, prior to the point at which
the threat is severe enough to be readily apparent. However, there could
also be political costs from intervening too late, especially if public
pressure in support of a specific intervention is high.52
Host-Nation Outcomes
There is very limited literature available on how intervention timing
affects the outcomes for the host nation, beyond what was already
described in the section on conflict outcomes (resolution, recurrence,
and intensity) that focused on how timing can affect the interests of
the intervening state. We can draw some inferences from this litera-
ture. For instance, an early intervention that is able to end a conflict
more quickly could have more-positive outcomes for the host nation.
As another example, early interventions into humanitarian crises could
be better able to prevent serious atrocities or limit the damage inflicted
on the host nation.53 Apart from these indirect observations, we know
little from previous work about whether intervention timing improves
the future political or economic trajectory of the host nation.
Table 2.2 summarizes the key insights from our assessment of the
prior literature regarding the relationship between intervention timing
51 Bruce Bueno de Mesquita, Alastair Smith, Randolph M. Siverson, and James D. Morrow,
The Logic of Political Survival, Cambridge, Mass.: MIT Press, 2005; Stephen Biddle, com-
ments to the authors, January 2019.
52 This pressure is hypothesized to come from media coverage and is known as the CNN
effect. Empirical support for the CNN effect is also mixed. See, for example, Piers Robin-
son, “The CNN Effect: Can the News Media Drive Foreign Policy?” Review of International
Studies, Vol. 25, No. 2, 1999; Babak Bahador, The CNN Effect in Action: How the News
Media Pushed the West Toward War in Kosovo, New York: Palgrave Macmillan, 2007; and
Bueno de Mesquita et al., 2005.
53 Zartman and Touval, 1985; Western and Goldstein, 2011, p. 57.
Table 2.2
Summary of Prior Research on Intervention Timing and Outcome Measures
Effect of Intervention Timing Found by

Category Outcome Prior Research
Conflict Duration and intensity Could reduce duration early in conflict or

late, after mutual stalemate.
Recurrence Limited evidence.
Intervening National interests Success most likely early or late in the

state (general) conflict. Type of intervention matters.
Economic costs and Limited evidence.

benefits
Strategic and Timing might matter little when

reputational costs and intervener acts in line with international
benefits preference or approval, but decisions to
intervene (or not) that run counter to
international expectations could have
strategic and reputational costs.
Domestic political costs Timing could be linked to electoral or

and benefits political cycle.
Host nation Political and economic Limited evidence.

trajectory
and outcomes. We highlight in gray those areas for which a lack of

empirical study limits our current understanding.54
The Size Decision: Should the Intervention Be Small or

Large?
The final decision that states face when considering an intervention is

that intervention’s size. In this section, we consider how intervention
size is associated with key outcomes.55
54 James Dobbins, Michele A. Poole, Austin Long, and Benjamin Runkle, After the War:
Nation-Building from FDR to George W. Bush, Santa Monica, Calif.: RAND Corporation,
MG-716-CC, 2008.
55 We consider large forces to be those in excess of a division of deployed U.S. forces and
small forces to be those of brigade size or smaller.
Conflict Outcomes
Research that considers the ability of smaller interventions to contain,
end, or deescalate conflict is relatively mixed in its findings. For exam-
ple, several RAND studies have found that small stabilization missions
can lead to better outcomes for the state actor but do not typically
shorten conflicts.56 Findings on the ability of larger interventions to
shorten conflict duration or to deescalate a conflict or crisis are also
mixed. There is some evidence that larger forces could be better able to
guarantee the terms of any settlement agreement, making one or both
sides more willing to come to the negotiating table.57 The extent to
which a large intervention force is able to help the party it is support-
ing militarily could speed progress toward a resolution of the conflict.58
Larger interventions are also associated with lower fatality levels (expe-
rienced and caused), at least during nation-building.59 However, past
experience in Afghanistan and elsewhere has underscored the fact that
having a large force size might be necessary for conflict termination
56 In addition to small ground forces, limited interventions might also rely on airpower
through drone strikes or limited bombing campaigns. Previous work on the use of drone
strikes to fight insurgency campaigns has mixed empirical results, with some research sug-
gesting it can reduce militant violence and other research suggesting it can provoke backlash.
The difference in effect could depend on the way the intervening state organizes and uses the
strikes. Stephen Watts, Caroline Baxter, Molly Dunigan, and Christopher Rizzi, The Uses
and Limits of Small-Scale Military Interventions, Santa Monica, Calif.: RAND Corporation,
MG-1226-RC, 2012, p. 45; Watts, Johnston, et al., 2017. See also Patrick B. Johnston and
Anoop K. Sarbahi, “The Impact of US Drone Strikes on Terrorism in Pakistan,” Interna-
tional Studies Quarterly, Vol. 60, No. 2, June 2016; and Javier Jordan, “The Effectiveness of
the Drone Campaign Against al Qaeda Central: A Case Study,” Journal of Strategic Studies,
Vol. 37, No. 1, 2014.
57 Walter, 1997; Watts, Johnston, et al., 2017.
58 Watts, Johnston, et al., 2017; Kavanagh, Frederick, Stark, et al., 2019.
59 James Dobbins, Seth G. Jones, Keith Crane, Andrew Rathmell, Brett Steele, Richard
Teltschik, and Anga R. Timilsina, The UN’s Role in Nation-Building: From the Congo to
Iraq, Santa Monica, Calif.: RAND Corporation, MG-304-RC, 2005; Stephen Biddle, Jef-
frey A. Friedman, and Jacob N. Shapiro, “Testing the Surge: Why Did Violence Decline in
Iraq in 2007?” International Security, Vol. 37, No. 1, Summer 2012; James T. Quinlivan,
“Force Requirements in Stability Operations,” Parameters: U.S. Army War College Quarterly,
Vol. 25, No. 4, Winter 1995–1996.
and deescalation but is by no means sufficient.60 Large-scale interven-

tions bring with them the potential for backlash from local popula-
tions, and even overwhelming military strength cannot guarantee the
effective governance needed for lasting deescalation.61
Recurrence
As already noted, the primary drivers of conflict recurrence are endog-
enous to the conflict itself and to the local, state, and regional con-
text in which the conflict occurs. However, although intervention size
might not be directly related to recurrence, there are two possible indi-
rect relationships. First, larger forces might be better able to guaran-
tee the terms of any settlement agreement, reducing the likelihood for
recurrence.62 Second, even if a larger intervention force cannot guaran-
tee a victory for the supported side, it might be able to raise the costs
of continued fighting in ways that highlight the value of stability and
encourage settlement.63
Intervening State Outcomes

Likelihood of Success
There has been a considerable amount of research exploring the link
between the size of an intervention and the likelihood that it is suc-
cessful in achieving the state’s political objectives (which are overlap-
ping with, but not identical to, the goals of terminating the conflict or
preventing recurrence). Previous RAND Arroyo Center work found
that the number of forces appears to matter most to intervention suc-
cess in combat interventions, which includes cases in which the United
States is able to overwhelm adversaries with the size and technological
60 Watts, Johnston, et al, 2017.

61 Kilcullen, 2009; Patrick T. Brandt, T. David Mason, Mehmet Gurses, Nicolai Petrovsky,
and Dagmar Radin, “When and How the Fighting Stops: Explaining the Duration and Out-
come of Civil Wars,” Defence and Peace Economics, Vol. 19, No. 6, 2008; Håvard Hegre and
Håvard Mokleiv Nygård, “Governance and Conflict Relapse,” Journal of Conflict Resolution,
Vol. 59, No. 6, 2015; Charles T. Call, Why Peace Fails: The Causes and Prevention of Civil War
Recurrence, Washington, D.C.: Georgetown University Press, 2012.
63 Watts, Baxter, et al., 2012.
superiority of its military forces.64 There also appears to be a relation-

ship between the number of forces and the success of peacekeeping and
nation-building efforts.65 Prior research does not find any clear rela-
tionship between troop size and the success of deterrent interventions.
For interventions of equal length, larger interventions of course have
higher direct monetary costs than smaller interventions. In fact, this
is one of the primary reasons that policymakers often prefer smaller
military responses to conflicts and crises.66 However, a yearslong or
decades-long small intervention intended to contain a threat could, in
some cases, be more costly than a short, larger intervention, especially
if the larger intervention more directly addresses the underlying issue
and leads to a better overall outcome. The stakes and level of risk could
matter as well. The smaller force could be more appealing when the
stakes for the United States are relatively lower or when a larger force
would be perceived as too provocative or aggressive. In high-risk situa-
tions and when the threat seems vital, however, the benefits of deploy-
ing a larger force might offset any additional costs.67
Strategic and Reputational Costs and Benefits
There has also been little direct consideration of the relationship
between intervention size and the strategic or reputational costs and
benefits for the intervening state. However, we can make some general
observations. The decision about how many forces to deploy is likely to
send a signal to both allies and adversaries about the intervening state’s
level of commitment and degree of military strength.68 When such
a response is in support of allies and partners, it could improve U.S.

65 Dobbins, Jones, et al., 2005; Kavanagh, Frederick, Stark, et al., 2019.
67 See, for example, U.S. Department of Defense, Summary of the 2018 National Defense
Strategy of the United States of America: Sharpening the American Military’s Competitive Edge,
Washington, D.C., 2018.
68 Schelling, 1960. See also Thomas C. Schelling, Arms and Influence, New Haven, Conn.:
Yale University Press, 1966.
credibility and strengthen bilateral ties. However, if the intervention is

not supported by allies or if the large commitment is seen as an over-
reach, large force size could damage the reputation of the intervener,
undermine the legitimacy of the intervention, and ultimately have stra-
tegic costs imposed by allies or adversaries that do not approve of the
original mission.69 However, there could be such costs from an under-
response as well, especially if the under-response is seen as a failure to
meet responsibilities.70
There has been little careful investigation of the relationship between
intervention size and domestic political costs. Most likely this rela-
tionship operates through the connection between intervention out-
comes and domestic political attitudes. The domestic public likely does
not have well-informed preferences about the size of an intervention.
Rather, any attitudes they have are likely tied to such factors as the per-
ceived success of the intervention and the perceived importance of the
intervention objective.71 Therefore, as long as the intervention is per-
ceived to be going well and to have an important objective, there seems
to be little reason to expect the domestic public to reward or punish
political representatives according to intervention size.72
However, there are three possible exceptions. First, it is possible
that small-footprint interventions (which the public might not hear
about or might know little about) could be used to reduce or even
eliminate the political costs associated with unpopular military activi-
ties. Second, large interventions could attract negative publicity and
trigger strong public reactions if they are seen as unnecessary and very
69 Dennis Florig, “Hegemonic Overreach vs. Imperial Overstretch,” Review of International

Studies, Vol. 36, No. 4, October 2010.
70 Berit Bliesemann de Guevara and Florian P. Kühn, “‘The International Community
Needs to Act’: Loose Use and Empty Signalling of a Hackneyed Concept,” International
Peacekeeping, Vol. 18, No. 2, 2011; James Pattison, “Outsourcing the Responsibility to Pro-
tect: Humanitarian Intervention and Private Military and Security Companies,” Interna-
tional Theory, Vol. 2, No. 1, 2010.
71 Larson, 1996, pp. 99–101.
72 Bueno de Mesquita et al., 2005; Larson, 1996, pp. 99–101.
costly, especially during periods of budget austerity.73 Third, large

interventions of certain types (specifically, combat) have a higher risk
than smaller ones do of having high U.S. military casualties, which can
have negative political costs for domestic politicians.74
As noted elsewhere, the key determinant of host-nation outcomes is
likely to be the context and characteristics of the host nation itself, such
as its level of political and economic development. As a result, interven-
tion size might have a fairly limited effect on the political, economic,
and development outcomes of the host nation, especially in the longer
term.75
However, intervention size could have an indirect effect in sev-
eral contexts, most already mentioned elsewhere in this chapter. First,
larger intervention forces could support better long-term outcomes if
they are able to facilitate a durable peace settlement and prevent con-
flict recurrence.76 Second, larger forces are likely necessary for the
nation-building tasks typically required in a postconflict environ-
ment.77 Other relevant factors include the level of political develop-
ment of the host nation and the application of nonmilitary resources.78
However, foreign intervention forces cannot stay forever, and having a
large force in a postconflict situation could build a dependency on or
trigger backlash against the intervening force within the host nation
that could interfere with the long-term stability and independence of
the host nation.79
Table 2.3 highlights the insights that emerge from a review of
research that considers the relationship between intervention size and

74 Mueller, 1973.
75 Pickering and Kisangani, 2006.
77 Dobbins, Jones, et al., 2005.
Table 2.3
Summary of Prior Research on Intervention Size and Outcome Measures
Effect of Intervention Size Found by Prior

Category Outcome Research
Conflict Duration and intensity Mixed. Small interventions might

contain conflict but are less effective at
terminating. Larger interventions might
speed conflict termination in some cases
but can also trigger backlash.
Recurrence Larger interventions might be able

to underwrite peace agreements and
achieve more-stable military and political
outcomes.
Intervening National interests Larger forces might more effectively

state (general) advance intervener interests during combat
and stabilization missions. Large forces can
be provocative and counterproductive in
some contexts.
Economic costs and Limited evidence.

benefits
Strategic and Limited evidence.

reputational costs and
benefits
Domestic political costs Larger interventions could have higher

and benefits political costs, to the extent that they are
more expensive and have higher fatality
rates.
Host nation Political and economic Effect of intervention size on host-nation

trajectory outcomes likely depends on the nature of
the conflict and intervention and on the
domestic context.
outcomes. We highlight in gray the areas in which empirical assess-

ment is lacking.
Synthesis and Limitations
In this chapter, we considered what previous research has to say about

the relationships between military interventions, their timing, and
their size and outcomes of the conflict (for the intervener and for the
host nation). Our review reveals that although the existing work is
both extensive and diverse, the focus of the most-rigorous empirical
work has been on the intervention decision: specifically, the effect of
interventions on conflict outcomes and, to a lesser extent, the interests
of the intervener. The effects of intervention size have also been stud-
ied, with this work focusing primarily on the relationship with conflict
outcomes. There has been limited attention paid to the implications of
different intervention timings.
The other key observation that emerges from our review of the lit-
erature is just how conditional the relationships between interventions
(their timing and size) and relevant outcomes appear to be. In most
cases, the direction and strength of the relationship depends on other
mediating factors, such as the domestic context of the host nation, the
side in the conflict that the intervention supports, and the presence of
an international coalition, among others. Understanding the role of
these mediating variables is important to our approach in the remain-
ing chapters in this report. In the next chapter, we investigate quantita-
tive patterns in our database of intervention and potential intervention
cases.
CHAPTER THREE
Quantitative Analysis: Studying Trends and

Patterns
To study the trade-offs between intervening and not intervening, inter-

vening early and intervening late, and intervening with larger or smaller
forces, it is useful to investigate general trends in the U.S. experience
since 1946 to consider whether there have been systematic differences
between different types of interventions. In this chapter, we describe
how we built a data set of intervention and potential intervention cases
and our approach to coding and categorizing these cases. Then we
compare conflict outcomes, outcomes for the United States, and out-
comes for the host nation following noninterventions and interventions
of different sizes and timings. Although our data set is comprehensive,
there remain a relatively small number of total cases from which it is
difficult to draw definitive conclusions. Instead, we use the descriptive
statistics presented in this chapter to explore trends and patterns. The
case study–based methodologies we employ in Chapters Four and Five
will be better able to isolate contextual factors and clarify any potential
causal relationships.
Data Set of Intervention and Nonintervention Cases
To facilitate this analysis, we needed to construct a data set of cases

that included not only historical U.S. military interventions but also
potential cases of U.S. intervention that did not result in an interven-
tion. For this data set, we needed to identify a set of cases in which U.S.
39
interests were plausibly sufficient to justify a U.S. intervention, regard-

less of whether one actually took place. Our approach was to construct
a data set of cases made up of wars and crises: specifically, those wars
and crises since 1946 for which the United States had a strong national
interest. We drew the cases from two primary sources. To identify rele-
vant wars, we used the Uppsala Conflict Data Project (UCDP) Armed
Conflict Dataset;1 for crises, we used the International Crisis Behavior
(ICB) data set.2 The UCDP data include major wars, such the Vietnam
War and the Korean War, and the ICB data include such events as the
Berlin Airlift Crisis of 1948 and other lower-level confrontations. Each
of these data sets considers each instance of a crisis or war as a separate
event. Neither combines recurring crises or wars into a single series. We
follow this coding.
To define those wars and crises in which the United States has
a strong national interest, we included four types of cases: (1) those
that posed a direct threat to the United States (e.g., the September 11,
2001, attacks or the Cuban Missile Crisis), (2) those that posed a direct
threat to a U.S. treaty ally, such as the North Atlantic Treaty Organi-
zation [NATO], Korea, or Japan (e.g., the Korean War), (3) those that
involved a security threat to the countries involved and occurred in an
area of high strategic importance to the United States (specifically, Cen-
tral America and the Caribbean, Europe, the Middle East and North
1 The UCDP Armed Conflict data include conflicts and wars. We include only wars
(annual battle deaths over 1,000) in our data. Therése Pettersson and Kristine Eck, “Orga-
nized Violence, 1989–2017,” Journal of Peace Research, Vol. 55, No. 4, 2018; Nils Petter
Gleditsch, Peter Wallensteen, Mikael Eriksson, Margareta Sollenberg, and Håvard Strand,
“Armed Conflict 1946–2001: A New Dataset,” Journal of Peace Research, Vol. 39, No. 5,
2002.
2 To be included in the ICB data, an interstate interaction must meet three criteria: “a
threat to one or more basic values, along with an awareness of finite time for response to the
value threat, and a heightened probability of involvement in military hostilities.” Michael
Brecher and Jonathan Wilkenfeld, A Study of Crisis, Ann Arbor, Mich.: University of Michi-
gan Press, 2000, p. 3.; Michael Brecher, Jonathan Wilkenfeld, Kyle Beardsley, Patrick James,
and David Quinn, International Crisis Behavior Data Codebook, Version 12, 2017.
Quantitative Analysis: Studying Trends and Patterns 41
Africa, and East and Southeast Asia), such as the 1991 Gulf War, and
(4) severe humanitarian crises (e.g., the 1994 Rwandan genocide).3
We made the decision to limit our data set to only those wars and
crises in which the United States had a strong national interest in order
to remove those cases in which there was simply no realistic possibil-
ity that the United States would intervene. It is possible that we elimi-
nated one or more crises or wars in which the United States intervened
or might have considered intervening in this process of narrowing the
universe of cases. However, the loss of any such case is likely balanced
out by the benefits that we get from eliminating the larger number of
cases that were not plausible intervention opportunities. We could also
have chosen a more restrictive set of inclusion criteria that would have
further narrowed the data. However, the more restrictive the inclusion
criteria, the higher the chance that we begin excluding cases in which
the United States could realistically have intervened. Furthermore, U.S.
intervention decisions are very idiosyncratic, often the product of a
confluence of events, personal interests, national interests, and timing,
so criteria that are too narrow might have resulted in a sample that was
not only missing meaningful nonintervention cases but also missing
intervention cases. Such a set of cases would then provide a misleading
and incomplete account of the trade-offs between intervening and not
intervening and the factors that influence this decision and its effects
on outcomes. After our attempt to balance these concerns, we were left
with a data set of 286 cases, including 130 crises and 156 wars.
Coding Interventions
With this data set of cases defined, the next step was to identify which
of the 286 cases had actually experienced a U.S. military interven-
tion. To do this, we relied on a database of such interventions that was
compiled and refined during a series of RAND Arroyo Center proj-
ects: the RAND U.S. Military Intervention Dataset (RUMID). The
3 We included large-scale humanitarian crises, defined as having greater than roughly

100,000 civilian deaths. It is worth noting that large refugee flows constituted an additional
potential motivation for U.S. intervention. However, we were unable to reliably identify
refugee data that would allow us to systematically assess whether cases should be included in
our universe on these grounds.
data include entries for each U.S. military intervention since 1946 that
meets our size threshold.4 We relied on the following size thresholds to
identify military interventions for each domain:
• Ground forces: To qualify as an intervention on the basis of the

ground forces involved, we required at least 100 person-years of
presence in another country, for each year during which the inter-
vention was ongoing. This requirement could be met by 100 sol-
diers for a full year, 200 soldiers for six months, etc., but not by 10
soldiers present for 10 years (that is, the size criteria must be met
in each year of the intervention).
• Naval forces: To qualify as an intervention on the basis of its naval
forces, we required the presence of a carrier strike group in the
post-1945 era, or an equivalent size force in the pre-1945 era (when
carriers were either less prevalent or did not yet exist).
• Air forces: To qualify as an intervention on the basis of the air
forces involved, we required either roughly a wing-year of aircraft
(roughly 80 planes) or an incidence of air-to-air or air-to-ground
combat or strikes, which are relatively rare but we believe are
always worthy of assessment.
For each intervention, the data capture information on duration,

number of forces, types of forces, primary activity types, political
objectives, and whether those objectives are achieved.5
To determine which of our 286 cases had experienced a U.S.
intervention, we matched the case universe to the intervention data,
using location, years, and (where needed) a description of each conflict,
crisis, and intervention. Our matching process identified three key
groups of cases. First, in 50 of the 286 cases, there were no substantial
U.S. forces present in the country, and the United States then initiated
a new intervention at some point in response to a war or crisis. Second,
in 64 other cases, the United States already had forces present in the
4 The data cover the years 1898 to 2016, but we use only the post-1945 data for this analy-
sis. As a reminder, we do not include covert missions or any special forces missions because
these are typically classified.
5 For more information on this data set, see Kavanagh, Frederick, Stark, et al., 2019.
country in question before the war or crisis started, often as part of an

ongoing deterrent force or as part of a previous intervention of some
kind. These forces are usually present because there was the expecta-
tion of such a crisis or conflict or because there had been some conflict
or crisis in the past. We treat these “forces already present” cases sepa-
rately from the new interventions launched in response to a crisis (that
is, no new intervention could be coded in cases where substantial U.S.
forces were already present). The focus of most of our analysis through-
out this report is on the new interventions and comparing them with
the noninterventions, since this is the key decision point of greatest
interest for this report. Finally, there are the noninterventions, or cases
in which the United States did not intervene or have forces already
present, of which we identify 172.
Categorizing Cases According to Timing and Size

As noted in previous chapters, we are interested in three key interven-
tion decisions: the decision to intervene, the timing of the intervention,
and the size of the intervention. These decisions highlight eight pos-
sible analytical categories of interest into which we sorted our cases:
• no intervention
• early, small intervention
• early, medium intervention
• early, large intervention
• late, small intervention
• late, medium intervention
• late, large intervention
• forces already present.6
6 These include cases where the United States already had troops in a location before the
conflict or crisis in that location broke out. The precise coding of these cases is situation
specific. In order to fall into this category, the United States had to have already forward
deployed U.S. forces to a specific region or country for the specific intention of countering
or deterring a given threat or security challenge. In some cases, this means U.S. forces were
deployed to the country in question to deter against internal threats. In others, U.S. forces
were deployed to a “threatened” country to deter an adversary. Forces deployed or stationed
at home that were not specifically tasked with countering a given threat are not counted as
already present.
We categorized our 286 cases into this taxonomy according to the

timing and size of the intervention. For intervention timing, we coded
interventions as either early or late. We set the threshold between early
and late interventions at six months, meaning that all interventions
that occurred within the first six months after the start of the crisis or
war are considered early and all others are considered late. This thresh-
old was determined by reading each of the intervention narratives to
determine what constitutes a policymaker decision to intervene that
could plausibly be understood to have been taken at the first oppor-
tunity or without substantial delay. Through this reading, we assessed
that, for most conflicts and crises, this early phase (without substantial
policymaker delay or hesitation) lasted about six months. After that
point, policymakers generally appeared to have passed on an initial
decision to intervene, such that subsequent interventions reflected a
reconsideration of a prior decision not to become involved. Using this
assessment, we coded early interventions as all those occurring within
the first six months. In practice, this means that there are no late inter-
ventions into crises, since crises typically last little more than a year.7
Within the 286 cases, we identified five that are ongoing as of the end
of 2018. In our analyses below, we exclude those that are ongoing when
comparing the changes over time in our intervention and noninterven-
tion cases because we cannot assess the postconflict status of an ongo-
ing case, for example. There are also six cases that ended less than five
years ago. These are also excluded from the comparative analysis.
To code the size of the U.S. military interventions, we used three
categories: small, medium, and large. When coding intervention size,
we took ground, air, and naval forces into account, but we weighted
ground forces more heavily than air and naval forces because of the
commitment implied. When considering only ground forces, small
interventions are those with fewer than 3,000 ground forces (about
the size of a brigade), medium interventions include interventions of
3,000 to 20,000 troops (roughly the size of a division), and large inter-
7 We code the intervention as starting at the time of deployment and follow coding the
ICB and COW datasets to determine timing and phasing of conflict.
ventions include anything involving more than 20,000 troops.8 For

independent air and naval interventions, we considered both the abso-
lute size of forces committed (number of carriers or other naval assets,
number of sorties or aircraft) as well as the relative size of forces com-
mitted compared with the total force. This relative size measure was an
effort to address the fact that what was a large commitment of naval
and air assets in the late 1940s might seem small in absolute terms by
the standards of the late 1980s. Intervention-size coding used the “typ-
ical” force size, defined as the size of the force for the majority of the
intervention (rather than the peak or minimum values). This defini-
tion means that if an intervention started off small but grew to 25,000
troops in the third year and remained at that size for the next seven
years, it would be coded as a large intervention, since the typical force
size was above the 20,000-troop threshold.
More challenging was how to code the size of interventions that
combine ground, air, and naval assets. Although assessing an aggregate
size across components necessarily requires an element of subjective
judgment, we aimed to capture in our assessment the aggregate level
of military effort or commitment to the intervention on the part of
the United States. As noted above, our criteria focused first on ground
personnel. Although the contributions of air and naval assets to U.S.
military interventions should not be minimized, it is also worth noting
that ground forces inherently involve and imply a greater commitment
because of the associated time, costs, and risk. We therefore required
a substantial commitment of air or naval forces before increasing an
intervention’s initial size categorization (which was made using its
ground component) from small to medium or medium to large. As an
example, an intervention with 15,000 ground troops and naval and air
commitments that were coded as medium in size independently would
remain a medium intervention, but a 15,000-troop intervention with
large air and naval components could increase to a large intervention.
As another example, consider the Cuban Missile Crisis. In our data, the
Cuban Missile Crisis involved about 2,000 ground troops, with very
8 We do not count naval or air personnel involved unless they are also active in ground
activities.
large air and naval components. We code this intervention as medium,

upgrading it from the small categorization that it would otherwise have
had because of its ground forces. In contrast, the 1992 intervention in
Bosnia involved about 4,500 ground troops and large air and naval
components, but because the air and naval forces involved represented
a smaller proportion of overall U.S. capabilities than in Cuba, this
intervention is coded as medium—no size upgrade occurs.9
Table 3.1 shows the number of cases that fall into each of the eight
categories and provides examples. As noted above, the most common
outcome across the universe of cases is that there is no U.S. interven-
tion. About 20 percent of all cases occur in places where the United
States already had forces. Of the six intervention categories, interven-
tions that occur early and are small are the most common. Late inter-
ventions are rare (there are only 12 out of 50 total), and there are no
large, late interventions. Table 3.2 reveals similar trends when we break
9 These size categories are obviously still broad and might group together interventions
that are quite different. For example, although our taxonomy will group interventions that
occur eight months after the conflict or crisis start with those that occur five years later, these
interventions could be different in important ways. Similarly, an intervention that involves
25,000 troops could be very different than one that involves 200,000. However, we decided
that a simpler categorization scheme, even with these inherently arbitrary break points, was
still preferable. In the case of size, commitments of 25,000 and 200,000 troops are signifi-
cant and meaningful commitments of forces that are more similar to each other than either
is to smaller commitments, which might represent a more-limited level of engagement or
interest by the United States. Similarly, interventions that include 100 troops and those that
include 2,990 troops are likely to be very different, but both are commitments that the Army
can more easily absorb without experiencing significant pressure on readiness. Again, both
are likely to be more similar to each other than to interventions of substantially larger sizes.
For the timing, a similar argument applies. An intervention that occurs immediately after a
crisis could be different than one that starts three months later, just as an intervention that
starts at month eight will be different from one that starts five years in. However, both sets
of cases are likely to share important similarities. The United States has undertaken very few
late interventions and has never undertaken a late intervention into a crisis. If the country is
going to intervene early, the intervention typically happens quickly. Interventions that occur
after six months are alike in that something prevented them from happening more immedi-
ately, whether it was a disagreement, an uncertainty, or simply a change in perceptions of the
importance of the cases for the United States. Of course, there will still be differences within
each category of cases that our quantitative analysis will not be able to capture. We will use
the focused and counterfactual case studies in Chapters Four and Five to help capture greater
nuance along these dimensions.
Table 3.1
Intervention Categories and Examples
Category Number of Cases Examples
No intervention 172 Hungary, 1956
Early, small 27 Libya, 2011
Early, medium 6 Grenada, 1983
Early, large 5 Kuwait, 1990
Late, small 7 Libya, 2014
Late, medium 5 Bosnia, 1992
Late, large — —
Forces already present 64 North Korea, 1994
Table 3.2
War and Crisis Interventions, by Timing and Size
Category Wars Crises
No intervention 103 69
Early, small 11 16
Early, medium 1 5
Early, large 1 4
Late, small 7 —
Late, medium 5 —
Late, large — —
Forces already present 28 36
the universe of cases into crises and wars. Early and small interventions
are still the most likely. There are nine interventions into crises that are
not small and none that are late. Late interventions do occur into wars
but slightly less frequently than those that occur early in the conflict.
Exploring Outcomes by Intervention Timing and Size
Before moving on to a discussion of how intervention outcomes vary

across these categories, it is worth pausing to consider the implications
of selection effects for this analysis and how selection effects might
affect our ability to identify cross-category outcome differences using
this quantitative approach. By selection effects we mean that some of the
same factors that shape the outcome of the intervention—specifically,
U.S. interests or the severity of the war or crisis—might also shape the
timing and the size of that intervention, so it will be difficult to dis-
entangle the effects of timing and size and those of U.S. interests. For
example, we might find that early, large interventions are associated
with the best postintervention outcomes for the host nation, but these
might also be the cases in which the United States perceives the high-
est stakes and so commits the most effort and resources. As another
example, we might find that large interventions have generally worse
outcomes, but they might be large interventions precisely because they
respond to more-severe wars or crises. A more sophisticated statistical
analysis could help to control for these selection effects, but we are
prevented from conducting such an analysis by the small number of
interventions in many of our timing and sizing categories. With only
five late, medium interventions, for instance, any attempt to use a more
sophisticated quantitative modeling approach would not yield mean-
ingful results.
Disentangling trends and patterns driven by intervention timing
and size and those driven by selection effects (especially given the small
number of cases that we have) will be challenging, so we will need
to interpret our results with caution. For this reason, we consider the
quantitative results in this chapter to be suggestive and useful in devel-
oping expectations for qualitative investigations (rather than using
them as research findings). In the discussion below, we describe pat-
terns and present hypotheses. We do not assess the statistical signifi-
cance of observed differences. We then build from the trends and pat-
terns observed here for our focused and counterfactual case studies to
help isolate meaningful patterns from selection effects.
Outcome Measures
In Chapter Two, we outlined three types of outcomes that are of inter-
est when considering the trade-offs between different intervention
decisions. In this section, we will use these same three groups of out-
comes: effects on the conflict, effects on the interests of the intervener,
and effects on the host nation. We consider what patterns we can dis-
cern from our quantitative comparison, how they match to the results
described in Chapter Two, and what expectations they set for our qual-
itative analysis.
First, we consider several conflict outcome measures:10
• Conflict duration: We measure conflict duration as the time

from the start to the end of the conflict. Ongoing conflicts do
not have a duration and therefore are omitted.
• Changes in conflict intensity: We measure conflict intensity
using an annualized measure of the battle deaths that occurred
on all sides of the conflict. We consider annual battle deaths in
the first year of the war or crisis, the last year of the war or crisis,
one year after the war or crisis ends, and five years after the war
or crisis ends. The battle death data we use comes from UCDP.11
• War or crisis recurrence: We consider the likelihood that the
war or crisis recurs, using the UCDP and ICB data sets.12 We
code for any recurrence, including recurrence of a war or crisis, a
10 We do not consider whether the United States or the side the United States supported
“won” in this analysis. First, in many cases, the United States did not intervene on one or the
other side but instead intervened to monitor the situation and deter the Soviets. Any assess-
ment of win/loss ratios would have to exclude those cases. This issue and the issue of which
wars and crises tend to escalate and which do not are covered in greater depth in Bryan Fred-
erick, Stephen Watts, Matthew Lane, Abby Doll, Ashley L. Rhoades, and Meagan L. Smith,
Understanding the Deterrent Impact of U.S. Overseas Forces, Santa Monica, Calif.: RAND
Corporation, RR-2533-A, 2020; and Watts, Johnston, et al., 2017.
11 Pettersson and Eck, 2018. The battle-death data is, however, affected by both missing
observations and uncertainty about included totals. Other measures of conflict intensity are
affected by similar limitations. We use this measure but note the uncertainty attached to
given estimates.
12 Brecher and Wilkenfeld, 2000; Brecher et al., 2017; Pettersson and Eck, 2018; Gleditsch
et al., 2002.
crisis that later recurs as war, or a war that later recurs in a crisis
phase.
Second, we consider outcomes that affect the interests of the

intervening state:
• Percentage of objectives successfully achieved: We consider the

percentage of total U.S. political objectives in each intervention
category that are successfully achieved according to past RAND
analysis of U.S. military interventions. The analysis uses coding
drawn from RUMID, which includes all political objectives and
the degree of success achieved.13 We include ongoing interven-
tions, since some have achieved some portion of their objectives
already.
• Military and economic assistance: We consider changes in the
percentage of total military and economic assistance allocated to
the host nation between the first year of the war or crisis and five
years after the end of the war or crisis. We use this as one mea-
sure of the longer-term economic and strategic implications of the
intervention for the intervening state.14 One consequence of an
intervention could be a long-term military and economic assis-
tance cost for the United States. Another possibility is that mili-
tary and economic assistance function in place of (or, conversely,
alongside) military force in certain contexts.
13 Political objectives are coded for success on a three-point scale. We only consider full
success here. The process for determining and coding both political objectives and degrees
of success can be found in Kavanagh, Frederick, Stark, et al., 2019. In that work, the authors
identified all political objectives for each intervention and then coded the degree of success.
They included multiple coders to increase robustness of the coding. Interventions could have
many objectives, and success for each was coded independently. For example, in the case of
the U.S. intervention in Afghanistan in 2001, the United States was successful in defeating
the Taliban but only partially successful at establishing stable governing institutions.
14 Compiled from U.S. Agency for International Development, US. Overseas Loans and
Grants, Obligations and Loan Authorizations, July 1, 1945–September 30, 2014, Washington,
D.C., 2015; and U.S. Agency for International Development, “The Greenbook,” webpage,
updated April 29, 2013.
Finally, we consider outcomes for the host nation:15
• Changes in polity scores: We consider changes in the polity

scores, a measure of the degree of democracy, in the host nation.
Using Polity IV data, we compare the values between the first
year of the war or crisis and five years after the end of the war or
crisis for the host nation.16
• Changes in gross domestic product (GDP) per capita: We con-
sider changes in GDP per capita between the first year of the war
or crisis and five years after the end of the war or crisis for the host
nation using Maddison GDP per capita.17 GDP per capita can
be a measure of economic well-being and is sometimes used as a
proxy measurement for the state’s institutional capacity.
• Changes in infant mortality: We consider changes in infant
mortality as a measure of social well-being and the country’s devel-
opment, comparing between the first year of the war or crisis and
five years after the end of the war or crisis for the host nation.18
15 Host nation is defined as “where the forces are located.” For naval and air deployments, we
consider the “host” nation to be the country whose airspace or nearby waters contains those
forces. Note that the Cuban Missile Crisis is unusual in that U.S. forces are coded as being
in Cuba because of their presence at Guantanamo, even though a large portion of forces were
deployed in U.S. waters or airspace.
It is also important to note that the intention of the United States regarding the host
nation varies in the cases in our data set. The United States was, for example, much more
motivated to improve the trajectory of Afghanistan after 2001 than it was to improve the tra-
jectory of Cuba after the Cuban Missile Crisis. This variation makes it difficult to interpret
the patterns in the data for host-nation trajectory. We discuss the implications of this when
we present those results. Although we are better able to account for such differences in our
focused case studies detailed in Chapter Four, this heterogeneity in U.S. goals highlights a
limitation of the quantitative analysis.
16 Monty G. Marshall, Ted Robert Gurr, and Keith Jaggers, Polity IV: “Political Regime
Characteristics and Transitions, 1800–2017,” Vienna, Va.: Center for Systemic Peace, 2018.
17 GDP per capita data from Angus Maddison, homepage, updated September 3, 2008.
Jutta Bolt, Robert Inklaar, Herman de Jong, and Jan Luiten van Zanden, “Rebasing ‘Mad-
dison’: New Income Comparisons and the Shape of Long-Run Economic Development,”
Maddison Project Database, Version 2018, Working Paper No. 10, 2018.
18 World Bank, “Mortality Rate, Infant (Per 1,000 Live Births), 1960–2017,” online data-
base, undated.
These data come from the World Bank and start in 1960, so we
are not able to use this metric for all cases.
The different indicators provide relatively rough operationaliza-

tions of the different outcome measures described in Chapter Two.
However, there are some outcomes of interest that we do not have
quantitative metrics for at all and therefore cannot assess. In particular,
the missing metrics affect our ability to consider the specific economic,
domestic political, and strategic and reputational costs and benefits for
the intervening state (we do have one economic measure, but it cap-
tures only a piece of the cost). With appropriate existing data sources
not available, the process of developing new, reasonable operationaliza-
tions of these types of outcomes for both intervention and noninter-
vention cases unfortunately lies outside the scope of the present effort.
Instead, we use the short, focused case studies and the longer counter-
factual cases summarized in Chapters Four and Five to provide insight
into these factors.
Comparisons
Conflict Outcomes
When considering conflict outcomes, we focus on conflict duration,
conflict intensity, and conflict recurrence. We first consider conflict
duration. Table 3.3 shows the duration of wars by intervention cat-
egory. Most wars are relatively short; as expected, the vast majority
of all wars in our data set last less than five years, with more than
half lasting less than two years. However, even in this context, U.S.
interventions appear to be associated with shorter wars. All wars with
early U.S. interventions last five years or less, though there are several
longer wars with late, small interventions. There are also several longer
wars in the “forces already present” category. However, selection effects
might be relevant to this association between U.S. interventions and
shorter wars. It is possible that the trend observed in Table 3.3 results,
for example, from an unwillingness by the United States to become
involved in wars that appear too messy, too difficult, or likely to last
more than a few years. Late interventions into wars, in contrast, are
somewhat more likely to be associated with longer wars, but that may
Table 3.3
Conflict Duration by Intervention Type for War Cases
Less than
Category 2 Years 2–5 Years 6–10 Years 10-Plus Years
No intervention 49 35 13 4
New U.S. 14 3 4 1
intervention
Early, small 10 1 — —
Early, medium 1 — — —
Early, large — 1 — —
Late, small 2 — 3 1
Late, medium 1 1 1 —
Late, large — — — —
Forces already 53 7 3 1
present
NOTE: Excludes ongoing crises and wars.
simply be because, in many cases, the war could have already been
lengthy by the time the United States became involved.
In crisis cases (shown in Table 3.4), about two-thirds of crisis
interventions occur in cases that resolve within less than one year, with
the rest occurring in crises that last roughly one full year or, in rela-
tively few cases, two years. For nonintervention cases, about 85 percent
resolve in less than one year, suggesting that new interventions into
crises do not seem to be associated with shorter crises than what occurs
in the absence of interventions. However, it is important to note that
the United States does not always withdraw forces at the end of a crisis,
even if the initial crisis resolves. In many cases, including Afghani-
stan 2001, the United States remains for a lengthy postcrisis period of
stabilization and other activities, which can last anywhere from a few
months to many years. The duration of the initial conflict or crisis,
then, does not necessarily tell us much about the cost or implications
of that intervention for the United States.
Table 3.4
Conflict Duration by Intervention Type for Crisis Cases
Category Less than 1 Year 1 Year 2 Years
No intervention 58 10 1
New U.S. 17 6 2
intervention
Early, small 11 3 2
Early, medium 5 — —
Early, large 1 3 —
Late, small — — —
Late, large — — —
Forces already 24 11 1
present
Overall, U.S. interventions might be associated with shorter wars

(particularly when those interventions occur early in the conflict) but
not necessarily with shorter crises. Although it is difficult to disentan-
gle these relationships from the selection effects that might be driving
them, trends in duration suggest at least one other interesting observa-
tion. Although recent experience has been dominated by several long
conflicts that have required extended commitments, U.S. experiences
with interventions since 1946 have predominantly included shorter
wars and crises.
The second conflict outcome metric that we consider is conflict
intensity, measured by annual battle deaths. We consider both the
number of battle deaths in the first year of the war or crisis and the
percentage change in battle deaths between the first year of the war or
crisis and five years after the end of the war or crisis (Table 3.5).19 The
first observation is that, with the exception of the late, small category
(driven by the relatively marginal U.S. intervention in the 1971 war in
19 All five-year postconflict comparisons in this chapter are calculated from the end year of
the crisis or conflict.
Table 3.5
Change in Average Annual Battle Deaths, by Intervention Category
Initial Average Percentage Change

Percentage of Battle Deaths in Battle Deaths
Number Cases That Are (First Year of (First Year to Five
Category of Cases Wars Conflict) Years Postconflict)a
No intervention 170 59% 5,744 −20%
New U.S. 47 50% 8,612 −36%

intervention
Early, small 27 41% 4,447 +21%
Early medium 6 17% 2,830 −68%
Early, large 5 20% 4,927 −99%
Late, small 6 100% 44,019 −100%
Late, medium 3 100% 4,793 −100%
Forces already 64 44% 5,589 +200%

present
a Excludes ongoing crises and wars and cases that ended less than five years ago.
Bangladesh), intervention and nonintervention cases look somewhat

similar in terms of their average battle deaths in the first year of the war
or crisis, with intervention cases being higher but not dramatically so.
Second, our exploration suggests that U.S. interventions might be asso-
ciated with larger decreases in battle deaths than either noninterven-
tions or cases in which U.S. forces were already present. The observed
decrease in battle deaths occurs across the size and timing categories
and is largest for late interventions in percentage terms.
The early, small category appears to be an exception, with an
increase in average battle deaths over time. Digging deeper, how-
ever, reveals that this increase is driven only by one case, in which
the increase in battle deaths is largely unrelated to U.S. activity: the
early, small intervention in response to the Iran hostage crisis. The
intervention had numerous problems, but it was also unrelated to the
subsequent initiation of the Iran-Iraq War, which is what produced the
spike in battle deaths in Iran.20 If we drop that single observation, the

decrease in fatalities for the early, small category is about 60 percent.
Notably, the results in terms of changes in conflict intensity are largely
similar across wars and crises once we address this one anomaly. Still,
the size of the reduction in conflict intensity appears to be greater for
late interventions than for early ones. The smaller number of late inter-
ventions means that we must interpret these patterns with caution. The
other result in this table that deserves explanation is the large increase
in battle deaths for the forces already present category. This, too, is
driven by only one observation: the conflict in Waziristan in 2004,
which experienced a rapid increase in battle deaths between 2004 and
five years after the end of the conflict because of regional escalation
in fighting between tribal groups and government forces. Dropping
that observation lowers the increase in battle deaths to only 15 percent.
Thus, this is a case in which estimates of changes over time could be
heavily affected by individual outlier cases; as a result, we should take
all trends in Table 3.5 with a large degree of caution.
As noted, even the patterns we see here are suggestive only. Selec-
tion effects are likely to complicate their interpretation. The United
States likely chooses to begin new interventions in cases where poli-
cymakers believe that the intervention can have a significant influ-
ence. They might avoid those cases where hope for conflict resolution
seems small or where the costs to bring it about seem extraordinarily
high. Nonintervention cases, therefore, could be especially likely to see
continuing increases in conflict intensity, but these increases (if antici-
pated) could be a reason for U.S. nonintervention rather than being
caused by it. On the other hand, the United States might also position
longer-term forces in locations where the threat of conflict and escala-
tion is greatest. This could explain why conflict intensity appears to
increase for “forces already present” cases in our analysis.21
The final conflict outcome measure we consider is war or crisis
recurrence. If one goal of U.S. military interventions is to promote sta-
20 We do not see strong effects by other outliers in our other intervention categories.
21 Further investigation reveals that this increase occurs only in war cases, and that average
annual battle deaths falls in crisis cases where forces were already present.
bility and prevent crisis and conflict recurrence, then we would hope
to see a clear association between U.S. interventions and lower rates
of conflict recurrence.22 However, Table 3.6 suggests little difference
at the aggregate level in the recurrence rates for cases in which the
United States does or does not intervene. Wars appear to recur about
85 percent of the time, while crises recur about 60 percent of the time,
regardless of whether there was a U.S. intervention. There are a few
differences when we consider individual intervention categories, but
the small number of cases and potential selection effects make it dif-
ficult to draw any real conclusions. In our universe for both wars and
crises, recurrence seems lower for early interventions regardless of size
(though the larger categories have few interventions). However, it is
worth underscoring that recurrence rates following U.S. interventions,
Table 3.6
Conflict Recurrence, by Intervention Category
Number of Recurrence Number of Recurrence

Category Wars Rate Crises Rate
No intervention 101 86% 69 58%
New U.S. 22 86% 25 56%

intervention
Early, small 11 90% 16 50%
Early, medium 1 100% 5 60%
Early, large 1 100% 4 75%
Late, small 6 67% — —
Late, medium 3 100% — —
Forces already 28 92% 36 22%

present
22 As already noted, even where others disagree, interventions always respond to the percep-
tion of a threat or crisis, even if the crisis or threat later proves to be nonexistent.
on average, remain as high as recurrence rates following noninterven-

tion wars and crises.
Taken together, these results provide little evidence that new U.S.
interventions have any notable effect on conflict recurrence, regard-
less of the characteristics of that intervention. The differences across
categories are small, and the number of cases on which the pattern
depends are limited. Interestingly, conflict seems substantially less
likely to recur in crises for which the United States already has forces
present. The relationship does not, however, extend to wars.
Next, we consider the effects of the intervention, its timing, and size
decisions on the intervening state (the United States). We focus on two
metrics: (1) the achievement of political objectives and (2) the medium-
term implications for military and economic assistance spending.
Table 3.7 shows the rate at which the United States achieves its politi-
cal objectives in interventions with different sizes and timings. This
Table 3.7
Percentage of Objectives Successfully Achieved, by Intervention Category
Percentage Percentage
of Objectives of Objectives
Number of Successfully Number of Successfully
Category Wars Achieved Crises Achieved
No Intervention 103 — 69 —
New U.S. 25 54% 25 54%

intervention
Early, small 11 72% 16 44%
Early, medium 1 100% 5 83%
Early, large 1 60% 4 62%
Late, small 7 38% — —
Late, medium 5 25% — —
Forces already 28 53% 36 82%

present
rate of success uses prior RAND research that identified the politi-
cal objectives of U.S. military interventions and coded the degree to
which they were achieved, according to in-depth research.23 These data
do not contain information on the political objectives of noninterven-
tions; at this point, we are not able to compare this outcome across
the intervention decision point. We will conduct an assessment of the
extent to which the United States achieves its objectives with a smaller
set of nonintervention cases in Chapter Four.
Table 3.7 suggests that rates of success are higher, in general, for
early interventions when compared with late interventions. This result
also seems consistent with previous research described in Chapter Two
that suggests that early interventions tend to be more likely to advance
the interests of the intervener. This result could be because interveners
have the most flexibility during the conflict’s early phases, before both
sides dig in and compromise becomes harder. The effects of size appear
to be somewhat more mixed, with no consistent patterns emerging.
Selection effects could also play a role in explaining these pat-
terns. It is possible that the United States intervenes early and contrib-
utes the most resources (not just forces, but also equipment, funding,
and support) to cases in which more-important interests are at stake.
This additional commitment, beyond what we can measure here, might
explain the higher success rate as much as the timing of the interven-
tion itself. We further explore the possible dynamics driving the higher
success rates of early interventions in Chapter Four. It is also worth
emphasizing that our analysis does not allow us to assess whether the
United States achieves a higher rate of success on intervention objec-
tives when it intervenes than when it does not. We will explore this
relationship qualitatively in Chapters Four and Five.
Ideally, we would use proxy metrics in addition to the level of suc-
cess to compare the strategic and reputational costs, domestic political
costs, and economic costs and benefits of the intervention borne by the
United States across categories. Practically, however, we lack the data
sources to comprehensively conduct such an investigation. Instead, we
23 See Kavanagh, Frederick, Stark, et al., 2019.

assess these dynamics in the short, focused case studies in Chapter Four
and the longer counterfactual case studies in Chapter Five.
However, one aspect of the costs and benefits to the intervener
for which we do have a relevant data source is the amount of U.S.
economic and military assistance spending. Increased levels of U.S.
assistance reflect, in part, an increasing economic burden, but they
might also reflect a degree of success in establishing a partner receptive
to receiving this assistance. Although they represent an economic cost,
increases in military or economic assistance could also reflect a strate-
gic benefit for the United States.
To assess changes in assistance, we first calculated the percent-
age of total military and economic assistance allocated to each country
during the first year of the conflict. We did the same for the five-year
postconflict period. Then we considered the change between the two.24
Table 3.8 shows the change in economic assistance and Table 3.9 shows
the change in military assistance. The two tables suggest a correlation
between U.S. intervention decisions and future levels of economic and
military assistance, but we believe that these patterns might not be
that meaningful as a reflection of advancing U.S. interests. Table 3.8
suggests a modest correlation between U.S. intervention (early U.S.
interventions in particular) and future increases in economic assis-
tance. Table 3.9 suggests a notably stronger correlation between U.S.
intervention (again, early U.S. interventions in particular) and future
increases in military assistance. It is also worth noting that we reran the
results of Tables 3.8 and 3.9 excluding two high-profile cases that were
known to involve very high levels of U.S. assistance (U.S. interventions
in Iraq post-2003 and Afghanistan post-2001) to see whether these
patterns were driven by outlier observations; although the increases in
military assistance correlated with U.S. interventions were somewhat
more modest, they were not dramatically different from those reported
in Tables 3.8 and 3.9.
24 We use this particular metric to deal with the dramatic year-over-year fluctuations in
military and economic assistance spending. The “noise” in these data makes it difficult to
identify meaningful trends. Furthermore, the size of the swings in certain countries and the
overall distribution of the data mean that outliers can significantly influence overall trends.
Table 3.8
Change in Economic Assistance to Host Nation, by Intervention Category
Initial Economic Economic Assistance

Assistance to to Host Country
Percentage of Host Country as as a Percentage of
Number of Cases That Are a Percentage of Total, Five Years
Category Cases Wars Total Postconflict
No intervention 170 59% 1.4% 2%
New U.S. 47 50% 1.5% 3%

intervention
Early, small 27 41% 1.3% 4%
Early, medium 6 17% 0.6% 2%
Early, large 5 20% 0.6% 4%
Late, small 6 100% 1.4% 1%
Late, medium 3 100% 0.9% 1%
Forces already 64 44% 1.9% 3%

present
On the one hand, it is perhaps not surprising that U.S. assistance

to a country would substantially increase in the aftermath of a U.S.
military intervention into that country. In many cases, U.S. forces
could still be present in the country five years after the end of the con-
flict that prompted the intervention, and the United States frequently
undertakes substantial capacity-building and other efforts with the
host nation as part of its postconflict stabilization efforts. Moreover,
a recent U.S. intervention and the possible continued presence of U.S.
troops could represent strong incentives for a country to accept U.S.
assistance and agree to work with the United States more broadly.
Iraq post-2003 represents a notable example, in which the pre-2003
regime under Saddam Hussein was swept away and replaced by one
that was willing to accept U.S. assistance and was at least relatively
more closely aligned with U.S. policy goals. Therefore, the possibility
Table 3.9
Change in Military Assistance to Host Nation, by Intervention Category
Initial Military Military Assistance

Assistance to to Host Country
Percentage of Host Country as as a Percentage of
Number of Cases That Are a Percentage of Total, Five Years
Category Cases Wars Total Postconflict
No intervention 170 59% 1.2% 1.7%
New U.S. 47 50% 1.1% 16%

intervention
Early, small 27 41% 1.9% 24%
Early, medium 6 17% 0.04% 6%
Early, large 5 20% 0.03% 7%
Late, small 6 100% 0.0% 0.02%
Late, medium 3 100% 0.3% 0.03%
Forces already 64 44% 3.8% 5%

present
that U.S. intervention could increase the willingness of states to accept

U.S. assistance seems plausible.
Complicating this assessment, however, is the fact that U.S. interest
in providing military and economic assistance is also likely to be trans-
formed by the intervention decision. Leaving aside the difficulties the
United States had with pre-intervention regimes in such places as Iraq,
Afghanistan, Bosnia, and Kosovo, the United States likely also became
more willing to provide assistance to these countries after conducting
an intervention there because it had become much more invested in
their success for political and strategic reasons. Higher assistance levels
might come to be seen as a small price to pay to increase stability or
build capacity and to avoid damaging perceptions of a failed interven-
tion in countries where the United States would have been unwilling to
provide such assistance before the intervention. Therefore, we should
be extremely cautious of interpreting increasing levels of assistance as
reflective only of the advancement of U.S. interests in the host nation.

The United States could be gaining additional leverage and access to
key countries, but it could also have simply become much more willing
to provide these funds. It is not clear how much leverage the United
States might be able to gain by providing increased assistance in post-
intervention situations. Because it is especially difficult to disentangle
these factors using only a quantitative approach, we will return to the
question regarding whether and under what circumstances U.S. inter-
ventions might advance U.S. interests in Chapters Four and Five.
Finally, we consider the relationship between intervention decisions
and host-nation outcomes using three metrics: (1) polity scores to assess
changes in the level of democracy, (2) GDP per capita to assess changes
in economic well-being and state capacity, and (3) infant mortality to
assess changes in well-being and health. In this section, we explore the
fundamental question of whether U.S. interventions of various char-
acteristics appear to help the host nation in the medium and longer
term.25
Table 3.10 shows the changes in polity score between the first
year of the war or crisis and five years after the end of the war or cri-
sis.26 If U.S. interventions increase the level of democracy of the host
nation, we would expect to see positive increases in the polity score. We
25 It is important to note when interpreting the results in this section that the U.S. com-
mitment to helping improve the host nation varied substantially across the interventions
in our data set. Although many cases involve clear commitments to nation-building and
assistance (such as post-2001 Afghanistan and post-2003 Iraq), others might have seen much
more limited involvement (such as naval missions to deter Soviet involvement in the 1971
India-Pakistan war over Bangladesh), while in still other cases, the United States might have
actually wished the host nation ill (such as Cuba following the Cuban Missile Crisis or Iran
following the Iran Hostage Crisis). Because the United States was, in most cases, at least
generally interested in improving outcomes for the host nation, we decided to include the
analysis showing overall patterns in these metrics. However, caution should be taken when
interpreting them, and they should not be understood to reflect the limits of how much
the United States might be capable of improving host-nation outcomes, if that were to be a
priority.
26 Polity scores rate the level of democracy of a state on a scale from −10 to 10, with 10 being
a full democracy and −10 being a full autocracy. Marshall, Gurr, and Jaggers, 2018.
Table 3.10
Changes in Polity Score, by Intervention Category
Change in Average
Percentage Polity Score (Initial
Number of of Cases That Average Initial Year to Five Years
Category Cases Are Wars Polity Score Postconflict)
No intervention 170 59% −2.5 +0.5
New U.S. 47 50% −2.5 −0.3

intervention
Early, small 27 41% −2.8 −0.6
Early, medium 6 17% −1.8 −1.3
Early, large 5 20% −2.3 −0.7
Late, small 6 100% −3.3 +3.0
Late, medium 3 100% 0.0 0.0
Forces already 64 44% 0.2 +0.4

present
include this measure because many U.S. interventions have increasing

the level of democracy or improving governance as either an overt or
unstated objective, but it is worth noting that democracy promotion is
not an objective of all U.S. interventions.
Table 3.10 shows that, in terms of starting point, intervention
and nonintervention cases have similar average polity scores. There are
modest differences according to the type of intervention, with early
interventions being associated with lower polity scores than are associ-
ated with late ones. However, there is little indication here that U.S.
interventions have consistently contributed to an increase in the level
of democracy of the host nation. Instead, changes are generally small
and often are in the negative direction. One exception is the late, small
intervention category. This differing result is driven by two cases: Ban-
gladesh and El Salvador, which experienced significant increases in their
level of democracy after small U.S. interventions. In the case of Ban-
gladesh, the observed increase in overall level of democracy occurred
following a bloody war between India and Pakistan that eventually led
to independence and the formation of a democratic Bangladesh. But
it would be misleading to argue that the U.S. intervention, limited to
the deployment of a carrier group to the Indian Ocean to monitor the
conflict, played any important role in this democratization.27 In the
case of El Salvador, the United States originally supported the Salva-
doran government in its fight against Leftist guerillas, despite the fact
that the government was less than democratic. It is true, however, that
the United States was also involved in the negotiation of the cease-
fire and supported the eventual holding of elections.28 Looking at the
data more generally, we see relatively few cases of significant increases
in polity score across intervention categories. Among those cases with
significant increases, such as Liberia in the 1990s, it can be hard to
trace the direct line between the military intervention and democrati-
zation.29 There are also cases in which U.S. interventions are followed
by large decreases in democracy, including in Indonesia, where U.S.
involvement in regime activities against Indonesian communists could
have contributed to the decline in democracy observed in the 1960s
in that country.30 Finally, a comparison of wars and crises suggests no
major differences between patterns in the two types of cases for this
metric. Taking a holistic view of all the intervention cases in the data
set, our analysis leads us to expect no substantial relationship between
U.S. interventions and democratization of the host nation.
Turning to economic outcomes, our analysis in Table 3.11 suggests
that, at the aggregate level, U.S. military interventions are not associ-
ated with any increase in GDP per capita in the five years after the end
27 Lorraine Boissoneault, “The Genocide the U.S. Can’t Remember, but Bangladesh Can’t
Forget,” Smithsonian, December 16, 2016.
28 William M. LeoGrande, Our Own Backyard: The United States in Central America,
1977–1992, Chapel Hill, N.C.: University of North Carolina Press, 1998.
29 Terrence Lyons, “Liberia’s Path from Anarchy to Elections,” Current History, Vol. 97,
No. 619, May 1998.
30 Vincent Bevins, “What the United States Did in Indonesia,” The Atlantic, October 20,
2017; Bradley R. Simpson, Economists with Guns: Authoritarian Development and U.S.-
Indonesian Relations, 1960–1968, Stanford, Calif.: Stanford University Press, 2008.
Table 3.11
Changes in GDP per Capita, by Intervention Category
Average Initial GDP Change in Average

Percentage per Capita (2011 GDP per Capita
Number of Cases That Geary-Khamis (Initial Year to Five
Category of Cases Are Wars dollars) Years Postconflict)
No intervention 170 59% $2,959 +13%
New U.S. 50 50% $2,483 +12%

intervention
Early, small 27 41% $2,615 +10%
Early, medium 6 17% $1,303 +18%
Early, large 5 20% $2,706 +41%
Late, small 6 100% $2,723 −20%
Late, medium 3 100% $2,709 +15%
Forces already 64 44% $5,575 +19%

present
NOTE: Excludes ongoing wars and crises as well as cases that ended after 2013.
of the crisis or war when compared with nonintervention cases or cases

in which the United States already had forces present. We do, how-
ever, see some differences within the intervention subcategories. First,
increases in GDP per capita following large interventions (though few
in number) appear to be somewhat greater on average than increases
following smaller interventions and nonintervention cases are. Second,
early interventions are also associated with better economic outcomes
than later ones are, although our evidence is suggestive only. We also
see one instance in which GDP per capita falls (specifically, for inter-
ventions that are late and small). This result is driven primarily by two
interventions: the limited intervention in the Persian Gulf associated
with the Iran-Iraq War and the intervention in Somalia in 1991. Taken
together, then, our findings suggest that larger, earlier interventions are
associated with greater economic growth, while later and smaller inter-
ventions are not, such that interventions overall appear to be broadly
similar to noninterventions. The number of cases involved in many
of these subcategories, however, is quite small, so these results should

again be interpreted with caution.
Our final outcome metric is infant mortality, which can be a
measure of a country’s level of development as well as its ability to
provide basic services. Falling rates of infant mortality would indicate
an improvement in development and the trajectory of the host nation.
Table 3.12 shows the level of infant mortality at the start of the crisis
or war for each intervention category and the percentage decline in the
period between the first year of the crisis or war and five years after its
end. With both metrics, there is little difference between the interven-
tion and nonintervention cases at the aggregate level, but there could
be some differences among intervention categories that are worth
investigating. Specifically, larger interventions appear to occur in cases
in which infant mortality rates are lower to start than is the case (in
our sample) for smaller interventions, and there is some suggestion that
postintervention reductions in infant mortality might be greater in late
Table 3.12
Changes in Infant Mortality Rate, by Intervention Category
Change in Infant
Percentage of Average Infant Mortality (Initial
Number of Cases That Are Mortality (per Year to Five Years
Category Cases Wars 1,000 Live Births) Postconflict)
No intervention 170 59% 78 −19%
New U.S. 47 50% 82 −19%

intervention
Early, small 27 41% 98 −18%
Early, medium 6 17% 73 −11%
Early, large 5 20% 43 −12%
Late, small 6 100% 92 −32%
Late, medium 3 100% 43 −22%
Forces already 64 44% 37 −21%

present
NOTE: Excludes ongoing wars and crises as well as cases that ended after 2013.
interventions. As elsewhere in this report, however, the influence of

selection effects, the diversity in U.S. goals in these countries, and the
small number of cases make it difficult to generalize.
Summary
Table 3.13 summarizes the results noted in this chapter. Because of

the small sample size (especially in the case of late interventions and
larger interventions) and selection effects, the quantitative evidence
presented in this chapter is limited. Although we found that the start-
ing conditions of conflicts that did and did not trigger U.S. interven-
tions appear to be similar in most ways, selection effects still likely play
a role in shaping intervention outcomes and decisions about whether,
when, and with how many forces to intervene. These limitations were
not unexpected; indeed, they were one of the key reasons this report
includes four different, parallel methodologies. However, we still iden-
tified some patterns of interest, which will be interesting to compare, as
we do in Chapter Six, with results from other methods that are better
equipped to more directly address questions regarding whether these
correlations might be causal.
We found very limited evidence that U.S. intervention decisions
are associated with changes in the dynamics of conflicts. Patterns
in conflict duration and conflict recurrence were not notably differ-
ent from nonintervention cases. The exception to this observation is
that we found an association between U.S. interventions (particularly,
interventions that occur later in a conflict) and a reduction in conflict
intensity.
We found more evidence that earlier interventions were associated
with the advancement of U.S. interests, in comparison with later inter-
ventions. Unfortunately, we did not have data available to assess how
intervention cases might have differed from nonintervention cases in
this regard. We also found evidence that U.S. assistance (and military
assistance in particular) increased after U.S. interventions, especially
with earlier interventions. However, it remains unclear whether this
increase in assistance should be interpreted as advancing U.S. interests.
Table 3.13
Summary of Quantitative Investigations into Relationship Between U.S.
Interventions and Outcomes
Category Outcome Patterns Observed Caveats and Concerns
Conflict Duration (conflict No clear patterns Interventions are

years) observed. associated with shorter
wars but longer crises. It
is difficult to anticipate
the direction of selection
effects.
Intensity (annual U.S. interventions The United States might

battle deaths) are associated with a be more likely to select the
somewhat greater decline most-feasible intervention
in conflict intensity than opportunities for ending
noninterventions are. the conflict, which would
Earlier interventions tend to overstate the
are associated with a intervention’s conflict-
lesser decline in conflict reducing effect.
intensity than later
interventions are.
Recurrence No clear patterns Intervention and

observed. nonintervention cases
have highly similar
recurrence rates.
Intervening National Earlier interventions are Potential selection effects

state interests (overall; associated with a higher include a higher likelihood
percentage likelihood of achieving of early intervention when
of objectives U.S. political objectives there are higher stakes
achieved) than later interventions for the United States
are. and therefore a greater
motivation to commit
sufficient resources
(beyond force size) to
succeed.
Access and Interventions are U.S. willingness to provide

leverage (military associated with a notable assistance to countries in
and economic increase in military which it has intervened
assistance to the assistance and a more might also increase
host nation) modest increase in dramatically. If so, then
economic assistance. this metric might not
Early interventions are reflect increased access
associated with increases to or leverage over the
in economic and military recipient country but
assistance in comparison instead a shift in U.S.
with late interventions. political and strategic
investment in the country.
Table 3.13—Continued
Category Outcome Patterns Observed Caveats and Concerns
Host nation Political (polity) No clear patterns Intervention and

observed. nonintervention cases
are broadly similar, with
some apparent differences
driven by cases that
appear to be unrelated
to democratization.
Democratization might
not always have been a
U.S. goal.
Economic (GDP Interventions are There is a limited number

per capita) broadly similar to of cases, particularly
noninterventions overall. in later and larger
However, earlier and intervention categories,
larger interventions reducing the level of
are associated with confidence in the results.
higher economic Economic improvements
growth than later and might not always have
smaller interventions or been a U.S. goal.
noninterventions.
Economic Interventions are similar The very limited number

(infant mortality) to noninterventions of later interventions
overall. However, later makes it difficult to
interventions might be generalize from those
associated with greater results.
decreases in infant
mortality rates than
earlier interventions or
nonintervention cases are.
Higher assistance levels might either produce or reflect greater U.S.

leverage over the host nation, but they might also reflect increased U.S.
willingness to provide assistance to a country in which it now has a
much larger political and strategic investment.
Finally, we again found no clear evidence that host-nation out-
comes differ substantially between intervention and noninterven-
tion cases. Within the set of intervention cases, however, earlier and
larger interventions were associated with larger increases in GDP per
capita, while later interventions might have been associated with larger
decreases in infant mortality. But these patterns, in particular, were
driven by a small number of cases.
Although we view these patterns as suggestive only, the hetero-
geneity among them suggests that there could be trade-offs for poli-
cymakers among the three main outcome measures that we focus on.
Earlier interventions, for example, could be those in which U.S. objec-
tives are more likely to be achieved but also could lead to higher con-
flict intensity and lower decreases in infant mortality. We do not feel
that the particulars of these trade-offs can be clearly established by this
analysis (the number of cases in each category is quite small), but these
patterns highlight the possibility that all outcome measures might not
be correlated with one another and that U.S. decisionmakers might
need to be aware that efforts to pursue one set of goals could make
pursuing another set more difficult. In Chapters Four and Five, we
will use case studies to explore relationships and trade-offs between
intervention decisions and characteristics and outcomes, and we will
attempt to provide greater insight into these relationships.
CHAPTER FOUR
Lessons from Focused Case Studies
The third method we employed in our investigation of the likely effects

of the decision to conduct a military intervention and of its timing
and size was a series of 45 short, focused case studies. The quantitative
investigations summarized in the previous chapter provided us with
the broadest available review of the historical context and were helpful
in identifying overall patterns in U.S. intervention decisions and out-
comes. However, only relatively few consistent patterns were identified,
and the interpretation of these patterns is complicated by the limita-
tions of the analyses because of a variety of data and methodological
issues previously discussed. These limitations highlight the potential
value of more-detailed studies of specific cases.
When deciding to conduct more-detailed investigations into par-
ticular cases, however, one faces a necessary trade-off between depth
and breadth. Conducting a smaller number of highly detailed cases can
be ideal for investigating specific pathways or hypothesized causal rela-
tionships, but it risks identifying relationships that might not be gener-
alizable if the cases selected are not representative. On the other hand,
conducting a larger number of more-limited case studies can help to
ensure a more representative set of cases and to consider different con-
texts but provides more-limited information on the precise operation of
potential causal pathways.
In this study, we use both approaches. In this chapter, we consider
a large number of case studies across different contexts in less depth.1
1 More information on the focused case studies is available from the authors on request.
73
Crucially, these focused studies allow us to explore cases that cover the
full breadth of differing intervention timing and size combinations and
to consider several nonintervention cases. In addition, this approach
allows us to assess a wider range of in-country and international char-
acteristics and their relationship to intervention decisions and interven-
tion outcomes. As a complement to this approach, we also conducted
four much more detailed counterfactual case studies, which are sum-
marized in the next chapter.
Methodology
The focused case studies were intended to serve several purposes

within our analysis. First, we used these case studies to explore the
U.S. decisionmaking process in each instance, including the interests
that motivated decisions to intervene and factors that shaped decisions
about intervention timing. Second, we considered the outcomes of
each intervention for the United States and for the host nation.2 Third,
we used the case studies to explore the dynamics of nonintervention
cases to better understand why interventions did not occur in those
cases and what the implication of the decision not to intervene was for
U.S. interests and the host-nation trajectory. This section summarizes
the methodology we employed to select and then conduct our focused
case studies.
Selecting Cases
When selecting the cases for our focused case studies, we considered
several factors. We wanted to ensure that the cases chosen included a
sufficient number of nonintervention cases to provide a useful baseline
and covered a wide range of intervention timings and sizes. Further,
we also wanted the cases to cover a representative set of activity types,
2 As discussed in Chapter One, we did not assess the third category of outcome measures
(effects on the conflict) in these case studies. The effects of interventions on conflict dynam-
ics were among the most robustly studied in the literature. For this reason, we decided to use
our case studies to focus on the least well-studied outcome measures, including a particular
emphasis on whether the interventions advanced the interests of the intervening state.
Lessons from Focused Case Studies 75
such that our cases were not unduly concentrated in, for example, all
combat missions or all deterrence missions. Finally, we wanted to con-
duct the largest number of cases that resources permitted to ensure that
our assessments of cases with particular timing and size characteristics
were as likely to be representative as possible.
With these criteria in mind, we selected 45 cases from the list of
286 potential cases of U.S. intervention discussed in Chapter Three.
During our selection process, we drew cases using the intervention cat-
egories described in Chapter Three: whether the intervention occurred
early or late in a crisis or war; whether the intervention involved small,
medium, or large U.S. forces; and combinations of these timing and
size dimensions, as well as cases in which U.S. forces were already pres-
ent before the onset of the crisis or war and cases in which the United
States did not intervene at all—there were no historical cases that were
late and large. After ensuring, to the best extent the universe of poten-
tial cases allowed, a sufficient number of cases from each category, we
then randomly selected cases from each category while making sure
that we also included a mix of activity types, including deterrence,
combat, and stability operations.3 During this process, we intentionally
oversampled intervention cases; in particular, we oversampled medium
and large intervention cases, of which there have historically been
fewer, given their greater substantive importance for policymakers. We
therefore undersampled cases in which U.S. forces were already pres-
ent or in which there was no U.S. intervention at all, although we still
included 20 nonintervention cases. A full list of cases selected is shown
in Table 4.1.
3 As shown in Chapter Three, most of these categories, or buckets, of intervention tim-

ings and sizes had sufficient numbers of potential cases to allow us to select a relatively large
number for investigation. However, it should be noted that the United States has not histori-
cally conducted any interventions that were large and began late in the timeline of the crisis
or war. The absence of cases in this category naturally limits our ability to assess the likely
outcomes of U.S. interventions with both of these characteristics. There have been several
cases that began late and involved medium-sized U.S. forces, however, and we conducted
focused case studies on three of these: Bosnia 1995, Somalia 1992, and Iraq 2014. In addi-
tion, we selected one of these cases (Bosnia) for use in a more detailed assessment in our
counterfactual case studies, discussed in Chapter Five.
Table 4.1
List of Cases Selected
Intervention Timing,
Case Name Years Size
Malagasy Uprising (France–Madagascar) 1947–1948 No intervention
Berlin Blockade 1948–1949 Forces already present
Costa Rican Civil War (Costa Rica–Nicaragua) 1948–1949 No intervention
La Violencia (Colombia civil war) 1948–1958 No intervention
Yugoslav-Soviet Crisis 1949–1951 Late, small
Hula Drainage Crisis (Israel–Syria) 1951 No intervention
Burma-Kuomintang Crisis 1953–1954 No intervention
Guatemalan Revolution 1953–1954 Early, small
Lebanon Crisis (Lebanon–United Arab 1958 Early, medium

Republic)
Rwandan Social Revolution 1959–1962 No intervention
Cuban Missile Crisis 1962 Early, medium
Dominican Civil War 1965–1966 Early, large
Biafra Crisis (Nigeria) 1967–1970 No intervention
Football War (El Salvador–Honduras) 1969 No intervention
Demilitarized Zone Poplar Tree Crisis (Korea) 1976 Forces already present
Entebbe Raid (Israel–Uganda) 1976 No intervention
Ogaden War (Ethiopia–Somalia) 1976–1977 No intervention
Litani Conflict (Israel–Lebanon) 1978 No intervention
Sandinista Rebellion (Nicaragua) 1978–1979 No intervention
Salvadoran Civil War 1979–1992 Late, small
Egypt-Libya Standoff 1980 No intervention
Gulf of Sirte Incident (Libya–United States) 1981 No intervention
Lebanon War (Israel–Palestinian Liberation 1982 Early, small

Movement [PLO])
Grenada Conflict (Grenada–United States) 1983 Early, medium

Intervention Timing,
Case Name Years Size
Lebanese Civil War 1983–1984 Forces already present
Libya Bombing (Libya–United States) 1986 Early, small
Panama Invasion (Panama–United States) 1989–1990 Early, large
Kashmir Nuclear Confrontation (India– 1990 No intervention

Pakistan)
Gulf War (Iraq–Kuwait) 1990–1991 Early, large
Shiite Uprising (Iraq) 1991 Early, medium
Kurdish Uprising (Iraq) 1991 Early, medium
Somali Civil War 1991–1997 Late, medium
Bosnian Civil War 1992–1994 Late, medium
Operation Accountability (Israel–Hezbollah) 1993 No intervention
North Korean Nuclear Crisis 1993–1994 Forces already present
Haitian Crisis (Haiti–United States) 1994 Early, medium
Afghanistan Invasion (Afghanistan–United 2001 Early, medium

States)
Parsley Island Crisis (Morocco–Spain) 2002 No intervention
Iraq War (Iraq–United States) 2002–2003 Early, large
Lebanon War (Israel–Hezbollah) 2006 No intervention
Iran Nuclear Crisis (Iran–United States) 2006–2007 No intervention
Houthi Insurgency (Yemen) 2007 No intervention
Libyan Civil War 2011 Early, small
Iraq–ISIL 2014–present Late, medium
Libya–ISIL 2015–2016 Late, small
NOTE: ISIL = Islamic State of Iraq and the Levant.

Taken together, these cases cover a wide variety of historical eras,

geographic regions, and activity types, from Latin American deter-
rence cases (the 1962 Cuban Missile Crisis) to Middle Eastern sta-
bility operations (in 1982 Lebanon) to Asian combat missions (in
2001 Afghanistan). The cases concentrate somewhat more heavily in
the Middle East and North Africa (20 cases) and Latin America (10
cases), with somewhat fewer cases in Asia (five cases) and Europe (three
cases). Although there was randomness in our draw of specific cases,
this distribution reflects our intention to avoid focusing on crises or
wars that occurred in places where U.S. troops were already present,
with the long-standing U.S. troop presence in both Western Europe
and East Asia after 1945 reducing the number of cases in those regions.
We view the decision regarding whether to “plus-up” forces in response
to a crisis or war as distinct from the decision regarding whether to
intervene in the first place; therefore, we deprioritized those cases for
this study, although we still conducted some cases in which U.S. forces
were already present for context.4
Guiding Questions
To ensure that the information from each of our 45 focused case stud-
ies was consistently collected by the research team and could be more
easily compared, we developed a set of specific questions that would be
answered across all of the cases. These questions were designed to focus
on the U.S. intervention decision and the resulting effects, if any, of the
intervention on U.S. interests. The questions asked varied depending
on whether the case in question included a U.S. military intervention,
although both sets of questions investigated similar, and parallel, fac-
tors. Box 4.1 lists the information collected in our set of focused case
studies.
These questions focused on identifying the U.S. decisionmaking
process in each case and the interests that motivated U.S. intervention
(or made it plausible). They further investigated the outcomes of the
case across four broad categories of U.S. interests: economic interests,
4 A more complete investigation of the effects of U.S. crisis deployments is undertaken in

Frederick et al., 2020.
Box 4.1
Focused Case Study Questions
Asked for every case:
• Describe the nature of the conflict or crisis.
• Describe the U.S. interests involved that made intervention plausible.
• Was the crisis or conflict one in a series?
• Discuss the size and scope of U.S. resources committed.
• Describe relevant other concurrent nonmilitary factors.
• Overall assessment of the U.S. intervention decision.
Asked for U.S. intervention cases:
• What was the timing of the intervention within the war or conflict?
• What factors most influenced the U.S. decision to intervene?
• If the intervention occurred late, was one considered earlier?
• If an earlier intervention was considered but not undertaken, why not?
• What casualties did U.S. forces incur during the intervention?
• What were the economic costs and benefits to the United States of the
intervention?
• What were the reputational costs and benefits to the United States of the
intervention?
• What were the strategic costs and benefits to the United States of the
intervention?
• What were the costs and benefits of the intervention for the host or target
state?
Asked for cases with no U.S. intervention:
• Was undertaking an intervention ever seriously considered?
• If an intervention was not seriously considered, why not?
• If an intervention was seriously considered, which factors were most
important in policymaker calculations?
• What were the economic costs and benefits to the United States of not
undertaking the intervention?
• What were the reputational costs and benefits to the United States of not
• What were the strategic costs and benefits to the United States of not
• What trajectory did the host nation take over the course of the war or crisis
and its aftermath?
reputational interests, strategic interests, and the extent of U.S. mili-

tary casualties when an intervention occurred. Taken together, these
questions were designed to enable us to assess the relationship between
U.S. military interventions of different types, timings, and sizes and
the outcomes of interest to U.S. policymakers. Extensive exploration
of nonintervention cases and the paths they took without a U.S. mili-
tary presence was equally crucial to better enable us to understand how
cases of U.S. intervention differ from the outset from cases in which
the U.S. chooses not to intervene. That is, the comparison between
intervention and nonintervention cases can help us understand the
selection effects (discussed in Chapter Two) that might complicate our

efforts to understand the effects that interventions of different timings
and sizes could have.
We use the focused case studies to explore two categories of out-
comes: those affecting the United States and those relevant to the host
nation’s trajectory. Although we were able to consider the relationship
between timing and conflict outcomes more completely in Chap-
ter Three using quantitative metrics, outcomes for the United States
and the host nation were more difficult to assess using this method.
The focused case studies give us the chance to more carefully consider
these outcomes using qualitative evidence. In Chapter Five, we will use
more-in-depth case studies to consider all three types of outcomes.
Assessment and Insights
There are several patterns of interest that emerge across the case stud-
ies. This section summarizes these patterns through two lenses. After
an initial discussion of the possibility that the size and timing of inter-
ventions are driven by the issues at stake in an intervention, this section
reviews overall patterns in the answers provided to the case study ques-
tions noted in Box 4.1. We then discuss our effort to isolate the effects
of U.S. intervention from other factors to identify its likely impact
and how this impact might have varied by the intervention’s timing
and size. This section concludes with a summary of results and their
implications.
Are the Timing and Size of Interventions Driven by the Issues

Motivating Them?
The issues that have motivated U.S. interventions could be a signpost
of the likely timing and size of an intervention if they reliably signal
the level of effort or commitment the United States is likely to apply
toward achieving its goals. For example, earlier and larger interven-
tions could reflect cases with categorically greater U.S. commitment
and interest, such that decisionmakers felt they had little choice but
to intervene as early and forcefully as possible. If so, then the timing
of interventions in particular would be less an important policy lever

and more a simple reflection of the issues involved. In two early inter-
vention cases—the Cuban Missile Crisis in 1962 and the invasion of
Afghanistan in 2001—policymakers perceived an imminent, direct
threat to U.S. national security, and waiting further to address the situ-
ation would have been a highly implausible decision.5 For intervention
cases with this motivation, questions regarding whether to intervene
later might be irrelevant.6
Generally speaking, however, this sort of clear link between the
interests at stake and the size and timing of the intervention is not what
the bulk of our focused case studies suggest. Apart from the two cases
just mentioned that involved direct, perceived threats to the United
States, there was no clear link between motivating issues and timing or
size. The types of issues that motivated earlier and larger interventions
were also found in interventions that were later or smaller and, to a cer-
tain extent, even in cases in which the United States did not intervene
at all. Regime change provides one example. For instance, there were
early and large cases in which the United States intervened with the
express intent of changing the regime of the country in question, such
as Iraq 2003 or Panama 1989.7 However, there were also cases that saw
early but small interventions for the same purpose, such as Libya 2011
or Guatemala 1953. Meanwhile, late interventions of both medium
and small size in Iraq and Libya to roll back the rise of ISIL had similar
5 Howard L. Burris, “Soviet Concern Over Cuba,” memorandum to Vice President

Lyndon B. Johnson, October 2, 1962; Congressional Research Service, Afghanistan: Back-
ground and U.S. Policy in Brief, Washington, D.C., R45122, May 1, 2018.
6 As we will discuss in greater detail later in this chapter, acute and pressing U.S. interests
might not necessarily lead to a commitment of the largest possible forces, given the trade-offs
that might arise between the size of a force and the speed with which it can be generated and
deployed. The U.S. interventions in Cuba 1962 and Afghanistan 2001 were both medium in
size in our typology, for example.
7 Richard L. Millet, “Panama and Haiti,” in Jeremy R. Azrael and Emil A. Payin, eds.,
U.S. and Russian Policymaking with Respect to the Use of Force, Santa Monica, Calif.: RAND
Corporation, CF-129-CRES, 1996; Joseph J. Collins, Choosing War: The Decision to Invade
Iraq and Its Aftermath, Washington, D.C.: Institute for National Strategic Studies, National
Defense University, April 2008.
motivations.8 Further, there were nonintervention cases in which the

United States would clearly have preferred to see a particular regime
prevail in a contested struggle for authority in a given country, such
as Nicaragua 1978, Costa Rica 1948, or Yemen 2007, but nonetheless
decided not to intervene militarily.9 The size of the force for interven-
tions to change regimes, then, might be driven more by other factors,
rather than indicating the importance of the case for U.S. interests.
Other motivations for intervention also show diverse patterns in
the size and timing with which the United States responded. Interven-
tions undertaken largely for humanitarian purposes (at least initially)
were present in many categories we assessed, from the early, small
intervention to prevent a massacre in Libya in 2011 to the late, medium
intervention to avert a famine in Somalia after 1991.10 Similarly,
humanitarian motivations were not necessarily less likely to prompt
intervention overall in our sample. In only a handful of our noninter-
vention cases were potential humanitarian motivations central (Mada-
gascar 1948, Rwanda 1959, and Nigeria 1968).11 Instead, humanitarian
8 Although ISIL was of course not fully in control of the recognized government of either
state, it did exercise effective territorial control in many areas, and U.S. efforts to reverse this
state of affairs are therefore similar to efforts to change regimes in other states.
9 U.S. National Security Council, “U.S. Policy to Nicaragua,” memorandum of conversa-
tion, Washington, D.C., September 4, 1978; Marcia Olander, “Costa Rica in 1948: Cold
War or Local War?” The Americas, Vol. 52, No. 4, April 1996; Barak Salmoni, Bryce Loidolt,
and Madeleine Wells, Regime and Periphery in Northern Yemen: The Huthi Phenomenon,
Santa Monica, Calif.: RAND Corporation, MG-962-DIA, 2010.
10 Christopher S. Chivvis, Toppling Qaddafi: Libya and the Limits of Liberal Intervention,
New York: Cambridge University Press, 2014; Maryann K. Cusimano, Operation Restore
Hope: The Bush Administration’s Decision to Intervene in Somalia, Washington, D.C.: Insti-
tute for the Study of Diplomacy, 1995.
11 Central Intelligence Agency, Office of National Estimates, “Implications of Madagas-
car’s Unfinished Revolution,” memorandum, July 27, 1972; Roger Hilsman, Director of the
Bureau of Intelligence and Research, “Basic Elements in the Ruanda-Urundi Situation,”
memorandum to Secretary of State Dean Rusk, June 8, 1962; Henry A. Kissinger, “U.S.
Options in Biafra Relief,” memorandum to President Richared Nixon, Washington, D.C.,
January 28, 1969, pp. 1–6.
motivations were concentrated most heavily in early interventions of

small or medium size, such as Haiti 1994 and Libya 2011.12
That said, we find some notable patterns in the issues at stake
between intervention and nonintervention cases. In most noninterven-
tion cases, the primary issues at stake for the United States reflected
more-diffuse geopolitical concerns related to the regional balance of
power or an interest in preventing escalation of conflict, such as India-
Pakistan 1990, Morocco-Spain 2002, or Ethiopia-Somalia 1976.13
Among intervention cases, such issues were less common but were
found in late, small interventions, such as Yugoslavia 1949.14 This pat-
tern could suggest that broader geopolitical or security interests could
be what policymakers are most willing to monitor and let play out,
sometimes prompting intervention later in a crisis or war but often not.
There are exceptions to this as well, of course, such as the early, large
intervention in Lebanon in 1958.15
In summary, although we find only limited evidence that the
type of issue most prominently at stake is associated with U.S. deci-
sions regarding whether to intervene and the timing and size with
which such an intervention might occur. Other associations do not
appear to be so determinative as to suggest that policymakers do not
retain substantial flexibility over the timing and size of interventions
in most cases. Only direct threats to the security of U.S. home territory
appear to be addressed early and with relatively sizable forces without
exception. For other issues, we observe some trends and patterns but
no ironclad rules. Humanitarian concerns might be more likely to be
12 Millet, 1996; Chivvis, 2014.

13 Michael Krepon and Mishi Faruqee, eds., Conflict Prevention and Confidence-Build-
ing Measures in South Asia: The 1990 Crisis, Washington, D.C.: Henry L. Stimson Center,
Occasional Paper No. 17, April 1994; Colin Powell, “Parsley Island,” The American Interest,
June 14, 2012; David D. Newsom, Director of the Office of Northern African Affairs, “U.S.
Policy Toward the Ogaden,” memorandum to Assistant Secretary of State for African Affairs
G. Mennen Williams, Washington, D.C., August 26, 1963.
14 “Dodging Armageddon: The Third World War That Almost Was, 1950,” Cryptologic
Quarterly, February 24, 1998.
15 Douglas Little, “His Finest Hour? Eisenhower, Lebanon, and the 1958 Middle East
Crisis,” Diplomatic History, Vol. 20, No. 1, January 1996, pp. 27, 40–44.
addressed with smaller forces, and broader geopolitical concerns might

be more likely to be addressed through military intervention later, if at
all. The relationship between the interests at stake and the timing and
size of the resulting intervention will be important to bear in mind as
we investigate differences in the outcomes of interventions of various
timings and sizes.
Patterns in Costs and Benefits to the United States for Different

Are there clear patterns in the outcomes for the United States of differ-
ent intervention size and timing decisions? Do larger or earlier inter-
ventions tend to lead to different outcomes than interventions that are
later or smaller? According to a summary of the focused case studies
we conducted, we find some evidence for such patterns, but they tend
to vary depending on the type of outcome measure being considered.
In this section, we survey the patterns in three main types of outcome
measures for the United States: economic costs and benefits, reputa-
tional costs and benefits, and strategic costs and benefits. We will ini-
tially focus just on describing the patterns we observe. In a subsequent
section, we will consider a broader synthesis of whether we see patterns
in the advancement of U.S. interests overall and will discuss in detail
the factors that appear to explain the observed patterns.
Interventions have typically had net economic costs for the United
States, sometimes dramatically so. U.S. economic interests were almost
never notably improved following interventions, with the possible
exceptions of Panama 1989 and Guatemala 1953.16 Instead, most inter-
ventions had either clearly negative effects on U.S. economic interests,
such as Iraq 2003 and Afghanistan post-2001, or limited or uncertain
16 Panama Canal Commission, Annual Report: Fiscal Year Ended September 30, 1990,
Washington, D.C.: U.S. Government Printing Office, 1990; Nicholas Cullather, Operation
PBSUCCESS: The United States and Guatemala, 1952–1954, Washington, D.C.: Central
Intelligence Agency, Center for the Study of Intelligence, 1994.
effects, such as Bosnia 1995 or Haiti 1994.17 But there do not appear
to be clear patterns in the timing or size of the interventions regarding
these effects. Interventions with clearly negative effects on U.S. eco-
nomic interests included those that were early (Iraq 2003) or late (Iraq
2014) as well as those that were small, medium, or large in size (Libya
2011, Afghanistan post-2001, and Iraq 2003, respectively).18 Noninter-
vention cases typically involved relatively low stakes for U.S. economic
interests, with little evidence of dramatic changes in either direction.
We therefore do not find evidence that U.S. decisions regarding the size
and timing of an intervention are overall likely to affect U.S. economic
interests, although of course individual circumstances can differ.
Reputational Costs and Benefits
Reputational costs and benefits appear to have been somewhat more
divergent across intervention size and timing categories. Early inter-
ventions such as Iraq 2003 and Guatemala 1953 had clearly negative
reputational effects for the United States, as did late interventions such
as El Salvador 1979.19 Clearly positive reputational effects were rare.
Those we identified (the 1990 Gulf War and the 1962 Cuban Missile
Crisis) were both early interventions, but there might not have been
17 Amy Belasco, The Cost of Iraq, Afghanistan, and Other Global War on Terror Opera-
tions Since 9/11, Washington, D.C.: Congressional Research Service, RL33110, December 8,
2014; Neta C. Crawford, United States Budgetary Costs of Post-9/11 Wars Through FY2018: A
Summary of the $5.6 Trillion in Costs for the US Wars in Iraq, Syria, Afghanistan, and Pakistan,
and Post-9/11 Veterans Care and Homeland Security, Providence, R.I.: Watson Institute of
International and Public Affairs, Brown University, November 2017; Kavanagh, Frederick,
Stark, et al., 2019, pp. 196–228; United Nations, “UN Comtrade Database,” online data-
base, undated b; U.S. Agency for International Development, 2015.
18 U.S. Department of Defense, “Combined Joint Task Force Operation Inherent Resolve,”
webpage, undated; Stefan Karlsson, “The Total Cost of the Libya Intervention,” Christian
Science Monitor, April 15, 2011.
19 Collins, 2008; David M. Barrett, “Sterilizing a ‘Red Infection’: Congress, the CIA, and
Guatemala, 1954,” Central Intelligence Agency, Center for the Study of Intelligence, May 8,
2007; Cullather, 1994; Raymond Bonner, “America’s Role in El Salvador’s Deterioration,”
The Atlantic, January 20, 2018; Paul P. Cale, “The United States Military Advisory Group
in El Salvador, 1979–1992,” Small Wars Journal, 1996; Kevin Sullivan and Mary Jordan, “In
Central America, Reagan Remains a Polarizing Figure,” Washington Post, June 10, 2004.
sufficient late intervention cases to know whether their absence in this

circumstance was meaningful.20
With respect to the size of interventions, we find both small and
large interventions that had clearly negative effects on the U.S. repu-
tation, including El Salvador 1979 and Guatemala 1953 (small) and
Iraq 2003 and Dominican Republic 1965 (large).21 However, in the
ten medium-sized interventions we assessed (excluding those in which
U.S. forces were already present), we did not identify any such cases of
clearly negative effects on the U.S. reputation. The two clearly positive
cases for the U.S. reputation, the 1990 Gulf War and the 1962 Cuban
Missile Crisis, were large and medium in size, respectively.
In sum, then, interventions that were medium in size were the
most likely to avoid clear damage to U.S. international reputation. Any
successes that occurred in enhancing the U.S. reputation came from
early interventions, which were also the interventions with the most
dramatically negative effects in this regard. We will return to this issue
in the next section.
Nonintervention cases, for the most part, had limited costs or
benefits for the U.S. reputation. Further, other countries did not
assume that the United States would intervene in most of the cases
we investigated, and we did not identify instances among our focused
case studies in which a U.S. failure to intervene damaged its reputa-
20 Lawrence Freedman and Efraim Karsh, The Gulf Conflict, 1990–1991: Diplomacy and
War in the New World Order, Princeton, N.J.: Princeton University Press, 1993; Tom Mahn-
ken, “The Gulf War in Retrospect,” Foreign Policy, January 20, 2011; Central Intelligence
Agency, “Implications of Khrushchev’s Message of 28 October,” Boston, Mass.: John F. Ken-
nedy Library, October 29, 1962a; David Coleman, “JFK’s Presidential Approval Ratings,”
webpage, undated; Central Intelligence Agency, “The Crisis: USSR/Cuba,” Boston, Mass.:
John F. Kennedy Library, October 29, 1962b.
21 Lawrence A. Yates, Power Pack: U.S. Intervention in the Dominican Republic, 1965–1966,
Fort Leavenworth, Kan.: Combat Studies Institute, U.S. Army Command and General Staff
College, 1988; Lawrence M. Greenberg, United States Army Unilateral and Coalition Opera-
tions in the 1965 Dominican Republic Intervention, Washington, D.C.: U.S. Army Center
of Military History, November 1986; David Coleman, ed., “LBJ Regretted Ordering U.S.
Troops into Dominican Republic in 1965, White House Tapes Confirm,” National Security
Archive, Electronic Briefing Book No. 513, April 28, 2015; Association for Diplomatic Stud-
ies and Training, “The Dominican Civil War of 1965,” Huffington Post, updated Decem-
ber 6, 2017.
tion. We did, however, identify at least two cases of nonintervention in

which the U.S. reputation improved: Iran 2006 and Costa Rica 1948.22
It is notable that there might have been a greater presumption among
other countries that the U.S. would intervene in these cases because of
prior U.S. actions toward Iraq and prior U.S. intervention behavior in
Latin America, such that avoiding the intervention improved the U.S.
reputation.
Strategic Costs and Benefits
The strategic effects on U.S. interests of the interventions we inves-
tigated varied notably across the timing and size categories. All cases
in which we observed notable deteriorations in U.S. strategic interests
involved early interventions (Iraq 2003, Libya 2011, and Guatemala
1953).23 Cases in which we found clear strategic benefits for the United
States were split between early (e.g., Dominican Republic 1965 and
Panama 1989–1990) and late (e.g., El Salvador 1979 and Libya 2015)24
22 Simond de Galbert, Back to the Future: Europe, the United States, and Iran Sanctions after
the Nuclear Deal, Washington, D.C.: Center for Strategic and International Studies, Novem-
ber 2015; Gideon Rose, “What Obama Gets Right,” Foreign Affairs, September/October
2015; Olander, 1996.
23 Collins, 2008; Daveed Gartenstein-Ross, “Successes and Failures of the U.S. and NATO
Intervention in Libya,” testimony presented before the Committee on Oversight and Gov-
ernment Reform, Washington, D.C.: Foundation for Defense of Democracies, May 1, 2014,
p. 4; Michael Clarke, “The Making of Britain’s Libya Strategy,” in Adrian Johnson and
Saqeb Mueen, eds., Short War, Long Shadow: The Political and Military Legacies of the 2011
Libya Campaign, London: Royal United Services Institute, 2012; Jonathan Marcus, “Why
China and Russia Rebuffed the West on Syria,” BBC News, October 5, 2011; Paul Lewis,
“Putin Insists Russia Wants to Repair US Relations – but Ill Will for Obama Is Clear,” The
Guardian, April 17, 2014; Stephen Schlesinger, “Ghosts of Guatemala’s Past,” New York
Times, June 3, 2011; Cullather, 1994; Barrett, 2007; Stephen M. Streeter, “Interpreting the
1954 U.S. Intervention in Guatemala: Realist, Revisionist, and Postrevisionist Perspectives,”
The History Teacher, Vol. 34, No. 1, November 2000.
24 Yates, 1988; Greenberg, 1986; Joseph B. Treaster, “20 Years After Dominican War,
Wounds Linger,” New York Times, May 1, 1985; Julio Yao, “Legacies of the U.S. Invasion
of Panama,” North American Congress on Latin America, March 22, 2012; Sullivan and
Jordan, 2004; Erick O. Granados, Impact of U.S. Foreign Policy and Assistance in El Salvador,
Quantico, Va.: U.S. Marine Corps Command and Staff College, May 2012; Christopher M.
Blanchard and Carla E. Humud, The Islamic State and U.S. Policy, Washington, D.C.: Con-
gressional Research Service, September 25, 2018.
interventions. Given the very small number of late interventions, how-

ever, the presence of such strategic successes and a lack of clear strategic
failures may or may not be notable.
With respect to size, we see a similar pattern to that observed with
respect to reputational costs and benefits. Although cases with clear
strategic benefits to the United States involved interventions that were
small (El Salvador 1979), medium (Grenada 1983), and large (Domini-
can Republic 1965), the same cannot be said for cases with clear stra-
tegic costs for the United States.25 These clearly negative cases for U.S.
strategic interests occurred in small interventions (Libya 2011 and
Guatemala 1953) as well as large interventions (Iraq 2003) but were not
found in the medium intervention category. Medium-sized interven-
tions appear to have avoided the most negative costs while maintaining
a broadly similar rate of cases with clear benefits for the United States.
Interpreting this finding is complex, however. In some instances,
medium-sized interventions could reflect caution and restraint on the
part of policymakers, who might have considered deploying large-sized
forces. In other instances, however, medium-sized interventions could
reflect a comparatively heavy investment of forces and resources into
a situation in which the realistic alternative might have been a much
smaller force. Medium size, then, is not an intervention characteristic
likely to be of value to policymakers per se. In the next section, we
describe the results of our analysis, which takes into account the con-
text in which interventions of different sizes take place, and isolate the
effects that the size of the intervention appears to have on the advance-
ment of U.S. interests.
Interestingly, nonintervention cases also vary in their effects on
U.S. strategic interests. The majority of nonintervention cases had no
25 Richard Stewart, Operation Urgent Fury: The Invasion of Grenada, October 1983,
Washington, D.C.: U.S. Army Center of Military History, 2008, pp 26–36; Ronald H.
Cole, Operation Urgent Fury: The Planning and Execution of Joint Operations in Grenada,
12 October–2 November 1983, Washington, D.C.: Joint History Office, Office of the Chair-
man of the Joint Chiefs of Staff, 1997, pp. 1–5; Daniel L. Haulman, “Crisis in Grenada:
Operation Urgent Fury,” in A. Timothy Warnock, ed., Short of War: Major USAF Contin-
gency Operations, 1947–1997, Maxwell Air Force Base, Ala.: Air University Press, 2000;
Ronald Reagan, “Remarks of the President and Prime Minister Eugenia Charles of Domi-
nica Announcing the Deployment of United States Forces in Grenada,” October 25, 1983.
clear effect on U.S. strategic interests. However, although there are

some cases in which not intervening led to an improved strategic situ-
ation from a U.S. perspective, such as Iran 2006, Colombia 1948, and
Costa Rica 1948, there were also cases in which avoiding intervention
was associated with a deterioration in U.S. strategic interests, such as
Yemen 2007 and Nicaragua 1978.26
Isolating the Effects of Intervention Decisions on U.S. National

Interests
The focused case studies we conducted allowed us to collect detailed
information regarding patterns in costs and benefits to the United
States. As discussed in the previous section, however, it remains pos-
sible that the patterns we observe in these costs and benefits are not
driven by U.S. intervention decisions but by other factors. Our detailed
narratives and discussions of the cases allow us to assess the extent to
which decisions about whether, when, and with how many forces to
intervene affected the likelihood that U.S. interventions advance U.S.
interests.27 That is, how would one assess the effects of U.S. interven-
tion decisions about size and timing in isolation from other factors?
In this section, we focus on whether the patterns we observe in
intervention outcomes were in fact driven by the U.S. interventions
(and their timing and size) themselves or whether they reflect other
26 de Galbert, 2015; Rose, 2015; U.S. Department of State, Communist Involvement in the
Colombian Riots of April 9, 1948, Washington, D.C., Report No. 4686, October 14, 1948;
U.S. National Security Council, “Current Policies of the Government of the United States of
America Relating to the National Security: Volume I, Geographical Area Policies: Part VII,
Latin America - Colombia,” Gale Group, November 1, 1952; Central Intelligence Agency,
“OCI Handbook: Colombia,” June 13, 1955; Olander, 1996; Andrew Exum, “What’s Really
at Stake for America in Yemen’s Conflict,” The Atlantic, April 14, 2017; Presidential Review
Committee, “Summary of a Presidential Review Committee (PRC) Meeting on Central
America. Issues Include Discussion on U.S. Policy and the Political Situation in the Follow-
ing Countries: Nicaragua; El Salvador; Guatemala; Honduras,” Gale Group, June 11, 1979;
“U.S. Aid to the Contras,” Washington Post, February 3, 1988.
27 In this context, by advance U.S. interests we mean increase the ability of the United States
to achieve its political objectives, enhance its security and strategic flexibility, and protect
access and assets globally.
unrelated factors. To assess this issue, we reviewed each focused case

study to answer five additional, related questions:28
1. Did the intervention itself substantially help improve U.S.

interests?29
2. Did the intervention itself substantially help worsen U.S. inter-
ests?
3. If yes to either 1 or 2, was the timing of the intervention (early
versus late) crucial to the effect that the intervention had?
4. If yes to either 1 or 2, was the size of the intervention (large,
medium, or small) crucial to the effect that the intervention
had?
5. If no intervention took place, did the absence of a U.S. military
intervention itself substantially help improve or worsen U.S.
interests?
In this effort, we did not assess whether the intervention or nonin-

tervention played the dominant role in the outcome of the case, nor
did we weigh its contributions against the actions of other actors or
structural factors. Instead, we focused on identifying cases in which
the intervention appeared to play an unambiguous, independent role
in affecting the outcome. We took into account selection effect issues,
such as the fact that in interventions against more-capable adversaries
(e.g., more-difficult cases), larger forces were more likely to be commit-
ted. Further, we attempted to isolate the initial decision regarding the
size and timing of the intervention from later decisions regarding how
to execute the intervention within these parameters, including deci-
sions regarding rules of engagement, targeting, with which domestic
28 Specifically, we had two researchers review the focused case studies and independently
provide answers to these questions. Across all cases, 94 percent of the answers they provided
to these questions were the same. The remaining 6 percent of questions were discussed and
adjudicated as a group to develop a consensus set of responses.
29 When answering this question, coders were instructed to consider the responses to the
questions about the specific types of U.S. interests detailed in Table 4.1 as a whole, and then
identify whether there appeared to be clear evidence of the intervention having played an
important role in the advancement of these interests, according to the history of the case.
groups to attempt to cooperate, and a host of similar potential issues.

Although initial intervention decisions and execution-of-mission deci-
sions both likely affected the outcome, our analysis was focused only
on isolating the effects of the former. The following section reviews the
results of these assessments.
Overall Trends
The overall rates at which interventions contribute to achieving U.S.
interests do not definitively vary by the timing or size of the interven-
tion, although there are some suggestive patterns. According to our
analysis, U.S. interventions helped advance overall U.S. interests in
roughly 60 percent of the cases we studied. This rate varies somewhat
by size, with larger interventions more likely to advance U.S. interests
(roughly 75 percent of large interventions and 80 percent of medium
interventions) than smaller interventions (roughly 50 percent), but
these differences may or may not be significant.30 Adding to our cau-
tion about identifying clear patterns, large interventions encompass
some of the clearest success cases (such as the 1991 Gulf War) as well
as some of the clearest failures (the 2003 U.S. invasion of Iraq and its
aftermath).
Early and late interventions show nearly identical rates of having
contributed to the advancement of U.S. interests (roughly 60 percent).
U.S. interventions, then, have the potential to contribute to the achieve-
ment of U.S. interests across a range of timings and sizes. Although
these rates of success might be similar across categories, with more pos-
sible divergence according to size, this potential still leaves open the
possibility that some timings and sizes of interventions might assist the
achievement of U.S. objectives more than others, given the possibility
that, for example, the cases in which the United States has historically
chosen to intervene early might be more difficult to achieve U.S. inter-
ests than those in which it chooses to intervene late. The possibility
that there could be some independent effect of the timing and size of
the intervention will be considered in the next two sections.
30 This difference is certainly not statistically significant in the relatively small number of
cases we considered.
U.S. Interests and the Timing Decision

When we assess not just whether the intervention contributed to achiev-
ing U.S. objectives but also whether the timing and size characteristics
of the intervention were themselves important in this regard, we see
clearer evidence that intervening early can assist the advancement of
U.S. interests. As noted above, early and late interventions advance
U.S. interests in roughly the same percentage of our cases. However,
we can identify several early cases in which the fact that the inter-
vention occurred early in the crisis or war specifically appears to have
contributed to the advancement of U.S. interests. These cases typically
involve circumstances that appear to directly threaten a U.S. interest
and in which substantial delay would likely mean at least the near-term
sacrifice of that interest. For example, in Lebanon 1958 (Operation
Blue Bat), the United States rapidly deployed some 14,000 troops to
Lebanon to stabilize the pro-Western government there.31 The deci-
sive intervention, at a time when the government was near collapse,
likely was essential to maintaining Lebanon’s pro-Western orienta-
tion and, potentially its independence from the newly formed United
Arab Republic.32 A U.S. intervention that was substantially delayed by
even a few additional months, beyond the end of the term of President
Camille Chamoun, might have led to a different outcome.33 Similarly,
after the Kurdish uprisings following the 1991 Gulf War threatened
to be met with violent Iraqi government repression, the United States
and other nations moved rapidly to provide humanitarian assistance
and to establish a safe haven in northern Iraq.34 The safe haven was
enforced by a no-fly zone and the introduction of ground troops.35 Had
this intervention been substantially delayed, a humanitarian tragedy
31 Little, 1996, p. 27.

32 Little, 1996, p. 27.
33 Little, 1996, pp. 40–44.
34 James Cockayne and David Malone, “Creeping Unilateralism: How Operation Provide
Comfort and the No-Fly Zones in 1991 and 1992 Paved the Way for the Iraq Crisis of 2003,”
Security Dialogue, Vol. 37, No. 1, March 2006, pp. 125–128.
35 Alexander Benard, “Lessons from Iraq and Bosnia on the Theory and Practice of
No-Fly Zones,” Journal of Strategic Studies, Vol. 27, No. 3, September 2004, pp. 463–464;
likely would have occurred among Kurdish refugees, and any attempts
to establish a similar safe haven in northern Iraq might have involved
fighting Iraqi ground forces.36 This is not to suggest that even in simi-
lar circumstances early interventions will be sufficient, as illustrated by
the 1991 U.S. intervention to establish a no-fly zone over southern Iraq
to provide protection to Shiite demonstrators there.37 Nor is an early
decision always necessary. In retrospect, the 2003 U.S. invasion of Iraq
likely would not have been less successful had the decision to under-
take it been delayed further, even if it had eventually still been under-
taken. But in cases where local circumstances are imminently threaten-
ing a U.S. interest and the U.S. intervention has the ability to prevent
this threat, an early intervention can be clearly preferable to a later one.
On the other hand, we only identified one case in which the
lateness of the U.S. intervention per se might have been beneficial,
although the extent to which it was beneficial is unclear. The 1995
U.S. intervention into the conflict in Bosnia occurred after several
years of fighting and ineffective U.S. efforts to terminate the conflict
through other forms of pressure, which helped increase the suscepti-
bility of the Bosnian Serbs to pressure.38 Had the U.S. intervention
occurred much earlier, it potentially still could have achieved a similar
outcome for the host nation while reducing the dramatic humanitarian
toll of the war, but an earlier intervention likely would have required
the United States to bear greater up-front costs, potentially including
higher casualties. This counterfactual scenario is explored in detail in
Appendix B and summarized in Chapter Five. It illustrates potential
circumstances under which a later intervention might be preferable,
Gordon W. Rudd, Humanitarian Intervention: Assisting the Iraqi Kurds in Operation PRO-
VIDE COMFORT, 1991, Washington, D.C.: Department of the Army, 2004, p. 226.
36 Benard, 2004, pp. 463–464; U.S. Government Accountability Office, Operation Desert
Shield/Storm: Costs and Funding Requirements, Washington, D.C., GAO/NSIAD-91-304,
September 1991, p. 3.
37 Benard, 2004, p. 464; Phillip Gibbons, “U.S. No-Fly Zones in Iraq: To What End?”
Washington, D.C: Washington Institute, July 1, 2002.
38 U.S. Department of State, The Road to Dayton: U.S. Diplomacy and the Bosnia Peace Pro-
cess, May–December 1995, Washington, D.C., May 1997; Ivo H. Daalder, Getting to Dayton
The Making of America’s Bosnia Policy, Washington D.C.: Brookings Institution Press, 2000.
such as the greater exhaustion of the combatants, which could enable

an easier or more feasible intervention. Whether these benefits that can
occur from waiting will be preferable to intervening earlier and avoid-
ing the costs that also result from potentially years of warfare will, of
course, depend on the circumstances.
In most late intervention cases, however, we did not see such
potential trade-offs at work. Instead, in cases such as the 2014 inter-
vention into Iraq to roll back the advance of ISIL or the similar 2015
intervention into Libya, an earlier intervention likely would have been
preferable for U.S. interests or, at worst, broadly similar.39 Although
late interventions are generally less common than early ones are, and
therefore represent a smaller set of cases from which to generalize, this
overall pattern is worth noting.
U.S. Interests and the Size Decision
As noted earlier, large, medium, and small interventions appear to
have some moderate differences in the rates with which they allow the
United States to advance its interests. When assessing whether the size
of the interventions specifically contributed to advancing U.S. inter-
ests, however, we find more evidence that the decision about the size of
forces committed to an intervention matters.
Larger intervention forces could be helpful for a wide variety of
reasons, including signaling a greater U.S. commitment to the success
of the mission and providing more forces to undertake a wider number
and variety of combat and noncombat tasks. One might assume, then,
that larger interventions are, generally speaking, better. We find some
evidence for this proposition. The large-scale U.S. intervention to
reverse the Iraqi conquest of Kuwait in 1991, for example, was essen-
tial.40 A smaller U.S. force would have sent a mixed signal to the inter-
national community regarding the seriousness with which the United
States viewed Iraqi aggression, perhaps limiting the size of the interna-
39 Cameron Glenn, “Timeline: The Rise, Spread, and Fall of the Islamic State,” Wilson
Center, July 5, 2016; Chivvis, 2014, pp. 25, 69.
40 Freedman and Karsh, 1993, p. 409.
tional coalition that could be assembled and putting at risk victory over
Iraqi forces on the battlefield.
However, our review of the focused case studies emphasized that
making an intervention force larger should not always be assumed to
improve the intervention’s likelihood of success. There appear to be at
least four circumstances under which limiting the size of U.S. forces
in an intervention might be preferable. First, larger forces could raise
political concerns for the host nation, with the potential to undercut
the success of the U.S. intervention and the effective cooperation of
the host government. As an example, the medium-sized U.S. inter-
vention into Iraq following the 2014 capture of Mosul by ISIL was
eventually successful in its objective, helping to roll back the territorial
gains of ISIL and reestablish Baghdad’s control over most of the coun-
try, although the long-term trajectory of the country remains unclear.41
A larger-scale U.S. invasion potentially could have achieved similar
successes on the battlefield against ISIL (and perhaps done so more
quickly), but it also would have risked a substantial popular backlash,
given the recent history of the United States in Iraq, that could have
undermined Iraqi stability more broadly and plunged the country and
intervening U.S. forces back into a renewed civil war. In that case,
restraint with the size of the U.S. force likely was a virtue.
Second, a more limited force could enable a more rapid U.S.
response to a crisis. During the 2001 U.S. invasion of Afghanistan,
only limited forces could be rapidly introduced into that country in the
aftermath of the 9/11 attacks.42 However, the advantages that such a
medium-sized force provided in terms of speed outweighed the poten-
tial benefits of waiting longer for a larger force to become available,
and (along with air and naval assets) still proved to be sufficient for the
initial objective of overthrowing the Taliban.43
41 Glenn, 2016.
42 Joseph J. Collins, “Initial Planning and Execution in Afghanistan and Iraq,” in Rich-
ard D. Hooker, Jr., and Joseph J. Collins, eds., Lessons Encountered: Learning from the Long
War, Washington, D.C.: National Defense University Press, September 2015.
43 Collins, 2015.
Third, a more limited force could help manage potential escala-

tion risks that might accompany a larger U.S. intervention. During
the 1962 Cuban Missile Crisis, the rapid deployment and mobilization
of very sizable air and naval forces played an essential role in convinc-
ing Soviet decisionmakers of U.S. resolve and encouraging them to
back down.44 The ground component involved in this intervention was
more restrained, however, reflecting concerns about potential escala-
tion (among other things) if the Soviets and Cubans were to believe
that a full-scale U.S. invasion of Cuba were imminent.45 Although it
was important for the United States to avoid a small, limited response
that could have risked miscommunicating whether Soviet missiles in
Cuba did, in fact, cross a U.S. red line, it was equally beneficial to
moderate the size of the U.S. ground forces to avoid perceptions of
potentially precipitous aggression.
Fourth, the larger intervention force might simply be unneces-
sary, not in terms of harming the achievement of U.S. objectives in the
intervention but by soaking up scarce resources that perhaps could be
better used elsewhere. In the Dominican Republic in 1965, for exam-
ple, the United States intervened with what was later determined to
be an overabundance of troops, expecting greater levels of resistance
and violence than were present.46 The size of the force committed was
excessive, though it did not notably hamper the accomplishment of the
mission, which did broadly advance U.S. interests. We did not assess
the financial and opportunity costs of the larger intervention in detail,
44 Adam B. Siegel, The Use of Naval Forces in the Post-War Era: U.S. Navy and U.S. Marine
Corps Crisis Response Activity, 1946–1990, Alexandria, Va.: CNA, 1991; John M. Young,
When the Russians Blinked: The U.S. Maritime Response to the Cuban Missile Crisis, Wash-
ington, D.C.: History and Museums Division, Headquarters, U.S. Marine Corps, 1990,
p. 94; Graham T. Allison and Philip Zelikow, Essence of Decision: Explaining the Cuban Mis-
sile Crisis, 2nd ed., New York: Longman, 1999, p. 123; Edward T. Russell, “Cuban Missile
Crisis,” in A. Timothy Warnock, ed., Short of War: Major USAF Contingency Operations,
1947–1997, Maxwell Air Force Base, Ala.: Air University Press, 2000.
45 Burris, 1962; Michael Dobbs, “One Minute to Midnight: Kennedy, Khrushchev and
Castro on the Brink of Nuclear War,” National Security Archive, George Washington Uni-
versity, June 4, 2008; Kavanagh, Frederick, Stark, et. al, 2019.
46 Yates, 1988; Greenberg, 1986; Coleman, 2015.
but certainly a blanket policy of oversizing U.S. interventions would

have clearly negative effects.
It is notable that the cases discussed above all discuss the decision
between committing large forces to an intervention and committing
medium forces. The decision between large and small interventions
is likely less salient. Committing several hundred troops to an inter-
vention compared with committing hundreds of thousands of troops
would necessarily involve a radical reshaping of the mission and its
goals; that is, policymakers are unlikely to need to decide between the
two for the same end. But what about deciding between medium and
small interventions? Do we see any evidence regarding how policy-
makers should approach this decision, and do they differ from the
dynamics discussed above between large and medium interventions?
There certainly were small intervention cases in which it appears
that a medium-sized force could have been better able to achieve U.S.
interests. For example, in the small U.S. intervention into the Libyan
Civil War in 2011, medium-sized forces could plausibly have improved
the outcome from a U.S. perspective, particularly if these larger forces
allowed for greater stabilization of the country in the postconflict
period. Therefore, there are some instances in which more forces could
be preferable at this lower end of the decision space as well.
It is more difficult to assess whether the characteristics noted above
regarding when a smaller force might actually be preferable to a larger
one also apply to cases in which U.S. policymakers might be choosing
between a small- or medium-sized intervention force. Among the small
intervention cases that were broadly successful, we did not identify any
in which the small size of U.S. forces itself was clearly linked with this
success. For example, the August 1982 U.S. intervention in Lebanon
as part of a multinational force to evacuate PLO forces was broadly
successful, though it is not clear that a larger, medium-sized U.S. force
would have been less successful (however, it likely would have been
unnecessary).47 Similarly, a small U.S. intervention in Libya in 2015–
2016 that was designed to counter gains made by ISIL in that country
47 John H. Kelly, “Lebanon: 1982–1984” in Jeremy R. Azrael and Emil A. Payin, eds.,
U.S. and Russian Policymaking with Respect to the Use of Force, Santa Monica, Calif.: RAND
was broadly successful, but a moderately larger U.S. force might have
been equally successful (if not more so).48 This is not to suggest that the
conditions discussed above regarding when the political context, speed
of deployment, or escalation risks might make a smaller intervention
preferable cannot apply to the decision regarding whether to employ
a medium or small force. But we do not find clear evidence for these
conditions in the cases we considered. There do appear to have been
relevant cases regarding whether the size of the forces committed were
unnecessary. The 1983 U.S. invasion of Grenada involved a substantial
medium-sized force that met with very limited resistance from the tiny
island nation of Grenada.49 In this case, as with the 1965 intervention
in the Dominican Republic, the forces committed likely were overkill,
though they do not appear to have hampered the achievement of U.S.
objectives.
To sum up, then, we find evidence that the question of what size
force should be committed to an intervention to advance U.S. objec-
tives should be considered carefully because it can have a notable effect
on the likelihood of success in achieving U.S. objectives or advancing
U.S. interests. But with respect to whether such forces should be large,
medium, or small, we find that contextual factors are likely to be quite
important. We will discuss the implications of these findings in greater
detail later in this chapter.
U.S. Interests and Nonintervention Cases
The most-striking pattern among our nonintervention cases was how
rarely they tended to evolve to either improve or worsen U.S. interests
to any notable degree (again, absent U.S. military intervention). Over-
all, roughly three-quarters of cases saw the aftermath of the case as
broadly similar to the status quo ante bellum from the perspective of
U.S. interests. The Biafra War in Nigeria, for example, involved hor-
Corporation, CF-129-CRES, 1996; Caspar W. Weinberger, Fighting for Peace: Seven Critical
Years in the Pentagon, New York: Warner Books, 1990, pp. 144–145.
48 Eric Schmitt, “Obama Is Pressed to Open Military Front Against ISIS in Libya”, New
York Times, February 4, 2016.
49 Stewart, 2008, pp. 3–8; “Leader of Grenada Is Reported Killed by Troops,” New York
Times, October 20, 1983.
rific numbers of civilian deaths—in the millions—and was a traumatic

event for that country, but after its end in 1970, Nigeria remained a
reliable U.S. partner in the Cold War and a producer of oil, just as it
had been before the conflict.50 Failing to intervene to halt the conflict
and save lives does not appear to have notably undermined U.S. inter-
national leadership at the time. The 1953 crisis in Burma over the pres-
ence of Kuomintang troops that had taken up looting and resistance to
the government led, through the application of U.S. diplomatic pres-
sure on Taipei, to a partial withdrawal of troops that was enough to
convince Burma that the United States was no longer complicit in their
presence but that also did not fully resolve the underlying situation or
instability, which persisted for years afterward.51 But the United States
was able to preserve its relationships with both Burma and Taiwan
in roughly the same shape they were before the crisis. In other cases,
such as Yemen 2007, U.S. interests did notably decline absent a U.S.
intervention, but it is also not clear that there was any plausible U.S.
intervention that could have led to a notably better outcome.52 Overall,
however, although the United States is rarely entirely absent from even
these nonintervention cases and often applies diplomatic pressure or
uses economic leverage in lieu of military forces, most of the noninter-
vention cases we considered did not involve any notable alteration in
the status quo for U.S. interests.
Circumstances in Which Nonintervention Decisions Mattered
We did, however, identify a handful of cases in which U.S. interests
were notably altered in the absence of a U.S. military intervention and
in which the lack of intervention itself was likely an important factor
50 Brian Edward McNeil, Frontiers of Need: Humanitarianism and the American Involve-
ment in the Nigerian Civil War, 1967–1970, dissertation, University of Texas at Austin, 2014,
pp. 68–70; Lasse Heerten and A. Dirk Moses, “The Nigeria-Biafra War: Postcolonial Con-
flict and the Question of Genocide,” Journal of Genocide Research, Vol. 16, Nos. 2–3, 2014,
pp. 169–176; Kissinger, 1969, pp. 1–6.
51 Kenton Clymer, “The United States and the Guomindang (KMT) Forces in Burma,
1949–1954: A Diplomatic Disaster,” Chinese Historical Review, Vol. 21, No. 1, 2014,
pp. 41–42.
52 Salmoni et al., 2010, pp. 273–274.
in leading to this change. Of these cases, there were more in which the
lack of intervention led to positive outcomes for the United States than
led to negative outcomes. Next, we will survey three such cases that
led to positive outcomes, as well as a fourth case in which the opposite
happened: U.S. interests declined, and a U.S. intervention might have
helped.
Following a disputed election in Costa Rica in 1948, a mixed
group of Costa Rican political exiles with communist ties and Nicara-
guans invaded Costa Rica from Nicaragua. The United States declined
to intervene in the crisis militarily to restore stability or prevent an
increase in communist influence in the region. Instead, it supported
a diplomatic effort through the nascent Organization of American
States that succeeded in resolving the crisis, led to a durable transi-
tion to democracy in Costa Rica, and strengthened the Organization
of American States, enabling it to play a constructive role in future
regional crises.53 A U.S. intervention to impose a friendly government
in Costa Rica might have been able to achieve that narrower goal but
likely would have undermined or put at risk these other positive devel-
opments, which is roughly what occurred in the later U.S. intervention
to overthrow a communist-linked government in Guatemala in 1954.54
In the 2006 crisis over Iran’s nuclear program, the United States,
deeply enmeshed in the conflict in neighboring Iraq, declined to take
military action to attempt to destroy Iran’s emerging nuclear capabili-
ties. Exercising restraint at this time gave space both for the interna-
tional community to build a broad-based sanctions regime to pressure
Iran over its nuclear program and for Iranian leaders to conclude that
a diplomatic deal with the international community might be a prefer-
able arrangement.55 These developments led to the conclusion of the
Joint Comprehensive Plan of Action (JCPOA) in 2015, which commit-
53 Olander, 1996.
54 Barrett, 2007; Cullather, 1994.
55 Peter Baker, “Bush Warns Against Nuclear-Armed Iran; U.S. Presses Allies for U.N.
Action,” Washington Post, January 14, 2006; Ivo H. Daalder and Philip H. Gordon, “We
Should Strike Iran, but Not with Bombs,” Washington Post, January 22, 2006; David E.
Sanger, “Why Not a Strike on Iran?” New York Times, January 22, 2006.
ted Iran to dramatically scaling back its nuclear program in exchange

for sanctions relief.56 A 2006 U.S. military intervention might have
been able to temporarily destroy Iranian enrichment facilities, but it
also likely would have convinced the Iranian government of the aggres-
sive intent of the United States and strengthened the determination of
Iran to restart a covert nuclear weapons program in the future.
As a third example, a destructive civil war began in Colombia in
1948, the end result of intense political tensions between Liberals and
Conservatives and the assassination of the Liberal Party leader, as Lib-
eral guerillas fought bloody battles with the Conservative government
and militias and plunged the country into chaos.57 Despite a close his-
torical relationship with the Colombian government, the United States
opted not to intervene to halt the violence directly, although it provided
the government with substantial military and economic aid.58 Beyond
a desire for stability and continued partnership with Colombia, the
United States was also concerned with a potential upsurge in support
for communism. After a decade of intense violence, the Liberals and
Conservatives reached a power-sharing arrangement that returned a
measure of stability to the country and kept it closely aligned with the
United States. The decision to limit U.S. involvement to diplomatic
efforts and aid, while doing little to limit the tremendous civilian toll
from the conflict, likely helped avoid encouraging anti-American com-
56 The Trump administration later decided to unilaterally withdraw from the JCPOA in
favor of renewed sanctions and an effort to modify Iranian regime behavior in other areas,
which has increased tensions. However, this later action does not alter the assessment that
the nonintervention in 2006 led to an improvement in U.S. interests in the 2006 to 2016
period because international sanctions coordination increased leverage over Iran’s decision
to pursue a nuclear weapons program. Whether that increased leverage was best used in the
negotiation of the 2015 JCPOA is a separate issue from an assessment of the likely effects of
having forgone a military strike against Iran in 2006. See Zachary Laub and Kali Robinson,
“What Is the Status of the Iran Nuclear Agreement?” Council on Foreign Relations, updated
January 7, 2020.
57 World Peace Foundation, “Colombia: La Violencia,” Fletcher School, Tufts University,
December 14, 2016.
58 U.S. Department of Defense, Security Cooperation Agency, “Fiscal Year Series: Foreign
Military Sales, Foreign Military Construction Sales and Other Security Cooperation His-
torical Facts,” September 30, 2016.
munist sentiment and helped keep Colombia a key U.S. ally during the
Cold War.
Although these cases illustrate how nonintervention can improve
U.S. interests, we also identified one case in which a U.S. interven-
tion could very plausibly have improved U.S. interests. In Nicaragua in
1978, decades of harsh dictatorship had produced widespread opposi-
tion to the government and a growing communist-linked insurgency
in the Sandinistas. The United States sought to apply diplomatic and
political pressure to force President Anastasio Somoza Debayle to
either reform or leave office, the latter of which he did in July 1979.59
Although the United States had made plans for a transition to a
national reconciliation government that included the moderate opposi-
tion, it decided not to commit any military forces to ensure stability,
and Sandinista forces were able to take advantage of the power vacuum
and seize power themselves.60 This marked a clear reversal for U.S.
efforts to contain communism in Latin America, and it was a rever-
sal that the United States proved unwilling to let stand. Throughout
the 1980s, the United States provided extensive military support to
the anti-Sandinista Contras, widening and prolonging the Nicaraguan
civil war and helping to destabilize the broader region.61 This aid con-
tinued illegally even when prohibited by Congress, eventually leading
to the 1987 Iran-Contra scandal. Although an intervention was not
seriously considered by the Carter administration in 1979, a relatively
modest U.S. intervention in that year could plausibly have helped to
stave off a Sandinista seizure of power in the first place and to ensure
that the transition to the moderate opposition that the United States
had envisioned actually took place.
We highlight two key lessons that can be drawn from these
notable nonintervention cases. First, the international context can be
vital when determining whether wars or crises are likely to worsen or
improve on their own. In 1940s Latin America, the United States held
59U.S. Department of State, “Milestones in the History of U.S. Foreign Relations: Central
America, 1977–1980,” webpage, undated.
60 Presidential Review Committee, 1979.
61 “U.S. Aid to the Contras,” 1988.
tremendous sway, and outside actors such as the Soviet Union were not
yet capable of dramatically influencing events. In both Costa Rica and
Colombia, the United States could feel reasonably confident that the
diplomatic and other policy levers at its disposal could help shape events
in the absence of well-funded spoilers. Similarly, there was broad-based
international concern with the Iranian nuclear weapons program, even
from some of Iran’s erstwhile partners, such as Russia, enabling the
United States to build up international pressure to apparently prompt a
change in Iranian thinking about the utility of its nuclear weapons pro-
gram. In late-1970s Latin America, by contrast, the Soviet Union had
well-established funding networks and proxies in the region, enabling
it to fund spoilers, such as the Sandinistas, that could divert events in
ways the United States proved unable to counter absent the application
of military force. Assessing the regional context, particularly the influ-
ence and motivations of potential third-party spoilers, can therefore
be an important step when determining whether a war or crisis might
require U.S. military intervention to safeguard U.S. interests.
Second, the relative strength and influence of the different fac-
tions on the ground is of course a vitally important factor in deter-
mining the outcome of a war or crisis, even absent any external inter-
ference. In Colombia 1948, for example, the United States remained
confident in the ability of government forces to ultimately hold on to
power throughout a decade of violence, making the decision not to
intervene on the government’s behalf easier.62 In Iran 2006, most U.S.
analysts assessed that the likely result of military strikes against the
Iranian regime would not have been instability and regime change but
instead a reinvigorated popular backing for an otherwise increasingly
unpopular regime, strengthening the case for pursuing diplomatic and
economic leverage.63 In Nicaragua, the United States misjudged the
capabilities and support of moderate groups, leaving a vacuum that
could be filled quickly by the Sandinistas. Accurate intelligence and
62Central Intelligence Agency, “The Current Situation in Colombia,” Situation Memoran-

dum 32-49, May 31, 1949.
63 Charles P. Blair, “War with Iran? Revisiting the Potentially Staggering Costs to the
Global Economy,” Federation of American Scientists, May 23, 2013.
assessments of the relative strengths of different actors can therefore be

vitally important to the intervention decision.
Intervention Timing, Size, and the Host-Nation Trajectory

To this point, we have focused on summarizing changes in U.S. inter-
ests that occurred in intervention and nonintervention cases. It is also
worth noting, however, that we see some notable trends in the tra-
jectory of the host or target nation that appear to correlate with the
intervention decision. Most notably, we observe many more positive
trajectories in intervention cases than in nonintervention cases. In the
20 nonintervention cases we considered, only one had a clearly posi-
tive trajectory following the end of the war or crisis: Costa Rica 1948.64
Most of the remainder followed a mixed or uneven trajectory, although
there was also a substantial minority of cases with clearly negative tra-
jectories for the host nation, such as Yemen 2007, Rwanda 1959, Soma-
lia 1976, and Nicaragua 1978.65
Although there were numerous intervention cases that also saw
negative trajectories for the host nation, such as Iraq 1991, Guate-
mala 1953, and Somalia 1991, there were also numerous cases that
saw positive trajectories, including Lebanon 1958, Panama 1989,
Kuwait 1990, and Bosnia 1995. These positive host-nation trajectories
tended to occur more often in large or medium interventions, though
of course this relationship was not universal, as well-known cases such
as Iraq 2003 illustrate. Nonetheless, conditions in host nations only
64 Olander, 1996; Fernando F. Sanchez, Democracy After War: Causes and Consequences of
the 1948 Civil War in Costa Rica, paper presented at the European Consortium for Political
Research Joint Sessions of Workshops, Uppsala, Sweden: Uppsala University, April 13–18,
2004; Desmond A. Turner, Civil War and (Re)Building the State: Recovery in Costa Rica,
Nicaragua, and El Salvador, thesis, Coral Gables, Fla.: University of Miami, May 2017.
65 Exum, 2017; Hilsman, 1962; United Nations, “Rwanda: A Brief History of the Country,”
webpage, undated a; U.S. Department of State, “Interagency Intelligence Memorandum,”
Foreign Relations of the United States, 1977-1980, Vol. XVII: Horn of Africa, Part 1, Washing-
ton, D.C., July 31, 1979; U.S. Department of State, “Memorandum from Paul B. Henze of
the National Security Council Staff to the President’s Assistant for National Security Affairs
(Brzezinski),” Foreign Relations of the United States, 1977-1980, Vol. XVII: Horn of Africa,
Part 1, Washington, D.C., June 3, 1980; John M. Goshko, “U.S. War by Proxy at an End,”
Washington Post, February 27, 1990.
rarely improved absent a U.S. intervention, at least in the crisis and war
cases we considered; when such interventions were conducted, positive
trajectories became more frequent, particularly when the intervention
was larger.
Implications and Conclusion
In the focused case studies we conducted, U.S. decisions regarding

whether, when, and with what size forces to conduct a military inter-
vention have not led to many consistent effects on outcome measures,
such as the advancement of U.S. interests. Both clear successes and
failures have historically accompanied nearly the full range of these
decisions. We therefore do not find simple rules of thumb regarding
how the United States should approach these decisions, always lean-
ing in one direction or the other. One exception to this is the find-
ing that U.S. military interventions typically have substantial net eco-
nomic costs for the United States, roughly in proportion to their size
and duration. Interventions with ancillary economic benefits in terms
of greater future investment or trade opportunities sufficient to offset
the direct costs of the intervention were rare.
What we have identified instead are a series of contextually depen-
dent patterns in which U.S. intervention decisions in certain local or
international contexts have tended to produce advantageous or disad-
vantageous results. Next, we summarize these patterns as they relate
to each of the three key guiding questions of this report: whether the
United States should intervene, whether it is preferable to intervene
early or late in a crisis or conflict, and with what size forces it might
most effectively intervene.
Insights into Intervention Decisions

The analysis in this chapter provides relevant insights for each of the
three key guiding questions of this report.
When Should the United States Choose Intervention over

Nonintervention?
Our analysis of the focused case studies suggests that military inter-
ventions are both highly likely and likely to occur early when the
United States feels its national security is directly threatened. Alter-
native intervention decisions in such cases seem improbable. Outside
of such unambiguous cases, however, our focused case studies suggest
that interventions are likely to be most beneficial in situations in which
a war or crisis is likely to deteriorate in a way that will harm U.S.
interests, and U.S. intervention can make a difference in avoiding that
outcome. Careful analysis of the local balance of power among key
actors and the preferences of key regional and international actors will
be essential in accurately making such a determination. The presence
of third-party spoilers can be an important indicator that a situation
could deteriorate to harm U.S. interests absent a military intervention.
However, when the balance of power is more ambiguous or currently
favors the U.S.-friendly side, and when third-party spoilers are absent
(or present but so influential that a U.S. intervention might not be able
to overcome them), the United States could benefit from deferring an
intervention. Some situations could be likely to deteriorate but still lack
a realistic pathway for a U.S. intervention to improve them.
If an intervention does not seem either necessary or promising on
these grounds, then U.S. decisionmakers could note that, in the cases
we assessed, a decision not to intervene was more frequently associated
with an improvement in U.S. interests than a decline. We did not,
for example, assess any nonintervention cases in which the U.S. repu-
tation (either for resolve or for its humanitarian responsibilities) was
substantially and negatively affected. There were some cases in which
U.S. strategic interests declined as a result of nonintervention, but they
typically were less frequent than cases in which U.S. strategic inter-
ests improved as a result of the decision not to intervene. Although of
course this finding does not necessarily suggest that the United States
should intervene less frequently and expect to see this pattern continue,
it does seem to suggest that the United States likely does not need to
intervene more frequently, or in more circumstances, than has been its

historical pattern in order to safeguard its interests.66
When Should the United States Intervene Early, and When Should It
Intervene Later?
The focused cases clearly suggest that the United States is likely to
respond immediately if its national security is directly threatened.
Absent such a condition, our focused cases suggest that interven-
ing early could provide some independent benefit to advancing U.S.
national interests in cases in which U.S. interests are imminently threat-
ened and in which the U.S. intervention has the potential to relieve this
threat. We only identified one potential case in which intervening later
in a conflict or crisis, regardless of the issues motivating the interven-
tion, might have provided an advantage (caused by the greater exhaus-
tion of the primary target of the intervention). Although the literature
suggests plausible additional theoretical circumstances in which wait-
ing to intervene later could be preferable, we did not see these circum-
stances reflected in the limited set of cases we assessed.
When threats to U.S. interests are clearly diffuse, however, the
benefits of intervening early are much less clear. The focused cases
suggest that when the stakes involved in a war or crisis are somewhat
lower (such as overall geopolitical concerns), the United States has
often waited without adversely affecting the overall rates of success,
even if it later decided an intervention was warranted. In our cases, the
United States rarely experienced substantial costs from waiting with
these types of issues at stake.
66 To be clear, there are substantial endogeneity concerns with interpreting this observed
pattern. The United States would seem much more likely to intervene in cases in which it
assessed acute threats to its interests, such that the remaining set of nonintervention cases
would not be expected to be replete with cases of sharp deterioration in these interests (absent
dramatic failures in U.S. analyses or limitations on forces available for intervention). We
therefore do not view this finding as speaking directly to whether the United States might be
“over-intervening.” This finding is, however, notable in that it does not suggest any evidence
that the United States could instead be “under-intervening.”
When Should the United States Use a Larger Force Rather than a
Smaller One?
The focused cases suggest that large or medium interventions might
more often advance U.S. interests than small interventions do. When
facing larger or stronger adversaries in combat or when the potential
intervention needs to affect a wide area or large population, the use of
a larger force could be essential. Even in other cases, however, a larger
force could still help achieve U.S. objectives and, at worst, might be
redundant or unnecessary.
Although the benefits of larger intervention forces are clear in
some cases, we also identified several circumstances in which more-
limited forces are likely to be preferable and to better support the
advancement of U.S. interests. These circumstances include when a
larger force might produce a popular backlash in the host nation, when
a larger force might enhance the risks of escalation with an adversary,
and when waiting to mobilize a larger force might limit the timeliness
(and therefore the effectiveness) of the U.S. intervention. Intervention
decisions should be informed by an analysis of whether these condi-
tions appear to be present in the particular circumstances being con-
sidered for an intervention.
Conclusion
In this chapter, we considered what can be learned about the costs and
benefits of different intervention decisions across a large set of shorter,
focused case studies. In the next chapter, we look in much greater
depth at a smaller number of cases, exploring not only the real world
intervention decisions the United States made but also possible coun-
terfactual scenarios. We use the counterfactual case studies to further
explore the observations made in Chapters Two through Four, with
a particular focus on identifying those key factors and characteristics
most likely to influence the relationship between intervention timing
and size and the ability of the United States to advance and protect its
national interests.
CHAPTER FIVE
Counterfactuals
The final method we employed to assess the effects of different U.S.

intervention timing and size decisions was a series of four in-depth
counterfactual case studies. In these counterfactual cases, we not only
provided more-detailed investigations of the historical effects of the
U.S. intervention decision than was available in the shorter focused
case studies detailed in Chapter Four, but also explored how the out-
come of each case could have been altered if the United States made
a different intervention decision. These counterfactual cases provide
useful, realistic illustrations of the dynamics involved in specific situ-
ations and the effects that different U.S. intervention decisions might
have. With only four cases, we must be cautious not to generalize
observations from these four examples to the universe of cases. Still,
there are some patterns we identify that appear to have broader impli-
cations. First, our counterfactual analyses suggest that, in some cases,
even radically different U.S. intervention decisions might not radically
alter the trajectory of the conflict or crisis. In other words, interven-
tions are only one factor among many in determining the outcome of
most crises or wars. Second, although the medium- and longer-term
implications of U.S. interventions appear to be more modest in our
four cases, the timing and size of the intervention might have more
substantial short-term effects, especially for the host nation. In this
chapter, we will also explore the conditions under which U.S. leverage
over outcomes through intervention appears to be relatively greater.
This chapter first provides an overview of the methodology we
employed when conducting the counterfactual cases, situating our
109
efforts within the political science literature on the method. We then

outline the specific cases we selected for study and the results of each
counterfactual investigation before concluding with a discussion of the
implications we identify from this method.
Methodology
Constructing Counterfactuals
Counterfactuals are imagined alternative versions of real-world events
in which key features are altered. For example, consider the trajectory
of the Vietnam War had John F. Kennedy not been assassinated or the
path of the United States over the past 15 years if the 2003 intervention
in Iraq had not occurred. Counterfactuals can be a useful tool to evalu-
ate causal theories because they allow researchers to explore how such
imagined changes might affect a real-world outcome.1 Counterfactual
logic is necessarily embedded in any research that aims to understand
causation: If a researcher finds that variable X is necessary for a cer-
tain outcome, then the absence of variable X (i.e., in the counterfac-
tual scenario) should also be sufficient to produce a different outcome.
This reasoning is built into most statistical analysis and qualitative case
study research in the social sciences.2
Political scientists and historians have identified several counter-
factual ideal types that can broadly be sorted into two baskets: sta-
tistical and computer-assisted approaches and qualitative approaches.
Statistical and computer-assisted methods are typically used to predict
outcomes that are measured numerically, such as the number of civil-
1 See James D. Fearon, “Counterfactuals and Hypothesis Testing in Political Science,”

World Politics, Vol. 43, No. 2, January 1991; Richard Ned Lebow, “What’s So Different
about a Counterfactual?” World Politics, Vol. 52, No. 4, July 2000; Jack S. Levy, “Counter-
factuals and Case Studies,” in Janet M. Box-Steffensmeier, Henry E. Brady, and David Col-
lier, eds., The Oxford Handbook of Political Methodology, Oxford, United Kingdom: Oxford
University Press, 2008; and Philip E. Tetlock and Richard Ned Lebow, “Poking Counterfac-
tual Holes in Covering Laws: Cognitive Styles and Historical Reasoning,” American Political
Science Review, Vol. 95, No. 4, December 2001.
2 Jack S. Levy and Gary Goertz, eds., Explaining War and Peace: Case Studies and Necessary
Condition Counterfactuals, New York: Routledge, 2007, p. 15.
Counterfactuals 111
ian casualties or the annual growth of a country’s GDP, while qualita-

tive counterfactual case studies are more useful in capturing nonnu-
merical outcomes.
Statistical and computer-assisted approaches alter one explana-
tory variable in a real-world case to estimate how the case might have
turned out differently. For instance, Abadie, Diamond, and Hainm-
ueller use statistical methods to create a synthetic control: essentially,
a manufactured control case to the case of interest.3 The control case
is constructed to have characteristics and properties drawn from the
broader universe of relevant cases so as to be statistically similar to the
case of interest, except for exposure to the treatment or the key vari-
able in the analysis. Analysts can then estimate the effect of a key vari-
able by comparing the outcome of the real-world case to the estimated
outcome of the synthetic case. For example, Abadie and Garbeazabal
use synthetic controls to estimate the economic effects of terrorism in
Basque Country, Spain, in the late 1960’s, comparing per capita GDP
in the region (which has suffered from extensive terrorism) to a syn-
thetic control region without terrorist conflict that is otherwise similar
to Basque Country.4 Computer-assisted, agent-based models have also
been used to generate quantitative estimates for counterfactual scenar-
ios. This method uses computer simulation to understand nonlinear,
networked relationships between the model’s inputs and outputs. For
example, Lustick, Garces, and McCauley build an agent-based model
to identify critical junctures in the Syrian civil war between 2011 and
2014 to understand how U.S. policy decisions affected the number of
civilian casualties.5
Qualitative counterfactual approaches draw on similar logic,
assessing whether the observed outcome of a case would likely differ
3 Alberto Abadie, Alexis Diamond, and Jens Hainmueller, “Comparative Politics and the
Synthetic Control Method,” American Journal of Political Science, Vol. 59, No. 2, April 2015.
4 Alberto Abadie and Javier Garbeazabal, “The Economic Costs of Conflict: A Case Study
of the Basque Country,” American Economic Review, Vol. 93, No. 1, March 2003.
5 Ian Lustick, Miguel Garces, and Thomas McCauley, “An Agent-Based Model of Counter-
factual Opportunities for Reducing Atrocities in Syria, 2011-2014,” Washington, D.C.: Simon-
Skjodt Center for the Prevention of Genocide, August 2017.
if the value of a key explanatory variable were different. Qualitative

approaches typically use within-case process tracing to identify criti-
cal junctures leading up to an empirically observed outcome. Process
tracing is a means of identifying causal mechanisms by closely inves-
tigating the relationships among sequential events within a particu-
lar case.6 Qualitative approaches often involve developing competing
counterfactual scenarios for key critical junctures within a case, then
evaluating the evidence for the competing counterfactuals. For exam-
ple, Frank P. Harvey evaluates the counterfactual claim that the 2003
Iraq War would not have happened had Al Gore been elected U.S.
president in 2000. He lays out three competing counterfactual sce-
narios, then traces multiple decisionmaking junctures on the path to
the 2003 war, exploring the evidence in favor of each counterfactual
scenario.7 Similarly, Mona Yacoubian identifies five critical junctures
in U.S. policy debates where policymakers considered multiple options
regarding the Syrian civil war. For each critical juncture, she identifies
plausible alternative decisions using these policy debates, then assesses
the likely effect of implementing the alternative policy.8
Our review of the literature highlights four accepted best prac-
tices for conducting rigorous counterfactual analysis:
1. Co-tenability (logical consistency). Researchers using coun-

terfactuals need to make sure that their counterfactual assump-
tions are logically consistent (or “co-tenable”) with the theories,
facts, and initial conditions that are used to render the causal
inferences.9 For example, Robert Fogel has argued that if rail-
roads had not existed, the U.S. economy in the 19th century
6 Alexander L. George and Andrew Bennett, Case Studies and Theory Development in the
Social Sciences, Cambridge, Mass.: MIT Press, 2005, p. 147.
7 Frank P. Harvey, “President Al Gore and the 2003 Iraq War: A Counterfactual Test of
Conventional ‘“W’”isdom,” Canadian Journal of Political Science, Vol. 45, No. 1, March
2012.
8 Mona Yacoubian, “Critical Junctures in United States Policy Toward Syria: An Assessment
of the Counterfactuals,” WashingUon, D.C.: Simon-Skjodt Center for the Prevention of Geno-
cide, August 2017.
9 Fearon, 1991, p. 193.
Counterfactuals 113
would have seen similar growth anyway because of a stronger

motivation to invent the internal combustion engine at an ear-
lier date.10 However, according to Jon Elster’s counterargument,
if the technology for inventing automobiles existed, it would
almost certainly have allowed the development of railroads;
therefore, Fogel’s counterfactual argument is not co-tenable
because the counterfactual that he imagines is not internally
logically consistent.11
2. Clarity. Every causal argument must be defined as clearly as
possible: What is to be explained (the consequent in the counter-
factual arguments), what accounts for this outcome (the anteced-
ent), and the causal path linking the consequent and antecedent
should each be carefully defined.12 Furthermore, the antecedent
should be realistic. Lebow gives an example: Historians have
raised the counterfactual argument that posits that if public
health measures were implemented in time, the 14th-century
Black Death pandemic mortality rate could have been lower.
However, to implement public health measures, a ban on travel
and trade would also need to be imposed. The wealthy and mer-
chant classes at that time would have been unlikely to allow
such measures to be implemented; the counterfactual argument
might therefore be unrealistic for the 14th century, when the
Black Death pandemic occurred.13
3. Theoretical consistency and relevance. Scholars using coun-
terfactual analysis should articulate credible causal mechanisms
between the antecedent and the hypothesized consequent.
10 Robert William Fogel, Railroads and American Economic Growth: Essays in Econometric
History, Baltimore, Md.: Johns Hopkins Press, 1964, cited in Lebow, 2000, p. 582.
11 Jon Elster, Logic and Society: Contradictions and Possible Worlds, New York: John Wiley,
1978, pp. 204–208, cited in Lebow, 2000, p. 582.
12 Philip E. Tetlock and Aaron Belkin, eds., Counterfactual Thought Experiments in World
Politics: Logical, Methodological, and Psychological Perspectives, Princeton, N.J.: Princeton
University Press, 1996.
13 Geoffrey Hawthorn, Plausible Worlds: Possibility and Understanding in History and the
Social Sciences, New York: Cambridge University Press, 1991, pp. 31–60, quoted in Lebow,
2000, p. 582.
“Well-established theoretical laws”14 should be a vital compo-

nent that drives the analysis because, without a sound theory,
there could be an infinite number of possible consequences (or
outcomes) of the counterfactual scenario.15
4. Minimum-rewrite (historical consistency). Weber argued
that counterfactual analysis should make as few changes as pos-
sible from the real world.16 One way to ensure plausibility and
minimal rewrites is to focus on the set of choices that politi-
cal leaders faced, especially when they came close to making
another choice. Different scholars called the same concept the
“closest possible world” test,17 the “minimal-rewrite of history”
rule,18 or the “horizon of possibility.”19 These rules apply to the
antecedent, not to the consequent.20 That is, they involve a min-
imal change to the preceding circumstances or actions, while
the later outcome might vary minimally, dramatically, or not
at all. Analysts must recognize the interconnectedness of causes
and outcomes and therefore consider what else in a historical
event might change if the explanatory event changes and how
such changes might interact to potentially affect the outcome.21
Counterfactual analysis should also refrain from several common

pitfalls. Lebow warns that if researchers make too many leaps from
hypothetical antecedent (explanatory variable) to hypothetical con-
sequent (outcome to be explained), the probability of the consequent
14 Tetlock and Belkin, 1996, pp. 25–27.

15 Levy, 2008, p. 395.
16Max Weber, On the Methodology of the Social Sciences, ed. and trans. Edward A. Shils and
Henry A. Finch, Glencoe, Ill.: Free Press, 1949, Ch. 3, quoted in Levy, 2008.
17 Hawthorn, 1991, pp. 31–60.
18 Tetlock and Belkin, 1996, pp. 23–25.
19 Martin Bunzl, “Counterfactual History: A User’s Guide,” American Historical Review,
Vol. 109, No. 3, June 2004, p. 857.
20 Levy, 2008, p. 390.
21 Lebow, 2000.
Counterfactuals 115
occurring becomes increasingly smaller (compound probability).22 That

is, counterfactuals should avoid overly long or complicated chains of
reasoning and focus on simpler, more-direct relationships. Addition-
ally, we cannot necessarily assume that one aspect of the past could
be changed while everything else is held constant.23 As noted above,
scholars therefore need to recognize the interconnectedness of causes and
outcomes and consider what else might change if the explanatory vari-
able were to change.24 Finally, scholars should be cognizant that history
does not stop when the hypothesized outcome is reached: Subsequent
events could return history to its original course. Scholars using coun-
terfactual analysis should therefore be careful to ask whether history
might revert to the course from which it was diverted regardless of a
change in one or a few explanatory variables.25
We used these best practices and endeavored to avoid these pit-
falls when conducting our counterfactual case studies. Broadly speak-
ing, each counterfactual case first develops a detailed model of the
key actors in each case, including their interests and capabilities, then
details how they interacted to produce historical events. Next, we spec-
ify the counterfactual being considered (specifically, the change in a
U.S. intervention decision) and the capabilities that the United States
would or could most realistically have brought to bear, as appropriate.
After stipulating the change in the U.S. intervention decision, the case
then explores the likely outcome of the case, given what has already
been established about the interests and capabilities of the other actors
involved. Each case concludes by assessing the similarities and differ-
ences between the historical and counterfactual events for a variety of
outcomes, including the effects on the conflict or crisis, the trajectory
of the host nation, and the effects on U.S. interests.
22 Lebow, 2000, p. 574.

23 Lebow, 2000, p. 576.
24 Tetlock and Belkin, 1996.
25 Tetlock and Belkin, 1996.
Case Selection
In total, we conducted four in-depth counterfactual case studies.26 To
select the specific four cases we would study, we applied several criteria.
First, we wanted to select cases in which a different U.S. intervention
decision could have been broadly plausible. On the one hand, a U.S.
decision not to intervene in the Cuban Missile Crisis would have been
relatively difficult to justify, given the stakes involved for the United
States. On the other hand, U.S. intervention in the ongoing (as of this
writing) civil war in the Central African Republic also appears highly
improbable, given the lack of clear U.S. interests involved. Selecting
cases in which both intervention and nonintervention (or interven-
tion of a different size and timing) could plausibly have occurred helps
increase the utility of the case. Second, we focused on cases whose
histories are relatively better known and better documented. This was
for pragmatic purposes, to ensure that we would be able to develop the
detailed understanding of the actors and their capabilities and interests
necessary to conduct a rigorous counterfactual investigation. Third, we
wanted to explore counterfactual cases that varied by their interven-
tion category. That is, we wanted a mixture of nonintervention cases,
intervention cases, and interventions with different timings and sizes,
to the extent feasible with only four cases. Fourth, we also wanted to
select cases that varied according to the issues that they presented for
U.S. policymakers. We wanted to include some cases that focused on
more-geostrategic issues, while others included an important humani-
tarian component. Finally, to the extent possible, we wanted to achieve
variety in the region and time in which the cases took place.27
26 Although more in-depth cases studies are always preferable, the scope of this report lim-
ited us to conducting four such cases.
27 Of course, with this many dimensions under consideration and only four cases to select,
identifying cases that would allow us to directly assess the causal effects of intervention deci-
sions on outcomes across each category was not possible. However, these counterfactual cases
were not intended or designed for the testing of novel hypotheses in this manner. Instead,
they were intended as an illustrative tool. They represent a plausibility demonstration but
should not, on their own, be taken to represent clear evidence of generalizable relationships.
Counterfactuals 117
We selected the following four cases, listed in Table 5.1: the

Korean War, the Iranian Revolution, the Bosnian War, and the Libyan
Civil War.28
Summary of the Counterfactual Cases
In this section, we briefly summarize each of the counterfactual cases

we conducted, including narratives of both the historical events and
what occurred in the counterfactual scenario. The full text of each of
these cases and our complete analysis of them are available in Appen-
dix B.
The Korean War

Historically, the Korean War erupted when North Korea invaded the
South early in the morning of Sunday, June 25, 1950. The United States
presented a resolution at the UN Security Council and condemned
North Korea for its attack. When North Korea did not withdraw to
the 38th parallel despite the UN’s request, the United States intervened
in the conflict as part of a multinational coalition under the banner
Table 5.1
List of Selected Counterfactual Cases
Historical U.S. Counterfactual U.S.

Case Name Year(s) Intervention Decision Intervention Decision
Korean War 1950–1953 Early, large intervention No intervention
Iranian Revolution 1978–1979 No intervention Early, medium

intervention
Bosnian War 1992–1995 Late, medium Early, medium

intervention intervention
Libyan Civil War 2011 Early, small intervention No intervention
28 It is worth noting that although two of these cases (Libya and Bosnia) were investi-
gated as focused case studies, as discussed in Chapter Four, we did not require that cases
either were or were not investigated using both methods. The degree of overlap, therefore, is
coincidental.
of the UN. Before these forces arrived en masse, the battle-hardened

North Korean forces, with many veterans of the Chinese Civil War,
achieved notable early successes. However, following a daring Septem-
ber amphibious landing behind North Korean lines at the port city of
Incheon, UN forces recaptured Seoul and began to roll back the invad-
ing North Korean forces. By November 1950, UN forces had reversed
the North’s successes and penetrated deep into North Korea, approach-
ing the border with China. By January 1951, the People’s Republic of
China had dispatched hundreds of thousands of “Chinese People’s Vol-
unteers” to back crumbling North Korean forces. The Chinese inter-
vention reversed the tide of the war once again, but battle lines soon
stabilized not far from the 38th parallel, the original line demarcating
the border between North and South Korea. The parties came to real-
ize that changing the front line further would be costly, and lengthy
negotiations for an armistice began. After lengthy delays, the major-
power participants finally reached an agreement essentially codifying
the battlefield results and signed the armistice on July 27, 1953.
In our counterfactual case, the United States decides not to inter-
vene to assist South Korea after the North Korean invasion. With-
out timely help from the United States and international forces, South
Korea was largely conquered by the North in a matter of weeks, as a
result of the North’s superior military capabilities. After the end of the
conflict, Korea’s trajectory in the immediate postwar years would have
substantially differed from the real-world case. The country would
have been unified under the leadership of Kim Il Sung and would
have formed close relations with other communist states, including the
Soviet Union and the People’s Republic of China. Economic condi-
tions in Korea following this much shorter war would have improved
to a higher level, at least initially, than seen in the real-world case, as a
result of having avoided the devastation wrought by the lengthy con-
flict in the historical case; however, the medium- and longer-term pros-
pects would have been more negative.
The Iranian Revolution

The Iranian revolution began in January 1978, when protests in the
Shi’ia holy city of Qom erupted after an article critical of the popu-
Counterfactuals 119
lar exiled Ayatollah Ruhollah Khomeini was published in a conserva-

tive newspaper. Over the course of 1978, the protest movement grew
and spread across the country. As late as October 1978, U.S. officials
believed that the Shah would be able to successfully put down the pro-
tests because his government had cracked down on opposition move-
ments before. But the Shah’s tepid response to the protests was not
sufficient to quell the revolution as it spread, and on January 16, 1979,
the Shah fled the country. On February 1, 1979, Ayatollah Khomeini
returned to Iran and was greeted by crowds of millions. Within ten
days of his return, Iran’s military leadership declared that the military
would subsequently remain neutral and withdrew to the barracks, leav-
ing open the ability for Khomeini to form a new regime. This regime
replaced a long-standing ally of the United States with an oppositional
regime that took U.S. embassy staff hostage and has since played an
antagonistic role toward U.S. allies and policy preferences in the region.
In the counterfactual scenario, President Carter decides to support
a military coup in Iran in December 1978 to forestall a revolution that
would likely yield an opposition government with an anti-American
orientation. At the same time, the United States deploys the 3rd bri-
gade of the 82nd Airborne and a battalion-sized Marine amphibious
unit to secure key Iranian oil facilities in the south of the country. An
additional Marine unit is also sent to Tehran to supervise the evacua-
tion of U.S. citizens, the Seventh Fleet Carrier Task Force is moved to
the Indian Ocean, and several U.S. Air Force F-111 Aardvark fighter-
bombers are deployed to the Cairo West airbase in Egypt as a sign of
U.S. support for the coup and as a deterrent signal to the Soviet Union.
The streets are quiet for a few months following the coup.
Although few Iranians return to work, and strikes are therefore still in
effect in practice, there are no protests. However, discontent with the
government persists, and discontent grows as reports of heavy-handed
actions from the military government grow and Khomeini denounces
the military government from exile. Army mutinies spread throughout
the country as many rank-and-file members of the military defect to
the side of the protestors. Senior military leadership is paralyzed by
indecision as the situation deteriorates, eventually deciding to flee the
country along with the Shah. Available U.S. support was ultimately
not sufficient to fundamentally alter underlying revolutionary social

currents in Iran. Khomeini returns to Iran within a month of their
departure. U.S. forces continue to occupy the southern oil fields, pre-
venting the Khomeini regime from gaining access to a significant
source of income. Then, several months after the constitutional refer-
endum, a group of Iranian students takes over the U.S. embassy com-
pound in Tehran. Although most U.S. citizens have been evacuated
from Tehran, a small staff remains to coordinate U.S. policy toward
the new regime. When the regime refuses to release the hostages, U.S.
and Iranian officials enter into secret negotiations. The United States
agrees to withdraw from the oil fields in exchange for the release of the
hostages. The next day, the U.S. embassy staff are driven to the air-
port and flown back to the United States, and the 82nd Airborne and
Marine amphibious unit withdraw.
From here, history proceeds much as it did in the real-world sce-
nario, albeit on a delayed timeline. Khomeini is even more popular
than he was before the coup because of anti-Western and antimilitary
sentiment, and this popularity helps his followers eliminate competing
opposition groups and consolidate their authority over the government.
A new constitution is drafted, and Iran votes overwhelmingly to estab-
lish an Islamic republic via referendum.
The Bosnian War

The end of the Cold War and the rise of ethnic nationalist leaders led to
the dissolution of the six republics of Yugoslavia, as Slovenia and Croa-
tia seceded after brief wars in 1991. On March 1, 1992, Bosnia voted to
secede from Yugoslavia, touching off a civil war when members of the
Serb minority organized to resist the establishment of an independent
state. Between 1992 and 1995, the ensuing fighting was accompanied
by ethnic cleansing, or the use of violence and intimidation against civil-
ians that is designed to create an ethnically homogenous territory. The
war generated the largest refugee crisis in Europe since World War II as
well as widely documented war crimes, including genocide, mass rape,
and the systematic mistreatment of civilians and prisoners of war. The
international community initially responded by providing humanitar-
ian assistance and deploying UN peacekeeping forces (the UN Provi-
Counterfactuals 121
sional Force [UNPROFOR]) to protect humanitarian relief convoys

and later to protect UN-designated “safe areas.” This approach pro-
vided considerable humanitarian relief but failed to stop the conflict
or prevent atrocities against civilians. NATO created and later began
to actively enforce a no-fly zone and engaged in occasional air strikes
in 1994 through mid-1995 that were derided as “pinprick” strikes
because they inflicted little damage on their Bosnian Serb military
targets. After Bosnian Serb forces overran the safe area of Srebrenica
and massacred some 7,000 civilians in July 1995, U.S. and European
leaders agreed that a new approach was warranted. At a conference
of foreign ministers later that July, the allies agreed to use extensive
air power in retaliation for future violations of safe areas and stream-
lined the authority to call in air strikes. After Bosnian Serbs shelled a
marketplace in Sarajevo in August 1995, NATO launched Operation
Deliberate Force, conducting airstrikes across the country on Bosnian
Serb military targets. In combination with an increasingly successful
Bosnian-Croat ground offensive against Bosnian Serb positions, the
NATO operation eventually helped convince Bosnian Serb leaders to
come to the table. On September 20, Bosnian Serb leaders agreed to
participate in negotiations indirectly through Serbian President Slobo-
dan Milosevic, concluding with the signing of the Dayton Accords in
December 1995. On December 20, 1995, most of the UNPROFOR
troops remaining in Bosnia changed helmets and became the Imple-
mentation Force, supplemented by additional NATO forces, bringing
the total force to 60,000.
In the counterfactual scenario, the United States uses the August
1992 London Conference to take the leadership role on the interna-
tional community’s response to Bosnia. U.S. officials engage in inten-
sive diplomacy with NATO allies to ensure that European allies agree
to steps to encourage Serbia, Croatia, and the warring parties in Bosnia
to comply with their commitments to protect civilians and deesca-
late the conflict. Instead of taking an impartial approach to deliver
humanitarian aid and encourage the parties to stop fighting, NATO
announces that it will use military force to enforce the London Con-
ference agreements, if necessary. Within a few days of the conference,
it becomes clear that the Bosnian Serbs do not plan to comply with
the London Conference agreement. Shelling of Bosnian cities (includ-

ing Sarajevo) continues, and an attack on a market in Sarajevo kills
more than 20 civilians, generating outrage in the international media.
In early October, Bosnian Serb forces appear to target humanitarian
convoys crossing through their territory to reach Bosniak enclaves, kill-
ing five peacekeepers and two civilian aid workers. The attack on the
peacekeepers and continued shelling of cities and towns triggers the
threatened NATO intervention.
In early October 1992, NATO begins an intensive air campaign
against Bosnian Serb positions around Sarajevo. NATO and UN
officials demand that the Bosnian Serbs lift the siege of Sarajevo and
remove heavy weapons from exclusion zones around cities and towns.
They make it clear that if the Bosnian Serbs do not comply with these
demands, NATO airstrikes will expand their targets. Bosnian Serb
forces are given 72 hours to comply with these demands. In response,
Bosnian Serb forces move to attack UNPROFOR positions, trigger-
ing an expansion of the bombing campaign to target key Bosnian Serb
massed forces, command-and-control, and war infrastructure targets
throughout the country. Although these strikes severely hamper Bos-
nian Serb military capabilities and mobility, they do not prove to be
sufficient to immediately bring an end to the conflict. The United
States decides to take several steps to increase pressure on the Bosnian
Serbs, including tightening sanctions and deploying its own ground
forces as part of an UNPROFOR Rapid Reaction force, with rules
of engagement that allow them to attack Bosnian Serb forces within
designated areas. U.S. forces become involved in substantial combat
with Bosnian Serb forces, taking casualties of their own. Under this
pressure, and beginning to see Bosnian and Croat forces alter the divi-
sion of territory on the ground among the groups, Bosnian Serb leaders
finally decide to accept the UN’s offer to host negotiations. Leaders
from Bosnian, Croat, and Serb factions travel to Geneva to partici-
pate in negotiations. UNPROFOR and rapid reaction forces, including
U.S. troops, are deployed along the front lines of the war to monitor
and verify the ceasefire, and NATO aircraft continue to monitor the
military no-fly zone and continue reconnaissance missions to deter-
mine that forces on the ground are staying in place.
Counterfactuals 123
In Geneva, UN officials present a peace plan that consists of

decentralized, ethnically based cantons or republics, united by a weak
central government delineated by ethnically based power-sharing, sim-
ilar to the Vance-Owen plan and others proposed over the course of
the conflict. The parties negotiate over the division of territory among
the three ethnic groups and eventually agree to a peace deal along these
lines. Once the settlement is reached, UNPROFOR and the rapid reac-
tion forces in place, including those from the United States, re-hat and
become a postconflict stabilization force tasked with the implementa-
tion of the terms of the negotiated agreement. The stabilization forces
remain in Bosnia for several years, drawing down gradually as the
agreement is implemented. The eventual outcome of this counterfac-
tual case therefore has many similarities with historical events, though
it differs in the process of getting there and many of the details. The
counterfactual case involves a much shorter, less destructive civil war
in Bosnia, and the country remains a more ethnically mixed territory
as a result. The United States avoids a period of apparent inaction that
was damaging for its reputation, but it does so by bearing much higher
direct costs of its own, including U.S. military casualties.
The Libyan Civil War

Following uprisings in Tunisia and Egypt, demonstrations broke out
in the Libyan city of Benghazi in February 2011, and “revolutionary
councils” emerged across the country. Protests soon turned violent
after heavy suppression from security forces, which included airstrikes
against civilians and the use of mercenary troops as rebel groups took
control of parts of the eastern provinces. Concern grew that large-scale
civilian massacres were imminent as government forces approached
rebel-held towns, especially in Benghazi, with a population of roughly
750,000. Although Libya and the United States had taken steps toward
a normalization of their relations in the previous decade, President
Muammar al-Qaddafi’s long history of brutality and support for ter-
rorism left him few friends in the region or beyond and contributed
toward an unusually broad willingness to see his departure. The United
States and key European allies initiated Operation Odyssey Dawn on
March 19, 2011, a civilian protection mission combining a no-fly zone
with strikes against the Libyan regime’s ground troops and command
and control centers, later continued as NATO Operation Unified Pro-
tector. With this international support, rebel forces gradually pushed
back Libyan government troops until they had captured all key cities
and killed Qaddafi by October 2011.
The intervention succeeded in averting a near-term humanitar-
ian catastrophe and eliminating Qaddafi from power, but it was also
accompanied by a variety of unintended consequences. In the after-
math of the intervention, Libyan politics remained fractured and cha-
otic, with a battered economy and persistent violence. The intervention
undermined the stability of neighboring countries, with jihadist groups
in neighboring countries benefiting from the instability in Libya and
weapons proliferation throughout the region. Further, the Russian
sense that the United States had violated an understanding regarding
the limits of force that would be used in Libya contributed to a wider
decline in relations between Russia and the West.
In the counterfactual scenario, President Barack Obama main-
tains the U.S. position of not seeking military involvement in Libya,
a country that presents limited strategic interest for the United States
and where the risks of entanglement are perceived to be high. In the
absence of U.S. support, European allies decide they are neither will-
ing nor able to conduct an operation on their own. Absent interna-
tional intervention, the Libyan rebellion ends up defeated and Qad-
dafi remains in power. A failure of the rebellion, however, does not
mean that rebel activity disappears in Libya. Instead, it morphs into a
low-intensity conflict consisting mainly of guerilla-type attacks against
Libyan government troops, infrastructure, and official buildings and
that is mostly confined to the Benghazi area, where protests started in
February 2011, and the Misrata area, which had shown particularly
strong resistance during its siege by Qaddafi’s troops from February
to May 2011. Regionally, Qaddafi’s continued hold on power results
in lower levels of conflict from Mali to Syria because Libya no longer
functions as a source of exported weaponry or fighters. U.S.-Russian
relations are improved, at least over the near term, while U.S. relations
with NATO allies are negatively affected following the U.S. refusal to
intervene alongside them in Libya.
Counterfactuals 125
Results
The four counterfactual cases we conducted provide a wealth of

detailed information regarding how U.S. military interventions did or
did not affect the various outcomes of interest discussed in this report.
These cases were not selected to be representative of general patterns
across all cases but instead allow us to explore a wide range of diverse
conditions. In our interpretation of the findings of these cases, then,
we will discuss how these cases do or do not illustrate relationships
identified by the other methods we have employed in order to shed
additional light on how they operate and their plausibility. To this end,
next we discuss two sets of results. We first explore the implications of
the counterfactual case studies for the key outcome variables of inter-
est: effects on the conflict, effects on U.S. interests, and effects on the
trajectory of the host nation. Then, we summarize the implications
according to the three key questions for policymakers in this report:
whether to intervene, whether it is beneficial to do so early or late, and
whether it is beneficial to do so with a large or small force.
Effects of Intervention on the Conflict

Each of our counterfactual cases illustrates how external interventions,
mediated by the balance of power among the local actors, can affect
conflict duration and intensity, sometimes lengthening its duration to
achieve a preferred outcome for the United States. In Korea, the U.S.
and UN intervention lengthened the conflict substantially, at first by
staving off a rapid defeat by the Republic of Korea (ROK). The UN
forces were of sufficient size to eventually provide an overwhelming
advantage and push the war toward an end point while still in its first
year, until the Chinese intervention shifted the conflict again into a
stalemate, adding two years to the conflict. In Libya, the U.S. and
allied intervention likely lengthened the conflict as well, preventing a
rapid rebel defeat though eventually also providing enough support for
a rebel victory.
In Bosnia, the historical U.S. and allied intervention was suf-
ficient to change the local balance of power, convince Bosnian Serb
forces that events were moving against them, and force an end to the
conflict. In the counterfactual scenario, in which the United States

and its allies intervene earlier in the conflict, the United States needed
to play a much more active role in ending the conflict, including the
introduction of ground troops, to compensate for the weaker Bosnian
and Croat forces also fighting the Bosnian Serbs and produce a similar
outcome.
In our counterfactual scenarios in Libya and Korea, however, the
U.S.-supported side loses the conflict in relatively short order absent the
U.S. intervention. This again illustrates how intervening on the side of
the weaker party could lengthen the conflict as the cost of achieving
other U.S. objectives. Korea also illustrates the tremendous potential
importance of third-party spoilers because the Chinese intervention
transformed the conflict, intensifying it, lengthening it, and altering
the eventual outcome.
The Iran counterfactual case, in which a U.S. intervention does
occur, differs from the others. In the counterfactual scenario, the U.S.
intervention was quite limited in scale for logistical and operational
reasons and was not sufficient to dramatically affect the local balance
of power when supporting a weak military regime in a country as large
as Iran. The intervention nonetheless modestly increases the duration
of the conflict without greatly affecting the ultimate result.
Overall, the counterfactual cases highlight the importance of the
local balance of power as a mediating variable when determining the
effect that a U.S. intervention should be expected to have on the dura-
tion and intensity of the conflict. These cases speak less to the potential
effects on conflict recurrence because we intentionally limited how far
into the future the counterfactual scenarios were projected to preserve
their reliability.
Effects of Intervention on U.S. Interests

In two of our cases, the historical U.S. intervention clearly helps
achieve U.S. interests. In Korea, the historical intervention helps avoid
the defeat of a partner, strengthens the U.S. strategic position in East
Asia, and provides a demonstration of U.S. credibility and commit-
ment. In Bosnia, the historical U.S. intervention helps end a destruc-
tive conflict in Europe and restore confidence in U.S. international
Counterfactuals 127
leadership. In both of these instances, it is reasonably clear that the

outcome would have been worse without an intervention, from the
U.S. perspective. In the two other cases, however, the extent to which
the intervention helped (or would have helped) advance U.S. interests
is more limited. In Libya, the intervention helps avoid a civilian massa-
cre and deposes a regime that had been unfriendly to the United States
in the past. However, partly because of the lack of postconflict engage-
ment by both the United States and its European allies, the interven-
tion (compared with the counterfactual) also helped increase local and
regional instability that worsened conflicts elsewhere, worsened rela-
tions with Russia, and eventually prompted a follow-up intervention
several years later to help contain the growing chaos. In the Iran coun-
terfactual scenario, the U.S. intervention provides short-term assistance
to help maintain a friendly regime but proves unsustainable, given that
partner’s weakness.
It is also worth considering the relationship between U.S. inter-
ventions and the more-specific U.S. outcomes of interest discussed ear-
lier, including any economic, strategic, and reputational costs and ben-
efits. Continuing a pattern noted in Chapter Four, U.S. interventions
were not associated with positive economic outcomes for the United
States. In all cases we considered for this chapter, U.S. interventions,
even those that achieved U.S. objectives, appear to have had net eco-
nomic costs. In the Iranian counterfactual scenario, efforts to stabilize
oil markets by controlling the oil fields were ultimately not successful
beyond the very short term. In the Bosnia case, the earlier intervention
hypothesized as our counterfactual likely had similar economic costs
in comparison with the historical experience, with the greater intensity
and costliness of the U.S. military effort in the counterfactual sce-
nario (including increased U.S. casualties) being at least partly offset
by several years of more-limited U.S. effort in the historical 1992–995
period.
Interventions have mixed reputational effects in our cases. In
Korea, the historical U.S. intervention conducted through the United
Nations demonstrated credibility and a strong willingness to defend
against the spread of communism, increasing perceptions of U.S. inter-
national leadership. Absent this intervention, abandoning the ROK
would likely have had reputational costs for the United States as it
sought to rally a coalition to contain communism. In Libya, the deci-
sion to intervene likely had mixed effects on the U.S. reputation. In
some circles, the decision to intervene, including with UN authori-
zation, suggested a commitment to protecting against human rights
atrocities. However, the perceived expansion of the goals of the inter-
vention beyond the short-term humanitarian imperatives to encompass
the fall of Qaddafi reinforced concerns with U.S. leadership in other
circles, most notably in Russia, that the United States was an incorrigi-
ble interventionist bent on regime change in non-allied states. In Iran, a
U.S. intervention might have had more-negative effects on reputation,
especially without international support. In Bosnia, the U.S. reputa-
tion for international leadership would likely have been improved with
an early intervention instead of the historical later approach, though we
did not assess the reputational effects that would likely have occurred
if the United States had decided not to intervene in the war altogether.
Finally, in only two of the counterfactual cases did interventions
have strategic benefits: Bosnia and Korea. In Bosnia, the U.S. inter-
vention restored stability in a region very close to key European allies,
though this benefit was realized in both the late (historical) and early
(counterfactual) interventions. In Korea, the intervention provided the
United States with an important regional partner and counterweight
to Russia and China. In Libya and Iran, in contrast, it is not clear that
an intervention, historical in the case of Libya and counterfactual in
the case of Iran, had any net strategic benefits. The Libyan intervention
helped increase regional instability and required a follow-up interven-
tion to help stabilize Libya, while the Iranian intervention failed to
save a key U.S. partner in the Middle East, forcing the United States to
rethink its strategic approach to the region.
A key insight from this discussion of U.S. interests in our coun-
terfactual cases is that those cases in which U.S. objectives were clearly
advanced tended to involve larger forces than those cases in which they
were not, though there was some overlap. Further, the large size of
the intervention in Korea and the relatively substantial application of
forces in Bosnia were an essential component of their ability to advance
U.S. interests. However, our cases also suggest the limits of larger inter-
Counterfactuals 129
ventions. In Libya, a larger intervention might have helped advance

U.S. interests in that country by providing troops and resources that
could have been used to move that country more quickly on a path to
stability and reconstruction. Such an intervention would likely have
been unpopular both in Libya and in the United States, not to men-
tion in the broader Middle East, and therefore the broader effectiveness
of such a mission in advancing U.S. objectives is not clear. This con-
straint echoes one of the concerns with larger force sizes (the potential
for popular backlash) identified in Chapter Four. In Iran, even a nota-
bly larger intervention (which was not logistically feasible at the time)
might not have been sufficient to indefinitely maintain an unpopular,
pro-Western regime in power, although we did not assess this option in
our counterfactual case. Nonetheless, we can safely note that Iranian
popular sentiment had turned resolutely in an anti-Western direction
by 1979, a turn that would only have been exacerbated by a large-scale
U.S. intervention to maintain an unpopular autocratic regime.
The effectiveness of larger interventions in advancing U.S. inter-
ests therefore also appears to be mediated by the effect that they have
on the local balance of power and politics in the host nation and
broader region, themes that we discussed previously in Chapters Two
and Four. In Korea, Bosnia, and Iran, the local balance of power after
the introduction of U.S. forces determines the degree of influence the
United States is able to exercise in the advancement of its interests.
Libya, meanwhile, illustrates the limits of this analysis because a large
intervention might have been sufficient to secure a better outcome in
the host nation itself but potentially at the cost of broader regional and
geopolitical factors.
Effects of Intervention on Host Nation

Across the cases considered in this chapter, the effects of U.S. inter-
ventions on the host nations are (perhaps surprisingly) mixed. Only in
Bosnia, where a substantial international intervention brought a deci-
sive end to the conflict and led to sustained efforts (albeit only partly
successful) to improve the economic and political trajectory of the host
nation, was the effect of the U.S. intervention clearly positive. In Korea,
the U.S. intervention led to a dramatic increase in the short-term dev-
astation experienced by the country, in comparison with a rapid Dem-

ocratic People’s Republic of Korea (DPRK) victory that would have
unified the country. In the longer run, of course, the ROK, linked with
the West, was able to begin a process of dramatic economic growth
and, decades later, political liberalization, while the DPRK stagnated
politically and economically in the war’s aftermath. It is difficult to
project the future trajectory of a unified Korea that had been spared
the destruction of the Korean War, but a middle path between the
historical ROK and DPRK trajectories seems plausible. Although this
path likely would have represented a worse outcome for the South, it
likely would have been an improved outcome for the North. In Libya,
the U.S. intervention helps to topple the Qaddafi regime and avert a
short-term humanitarian catastrophe. But it also creates instability and
a vacuum of power that is later filled in part by extremists, such as
ISIL. In the counterfactual scenario, in which the United States does
not intervene, Qaddafi retains power, brutally repressing the opposi-
tion while remaining unable to fully stamp out a low-grade insurgency.
The intervention is too recent for us to have a full sense of how the his-
torical scenario will play out. Although the U.S. intervention has the
potential to have improved the trajectory of the host nation by remov-
ing a regime that had stagnated political development, this possibil-
ity has been accompanied by chronic instability. In the counterfactual
scenario in Libya, meanwhile, the absence of a U.S. intervention had
limited effects on the trajectory of the host nation one way or the other.
The local balance of power with and without a U.S. intervention
and the broader political and economic context of the host nation help
explain the outcomes observed in our counterfactual cases. In Bosnia,
the dominance of the U.S. and allied intervention gives the interven-
ers the leverage to reshape Bosnian institutions and guarantee secu-
rity, helping to facilitate an overall positive trajectory for the country.
In Korea, the UN intervention—once the Chinese intervened—was
only able to achieve this level of control over half of the country; the
intervention did help set the ROK on a positive trajectory, although
first the ROK needed to overcome a massive wartime toll. In Iran, the
U.S. intervention does not notably alter the local balance of power or
the host nation’s trajectory. In part, this was due to the large number
Counterfactuals 131
of competing actors within the Iranian political system and the weak-
ness of groups willing to ally with the United States. In Libya, the U.S.
intervention was quite limited. Although it might have been feasible
for a larger U.S. and allied intervention to provide greater postconflict
security and facilitate a political settlement in the country, the Libyan
transitional government was opposed to this option; instead, the lim-
ited U.S. intervention led to instability, characterized by competing
militias.
The effects of U.S. interventions on the outcome categories that
are illustrated by our counterfactual cases are summarized in Table 5.2.
Table 5.3 summarizes the effects of intervention across the three coun-
terfactual cases that consider the intervention decisions. Table 5.4 sum-
marizes the effects of timing in the Bosnia case.
Implications
When deciding whether to intervene in a war or crisis in which impor-
tant U.S. interests are at stake, our cases again illustrate the importance
of considering the likely effect that a U.S. intervention could have on
the local balance of power and, in turn, the leverage that the U.S.
intervention would provide for the United States to alter the outcome.
When the United States is able to decisively and sustainably intervene,
Table 5.2
Summary of Effects on Outcomes Illustrated by Counterfactual Cases
Category Effect of U.S. Military Interventions
Conflict Dependent on how U.S. intervention affects the local balance of

outcomes power as well as the behavior and effects of third-party spoilers.
Intervener Larger interventions, for the most part, provided greater leverage
interests to advance U.S. interests. Local context and balance of power are
also crucial.
Host nation Local conditions of the host nation, including the local balance
of power and the broader political and economic environment,
shape outcomes. The feasibility of sufficiently larger or persistent
interventions is dependent on international context.
Table 5.3
Summary of Effects of Intervention for Counterfactual Cases
Category Korea Libya Iran
Intervention type Early, large Early, small Early, medium
Historical or Historical Historical Counterfactual

counterfactual U.S.
intervention
Conflict duration Longer Longer Longer
Intervener interests Improved Mixed No change
Economic interests Costs because of Costs because of Costs because of

conflict conflict conflict
Strategic interest Improved Mixed Mixed
Reputational Improved Mixed Possibly negative

interests
Host nation Short-term costs Mixed No change

but longer-term
improvement
Table 5.4
Summary of Effects of Timing in Bosnia Counterfactual Case
Category Outcome
Conflict duration Shorter
Intervener interests Notably higher direct costs, but objectives achieved

more quickly
Economic interests Likely similar overall costs
Strategic interest Likely similar, but objectives achieved more quickly
Reputational interests Likely similar
Host nation Reduced costs and faster conflict cessation

Counterfactuals 133
as in Bosnia, it has a greater chance of making the intervention, from

its perspective, a success. At the other extreme, as in Iran, when no U.S.
intervention that could feasibly have been conducted quickly could
have sufficiently altered the local balance of power in a sustained way,
a U.S. intervention is likely to provide limited benefits.
However, two other cases illustrate the challenges in making a
proper assessment ahead of time regarding the effect that a U.S. inter-
vention might be able to have. In Korea, the United States was highly
motivated to intervene, given the strategic interests at stake, but it also
did not foresee the effect that the Chinese intervention would have
when deciding to expand the goals of the intervention to include
taking the entire Korean peninsula. The result was a bloody stalemate
that achieved many U.S. goals in comparison with a U.S. decision not
to intervene at all but at a high cost. In Libya, the U.S. intervention was
kept quite limited. Although it proved sufficient to achieve the proxi-
mate U.S. goal of toppling the Qaddafi regime, the failure of either the
Europeans or the United States to remain engaged in the aftermath of
the conflict was not anticipated at the time of the intervention deci-
sion and helped undermine the overall success of the intervention. An
improved understanding of the role and the intentions of third parties,
whether allies or adversaries, could have better informed U.S. decision-
making, although such an understanding might have been more likely
to alter the execution of the intervention than the decision regarding
whether to intervene at all in our cases. For example, this improved
understanding could have led to a U.S. decision to stop at the 38th
parallel in Korea or led to a determination to extract guarantees from
European allies regarding their commitment to the postconflict recon-
struction in Libya before deciding to support that intervention.

Early interventions appear to have been preferable in all the cases we
considered. In Korea and Libya, an early intervention was likely the
only decision that was feasible, given the rapidity with which adversary
forces would likely have won the conflict in both cases. That is, if the
United States had opted against an early intervention, it likely would
not have intervened at all. In our Iran counterfactual scenario, an early
intervention was also the most feasible alternative in comparison with

a decision to intervene later, once Khomeini had returned to Iran and
begun to consolidate power, although this early intervention did not
dramatically alter the outcome.
Bosnia was our one counterfactual case to make an explicit com-
parison of the benefits and costs of intervening early in the conflict
as opposed to waiting to intervene at a later point. The analysis of
the early, counterfactual intervention scenario illustrates how waiting
for a later moment in the conflict might only have provided limited
additional benefits because, in its hypothesized early intervention, the
United States is still able to end the conflict in relatively short order
and stabilize the country while avoiding much of the devastation that
occurred in the country throughout the 1992–1995 Bosnian War. Off-
setting these benefits are a substantially increased level of risk to U.S.
forces and the more-immediate direct costs of intervention paid by the
United States, which were required to offset the lesser degree of exhaus-
tion on the part of Bosnian Serb forces and the lesser degree of success
from Bosnian and Croat forces on the ground. This counterfactual
illustrates a potential pathway by which later interventions might be
preferable in some cases, in that longer conflicts could allow time for
the parties to reach a point when forcing a settlement becomes easier,
and fewer costs might be paid by the interveners. If the interveners are
willing to pay these costs, which of course cannot be taken for granted,
then they might be able to achieve other goals more quickly and at
lower cost to the host nation.
Taken together, though, the cases highlight the value of inter-
vening early in a crisis or war in many circumstances, but they do
not suggest that the United States should always intervene early. The
alternative to intervening early, of course, is not intervening, at least for
the time being, allowing policymakers to gather more information and
see how events develop before committing U.S. forces. Avoiding an
early intervention in Iran, for example, appears to have been preferable
because it avoided an intervention that would likely have done little
to advance U.S. interests. The difficulty for policymakers is properly
assessing whether a U.S. intervention is an inevitability, given the inter-
ests at stake (e.g., the United States is highly likely to be drawn into
Counterfactuals 135
the conflict at some point), or whether it could plausibly be avoided

altogether depending on how the situation develops. If the former is
true, then an early intervention could be preferable, depending on how
the intervention would affect the local balance of power (as illustrated
in the Bosnia counterfactual). If the latter is true, then a wait-and-see
approach could be more appropriate. But making the assessment of
which category the case falls into requires a clear analysis of the inter-
ests at stake, the local balance of power, and the likely future trajectory
of the situation.

Two of our counterfactual cases shed perhaps less light than the other
two on the question of the size of the force U.S. policymakers should
commit to an intervention and under what conditions. In Korea, the
need for a large force was obvious, given the scale of the military chal-
lenge. In Iran, a more-limited, medium force was all that feasibly could
have been generated in the counterfactual scenario, and so we did not
assess the potential effects of a larger force for comparison.
In Bosnia and Libya, the lessons that can be drawn from these
cases regarding the size of the intervention are more nuanced. In the
historical case in which the United States intervened in Bosnia in
1995, the need for a substantial force was clear, given repeated earlier
attempts to alter Bosnian Serb behavior using UN peacekeepers and
the threat of limited air strikes as well as the comprehensive under-
standing by then of the strength of Bosnian Serb forces. In our coun-
terfactual scenario in which the United States instead intervenes in
1992, we assessed that a larger intervention force ultimately would have
been required for the United States and its allies in order to achieve the
outcomes they sought. The United States does not know this immedi-
ately, however, and initially attempts in 1992 to employ more-limited
forces before increasing the size, and belligerence, of the intervention to
ensure greater success. Again, larger force sizes appear to have provided
some benefits to the advancement of U.S. interests.
That said, our Libya case might illustrate the limits of that pat-
tern. Although a larger or a more-sustained U.S. intervention in Libya
likely would have better helped stabilize the country in the short term,
its long-term feasibility and advisability might have differed. On the

heels of the U.S. experience in Iraq, a large 2011 U.S. intervention
into Libya would likely have been immensely controversial, both in the
United States and in the region. Although not as acute as the concerns
regarding a large intervention in the post-2014 Iraq case discussed in
Chapter Four, a large or more-sustained intervention in Libya in 2011
might have involved trade-offs among our outcomes of interest, likely
shortening the conflict and increasing short-term stability while having
more-uncertain effects on broader U.S. political and strategic interests.
The discussion for each of these three questions is summarized
in Table 5.5.
Table 5.5
Summary of Implications for Intervention Decisions Illustrated by
Counterfactual Cases
Question Effect of U.S. Military Interventions
Whether to The United States is likely to advance its interests by intervening

intervene when doing so will shift the local balance of power in favor of the
supported side. However, assessing whether the intervention is
likely to produce this effect could be difficult, particularly when
assessing the behavior of third parties is crucial. Good, locally specific
intelligence and analysis is therefore critical.
When to Early interventions can help advance U.S. interests in many

intervene circumstances. Exceptions can include cases in which potential
adversaries are strong but might weaken in the future or in which the
United States might not have sufficient forces available to decisively
alter the local balance of power. But the decision to intervene early
is as much a decision about whether to intervene at all: When the
interests at stake mean eventual U.S. intervention is highly likely,
then early intervention might bring fewer risks. If eventual U.S.
intervention is less certain, however, then keeping open the option
to intervene later or not at all might be more advisable. Not every
situation is improved by a U.S. intervention, early or not.
With what Larger interventions are able to help advance U.S. interests in
size force to certain circumstances. This is not universal, however. In some cases,
intervene depending on the local balance of power, even large forces might
be insufficient to advance U.S. interests. In others, the presence of
larger U.S. forces would be likely to generate substantial backlash,
suggesting that smaller forces might be more advisable.
CHAPTER SIX
Conclusion and Implications
In this report, we have considered the role of intervention decisions

(whether to intervene, intervention timing, and intervention size)
through four lenses: (1) using past research, (2) using quantitative data,
(3) using a set of short, focused case studies, and (4) using a set of
longer counterfactual case studies. In this chapter, we will synthesize
and summarize the insights from across these four lenses and discuss
the implications and recommendations for the Army.
The Effect of Interventions on Outcomes
Throughout this report, we have considered the effects of different

intervention decisions on three main types of outcomes of relevance
to policymakers: the effect of the intervention on the course of the
conflict, on the interests of the intervening state, and on the trajectory
of the host nation where the intervention takes place. Table 6.1 sum-
marizes the findings from each of our four methodologies for each of
these categories of outcomes.
Next, we will summarize these results regarding the effect of
intervention decisions on conflict outcomes and the host nation’s tra-
jectory. The final category of outcomes, intervener interests (in this
case, the interests of the United States), will be assessed in greater detail
below, with our findings organized to answer the three main policy
questions noted throughout this report: whether to intervene, whether
it is better to intervene early, and whether it is better to intervene with a
larger force. Although of course outcomes for the conflict and the host
137
138
Table 6.1
Summary of Findings for Outcome Measures
Assessing Trade-Offs in U.S. Military Intervention Decisions

Outcome
Measures Literature Review Quantitative Analysis Focused Case Studies Counterfactuals
Conflict • Interventions can help • No clear patterns with • Not assessed. • Effects on conflict are
outcomes terminate conflicts and respect to conflict dura- highly dependent on
reduce conflict duration or recurrence. how U.S. intervention
tion, especially if the • U.S. interventions, affects the local bal-
intervention is larger especially when later ance of power, as well
and earlier. in a conflict, could as the behavior and
• But the intervention’s be associated with a effects of third-party
effect depends on reduction in conflict spoilers.
its influence on the intensity in comparison • For example, early
local balance of power with nonintervention interventions can
and the potential for cases. reduce conflict duration
backlash. • However, selection if they lead to a quick
effects likely play a victory or dramatically
role, particularly in the increase it if they lead
case of timing. to a protracted stale-
mate. In some cases,
the longer conflict is
required to advance
other U.S. interests.
Outcome
Intervener • Earlier interventions • Earlier interventions • The advisability of an • Interventions had a

interests are advisable for some are associated with a intervention is depen- mixed track record
issues, such as disaster higher likelihood of dent on its potential in advancing U.S.
relief or atrocity pre- achieving U.S. objec- effect on the local interests.
vention; for others, tives than later inter- balance of power, • Where interventions
waiting might be ventions are. willingness to commit were advantageous,
preferable. • Interventions are asso- resources, and presence they were larger in size,
• Larger interventions ciated with a notable of spoilers. and a large number of
can advance interests, increase in military • Nonintervention cases troops directly helped
especially in combat assistance and a more improved more fre- achieve U.S. goals.
and stabilization modest increase in eco- quently than they • In other cases, where
missions. nomic assistance. deteriorated. large interventions
• But these larger forces • Early interventions are • Early interventions were not possible or
can also be counter- associated with greater are more likely when attempted, our analysis
productive if they trig- increases in economic homeland security of cases suggests large
ger a backlash. and military assistance or vital interests are interventions would
in comparison with late directly threatened. In have been insufficient
Conclusion and Implications

interventions. other cases, waiting or led to backlash
• However, interpret- was not associated with because of entrenched
ing the relationship worse outcomes. local adversaries and
between levels of • Larger interventions popular resentment of
U.S. assistance and help advance interests U.S. presence.
U.S. interests remains overall, but smaller • Large force size can
difficult. forces could be prefer- therefore help in some
• Selection effects are able if there are con- circumstances but not
likely relevant. cerns about backlash, universally.
escalation risks, or
deployment timelines.
139
140
Assessing Trade-Offs in U.S. Military Intervention Decisions

Outcome
Host- • Mixed or limited inter- • No clear differences • Although results varied, • Interventions, par-
nation ventions tend to have between intervention interventions are asso- ticularly when larger,
trajectory highly contingent and nonintervention ciated with positive have the potential to
effects on host nations. cases. host-nation trajectories positively alter the host
• Earlier and larger inter- more often than non- nation’s trajectory.
ventions are associated interventions are. • However, their ability
with higher economic • Positive trajectories are to do so depends on
growth, while later more likely in larger local conditions, includ-
interventions are asso- interventions, with ing the local balance of
ciated with greater notable exceptions. power and the broader
decreases in infant political and economic
mortality. environment.
• Patterns within inter-
vention subcategories
are driven by a very
small number of cases.
Conclusion and Implications 141
nation relate to or overlap with U.S. interests in some cases, we none-

theless highlight the results with respect to U.S. interests in particular
because of their direct relevance to policymakers.
Effects of Intervention Decisions on Conflict Outcomes

Our research suggested that there could be contingent relationships
between the U.S. decision to intervene in a conflict and the subse-
quent course of that conflict but that this relationship is not always
in one direction. Although U.S. intervention might shorten conflicts
or make them less likely to recur in some circumstances, it could have
the opposite effect in others. Although we do not observe an overall
pattern in the effects of U.S. interventions on conflict outcomes, U.S.
interventions have substantial potential to affect these outcomes. How-
ever, it appears that their effects are dependent on other factors, such
as how these interventions might affect the local balance of power in
the conflict or how they will affect the calculations of third parties,
including the local population in the host nation and other regional or
global powers. Where a U.S. intervention establishes a clear balance of
power in favor of the side on which it intervenes, or where an interven-
tion incentivizes other states to support U.S. efforts, it can shorten or
limit the conflict. Where an intervention merely forestalls defeat for
the supported side or leads to a stalemate, or where an intervention
causes a popular backlash or incentivizes other states to intervene to
counter the U.S. effort, it could prolong or intensify the conflict. Care-
ful analysis of these factors therefore appears to be crucial for anticipat-
ing the effect that U.S. intervention might have on conflict outcomes.
We did not find substantial evidence that U.S. decisions regarding the
timing or size of the intervention would be expected to affect conflict
dynamics directly, though we would expect that timing and size could
well affect the course of the conflict through these same intermediary
variables of the local balance of power and third-party reactions. For
example, a larger intervention force would be expected to alter the local
balance of power in a different way than a smaller force would, and it

might affect the calculations of third parties in different ways as well.1
Effects of Intervention Decisions on the Host-Nation Trajectory

We found a mixed overall pattern in our investigations into the relation-
ship between U.S. intervention decisions and the trajectory of the host
nation. Across our full data set of U.S. intervention and noninterven-
tion cases, we did not identify clear differences in the later trajectory of
the host nation. In the subset of cases for which we conducted focused
case studies, however, we did identify many more instances in which
U.S. interventions appear to have improved the economic and political
fortunes of the host nation in comparison with nonintervention cases,
in which the trajectory of the host nation tended to change in any
notable way less frequently. That is, the U.S. intervention appeared to
function like an experimental medical therapy: Prospects for improve-
ment in its absence might be limited, but its application will not always
be beneficial.
Our analysis also suggests that there could be trade-offs in the
types of improvements in the host nation’s trajectory because different
aspects of these outcomes often do not appear to progress in tandem.
Efforts to improve political liberalization, for example, might conflict
in the short term with efforts to build state capacity or promote eco-
nomic development. U.S. policymakers might therefore need to decide
what aspect of the host nation’s trajectory they wish to prioritize to
increase their odds for success.
We also found suggestive evidence in both our qualitative and
quantitative methods that larger interventions might be more likely
to improve the host nation’s trajectory than smaller interventions or
noninterventions. Although this relationship is complex, our analysis
highlights at least one important contingent factor: the degree of popu-
lar backlash prompted by the larger intervention. Where U.S. forces
are generally accepted by the local population, their greater size can
1 Further, an early intervention naturally has greater potential to affect conflict duration
than a later intervention, given that the conflict has already persisted for some time when the
later intervention takes place.
give them greater capacity to reshape host-nation institutions or build

economic and state capacity. Where this is not the case, however, any
improvements they attempt are more likely to be transitory.
Although the effects of U.S. interventions on conflict outcomes
and the trajectory of the host nation are important considerations, ulti-
mately this study aimed to inform the decisions of U.S. policymakers
for whom the advancement of U.S. interests remains the paramount
concern—although, as mentioned earlier, these other types of out-
comes could be closely related to U.S. interests in many cases. Next,
we summarize our results regarding the effects of U.S. intervention
decisions on the advancement of U.S. interests for three main policy
questions: whether the United States should intervene, whether it is
preferable to intervene early, and whether it is preferable to intervene
with larger forces.
Effects of Intervention Decisions on U.S. Interests

Although military interventions can be very effective in some contexts
and situations at advancing U.S. interests, they could have the opposite
effect in other contexts, creating long-term entanglements and increas-
ing economic and strategic costs without realizing U.S. objectives. Our
analysis suggested that the United States achieves its political objec-
tives in military interventions about 60 percent of the time. More-
over, U.S. intervention is not a requirement for protecting U.S. inter-
ests in all cases. Many nonintervention cases resolve in ways that do
not have significant implications for U.S. interests; historically, more
nonintervention cases resulted in improved outcomes for U.S. interests
than resulted in deteriorated outcomes. Rather than suggesting a bias
toward intervening to solve problems, then, this study highlights for
decisionmakers the importance of carefully considering local condi-
tions to assess whether conducting a military intervention is likely to
advance U.S. goals.
The nature of the U.S. interests at stake can also play an impor-
tant role in determining whether an intervention occurs. Direct threats
to U.S. national security, such as the September 11, 2001, attacks or
the Cuban Missile Crisis, provoke immediate and decisive responses.
Other than instances of direct threat, however, the types of interests

involved are no guarantee of intervention or nonintervention, reflecting
how policymakers have historically made decisions on a case-by-case
basis. Factors such as the geographic region, the presence or absence of
international support, the number of other ongoing engagements, and
the broader policy context at the time all play a role in determining
whether the United States has historically chosen to intervene.
Separate from the question of when the United States has inter-
vened, of course, is when should it intervene: That is, when will an inter-
vention advance U.S. interests? The keys to distinguishing between
intervention opportunities that advance U.S. interests and those that
have the opposite effect appear to be contextual factors. In particular,
our focused case studies, counterfactual case studies, and review of the
literature all highlight the centrality of understanding how an outside
intervention could shift the local balance of power when determining
how the intervention might affect the course of the conflict and the
leverage available to the United States to achieve its interests. If the
U.S. intervention would serve to shift the power of the favored side
in a conflict from either a stalemated or a disadvantaged position to a
clearly advantaged position, then we would expect the intervention to
help end the conflict and increase the likelihood that the United States
is able to advance its other policy goals.
Historical examples of this dynamic can be seen in the 1991 Gulf
War and the 1995 Bosnian intervention. Other dynamics, however,
could force policymakers to make explicit trade-offs across different
outcomes. In the 1950 intervention into the Korean War, for exam-
ple, U.S. intervention helped avoid the defeat of the ROK, but it also
served to substantially prolong the conflict, particularly once Chinese
intervention pushed the conflict into a lengthy stalemate. Although the
end result helped advance U.S. strategic interests and placed the ROK
(though not the DPRK) on a positive developmental trajectory, these
achievements came at the cost of involvement in a lengthy conflict
with a substantial human toll. In still other cases, illustrated by our
counterfactual study of the 1979 Iranian Revolution and our analysis
of the 2007 war in Yemen, any plausibly sized U.S. intervention would
likely be unable to dramatically alter the local balance of power in
the desired direction, and intervention is therefore likely best avoided.

Finally, a U.S. intervention could be largely superfluous if it occurs
in a context in which the supported side already has the upper hand.
This was partly the case in the 1965 Dominican Republic intervention,
where a combination of other local and regional actors likely could
have achieved the same generally positive outcome without U.S. inter-
vention (or at least with a much smaller U.S. force). In the 1948 Costa
Rican civil war, the U.S. decision not to undertake a military interven-
tion at all helped facilitate a regional response that settled the immedi-
ate conflict and put the country on a sustainable, positive trajectory for
decades.
Our analysis also highlights that despite the centrality of deter-
mining the local balance of power for anticipating intervention out-
comes, this determination can be difficult in practice, particularly if
the motivations and capabilities of third-party spoilers are not properly
understood. The Chinese intervention to reverse U.S. advances in the
Korean War, for example, was not anticipated, although it proved deci-
sive for shifting the outcome of the war. The role Pakistan has played
in the post-2001 conflict in Afghanistan provides another example.
On the question of timing, our analysis suggests that early interven-
tions are associated with both the highest rewards and increased risks.
In our quantitative analysis, early interventions are somewhat more
likely to achieve U.S. political objectives than later ones are. The short,
focused case studies suggest more-equal rates of success between early
and late interventions in terms of achieving U.S. objectives, but we also
find many more instances in which the early timing of an intervention
was vitally important to those successes and only one potential case in
which the lateness of the intervention might have been beneficial. For
example, quick U.S. responses to the 1991 Kurdish humanitarian crisis
in northern Iraq and the rapid mobilizations and deployments under-
taken in response to the 1962 Cuban Missile Crisis were successful in
good part because they occurred early in the crises. Although some
later interventions, such as the 2014 intervention into Iraq to combat
ISIL, were ultimately successful, our research suggests that most of the
ways in which these interventions were not successful could be attrib-

uted to them beginning late in these conflicts.2 One potential excep-
tion to this trend was the U.S. intervention in Bosnia. Although our
counterfactual analysis suggests that an earlier U.S. intervention could
have led to a similar outcome while decreasing the humanitarian toll
of the conflict, it would only have done so if the United States had
been willing to pay a higher price itself, including increased risk to its
ground forces. The later intervention that occurred historically in 1995
required fewer trade-offs for the United States because the local bal-
ance of power was already moving away from the Bosnian Serbs before
robust U.S. intervention began. Waiting to undertake an intervention
later in a conflict could therefore provide a local balance of power that
is more conducive to advancing U.S. interests, thereby limiting U.S.
costs or increasing the feasibility of intervention.
More broadly, however, it is important to emphasize that the deci-
sion for policymakers about timing is really an iterative one that is
closely intertwined with the decision regarding whether to intervene at
all. Early interventions, from this perspective, simply reflect instances
when policymakers decided to intervene at their first opportunity to
do so, while later interventions occur when they decide to intervene
at subsequent points in the conflict. But policymakers do not typi-
cally make a decision at one point to intervene at some later point,
so there is no actual decision to “intervene late.” The decision that
policymakers face is therefore similar regardless of when the decision
occurs during the conflict. The analysis in the prior section regarding
whether the United States should intervene at all therefore remains rel-
evant regardless of when in the course of the conflict an intervention
is being contemplated.
There is a tendency to assume that “more is always better” when decid-
ing how many forces to commit to an intervention. Certainly, robust
2 To clarify, we assessed that a U.S. intervention prior to the fall of Mosul could have been
more effective (and at lower cost) than the historical intervention that took place afterward.
Once Mosul had fallen, however, U.S. patience in rebuilding Iraqi security forces before
mounting a sustained counteroffensive appears to have paid dividends.
interventions appear to be both warranted and highly likely in response

to direct threats posed to U.S. national security, such as the invasion
of Afghanistan following the September 11, 2001, attacks and the
response to the 1962 Cuban Missile Crisis.
We also have evidence suggesting that large or medium interven-
tions might be associated with better outcomes more generally. Our
focused cases, for example, find that large and medium interventions
more often achieve U.S. political objectives. Further, we observe that
there are more instances in which the size of a larger intervention is
central to its success than is the case for smaller interventions. Large
interventions such as the 1991 Gulf War highlight the value that larger
forces can provide, particularly in cases of traditional combat and when
facing large, powerful adversaries.
That said, we also find reasons for caution with the employment
of larger forces. To begin with, a blanket application of large forces
in interventions could often prove to be unnecessary, adding addi-
tional costs without realizing additional benefits, as was the case in the
large U.S. intervention in the Dominican Republic in 1965. But the
potential concerns with large forces extend beyond a lack of efficiency.
Our analysis indicates at least three circumstances in which forces of
a more-limited size could be preferable. First, a large U.S. force might
produce a substantial backlash among the local population or impor-
tant regional actors, undermining the intervention’s ability to achieve
its objectives and potentially strengthening adversary forces. Second, a
large U.S. force could appear highly threatening to an adversary and
prompt unwanted escalation in the conflict or crisis. Third, a large
force, particularly a large ground force, could take substantial time to
mobilize and deploy, such that policymakers might prefer to send a
smaller force that can be ready more quickly.
Even beyond these conditions that could suggest a preference for
a more-limited-size force, the introduction of large forces effectively
increases the “bet” the United States is placing on the outcome of the
intervention. As noted above, large forces can help increase success, but
in other instances, such as the post-2001 intervention in Afghanistan,
even greatly increased forces might prove unable to fundamentally alter
the dynamics of the conflict, increasing the costs of the intervention
without substantially increasing its benefits. It is notable that the large

interventions in our focused case studies encompass some of the most
notable successes and the most notable failures for U.S. interventions.
Small interventions, by contrast, tend to have less effect overall on U.S.
interests: For the most part, they are associated with more-limited gains
and, in other circumstances, more-circumscribed losses.
What determines when larger forces are a good bet to make in an
intervention? Although this determination is, in part, affected by the
U.S. interests involved, with direct threats to national security warrant-
ing a decisive response, how large forces would be expected to affect
the local balance of power and the degree of leverage the United States
can exercise in the host nation appear to be central. Where large forces
allow the United States to tip the balance of power clearly in favor of
the side it supports, it can generally expect these larger forces to help
lead to better outcomes. The United States has tremendous military
capabilities, so situations where this dynamic is salient are likely to
occur frequently. There are, however, exceptions, in which the intro-
duction of large forces could prompt local backlash, unwanted escala-
tion, inadvisable delays, or third-party intervention to counterbalance
U.S. forces (such as the Chinese intervention during the Korean War)
or could simply be insufficient to decisively alter the local balance of
power. The decision to intervene with large forces therefore represents a
higher-risk, higher-reward strategy, the advisability of which should be
carefully assessed according to the local context and the criteria noted
above.
Recommendations
Using the key insights of our analysis, we make three recommenda-

tions to inform U.S. and Army planning and decisionmaking regard-
ing future interventions.
Select Interventions Judiciously: Context Is Likely to Be Decisive

Our analysis clearly shows that military interventions can advance U.S.
national interests in certain circumstances. But we also find suggestive
evidence that, since 1945, the United States could be over-intervening

in aggregate. Instances of U.S. interventions that failed to achieve their
objectives and where U.S. interests might have been better served by
not undertaking the intervention in the first place, such as the post-
2003 intervention in Iraq, are often well known. Equally interesting,
however, is that we found few cases in which the United States did not
intervene but likely would have been better off if it had, such as the
1978 Nicaraguan Civil War. More numerous were cases in which the
United States did not intervene and saw its interests advanced partly
because of its restraint, such as the 2006 Iranian nuclear crisis, and the
1948 civil wars in Costa Rica and Colombia. In general, the costs of not
intervening may have been overemphasized in analyses of U.S. inter-
vention policy in comparison with the potential benefits of restraint.
However, this general observation is of more-limited value for
policymakers charged with identifying which specific interventions
are most essential to undertake and which can be safely avoided. Our
analysis identifies four key criteria that can be assessed when deciding
whether a military intervention is likely to advance U.S. interests or
prove counterproductive or wasteful:
• The local balance of power: In crises and wars where the balance of
power already tilts toward the side favored by the United States,
intervention might be unnecessary. However, where the United
States’ interests are under threat, intervention could make more
sense, provided the U.S. intervention would be capable of turning
the tide.
• The role of spoilers: If external actors might respond to a U.S.
intervention by increasing their involvement in a crisis or war
sufficiently to counteract the introduction of U.S. forces, then
intervention could prove ineffective. U.S. adversaries might also
respond to the intervention by taking actions in other theaters
that undermine U.S. interests in a manner that a narrow analysis
of the host nation might not account for. During the Cold War,
for example, the United States had to carefully consider how the
Soviet Union would react to its overseas activities in order to pre-
vent unintentional escalation.
• The potential risks of escalation: If a U.S. intervention has the poten-

tial to directly threaten, or to be perceived as directly threatening,
the security of a near-peer or other highly capable U.S. adversary,
then the intervention has the potential to draw the United States
into a wider or more-dangerous confrontation. Although such cir-
cumstances were limited in the cases we assessed, their potential
consequences highlight the importance of considering this factor
when making intervention decisions.
• The degree of legitimacy: Interventions taken without international
legitimacy can affect U.S. relations with allies and the U.S. repu-
tation more broadly. Legitimacy can also be vital for avoiding a
backlash in the host nation; such a backlash could undermine
the ability of the intervention to achieve its goals. This concern is
most acute for interventions with large numbers of U.S. forces. In
other cases (most notably, those involving humanitarian concerns
for which a U.S. or multilateral intervention might be increas-
ingly expected), it could be that a failure to intervene to prevent
civilian suffering could sap the legitimacy of the United States as
a global leader.
Earlier, Larger Interventions Are Higher Risk, Higher Reward

Our analyses suggest that earlier and larger interventions are overall
more likely to result in positive outcomes for the United States. More-
over, our case studies illustrate clearly how both the timing and size
of these interventions can be essential factors in their success in this
regard. However, earlier and larger interventions are also associated
with some of the most negative outcomes that we observe and, even
when successful, involve substantial economic costs for the United
States. These interventions therefore constitute “betting big.” They
have a clear potential to achieve U.S. objectives, and in some instances,
they might be the only way to do so. But there are several circum-
stances, most notably including the presence of highly capable third-
party spoilers, where such interventions might be apt to fail nonethe-
less. When such interventions fail to achieve their objectives or when
they generate additional, unforeseen complications, the strategic, repu-
tational, and economic costs they can impose on the United States are
enormous. Earlier, larger interventions should be undertaken only after

a careful analysis of their likelihood of success, informed by an assess-
ment of the factors noted above and an assessment of potential alterna-
tives to military intervention.
Prioritize Collection and Analysis of Key Intelligence to Inform

The above two recommendations emphasize the importance of under-
standing the local context in order to inform intervention decisions.
These factors, as much as anything the United States does in the actual
conduct of the intervention, can determine whether an intervention
advances or harms U.S. interests. This is an area in which focused
intelligence and data gathering and subsequent careful analysis can
have a significant impact. Intelligence collection to inform interven-
tion decisions should focus on the following questions:
• Warring parties and the balance of power: Who are the warring
parties? What is the balance of power between relevant parties?
Would the balance be changed in a way favorable to U.S. interests
with an intervention?
• Motivations of the parties: Is the war or crisis one that seems ame-
nable to settlement? How long is each side willing to fight? Are
the issues at stake likely to remain intractable, or could they be
resolved through negotiation?
• External spoilers and influencers: Which external actors could
become involved in ways that counter U.S. interests? What are
their goals and capabilities? Are they likely to intervene with mili-
tary forces or to exert other types of influence?
• Local populations: Will the local population of the host nation be
supportive of a U.S. intervention? If not, what type of backlash
is likely, and how severe will it be? What are the relationships
between the other parties to the conflict and the local population?
• Adversary responses: How will adversaries respond to a U.S. inter-
vention? Are they likely to feel threatened by it? Are there impor-
tant escalation risks to consider? Do adversaries have the ability
and motivation to undermine U.S. interests elsewhere in response

to the intervention?
• International community: What will be the response of the inter-
national community to the intervention? Will the United States
face backlash from allies for intervening or not intervening?
Ensuring high quality and up-to-date information on these ques-

tions can support better decisions about whether and when to intervene
and with what size force. Of course, the U.S. military, and the U.S.
government more broadly, might already be collecting much of this
information for some locations. We make this recommendation partly
to emphasize the importance of filling any gaps that might exist with
regard to countries where an intervention is being contemplated but
also to emphasize the information that decisionmakers will need in
order to apply the framework developed in this report to the interven-
tion decision questions we explore.
The theme running throughout this report is that decisions about
whether, when, and with what size force to intervene can have impor-
tant implications for U.S. interests. These decisions can be optimized
to maximize benefits to the United States but only if contextual factors
are carefully weighed and analyzed. Simplistic advice for these com-
plex intervention decisions is not possible because there is no single
approach that will work every time. Instead, each intervention decision
must be made on its own terms. Our aim with this report is to reduce
the complexity of this decision for senior leaders by identifying the key
factors that should be weighed when making this decision, describ-
ing the mechanisms that link these factors to intervention outcomes,
providing detailed historical examples, and assessing systematically the
likely implications of specific intervention decisions executed in par-
ticular contexts.
APPENDIX A
List of Cases of Potential U.S. Intervention
Table A.1 provides additional detail on our list of potential U.S. inter-
vention cases.
153
Table A.1
List of Cases of Potential U.S. Intervention
Country Intervention
Location Crisis or War Start Year End Year Type
Poland Polish Ukrainians 1945 1947 No intervention
Indonesia Indonesian Civil War 1945 1946 No intervention
Union of Forest Brethren 1945 1951 No intervention

Soviet Socialist
Republics
India Partition Communal War 1946 1947 No intervention
Turkey Turkish Straits 1946 1946 Early, small
China Chinese Civil War, Phase 2 1946 1950 Forces already

present
Vietnam French Indochina War 1946 1954 Late, small
Madagascar Third Franco–Madagascan 1947 1948 No intervention
Dominican Cuba–Dominican Republic 1947 1947 No intervention

Republic
China Taiwan Revolt 1947 1947 No intervention
Greece Truman Doctrine 1947 1947 Early, small
Israel Palestine War 1947 1948 Early, small
South Korea Cheju Rebellion 1948 1949 Forces already

present
Czechoslovakia Communism in 1948 1948 No intervention

Czechoslovakia
Costa Rica Costa Rica 1948 1948 No intervention
Yemen Arab Yemeni Imamate 1948 1948 No intervention

Republic
Costa Rica Costa Rica–Nicaraguan I 1948 1949 No intervention
Israel Arab–Israeli Conflict 1948 1948 Forces already

present
Malaysia Malayan Rebellion 1948 1957 No intervention
German Federal Berlin Blockade 1948 1949 Forces already

Republic present
List of Cases of Potential U.S. Intervention 155
Table A.1—Continued
Colombia Seventh Colombian 1948 1958 No intervention

“La Violencia”
Yugoslavia Soviet Bloc–Yugoslavia 1949 1951 Early, small
China Third Sino–Tibetan 1950 1950 No intervention
Philippines Hukbalahap Rebellion 1950 1954 Forces already

present
Indonesia South Moluccas 1950 1950 No intervention
South Korea Korean War 1950 1953 Early, large
Egypt Suez Canal 1951 1952 No intervention
Israel Hula Drainage 1951 1951 No intervention
Tunisia Franco–Tunisian 1952 1954 No intervention
Morocco Moroccan Independence 1953 1956 Forces already

present
Guatemala Guatemala War 1953 1954 Early, small
Italy Trieste II 1953 1953 Forces already

present
German East German Uprising 1953 1953 No intervention

Democratic
Republic
Myanmar Burma Infiltration 1953 1954 No intervention
Indonesia Indonesia Darul Islam 1953 1953 No intervention
Taiwan Offshore Islands 1954 1955 Forces already

present
Algeria Third Franco–Algerian 1954 1962 No intervention
Costa Rica Costa Rica–Nicaragua II 1955 1955 No intervention
Israel Gaza Raid/Czechoslovakia 1955 1956 No intervention

Arms
China Tibetan Khamba Rebellion 1956 1959 No intervention
Egypt Sinai War 1956 1956 Early, small

Indonesia Indonesian Leftists 1956 1962 Late, small
Hungary Soviet Invasion of Hungary 1956 1956 No intervention
Jordan Jordan Regime 1957 1957 No intervention
Tunisia France–Tunisia 1957 1958 No intervention
Syria Syria-Turkey Confrontation 1957 1957 Early, small
Morocco Ifni War 1957 1958 Forces already

present
Honduras Mocoron Incident 1957 1957 No intervention
Taiwan Taiwan Straits 1958 1958 Forces already

present
Lebanon Iraq/Lebanon Upheaval 1958 1958 Early, large
Cuba Cuban Revolution 1958 1959 No intervention
German Federal Berlin Deadline 1958 1959 Forces already

Republic present
Iraq Iraq–Shammar 1959 1959 No intervention
Iran Shatt Al-Arab I 1959 1960 Forces already

present
Panama Central America/Cuba I 1959 1959 Forces already

present
Rwanda Rwandan Social Revolution 1959 1962 No intervention
Israel Rottem 1960 1960 No intervention
Laos First Laotian 1960 1962 No intervention
Republic of Vietnam, Phase 1 1960 1965 Forces already

Vietnam present
Nicaragua Central America/Cuba II 1960 1960 Early, small
Angola Angolan–Portuguese 1961 1974 No intervention
Kuwait Kuwait Independence 1961 1961 No intervention
Tunisia Bizerta 1961 1961 No intervention

German Federal Berlin Wall 1961 1961 Forces already

Republic present
Iraq First Iraqi Kurd 1961 1963 No intervention
Cuba Bay of Pigs 1961 1961 Early, small
Syria Breakup of United Arab 1961 1961 No intervention

Republic
Taiwan Taiwan Strait III 1962 1962 Forces already

present
Saudi Arabia Yemen War I 1962 1963 Forces already

present
Cuba Cuban Missiles 1962 1962 Early, large
India China-India Border II 1962 1963 No intervention
Algeria Algerian Revolutionaries 1962 1963 No intervention
Laos Second Laotian, Phase 1 1963 1968 No intervention
Cyprus Cyprus I 1963 1964 Early, small
Haiti Dominican Republic–Haiti II 1963 1963 No intervention
Sudan First South Sudan 1963 1972 No intervention
Morocco Algeria-Morocco Border 1963 1963 Forces already

present
Rwanda First Rwanda 1963 1964 No intervention
Panama Panama Flag 1964 1964 Forces already

present
Democratic Congo II 1964 1964 Forces already

Republic of the present
Congo
Republic of Vietnam War, Phase 2 1965 1975 Forces already

Vietnam present
Iraq Second Iraqi Kurds 1965 1966 No intervention
Indonesia First West Papua 1965 1969 No intervention

Dominican Dominican Intervention 1965 1965 Early, large

Republic
Guatemala First Guatemala 1966 1968 Early, small
Jordan El Samu 1966 1966 No intervention
Nigeria Biafra 1967 1970 No intervention
Egypt Six Day War 1967 1967 Early, small
Cyprus Cyprus II 1967 1967 No intervention
China Cultural Revolution, 1967 1967 No intervention

Phase 1
China Cultural Revolution, 1967 1968 No intervention

Phase 2
Jordan Karameh 1968 1968 No intervention
South Korea Pueblo 1968 1968 Forces already

present
Oman Dhofar Rebellion, Phase 1 1968 1971 No intervention
Laos Second Laotian, Phase 2 1968 1973 No intervention
Czechoslovakia Prague Spring 1968 1968 No intervention
South Korea EC-121 Spy Plane 1969 1969 Forces already

present
Honduras Football War 1969 1969 No intervention
Iran Shatt Al-Arab II 1969 1969 Forces already

present
Egypt War of Attrition 1969 1970 No intervention
Iraq Third Iraqi Kurds 1969 1970 No intervention
Lebanon Cairo Agreement–PLO 1969 1969 No intervention
China Ussuri River 1969 1969 No intervention
Cambodia Communist Coalition 1970 1971 Forces already

present
Jordan Black September 1970 1970 Early, small

Cuba Cienfuegos Submarine Base 1970 1970 No intervention
Guatemala Second Guatemala 1970 1971 No intervention
Pakistan Bangladesh 1971 1971 Late, small
Cambodia Khmer Rouge 1971 1975 No intervention
Yemen Arab North–South Yemen I 1972 1972 No intervention

Republic
Burundi First Burundi 1972 1972 No intervention
Thailand Communist Insurgency 1972 1973 Forces already

present
Philippines First Philippine–Moro 1972 1981 Forces already

present
Philippines Philippines–New People’s 1972 1992 Forces already

Army present
Egypt Yom Kippur War 1973 1973 Early, small
Oman Dhofar Rebellion, Phase 2 1973 1975 No intervention
Israel Israel Mobilization 1973 1973 No intervention
Kuwait Iraq Invasion of Kuwait 1973 1973 No intervention
Turkey Turco–Cypriot 1974 1974 No intervention
Iraq Fourth Iraqi Kurds 1974 1975 No intervention
Indonesia East Timorese, Phase 2 1975 1976 No intervention
Lebanon Second Lebanese 1975 1976 No intervention
Ethiopia Eritrean War 1975 1978 No intervention
Indonesia East Timorese War, Phase 1 1975 1975 No intervention
Belize Belize I 1975 1975 No intervention
Cambodia Mayaguez 1975 1975 Early, small
Morocco Moroccan March 1975 1976 Forces already

present
Lebanon Syria Mobilization 1976 1976 No intervention

Greece Aegean Sea I 1976 1976 Forces already

present
Angola Angolan Control 1976 1991 No intervention
Uganda Entebbe Raid 1976 1976 No intervention
Syria Iraqi Threat 1976 1976 No intervention
South Korea Poplar Tree 1976 1976 Forces already

present
Indonesia Second West Papua 1976 1978 No intervention
Ethiopia Second Ogaden War, 1976 1977 No intervention

Phase 1
Indonesia East Timorese War, Phase 3 1976 1979 No intervention
Laos Third Laotian 1976 1979 No intervention

Phase 2
Egypt Egypt–Libya Clashes 1977 1977 No intervention
Vietnam Vietnamese–Cambodian 1977 1979 No intervention
Cambodia Vietnam Invasion of 1977 1979 No intervention

Cambodia
Belize Belize II 1977 1977 No intervention
Iran Overthrow of the Shah 1978 1979 Forces already

present
Nicaragua Sandinista Rebellion 1978 1979 No intervention
Afghanistan First Afghan Mujahideen 1978 1980 No intervention

Uprising

Phase 3
Guatemala Third Guatemala 1978 1984 No intervention
Lebanon Third Lebanese 1978 1978 No intervention
Lebanon Litani Operation 1978 1978 No intervention

Vietnam Khmer Insurgency 1979 1989 No intervention
Iran Anti-Khomeini Coalition 1979 1984 No intervention
Yemen Arab North–South Yemen II 1979 1979 Early, small

Republic
Iran U.S. Hostages in Iran 1979 1981 Early, small
Morocco Goulimime–Tarfaya Road 1979 1979 No intervention
Mozambique Mozambique 1979 1992 No intervention
El Salvador El Salvador 1979 1992 Late, small
China Sino-Vietnamese Punitive 1979 1979 No intervention
Morocco Tan Tan 1979 1979 No intervention
Poland Solidarity 1980 1981 No intervention
Morocco Operation Iman 1980 1980 No intervention
Tunisia Raid on Gafsa 1980 1980 No intervention
Uganda Second Uganda 1980 1986 No intervention
Afghanistan Soviet Quagmire 1980 1989 No intervention
Iraq Iran–Iraq 1980 1988 Late, small
Jordan Jordan–Syria Confrontation 1980 1980 No intervention
Egypt Libya Threat/Sadat 1980 1980 No intervention
Syria Hama 1981 1982 No intervention
Libya Gulf of Syrte I 1981 1981 Early, small
Iraq Iraq Nuclear Reactor 1981 1981 No intervention
United Falkland Islands 1982 1982 No intervention

Kingdom
Nicaragua Contra War 1982 1990 No intervention
Ethiopia Tigrean and Eritrean 1982 1991 No intervention
Israel War Over Lebanon 1982 1982 Early, small

Libya Libya Threat/Sudan 1983 1983 Early, small
Lebanon Fourth Lebanese Civil 1983 1984 Forces already

present
Sudan Second South Sudan 1983 1991 No intervention
German Federal Able Archer 83 1983 1983 Forces already

Republic present
Grenada Invasion of Grenada 1983 1983 Early, small
Greece Aegean Sea II 1984 1984 Forces already

present
Thailand Three Village Border I 1984 1984 No intervention
Thailand Vietnam Incursion into 1984 1984 No intervention

Thailand
Turkey First Turkish Kurds 1984 1986 Forces already

present
Vietnam Sino-Vietnamese Clashes 1984 1984 No intervention
Iraq Fifth Iraqi Kurds 1985 1988 No intervention
Tunisia Expulsion of Tunisians 1985 1985 No intervention
Libya Gulf of Syrte II 1986 1986 Early, small
Yemen People’s South Yemen 1986 1986 No intervention

Republic
Vietnam Sino-Vietnamese Border 1987 1987 No intervention

War
Greece Aegean Sea III 1987 1987 Forces already

present
Lebanon Syrian Intervention/ 1987 1987 No intervention

Lebanon
Thailand Three Village Border II 1987 1988 No intervention
Libya Libyan Jets 1988 1989 Early, small
Vietnam Spratly Islands 1988 1988 No intervention
Somalia First Somalia 1988 1991 No intervention

Panama Invasion of Panama 1989 1990 Early, large
Romania Romania 1989 1989 No intervention
Morocco Galtat Zemmour II 1989 1989 No intervention
Papua New Bougainville Secession 1989 1992 No intervention

Guinea
Cambodia First Cambodian Civil War 1989 1991 No intervention
Colombia Eighth Colombia 1989 2016 Late, small
Indonesia First Aceh 1989 1991 No intervention
Lebanon Fifth Lebanese 1989 1990 No intervention
Afghanistan Second Afghan 1989 2001 No intervention

Mujahideen Uprising
Liberia First Liberia 1989 1990 Early, small
India Kashmir III–Nuclear 1990 1990 No intervention

Confrontation
Kuwait Gulf War 1990 1991 Early, large
Turkey Second Turkish Kurds 1991 1999 Forces already

present
Yugoslavia Croatian Independence 1991 1992 No intervention
Kuwait Bubiyan 1991 1991 Forces already

present
Somalia Second Somalia 1991 1997 Late, small
Turkey First Kurdistan Workers’ 1991 1992 Forces already

Party In Iraq present
Iraq Shiite and Kurdish 1991 1991 Early, large

Uprisings
Bosnia and Bosnian Serb Rebellion 1992 1994 Late, large

Herzegovina
Liberia Second Liberia 1992 1995 No intervention
Angola Angolan War of the Cities 1992 1994 No intervention

Algeria Algerian Islamic Front 1992 1999 No intervention
Iraq Iraq No-Fly Zone 1992 1992 Forces already

present
Bosnia and Bosnian Independence 1992 1992 No intervention

Herzegovina
Burundi Second Burundi 1993 1998 No intervention
Cambodia Second Cambodia Civil 1993 1997 No intervention
South Korea North Korea Nuclear I 1993 1994 Forces already

present
Israel Operation Accountability 1993 1993 No intervention
Iraq Iraqi Kurd Internecine 1994 1995 Forces already

present
Rwanda Second Rwanda 1994 1994 Late, small
Haiti Haiti Military Regime 1994 1994 Early, small
Kuwait Iraq Deployment/Kuwait 1994 1994 Forces already

present
Yemen South Yemeni Secessionist 1994 1994 No intervention
Philippines Spratly Islands 1995 1995 No intervention
Croatia Croatia–Krajina War 1995 1995 Forces already

present
Eritrea Red Sea Islands 1995 1995 No intervention
Taiwan Taiwan Strait IV 1995 1996 Forces already

present
Iraq Sixth Iraqi Kurds 1996 1996 Forces already

present
Greece Aegean Sea IV 1996 1996 No intervention
Israel Operation Grapes of Wrath 1996 1996 No intervention
Liberia Third Liberia 1996 1996 Early, small
South Korea North Korean Submarine 1996 1996 Forces already

present
Democratic Fifth Democratic Republic 1996 1997 No intervention

Republic of the of the Congo
Congo
Turkey Second Kurdistan Workers’ 1997 1997 Forces already

Party In Iraq present
Iraq UN Special Commission I 1997 1998 Forces already

present
Yugoslavia Kosovo Independence 1998 1999 No intervention
Afghanistan U.S. East Africa Embassy 1998 1998 Early, small

Bombings
Angola Third Angolan 1998 2002 No intervention
Cyprus Cyprus/Turkey Missile 1998 1998 No intervention
Iraq UN Special Commission II/ 1998 1998 Forces already

Operation Desert Fox present
India India/Pakistan Nuclear 1998 1998 No intervention

Tests
Democratic Africa’s World War 1998 2002 No intervention

Republic of the
Congo
Indonesia Second Aceh 1999 2002 No intervention
Indonesia Moluccas Sectarian 1999 2000 No intervention
East Timor East Timor II 1999 1999 No intervention
India Kargil War 1999 1999 No intervention
Yugoslavia War for Kosovo 1999 1999 Late, large
Israel Al Aqsa Intifada 2000 2003 No intervention
Philippines Second Philippine–Moro 2000 2001 Forces already

present
Afghanistan Afghanistan–United States 2001 Ongoing Early, large
Iraq Iraq Regime Change 2002 2011 Early, large
Thailand Myanmar–Thailand 2002 2002 No intervention

Morocco Parsley Island 2002 2002 No intervention
South Korea North Korea Nuclear II 2002 2004 Forces already

present
Iran Iran Nuclear I 2003 2004 No intervention
Sudan Darfur 2003 2006 No intervention
Indonesia Third Aceh 2003 2003 No intervention
Philippines Third Philippine–Moro 2003 2003 Forces already

present
Pakistan Waziristan 2004 2006 Forces already

present
Yemen First Yemeni Cleric 2004 2005 No intervention
Philippines Philippine Joint Offensive 2005 2006 Forces already

present
Israel Israel Lebanon War II 2006 2006 No intervention
South Korea North Korea Nuclear III 2006 2007 Forces already
present
Iran Iran Nuclear II 2006 2007 No intervention
Yemen Second Yemeni Cleric 2007 2007 No intervention
Yemen Yemen–Aqap 2009 Ongoing No intervention
South Korea North Korea Nuclear IV– 2009 2009 Forces already
Satellite Launch present
South Korea Yeonpyeong Island 2010 2011 Forces already

present
South Korea Cheonan Sinking 2010 2010 Forces already

present
Libya Libyan Civil War 2011 2011 Early, small
Syria Syrian Civil War 2011 Ongoing Late, small
Philippines Scarborough Shoal 2012 2012 No intervention
Turkey Syria-Turkey Border 2012 2012 No intervention

Incidents
South Korea North Korea Nuclear V 2013 2013 Forces already

present
Israel Israel–Gaza 2014 2014 No intervention
Iraq Iraq–ISIL 2014 Ongoing Late, small
Ukraine Donbass Conflict 2014 Ongoing No intervention
India India-Pakistan Border 2014 2014 No intervention

Firing
Turkey Turkey–Kurdistan Workers’ 2015 2016 Forces already

Party present
Libya Libya–ISIL 2015 2016 Late, small
South Korea Korean Land Mine 2015 2015 Forces already

present
Turkey Turkey–Russia Jet Incident 2015 2016 Forces already

present
References
Abadie, Alberto, Alexis Diamond, and Jens Hainmueller, “Comparative Politics

and the Synthetic Control Method,” American Journal of Political Science, Vol. 59,
No. 2, April 2015, pp. 495–510.
Abadie, Alberto, and Javier Garbeazabal, “The Economic Costs of Conflict: A
Case Study of the Basque Country,” American Economic Review, Vol. 93, No. 1,
March 2003, pp. 113–132.
Allison, Graham T., and Philip Zelikow, Essence of Decision: Explaining the Cuban
Missile Crisis, 2nd ed., New York: Longman, 1999.
Association for Diplomatic Studies and Training, “The Dominican Civil War of
1965,” Huffington Post, updated December 6, 2017. As of February 8, 2018:
https://www.huffingtonpost.com/adst/
the-dominican-civil-war-o_b_8478570.html
Bahador, Babak, The CNN Effect in Action: How the News Media Pushed the West
Toward War in Kosovo, New York: Palgrave Macmillan, 2007.
Baker, Peter, “Bush Warns Against Nuclear-Armed Iran; U.S. Presses Allies for
U.N. Action,” Washington Post, January 14, 2006.
Barrett, David M., “Sterilizing a ‘Red Infection’: Congress, the CIA, and
Guatemala, 1954,” Central Intelligence Agency, Center for the Study of
Intelligence, May 8, 2007. As of November 12, 2018:
https://www.cia.gov/library/center-for-the-study-of-intelligence/kent-csi/vol44no5/
html/v44i5a03p.htm
Belasco, Amy, The Cost of Iraq, Afghanistan, and Other Global War on Terror
Operations Since 9/11, Washington, D.C.: Congressional Research Service,
RL33110, December 8, 2014. As of November 12, 2018:
https://fas.org/sgp/crs/natsec/RL33110.pdf
Benard, Alexander, “Lessons from Iraq and Bosnia on the Theory and Practice
of No-Fly Zones,” Journal of Strategic Studies, Vol. 27, No. 3, September 2004,
pp. 454–478.
169
Betts, Richard K., American Force: Dangers, Delusions, and Dilemmas in National
Security, New York: Columbia University Press, 2012.
Bevins, Vincent, “What the United States Did in Indonesia,” The Atlantic,
October 20, 2017.
Biddle, Stephen, Military Power: Explaining Victory and Defeat in Modern Battle,
Princeton, N.J.: Princeton University Press, 2004.
Biddle, Stephen, Jeffrey A. Friedman, and Jacob N. Shapiro, “Testing the Surge:
Why Did Violence Decline in Iraq in 2007?” International Security, Vol. 37, No. 1,
Summer 2012, pp. 7–40.
Blair, Charles P., “War with Iran? Revisiting the Potentially Staggering Costs to
the Global Economy,” Federation of American Scientists, May 23, 2013.
Blanchard, Christopher M., and Carla E. Humud, The Islamic State and U.S.
Policy, Washington, D.C.: Congressional Research Service, R43612, September 25,
2018. As of November 12, 2018:
https://fas.org/sgp/crs/mideast/R43612.pdf
Bliesemann de Guevara, Berit, and Florian P. Kühn. “‘The International
Community Needs to Act’: Loose Use and Empty Signalling of a Hackneyed
Concept,” International Peacekeeping, Vol. 18, No. 2, 2011, pp. 135–151.
Boissoneault, Lorraine, “The Genocide the U.S. Can’t Remember, but Bangladesh
Can’t Forget,” Smithsonian, December 16, 2016.
Bolt, Jutta, Robert Inklaar, Herman de Jong, and Jan Luiten van Zanden,
“Rebasing ‘Maddison’: New Income Comparisons and the Shape of Long-Run
Economic Development,” Maddison Project Database, Version 2018, Working
Paper No. 10, 2018. As of January 29, 2020:
https://www.rug.nl/ggdc/historicaldevelopment/maddison/releases/
maddison-project-database-2018
Bonner, Raymond, “America’s Role in El Salvador’s Deterioration,” The Atlantic,
January 20, 2018.
Brandt, Patrick T., T. David Mason, Mehmet Gurses, Nicolai Petrovsky, and
Dagmar Radin, “When and How the Fighting Stops: Explaining the Duration
and Outcome of Civil Wars,” Defence and Peace Economics, Vol. 19, No. 6, 2008,
pp. 415–434.
Brecher, Michael, and Jonathan Wilkenfeld, A Study of Crisis, Ann Arbor, Mich.:
University of Michigan Press, 2000.
Brecher, Michael, Jonathan Wilkenfeld, Kyle Beardsley, Patrick James, and David
Quinn, International Crisis Behavior Data Codebook, Version 12, 2017.
Bueno de Mesquita, Bruce, and George W. Downs, “Intervention and
Democracy,” International Organization, Vol. 60, No. 3, 2006, pp. 627–649.
References 171
Bueno de Mesquita, Bruce, and Randolph M. Siverson, “War and the Survival
of Political Leaders: A Comparative Study of Regime Types and Political
Accountability,” American Political Science Review, Vol. 89, No. 4, December 1995,
pp. 841–855.
Bueno de Mesquita, Bruce, Alastair Smith, Randolph M. Siverson, and James D.
Morrow, The Logic of Political Survival, Cambridge, Mass.: MIT Press, 2005.
Bunzl, Martin, “Counterfactual History: A User’s Guide,” American Historical
Review, Vol. 109, No. 3, June 2004, pp. 845–858.
Burk, James, “Public Support for Peacekeeping in Lebanon and Somalia: Assessing
the Casualties Hypothesis,” Political Science Quarterly, Vol. 114, No. 1, 1999,
pp. 53–78.
Burris, Howard L., “Soviet Concern Over Cuba,” memorandum to Vice President
Lyndon B. Johnson, October 2, 1962.
Cale, Paul P., “The United States Military Advisory Group in El Salvador, 1979–
1992,” Small Wars Journal, 1996. As of November 11, 2018:
http://smallwarsjournal.com/documents/cale.pdf
Call, Charles T., Why Peace Fails: The Causes and Prevention of Civil War
Recurrence, Washington, D.C.: Georgetown University Press, 2012.
Central Intelligence Agency, “The Current Situation in Colombia,” Situation
Memorandum 32-49, May 31, 1949.
———, “OCI Handbook: Colombia,” June 13, 1955.
———, “Implications of Khrushchev’s Message of 28 October,” Boston, Mass.:
John F. Kennedy Library, October 29, 1962a.
———, “The Crisis: USSR/Cuba,” Boston, Mass.: John F. Kennedy Library,
October 29, 1962b.
———, Office of National Estimates, “Implications of Madagascar’s Unfinished
Revolution,” memorandum, July 27, 1972.
Chivvis, Christopher S., Toppling Qaddafi: Libya and the Limits of Liberal
Intervention, New York: Cambridge University Press, 2014.
Clarke, Michael, “The Making of Britain’s Libya Strategy,” in Adrian Johnson and
Saqeb Mueen, eds., Short War, Long Shadow: The Political and Military Legacies
of the 2011 Libya Campaign, London: Royal United Services Institute, 2012,
pp. 7–14. As of November 12, 2018:
https://rusi.org/publication/whitehall-reports/
short-war-long-shadow-political-and-military-legacies-2011-libya
Clymer, Kenton, “The United States and the Guomindang (KMT) Forces in
Burma, 1949–1954: A Diplomatic Disaster,” Chinese Historical Review, Vol. 21,
No. 1, 2014, pp. 24–44.
Cockayne, James, and David Malone, “Creeping Unilateralism: How Operation

Provide Comfort and the No-Fly Zones in 1991 and 1992 Paved the Way for the
Iraq Crisis of 2003,” Security Dialogue, Vol. 37, No. 1, March 2006, pp. 123–141.
Cole, Ronald H., Operation Urgent Fury: The Planning and Execution of Joint
Operations in Grenada, 12 October–2 November 1983, Washington, D.C.: Joint
History Office, Office of the Chairman of the Joint Chiefs of Staff, 1997.
Coleman, David, “JFK’s Presidential Approval Ratings,” webpage, undated. As of
November 12, 2018:
https://jfk14thday.com/jfk-presidential-approval-ratings/
Coleman, David, ed., “LBJ Regretted Ordering U.S. Troops into Dominican
Republic in 1965, White House Tapes Confirm,” National Security Archive,
Electronic Briefing Book No. 513, April 28, 2015. As of February 8, 2018:
https://nsarchive2.gwu.edu/NSAEBB/NSAEBB513/
Collins, Joseph J., Choosing War: The Decision to Invade Iraq and Its Aftermath,
Washington, D.C.: Institute for National Strategic Studies, National Defense
University, April 2008.
———, “Initial Planning and Execution in Afghanistan and Iraq,” in Richard D.
Hooker, Jr., and Joseph J. Collins, eds., Lessons Encountered: Learning from the
Long War, Washington, D.C.: National Defense University Press, September 2015,
pp. 21–88.
Congressional Research Service, Afghanistan: Background and U.S. Policy in Brief,
Washington, D.C., R45122, May 1, 2018. As of November 12, 2018:
https://www.everycrsreport.com/files/
20180501_R45122_aa2ec56eb7b9eea1d 00276f5c3d05d3bd4c5bd5c.pdf
Crawford, Neta C., United States Budgetary Costs of Post-9/11 Wars Through
FY2018: A Summary of the $5.6 Trillion in Costs for the US Wars in Iraq, Syria,
Afghanistan, and Pakistan, and Post-9/11 Veterans Care and Homeland Security,
Providence, R.I.: Watson Institute of International and Public Affairs, Brown
University, November 2017. As of November 12, 2018:
http://watson.brown.edu/costsofwar/files/cow/imce/papers/2017/
Costs%20of%20U.S.%20Post-9_11%20NC%20Crawford%20FINAL%20.pdf
Cullather, Nicholas, Operation PBSUCCESS: The United States and Guatemala,
1952–1954, Washington, D.C.: Central Intelligence Agency, Center for the Study
of Intelligence, 1994.
Cusimano, Maryann K., Operation Restore Hope: The Bush Administration’s
Decision to Intervene in Somalia, Washington, D.C.: Institute for the Study of
Diplomacy, 1995.
Daalder, Ivo H., Getting to Dayton: The Making of America’s Bosnia Policy,
Washington, D.C.: Brookings Institution Press, 2000.
References 173
Daalder, Ivo H., and Philip H. Gordon, “We Should Strike Iran, but Not with
Bombs,” Washington Post, January 22, 2006.
de Galbert, Simond, Back to the Future: Europe, the United States, and Iran
Sanctions After the Nuclear Deal, Washington, D.C.: Center for Strategic and
International Studies, November 2015. As of March 5, 2020:
https://www.csis.org/analysis/back-future
Dobbins, James, Seth G. Jones, Keith Crane, Andrew Rathmell, Brett Steele,
Richard Teltschik, and Anga R. Timilsina, The UN’s Role in Nation-Building:
From the Congo to Iraq, Santa Monica, Calif.: RAND Corporation, MG-304-RC,
2005. As of January 29, 2020:
https://www.rand.org/pubs/monographs/MG304.html
Dobbins, James, Michele A. Poole, Austin Long, and Benjamin Runkle, After the
War: Nation-Building from FDR to George W. Bush, Santa Monica, Calif.: RAND
Corporation, MG-716-CC, 2008. As of January 29, 2020:
Dobbs, Michael, “One Minute to Midnight: Kennedy, Khrushchev and Castro
on the Brink of Nuclear War,” National Security Archive, George Washington
University, June 4, 2008. As of November 12, 2018:
http://nsarchive.gwu.edu/nsa/cuba_mis_cri/dobbs/gitmo.htm
“Dodging Armageddon: The Third World War That Almost Was, 1950,”
Cryptologic Quarterly, February 24, 1998. As of May 1, 2020:
https://www.nsa.gov/Portals/70/documents/news-features/declassified-documents/
cryptologic-quarterly/Dodging_Armageddon.pdf
Downes, Alexander B., and Jonathan Monten, “Forced to Be Free? Why Foreign-
Imposed Regime Change Rarely Leads to Democratization,” International Security,
Vol. 37, No. 4, 2013, pp. 90–131.
Doyle, Michael W., and Nicholas Sambanis, “International Peacebuilding: A
Theoretical and Quantitative Analysis,” American Political Science Review, Vol. 94,
No. 4, December 2000, pp. 779–801.
Easterly, William, Shanker Satyanath, and Daniel Berger, Superpower Interventions
and Their Consequences for Democracy: An Empirical Inquiry, Cambridge, Mass.:
National Bureau of Economic Research, Working Paper No. 13992, May 2008.
Edelstein, David M., “Occupational Hazards: Why Military Occupations Succeed
or Fail,” International Security, Vol. 29, No. 1, Summer 2004, pp. 49–91.
Edelstein, David M., “Foreign Militaries, Sustainable Institutions, and Postwar
Statebuilding,” in Roland Paris and Timothy D. Sisk, eds., The Dilemmas of
Statebuilding: Confronting the Contradictions of Postwar Peace Operations, New
York: Routledge, 2009, pp. 81–103.
Elbadawi, Ibrahim A., and Nicholas Sambanis, External Interventions and the
Duration of Civil Wars, Washington, D.C.: World Bank, September 2000.
Elster, Jon, Logic and Society: Contradictions and Possible Worlds, New York: John
Wiley, 1978.
Exum, Andrew, “What’s Really at Stake for America in Yemen’s Conflict,” The
Atlantic, April 14, 2017. As of March 5, 2020:
https://www.theatlantic.com/international/archive/2017/04/
yemen-trump-aqap/522957/
Fearon, James D., “Counterfactuals and Hypothesis Testing in Political Science,”
World Politics, Vol. 43, No. 2, January 1991, pp. 169–195.
Fettweis, Christopher J., “Credibility and the War on Terror,” Political Science
Quarterly, Vol. 122, No. 4, 2007/2008, pp. 607–633.
Florig, Dennis, “Hegemonic Overreach vs. Imperial Overstretch,” Review of
International Studies, Vol. 36, No. 4, October 2010, pp. 1103–1119.
Fogel, Robert William, Railroads and American Economic Growth: Essays in
Econometric History, Baltimore, Md.: Johns Hopkins Press, 1964.
Fortna, Virginia Page, “Does Peacekeeping Keep Peace? International Intervention
and the Duration of Peace After Civil War,” International Studies Quarterly,
Vol. 48, No. 2, June 2004, pp. 269–292.
Frederick, Bryan, Stephen Watts, Matthew Lane, Abby Doll, Ashley L. Rhoades,
and Meagan L. Smith, Understanding the Deterrent Impact of U.S. Overseas Forces,
Santa Monica, Calif.: RAND Corporation, RR-2533-A, 2020. As of February 10,
2020:
https://www.rand.org/pubs/research_reports/RR2533.html
Freedman, Lawrence, and Efraim Karsh, The Gulf Conflict, 1990–1991: Diplomacy
and War in the New World Order, Princeton, N.J.: Princeton University Press,
1993.
Gartenstein-Ross, Daveed, “Successes and Failures of the U.S. and NATO
Intervention in Libya,” testimony presented before the Committee on Oversight
and Government Reform, Washington, D.C.: Foundation for Defense of
Democracies, May 1, 2014.
Gartner, Scott Sigmund, and Gary M. Segura, “War, Casualties, and Public
Opinion,” Journal of Conflict Resolution, Vol. 42, No. 3, June 1998, pp. 278–300.
George, Alexander L., and Andrew Bennett, Case Studies and Theory Development
in the Social Sciences, Cambridge, Mass.: MIT Press, 2005.
Gibbons, Phillip, “U.S. No-Fly Zones in Iraq: To What End?” Washington, D.C:
Washington Institute, July 1, 2002. As of November 12, 2018:
https://www.washingtoninstitute.org/policy-analysis/
view/u.s.-no-fly-zones-in-iraq-to-what-end
References 175
Gleditsch, Nils Petter, Lene Siljeholm Christiansen, and Håvard Hegre,

Democratic Jihad? Military Intervention and Democracy, Washington, D.C.: World
Bank, Working Paper No. WPS4242, June 2007.
Gleditsch, Nils Petter, Peter Wallensteen, Mikael Eriksson, Margareta Sollenberg,
and Håvard Strand, “Armed Conflict 1946–2001: A New Dataset,” Journal of
Peace Research, Vol. 39, No. 5, 2002, pp. 615–637.
Glenn, Cameron, “Timeline: The Rise, Spread, and Fall of the Islamic State,”
Wilson Center, July 5, 2016. As of November 11, 2018:
https://www.wilsoncenter.org/article/
timeline-the-rise-spread-and-fall-the-islamic-state
Goshko, John M., “U.S. War by Proxy at an End,” Washington Post, February 27,
2019. As of March 9, 2020:
https://www.washingtonpost.com/archive/politics/1990/02/27/
us-war-by-proxy-at-an-end/770483d0-c355-4288-8819-9b0dcc928aee/
Granados, Erick O., Impact of U.S. Foreign Policy and Assistance in El Salvador,
Quantico, Va.: U.S. Marine Corps Command and Staff College, May 2012. As of
November 12, 2018:
http://www.dtic.mil/dtic/tr/fulltext/u2/a601169.pdf
Greenberg, Lawrence M., United States Army Unilateral and Coalition Operations
in the 1965 Dominican Republic Intervention, Washington, D.C.: U.S. Army
Center of Military History, November 1986.
Grossman, Marc, Executive Secretary, U.S. Department of State, “Discussion
Paper for the Deputies Committee on Peacekeeping Options in Rwanda,”
memorandum to William H. Itoh, Executive Secretary, National Security Council,
Washington, D.C., May 16, 1994.
Haass, Richard N., “Military Force: A User’s Guide,” Foreign Policy, No. 96, 1994,
pp. 21–37.
Haass, Richard N., and Martin Indyk, “Beyond Iraq: A New U.S. Strategy for the
Middle East,” Foreign Affairs, Vol. 88, No. 1, 2009, pp. 41–58.
Hartzell, Caroline, Matthew Hoddie, and Donald Rothchild, “Stabilizing the
Peace After Civil War: An Investigation of Some Key Variables,” International
Organization, Vol. 55, No. 1, Winter 2001, pp. 183–208.
Harvey, Frank P., “President Al Gore and the 2003 Iraq War: A Counterfactual
Test of Conventional ‘W’isdom,” Canadian Journal of Political Science, Vol. 45,
No. 1, March 2012, pp. 1–32.
Haulman, Daniel L., “Crisis in Grenada: Operation Urgent Fury,” in A. Timothy
Warnock, ed., Short of War: Major USAF Contingency Operations, 1947–1997,
Maxwell Air Force Base, Ala.: Air University Press, 2000, pp. 135–144.
Hawthorn, Geoffrey, Plausible Worlds: Possibility and Understanding in History and
the Social Sciences, New York: Cambridge University Press, 1991.
Heerten, Lasse, and A. Dirk Moses, “The Nigeria-Biafra War: Postcolonial

Conflict and the Question of Genocide,” Journal of Genocide Research, Vol. 16,
Nos. 2–3, 2014, pp. 169–203.
Hegre, Håvard, and Håvard Mokleiv Nygård. “Governance and Conflict
Relapse,” Journal of Conflict Resolution, Vol. 59, No. 6, 2015, pp. 984–1016.
Hermann, Margaret G., and Charles W. Kegley, Jr., “The U.S. Use of Military
Intervention to Promote Democracy: Evaluating the Record,” International
Interactions, Vol. 24, No. 2, 1998, pp. 91–114.
Hilsman, Roger, Director of the Bureau of Intelligence and Research, “Basic
Elements in the Ruanda-Urundi Situation,” memorandum to Secretary of State
Dean Rusk, June 8, 1962.
James, Patrick, and Jean Sébastien Rioux, “International Crises and Linkage
Politics: The Experiences of the United States, 1953–1994,” Political Research
Quarterly, Vol. 51, No. 3, September 1998, pp. 781–812.
Jentleson, Bruce W., “The Pretty Prudent Public: Post Post-Vietnam American
Opinion on the Use of Military Force,” International Studies Quarterly, Vol. 36,
No. 1, 1992, pp. 49–73.
Johnston, Patrick B., and Anoop K. Sarbahi, “The Impact of US Drone Strikes on
Terrorism in Pakistan,” International Studies Quarterly, Vol. 60, No. 2, June 2016,
pp. 203–219.
Jordan, Javier, “The Effectiveness of the Drone Campaign Against al Qaeda
Central: A Case Study,” Journal of Strategic Studies, Vol. 37, No. 1, 2014, pp. 4–29.
Karlén, Niklas, “The Legacy of Foreign Patrons: External State Support and
Conflict Recurrence,” Journal of Peace Research, Vol. 54, No. 4, 2017, pp. 499–512.
Karlsson, Stefan, “The Total Cost of the Libya Intervention,” Christian Science
Monitor, April 15, 2011.
Kavanagh, Jennifer, Bryan Frederick, Matthew Povlock, Stacie L. Pettyjohn,
Angela O’Mahony, Stephen Watts, Nathan Chandler, John Speed Meyers,
and Eugeniu Han, The Past, Present, and Future of U.S. Ground Interventions:
Identifying Trends, Characteristics, and Signposts, Santa Monica, Calif.: RAND
Corporation, RR-1831-A, 2017. As of January 29, 2020:
Kavanagh, Jennifer, Bryan Frederick, Alexandra Stark, Nathan Chandler,
Meagan L. Smith, Matthew Povlock, Lynn E. Davis, and Edward Geist,
Characteristics of Successful U.S. Military Interventions, Santa Monica, Calif.:
RAND Corporation, RR-3062-A, 2019. As of January 29, 2020:
References 177
Kelly, John H., “Lebanon: 1982–1984,” in Jeremy R. Azrael and Emil A. Payin,
eds., U.S. and Russian Policymaking with Respect to the Use of Force, Santa Monica,
Calif.: RAND Corporation, CF-129-CRES, 1996. As of May 1, 2020:
https://www.rand.org/pubs/conf_proceedings/CF129.html
Kilcullen, David, The Accidental Guerrilla: Fighting Small Wars in the Midst of a
Big One, New York: Oxford University Press, 2009.
Kissinger, Henry A., “U.S. Options in Biafra Relief,” memorandum to President
Richard Nixon, Washington, D.C., January 28, 1969.
Kocher, Matthew Adam, Adria Lawrence, and Nuno P. Monteiro, The Rabbit in
the Hat: Nationalism and Resistance to Foreign Occupation, Chicago, Ill.: American
Political Science Association 2013 Annual Meeting, August 20, 2013.
Krepon, Michael, and Mishi Faruqee, eds., Conflict Prevention and Confidence-
Building Measures in South Asia: The 1990 Crisis, Washington, D.C.: Henry L.
Stimson Center, Occasional Paper No. 17, April 1994. As of November 12, 2018:
https://www.stimson.org/sites/default/files/file-attachments/
occasionalpaper17-web_1.pdf
Kreps, Sarah E., “Multilateral Military Interventions: Theory and
Practice,” Political Science Quarterly, Vol. 123, No. 4, Winter 2008/2009,
pp. 573–603.
Larson, Eric V., Casualties and Consensus: The Historical Role of Casualties in
Domestic Support for U.S. Military Operations, Santa Monica, Calif.: RAND
Corporation, MR-726-RC, 1996. As of January 29, 2020:
https://www.rand.org/pubs/monograph_reports/MR726.html
Laub, Zachary, and Kali Robinson, “What Is the Status of the Iran Nuclear
Agreement?” Council on Foreign Relations, updated January 7, 2020. As of
June 3, 2020:
https://www.cfr.org/backgrounder/impact-iran-nuclear-agreement
“Leader of Grenada Is Reported Killed by Troops,” New York Times, October 20,
1983.
Lebovic, James H., and Ashfaq Ishaq, “Military Burden, Security Needs, and
Economic Growth in the Middle East,” Journal of Conflict Resolution, Vol. 31,
No. 1, March 1987, pp. 106–138.
Lebow, Richard Ned, “What’s So Different About a Counterfactual?” World
Politics, Vol. 52, No. 4, July 2000, pp. 550–585.
LeoGrande, William M., Our Own Backyard: The United States in Central
America, 1977–1992, Chapel Hill, N.C.: University of North Carolina Press,
1998.
Levy, Jack S., “Counterfactuals and Case Studies,” in Janet M. Box-Steffensmeier,

Henry E. Brady, and David Collier, eds., The Oxford Handbook of Political
Methodology, Oxford, United Kingdom: Oxford University Press, 2008,
pp. 627–644.
Levy, Jack S., and Gary Goertz, eds., Explaining War and Peace: Case Studies and
Necessary Condition Counterfactuals, New York: Routledge, 2007.
Lewis, Paul, “Putin Insists Russia Wants to Repair US Relations – but Ill Will for
Obama Is Clear,” The Guardian, April 17, 2014. As of November 12, 2018:
https://www.theguardian.com/world/2014/apr/17/
putin-russia-relations-obama-us-ukraine
Little, Douglas, “His Finest Hour? Eisenhower, Lebanon, and the 1958 Middle
East Crisis,” Diplomatic History, Vol. 20, No. 1, January 1996, pp. 27–54.
Lounsbery, Marie Olson, Frederic Pearson, and Andrea Kathryn Talentino,
“Unilateral and Multilateral Military Intervention: Effects on Stability and
Security,” Democracy and Security, Vol. 7, No. 3, 2011, pp. 227–257.
Lustick, Ian, Miguel Garces, and Thomas McCauley, An Agent-Based Model
of Counterfactual Opportunities for Reducing Atrocities in Syria, 2011-2014,
Washington, D.C.: Simon-Skjodt Center for the Prevention of Genocide, August
2017.
Lyons, Terrence, “Liberia’s Path from Anarchy to Elections,” Current History,
Vol. 97, No. 619, May 1998, pp. 229–233.
Maddison, Angus, homepage, updated September 3, 2008. As of February 6,
2020:
http://www.ggdc.net/maddison/oriindex.htm
Mahnken, Tom, “The Gulf War in Retrospect,” Foreign Policy, January 20, 2011.
Mandelbaum, Michael, “The Reluctance to Intervene,” Foreign Policy, No. 95,
Summer 1994, pp. 3–18.
Marcus, Jonathan, “Why China and Russia Rebuffed the West on Syria,” BBC
News, October 5, 2011. As of November 12, 2018:
http://www.bbc.com/news/world-middle-east-15180732
Marshall, Monty G., Ted Robert Gurr, and Keith Jaggers, Polity IV: Political
Regime Characteristics and Transitions, 1800–2017, Vienna, Va.: Center for
Systemic Peace, 2018.
Mason, T. David, Joseph P. Weingarten, Jr., and Patrick J. Fett, “Win, Lose,
or Draw: Predicting the Outcome of Civil Wars,” Political Research Quarterly,
Vol. 52, No. 2, June 1999, pp. 239–268.
McCain, John, “Obama Has Made America Look Weak,” New York Times,
March 14, 2014.
References 179
McNeil, Brian Edward, Frontiers of Need: Humanitarianism and the American

Involvement in the Nigerian Civil War, 1967–1970, dissertation, University of
Texas at Austin, 2014.
Meernik, James, “United States Military Intervention and the Promotion of
Democracy,” Journal of Peace Research, Vol. 33, No. 4, 1996, pp. 391–402.
Millet, Richard L., “Panama and Haiti,” in Jeremy R. Azrael and Emil A.
Payin, eds., U.S. and Russian Policymaking with Respect to the Use of Force, Santa
Monica, Calif.: RAND Corporation, CF-129-CRES, 1996, pp. 137–162. As of
November 16, 2018:
https://www.rand.org/pubs/conf_proceedings/CF129.html
Mitchell, Christopher R., The Structure of International Conflict, London:
Macmillan, 1981.
———, Cutting Losses: Reflections on Appropriate Timing, Fairfax, Va.: Institute for
Conflict Analysis and Resolution, Working Paper No. 9, 1995.
Mitton, John, “Selling Schelling Short: Reputations and American Coercive
Diplomacy After Syria,” Contemporary Security Policy, Vol. 36, No. 3, 2015,
pp. 408–431.
Mueller, John E., “Presidential Popularity from Truman to Johnson,” American
Political Science Review, Vol. 64, No. 1, March 1970, pp. 18–34.
———, War, Presidents, and Public Opinion, New York: Wiley, 1973.
NATO—See North Atlantic Treaty Organization.
Newsom, David D., Director of the Office of Northern African Affairs, “U.S.
Policy Toward the Ogaden,” memorandum to Assistant Secretary of State for
African Affairs G. Mennen Williams, Washington, D.C., August 26, 1963.
Olander, Marcia, “Costa Rica in 1948: Cold War or Local War?” The Americas,
Vol. 52, No. 4, April 1996, pp. 465–493.
Oneal, John R., and Anna Lillian Bryan, “The Rally ‘Round the Flag Effect
in U.S. Foreign Policy Crises, 1950–1985,” Political Behavior, Vol. 17, No. 4,
December 1995, pp. 379–401.
Oneal, John R., Brad Lian, and James H. Joyner, Jr, “Are the American
People ‘Pretty Prudent’? Public Responses to U.S. Uses of Force, 1950–1988,”
International Studies Quarterly, Vol. 40, No. 2, June 1996, pp. 261–279.
Ostrom, Charles W., Jr., and Dennis M. Simon, “Promise and Performance: A
Dynamic Model of Presidential Popularity,” American Political Science Review,
Vol. 79, No. 2, June 1985, pp. 334–358.
Panama Canal Commission, Annual Report: Fiscal Year Ended September 30, 1990,
Washington, D.C.: U.S. Government Printing Office, 1990.
Pattison, James, “Outsourcing the Responsibility to Protect: Humanitarian

Intervention and Private Military and Security Companies,” International Theory,
Vol. 2, No. 1, 2010, pp. 1–31.
Pearson, Frederic S., Scott Walker, and Stephanie Stern, “Military Intervention
and Prospects for Democratization,” International Journal of Peace Studies, Vol. 11,
No. 2, 2006, pp. 63–86.
Peceny, Mark, “Forcing Them to Be Free,” Political Research Quarterly, Vol. 52,
No. 3, September 1999, pp. 549–582.
Peksen, Dursun, “Does Foreign Military Intervention Help Human Rights?”
Political Research Quarterly, Vol. 65, No. 3, 2012, pp. 558–571.
Pettersson, Therése, and Kristine Eck, “Organized Violence, 1989–2017,” Journal
of Peace Research, Vol. 55, No. 4, 2018, pp. 535–547.
Pettyjohn, Stacie L., and Jennifer Kavanagh, Access Granted: Political Challenges
to the U.S. Overseas Military Presence, 1945–2014, Santa Monica, Calif.: RAND
Corporation, RR-1339-AF, 2016. As of January 29, 2020:
Pickering, Jeffrey, and Emizet F. Kisangani, “Political, Economic, and Social
Consequences of Foreign Military Intervention,” Political Research Quarterly,
Vol. 59, No. 3, September 2006, pp. 363–376.
Pickering, Jeffrey, and Mark Peceny, “Forging Democracy at Gunpoint,”
International Studies Quarterly, Vol. 50, No. 3, September 2006, pp. 539–559.
Posen, Barry R., Restraint: A New Foundation for U.S. Grand Strategy, Ithaca,
N.Y.: Cornell University Press, 2014.
Powell, Colin, “Parsley Island,” The American Interest, June 14, 2012. As of
April 30, 2020:
https://www.the-american-interest.com/2012/06/14/parsley-island/
Power, Samantha, “Bystanders to Genocide,” The Atlantic, September 2001.
Presidential Review Committee, “Summary of a Presidential Review Committee
(PRC) Meeting on Central America. Issues Include Discussion on U.S. Policy
and the Political Situation in the Following Countries: Nicaragua; El Salvador;
Guatemala; Honduras,” Gale Group, June 11, 1979.
Quackenbush, Stephen L., and Jerome F. Venteicher, “Settlements, Outcomes, and
the Recurrence of Conflict,” Journal of Peace Research, Vol. 45, No. 6, November
2008, pp. 723–742.
Quinlivan, James T., “Force Requirements in Stability Operations,” Parameters:
U.S. Army War College Quarterly, Vol. 25, No. 4, Winter 1995–1996, pp. 59–69.
References 181
Reagan, Ronald, “Remarks of the President and Prime Minister Eugenia Charles
of Dominica Announcing the Deployment of United States Forces in Grenada,”
October 25, 1983. As of May 1, 2020:
https://www.reaganlibrary.gov/research/speeches/102583a
Regan, Patrick M., “Conditions of Successful Third-Party Intervention in
Intrastate Conflicts,” Journal of Conflict Resolution, Vol. 40, No. 2, June 1996,
pp. 336–359.
Regan, Patrick M., and Aysegul Aydin, “Diplomacy and Other Forms of
Intervention in Civil Wars,” Journal of Conflict Resolution, Vol. 50, No. 5, October
2006, pp. 736–756.
Regan, Patrick M., and Allan C. Stam, “In the Nick of Time: Conflict
Management, Mediation Timing, and the Duration of Interstate
Disputes,” International Studies Quarterly, Vol. 44, No. 2, June 2000,
pp. 239–260.
Rice, Susan E., National Security Council, “Updated Talking Points on Rwanda,”
memorandum to W. Anthony K. Lake et al., Washington, D.C., May 7, 1994.
Rice, Susan E., and Nick Rasmussen, “Deputies Committee Meeting on African
Peacekeeping Issues,” memorandum to Samuel R. Berger, Washington, D.C.,
September 20, 1993.
Robinson, Piers, “The CNN Effect: Can the News Media Drive Foreign
Policy?” Review of International Studies, Vol. 25, No. 2, 1999, pp. 301–309.
Rose, Gideon, “What Obama Gets Right,” Foreign Affairs, September/October
2015. As of March 5, 2020:
https://www.foreignaffairs.com/articles/2017-07-05/what-obama-gets-right
Rudd, Gordon W., Humanitarian Intervention: Assisting the Iraqi Kurds in
Operation PROVIDE COMFORT, 1991, Washington, D.C.: Department of the
Army, 2004.
Russell, Edward T., “Cuban Missile Crisis,” in A. Timothy Warnock, ed., Short of
War: Major USAF Contingency Operations, 1947–1997, Maxwell Air Force Base,
Ala.: Air University Press, 2000, pp. 33–42.
Russett, Bruce, “Economic Decline, Electoral Pressure, and the Initiation of
Interstate Conflict,” in Charles S. Gochman and Alan Ned Sabrosky, eds.,
Prisoners of War? Nation-States in the Modern Era, Lexington, Mass.: Lexington
Books, 1990, pp. 123–140.
Salmoni, Barak, Bryce Loidolt, and Madeleine Wells, Regime and Periphery
in Northern Yemen: The Huthi Phenomenon, Santa Monica, Calif.: RAND
Corporation, MG-962-DIA, 2010. As of April 30, 2020:
Sanchez, Fernando F., Democracy After War: Causes and Consequences of the 1948
Civil War in Costa Rica, paper presented at the European Consortium for Political
Research Joint Sessions of Workshops, Uppsala, Sweden: Uppsala University,
April 13–18, 2004.
Sanger, David E., “Why Not a Strike on Iran?” New York Times, January 22,
2006.
Schelling, Thomas C., The Strategy of Conflict, Cambridge, Mass.: Harvard
University Press, 1960.
———, Arms and Influence, New Haven, Conn.: Yale University Press, 1966.
Schlesinger, Stephen, “Ghosts of Guatemala’s Past,” New York Times, June 3, 2011.
Schmitt, Eric, “Obama Is Pressed to Open Military Front Against ISIS in Libya,”
New York Times, February 4, 2016.
Seybolt, Taylor B., Humanitarian Military Intervention: The Conditions for Success
and Failure, Oxford, United Kingdom: Stockholm International Peace Research
Institute and Oxford University Press, 2007.
Siegel, Adam B., The Use of Naval Forces in the Post-War Era: U.S. Navy and U.S.
Marine Corps Crisis Response Activity, 1946–1990, Alexandria, Va.: CNA, 1991.
Simpson, Bradley R., Economists with Guns: Authoritarian Development and U.S.-
Indonesian Relations, 1960–1968, Stanford, Calif.: Stanford University Press,
2008.
Stedman, Stephen John, Peacemaking in Civil War: International Mediation in
Zimbabwe, 1974–1980, Boulder, Colo.: Lynne Reinner, 1991.
Stewart, Richard, Operation Urgent Fury: The Invasion of Grenada, October 1983,
Washington, D.C.: U.S. Army Center of Military History, 2008.
Streeter, Stephen M., “Interpreting the 1954 U.S. Intervention in Guatemala:
Realist, Revisionist, and Postrevisionist Perspectives,” The History Teacher, Vol. 34,
No. 1, November 2000, pp. 61–74.
Sullivan, Kevin, and Mary Jordan, “In Central America, Reagan Remains a
Polarizing Figure,” Washington Post, June 10, 2004.
Sullivan, Patricia L., and Johannes Karreth, “The Conditional Impact of Military
Intervention on Internal Armed Conflict Outcomes,” Conflict Management and
Peace Science, Vol. 32, No. 3, 2015, pp. 269–288.
Tetlock, Philip E., and Aaron Belkin, eds., Counterfactual Thought Experiments
in World Politics: Logical, Methodological, and Psychological Perspectives, Princeton,
N.J.: Princeton University Press, 1996.
Tetlock, Philip E., and Richard Ned Lebow, “Poking Counterfactual Holes in
Covering Laws: Cognitive Styles and Historical Reasoning,” American Political
Science Review, Vol. 95, No. 4, December 2001, pp. 829–843.
References 183
Tierney, Dominic, “Syria and the Cycle of American Intervention,” The Atlantic,
October 24, 2016.
Treaster, Joseph B., “20 Years After Dominican War, Wounds Linger,” New York
Times, May 1, 1985. As of February 8, 2018:
http://www.nytimes.com/1985/05/01/world/
20-years-after-dominican-war-wounds-linger.html
Turner, Desmond A., Civil War and (Re)Building the State: Recovery in Costa Rica,
Nicaragua, and El Salvador, thesis, Coral Gables, Fla.: University of Miami, May
2017.
United Nations, “Rwanda: A Brief History of the Country,” webpage, undated a.
As of May 1, 2020:
https://www.un.org/en/preventgenocide/rwanda/historical-background.shtml
———, “UN Comtrade Database,” online database, undated b. As of April 15,
2018:
https://comtrade.un.org/data/
U.S. Agency for International Development, “The Greenbook,” webpage, updated
April 29, 2013. As of October 12, 2018:
https://www.usaid.gov/developer/greenbookapi
———, U.S. Overseas Loans and Grants: Obligations and Loan Authorizations,
July 1, 1945–September 30, 2014, Washington, D.C., 2015.
“U.S. Aid to the Contras,” Washington Post, February 3, 1988. As of May 1, 2020:
https://www.washingtonpost.com/archive/politics/1988/02/03/
us-aid-to-the-contras/842e041b-e9b3-4741-a787-6eae1f3a34da/
U.S. Department of Defense, “Combined Joint Task Force Operation Inherent
Resolve,” webpage, undated. As of May 1, 2020:
https://www.defense.gov/OIR/
———, Summary of the 2018 National Defense Strategy of the United States of
America: Sharpening the American Military’s Competitive Edge, Washington, D.C.,
2018. As of January 30, 2020:
https://dod.defense.gov/Portals/1/Documents/pubs/
2018-National-Defense-Strategy-Summary.pdf
U.S. Department of Defense, Security Cooperation Agency, “Fiscal Year Series:
Foreign Military Sales, Foreign Military Construction Sales and Other Security
Cooperation Historical Facts,” September 30, 2016. As of November 11, 2018:
http://www.dsca.mil/sites/default/files/
fiscal_year_series_-_30_september_2016.pdf
U.S. Department of State, “Milestones in the History of U.S. Foreign Relations:
Central America, 1977–1980,” webpage, undated. As of May 1, 2020:
https://history.state.gov/milestones/1977-1980/central-america-carter
———, Communist Involvement in the Colombian Riots of April 9, 1948,

Washington, D.C., Report No. 4686, October 14, 1948.
———, “Interagency Intelligence Memorandum,” Foreign Relations of the United
States, 1977-1980, Vol. XVII: Horn of Africa, Part 1, Washington, D.C., July 31,
1979, pp. 264–271.
———, “Memorandum from Paul B. Henze of the National Security Council
Staff to the President’s Assistant for National Security Affairs (Brzezinski),”
Foreign Relations of the United States, 1977-1980, Vol. XVII: Horn of Africa, Part 1,
Washington, D.C., June 3, 1980, pp. 288–291.
———, The Road to Dayton: U.S. Diplomacy and the Bosnia Peace Process, May–
December 1995, Washington, D.C., May 1997.
U.S. Government Accountability Office, Operation Desert Shield/Storm: Costs and
Funding Requirements, Washington, D.C., GAO/NSIAD-91-304, September 1991.
U.S. National Security Council, “Current Policies of the Government of
the United States of America Relating to the National Security: Volume I,
Geographical Area Policies: Part VII, Latin America - Colombia,” Gale Group,
November 1, 1952.
———, “U.S. Policy to Nicaragua,” memorandum of conversation, Washington,
D.C., September 4, 1978.
van der Maat, Eelco, “Sleeping Hegemons: Third-Party Intervention Following
Territorial Integrity Transgressions,” Journal of Peace Research, Vol. 48, No. 2,
2011, pp. 201–215.
Walt, Stephen M., “The Credibility Addiction,” Foreign Policy, January 6, 2015.
Walter, Barbara F., “The Critical Barrier to Civil War Settlement,” International
Organization, Vol. 51, No. 3, Summer 1997, pp. 335–364.
———, “Does Conflict Beget Conflict? Explaining Recurring Civil War,” Journal
of Peace Research, Vol. 41, No. 3, May 2004, pp. 371–388.
Watts, Stephen, Caroline Baxter, Molly Dunigan, and Christopher Rizzi, The
Uses and Limits of Small-Scale Military Interventions, Santa Monica, Calif.: RAND
Corporation, MG-1226-RC, 2012. As of January 30, 2020:
Watts, Steven, Patrick B. Johnston, Jennifer Kavanagh, Sean M. Zeigler, Bryan
Frederick, Trevor Johnston, Karl P. Mueller, Astrid Stuth Cevallos, Nathan
Chandler, Meagan L. Smith, Alexander Stephenson, and Julia A. Thompson,
Limited Intervention: Evaluating the Effectiveness of Limited Stabilization, Limited
Strike, and Containment Operations, Santa Monica, Calif.: RAND Corporation,
RR-2037-A, 2017. As of January 30, 2020:
References 185
Weber, Max, On the Methodology of the Social Sciences, ed. and trans. Edward A.
Shils and Henry A. Finch, Glencoe, Ill.: Free Press, 1949.
Weinberger, Caspar W., Fighting for Peace: Seven Critical Years in the Pentagon,
New York: Warner Books, 1990.
Western, Jon, “Sources of Humanitarian Intervention: Beliefs, Information, and
Advocacy in U.S. Decisions on Somalia and Bosnia,” in James M. McCormick,
ed., The Domestic Sources of American Foreign Policy: Insights and Evidence, 6th ed.,
New York: Rowman and Littlefield Publishers, 2012, pp. 399–416.
Western, Jon, and Joshua S. Goldstein, “Humanitarian Intervention Comes of
Age: Lessons from Somalia to Libya,” Foreign Affairs, Vol. 90, No. 6, November/
December 2011, pp. 48–59.
Wittman, Donald, “How a War Ends: A Rational Model Approach,” Journal of
Conflict Resolution, Vol. 23, No. 4, December 1979, pp. 743–763.
Woodward, Bob, Plan of Attack, New York: Simon and Schuster, 2004.
World Bank, “Mortality Rate, Infant (Per 1,000 Live Births), 1960–2017,” online
database, undated. As of January 30, 2020:
https://data.worldbank.org/indicator/SP.DYN.IMRT.IN
World Peace Foundation, “Colombia: La Violencia,” Fletcher School, Tufts
University, December 14, 2016. As of May 2018:
https://sites.tufts.edu/atrocityendings/2016/12/14/colombia-la-violencia-2/
Yacoubian, Mona, Critical Junctures in United States Policy Toward Syria: An
Assessment of the Counterfactuals, Washington, D.C.: Simon-Skjodt Center for the
Prevention of Genocide, August 2017.
Yao, Julio, “Legacies of the U.S. Invasion of Panama,” North American Congress
on Latin America, March 22, 2012.
Yates, Lawrence A., Power Pack: U.S.. Intervention in the Dominican Republic,
1965–1966, Fort Leavenworth, Kan.: Combat Studies Institute, U.S. Army
Command and General Staff College, 1988.
Yildirim, Jülide, and Nadir Öcal, “Military Expenditures, Economic Growth and
Spatial Spillovers,” Defence and Peace Economics, Vol. 27, No. 1, 2016, pp. 87–104.
Young, John M., When the Russians Blinked: The U.S. Maritime Response to
the Cuban Missile Crisis, Washington, D.C.: History and Museums Division,
Headquarters, U.S. Marine Corps, 1990.
Zartman, I. William, and Saadia Touval, “International Mediation: Conflict
Resolution and Power Politics,” Journal of Social Issues, Vol. 41, No. 2, 1985,
pp. 27–45.
AR R OY O C ENTER
W
hen faced with the outbreak of a war or the onset of an
international crisis that affects important U.S. interests,
U.S. policymakers must consider how best to respond.
One option they may consider is to directly intervene
militarily in the war or crisis. In this report, the authors
create a framework that can be used to rigorously consider the trade-offs
between intervening militarily early in a war or crisis, intervening later, and not
intervening at all, as well as the trade-offs involved in decisions regarding the
size of the potential intervention force to be employed. This framework can
provide a better understanding of the relationships between intervention timing,
intervention size, and intervention outcomes to inform future debates about
whether and when to undertake a military intervention.
The authors approached these issues in four ways. They conducted (1) an
extensive review of academic and policy literature on relevant topics, (2) a
quantitative analysis using a database of 286 crises and wars since 1945 in
which important U.S. interests were at stake, (3) a set of 45 short, focused
case studies, including both interventions and noninterventions, and (4) a set
of four in-depth counterfactual illustrations, in which U.S. intervention decisions
were altered from historical events to explore the implications. The findings
from these four research approaches lead to recommendations regarding the
contextual factors that are most essential for policymakers to consider when
making intervention decisions.
$37.00
ISBN-10 1-9774-0506-1
ISBN-13 978-1-9774-0506-7
53700
www.rand.org 9 781977 405067
RR-4293-A

Rand Rr4293

Uploaded by

Copyright:

Available Formats

Rand Rr4293

Uploaded by

Document Information

Original Description:

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

Rand Rr4293

Uploaded by

Copyright:

Available Formats

C O R P O R AT I O N

BRYAN FREDERICK, JENNIFER KAVANAGH, STEPHANIE PEZARD,

This report documents research and analysis conducted as part of a

S.1. The Intervention Decision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii

1.1. Intervention Outcomes of Interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

5.1. List of Selected Counterfactual Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

The research reported here was completed in November 2019, followed by

When faced with the outbreak of a war or the onset of an interna-

To this end, the objective of this report is to create a framework that

stylize these choices down to three: the commitment of small, medium,

Early, large Early, medium Early, small

Late, large Late, medium Late, small

We assessed that it would be unlikely that a single approach would

vention (if it occurred) was large, medium, or small in size. We then

The findings from these four research approaches provided insights

greatest interest to policymakers (the effects of intervention decisions

The Intervention Decision: Should the United States Intervene?

could force policymakers to make explicit trade-offs across different

The Timing Decision: Is It Better to Intervene Early?

early interventions were somewhat more likely to achieve U.S. political

all. Early interventions, from this perspective, simply reflect instances

The Size Decision: Is It Better to Intervene with Larger Forces?

produce a substantial backlash among either the local population or

large forces therefore represents a higher-risk, higher-reward strategy,

Using the key insights of our analysis, we make three recommenda-

Select Interventions Judiciously: Context Is Likely to Be Decisive

intervention might be unnecessary. However, where the United

Earlier, Larger Interventions Are Higher Risk, Higher Reward

Prioritize Collection and Analysis of Key Intelligence to Inform

Ensuring high-quality and up-to-date information on these ques-

plex intervention decisions is not possible because there is no single

DPRK Democratic People’s Republic of Korea

When faced with the outbreak of a war or the onset of an interna-

calculations in this regard to an increasing degree in the future, in ways

2011 before initiating a military intervention in the country and, even

tation and diplomatic influence; the strengthening of other regional

made to send military forces. Counterfactual comparisons are inher-

Although policymakers might decide to undertake a military interven-

sion of a middle category that is more plausible in a wider range of

Early, large Early, medium Early, small

Late, large Late, medium Late, small

shown in Figure 1.1.5 We identify three categories of outcomes as being

Conflict Outcomes Intervening State Interests Host-Nation Outcomes

Duration • Economic costs and • Economic trajectory

Intensity • Casualties • Political trajectory

Recurrence • Reputational costs and • Costs imposed by the

flict will recur. Host-nation outcomes might focus on that country’s

We assessed that it would be unlikely that a single approach would

Organization of This Report

We approach the question of how intervention decisions are associ-

takeaways from our four longer counterfactual case studies, including

What Do We Know? Reviewing Existing

As noted in Chapter One, states must make a series of decisions when

The Intervention Decision: When Do Interventions

duration of conflict if they make the conflict more difficult to resolve

operations can prevent recurrence, perhaps because such operations

Intervening State Interests

8 Michael Doyle and Nicholas Sambanis, “International Peacebuilding: A Theoretical and

democracy in the host nation13 and some research finding little or no