Paths to Peace: Domestic Coalition Shifts, War Termination and the Korean War

Paths to Peace

Domestic Coalition Shifts, War Termination and the Korean War

Elizabeth A. Stanley

Stanford University Press

Stanford, California

© 2009 by the Board of Trustees of the

Leland Stanford Junior University

All rights reserved

No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying and recording, or in any information storage or retrieval system without the prior written permission of Stanford University Press.

Printed in the United States of America

Library of Congress Cataloging-in-Publication Data

Stanley, Elizabeth A.

Paths to peace : domestic coalition shifts, war termination and the Korean war / Elizabeth A. Stanley.

p. cm.

Includes bibliographical references and index.

9780804772372

1. Korean War, 1950–1953—Peace. 2. Korean War, 1950–1953—Diplomatic history. 3. War—Termination—Decision making—Case studies. 4. Coalitions—Case studies. I. Title.

DS921.75.S73 2009

951.904’.21—dc22

2009007182

This book is printed on acid-free, archival-quality paper.

Typeset at Stanford University Press in 10/13.5 Minion

For John

Table of Contents

Title Page

Copyright Page

Dedication

Table of Figures

List of Tables

Acknowledgments

Abbreviations

PART I - The Domestic Politics of War Termination

1 - Old Baldy

2 - Domestic Coalition Shifts in War Termination

PART II - The Korean War

3 - The Korean States Powerless Players in the Forgotten War

4 - The Soviet Union Outlasting Stalin’s Preferences

5 - The United States, Part 1: Trapped by the NSC-68 Mindset

6 - The United States, Part 2: Trapped by Voluntary Prisoner Repatriation

7 - China Trapped by a Hawkish Ally

8 - Interacting Domestic Coalitions in Bargains for Peace

PART III - Testing and Extending the Argument

9 - Domestic Coalition Shifts in War Termination since 1862

10 - War Termination in Theory and Practice

Epilogue

APPENDIX A - Interstate Wars Included in the Post–World War II Dataset

APPENDIX B - Domestic Governing Coalition Shifts since World War II

Notes

Bibliography

Index

Table of Figures

FIG. 1.1

FIG. 2.1

FIG. 2.2

FIG. 2.3

List of Tables

TABLE 2.1

TABLE 8.1

TABLE 9.1

TABLE 9.2

TABLE 9.3

TABLE 9.4

TABLE 9.5

TABLE 9.6

TABLE 9.7

TABLE 9.8

Acknowledgments

Completing a book about war termination is tough business. Like the wars themselves, recent experience suggests that such books, once begun, are very difficult to end—lasting much longer and costing much more than the author expected at the outset.

This endeavor, which began as a Ph.D. thesis and has since gone through repeated changes, would not have been possible without the support of many wonderful people. My mentor at Harvard, Stephen Peter Rosen, and Iain Johnston, Theda Skocpol and Stephen Walt provided very useful early thoughts and recommendations. Financial support to conduct this research and revise the manuscript came from a graduate research fellowship from the National Science Foundation, a post-doctoral fellowship from Harvard’s John M. Olin Institute for Strategic Studies and a junior faculty research leave from Georgetown University. I am especially grateful to my colleagues and students at Georgetown for helping me see this project through to completion and to Georgetown’s Security Studies Program for providing the financial support for research assistance, from Saltanat Berdikeeva, Mindy Chan, Hailey Hoffman, Unyoung Lee, Ethan Marsh, Alexis Pierce and Michael Stanisich.

I thank the many participants of various seminars and conferences where I have presented portions of this manuscript over the years, including the American Political Science Association, the International Studies Association, Harvard’s Olin Institute and Georgetown’s Department of Government and Center for German and European Studies. I have benefited greatly from the comments these presentations elicited.

I also thank a number of friends and colleagues who read and provided insightful suggestions regarding earlier versions or portions of this manuscript, or supported me in other valuable ways: Gary Bass, Audrey Slesinger Beldio, Nora Bensahel, Beth Blaufuss, Risa Brooks, Michael Brown, Dan Byman, Victor Cha, Audrey Kurth Cronin, Colin Dueck, Robert Dujarric, Bernard Finel, Ben Fordham, Page Fortna, Bob Gallucci, Hein Goemans, Lori Gronich, Kristen Gunness, Lise Morjé Howard, Tim Hoyt, Macartan Humphreys, Dominic Johnson, Gary Kaplan, Charles King, Greg Koblentz, Mark Kramer, Alan Kuperman, Brad Lee, Phyllis Jacobson-Kram, Becky Johnson, Suzanne Neilson, Mariellen O’Hara, Olya Oliker, David Post, Jenny Lind Press, Kristen Quigley, Dani Reiter, Tim Rosenberg, George Shambaugh, Jeremy Shapiro, Sara Sievers, Jennifer Sims, Naunihal Singh, Jack Snyder, Al Stam, Monica Toft, Barbara Walter, Kathryn Weathersby and Jeremy Weinstein.

I am especially indebted to Kristin Goss, Brent Mitchell and Tammy Schultz for reading the entire draft of earlier versions of the manuscript and providing critical moral support along the way. I am humbled by the debt I owe to my colleague and friend John P. Sawyer for helping me complete the quantitative analysis in Chapter 9 and for listening to me refining my argument, over and over again. His generous and invaluable support was absolutely essential for completing this project.

I thank Geoffrey Burn, my editor at Stanford University Press, for his extraordinary efforts to help me bring this book to publication. I also thank Jessica Walsh and John Feneron for their detailed assistance.

I owe a great deal to my undergraduate mentor at Yale, the late Lt. Gen. (Ret.) William E. Odom. In many ways, he is largely responsible for my serving as a military intelligence officer in Korea in 1993—when I first developed an interest in this war—and for my becoming a political scientist after my service in the US Army. I am sad he was unable to see me finally complete this book.

Special thanks go to my parents, Deane and Cissie, and my sisters, Alison and Karalyn, for their love and encouragement throughout the process. I also thank Finley and the late Duffy for providing important lap-warmth and love during many hours at the computer. Finally, my deepest gratitude is to my partner, John Schaldach, whose love makes my life more meaningful and balanced, and whose contribution to this book is impossible to describe. This book is dedicated to him.

E.A.S

Abbreviations

PART I

The Domestic Politics of War Termination

1

Old Baldy

My insides felt like a thousand butterflies were still jumping around in there, and my legs felt like rubber. My mind was still seeing the picture at the top of Old Baldy as the shells burst among us, and hearing the sounds of Chinese bugles blowing as they came charging into our lines.... We would bleed and die for an ugly old hill that to the men in the trenches wasn’t worth throwing lives away for, just so we could see Chinese movement to the north....

—Corporal Rudolph W. Stephens, 23rd Regiment, 2nd Infantry Division (1995: 114–15)

Old Baldy is a mountain in central Korea, currently located within the demilitarized zone between North and South Korea. Old Baldy is no longer bald. Like many hills in Korea, it is covered with relatively young deciduous trees and evergreens, all grown in the last fifty years. Old Baldy earned its name during the Korean War, when intensive shelling obliterated all the trees and foliage on the hillside. At that time, Old Baldy’s landscape looked like the surface of the moon, with many old craters from artillery and mortar rounds that had fallen there (see Figure 1.1). As Jerry Ciaravino, who fought there in 1951, remembers, You knew why it was given that name. It was a big mountain of dirt with no greenery. There was just so much thrown at it that it was just dirt on the top. You were just glad to come back down after you were up there (Nisbet and Moore 2002: 5). Similarly Corporal Stephens observes, I believe the only things living on Old Baldy were men in the trenches (1995: 141).

Despite its hard-won name, Old Baldy had no strategic value. There were no natural resources buried inside it, no major communications lines it overlooked, no major industrial center anywhere nearby. In fact, Old Baldy’s only real value—a dubious one, at that—was height. At 266 meters, it was the tallest hill in the immediate area, dominating the terrain to the north, west and south, and it provided a good look-out over another infamous Korean War mountain, Pork Chop Hill. As such, the belligerent that controlled the crest of Old Baldy controlled the tactical situation in the valley below (see Map 1.1).

Except for this view down, Old Baldy was not important—not to Korea, and not to the ultimate course of the Korean War. Yet, for this view down, from July 1951 until July 1953, China and the United States fought so hard for this blasted, barren hill that, at the peak of the fighting, more than 1,500 men died there every week. During that time, the two sides also faced off in the negotiating tent, ultimately signing an agreement in 1953 that looked remarkably similar to the one proposed two years earlier.

FIG. 1.1. Unloading Logs for Bunker Construction at the 2nd Infantry Division Supply Point on Old Baldy, October 1, 1952.

Source: Photo by SFC Charles M. Roberts. Available at Wikimedia Commons, http://commons.wikimedia.org/wiki/Image:M39-AUV-Korea-19521001.jpg.

Old Baldy changed hands eleven times during the Korean War, with units on both sides suffering massive casualties—often in fierce hand-to-hand combat. As Stephens recounts, the Chinese troops fought with human-wave attack[s]. . . [they] were dying like flies as they came on. They had the troops to throw away, and they would sacrifice them to win the battle (1995: 119; also Hermes 1966: 293–95). As a result, as Ming Rose (1999) recalls, the enemy was killing so many Americans that they had to be piled up like cordwood. Photographs of them were so gruesome that they could not be shown back home. Indeed, units were sometimes reduced to one-sixth of their normal fighting strength. For example, in a particularly costly battle in March 1953, both sides expended 130,000 rounds of ammunition in one night, killing 300 Americans and 800 Chinese (Hermes 1966: 395; Allen 2002: 1).¹

MAP 1.1. Map of the Old Baldy Area.

Source: Wikimedia Commons. http://commons.wikimedia.org/wiki/Image:Old_Baldy_Area.gif

The fighting on Old Baldy was not unusual. For two years, from July 1951 until July 1953, this war of the hills was waged all around the 38th Parallel while negotiations to end the war stagnated.² These battles represented some of the most intense fighting in one of the most agonizing and miserable chapters of the Korean War and the casualties were staggering. While no complete tally of battle-deaths exists for Old Baldy, numbers for neighboring hills highlight the costs. For example, the United States and South Korea lost 6,400 men in the battles of Heartbreak Ridge and Bloody Ridge, while the Chinese suffered 10,000 killed in the battle of White Horse Hill. In between these costly battles, fighting tapered off to patrol clashes, raids and ambushes for possession of outposts in no-man’s land. Adding to the misery were the bone-numbing cold, the relentless rain, the inadequate food rations, the pounding artillery duels, the invariable sleep deprivation and the inevitable fear. Holding these hills was no walk in the park.

To put the costs into perspective, US statistics show that 45 percent of its casualties—22, 000 killed, 63,200 wounded—occurred during the war of the hills (Foot 1990: 208). The United Nations Command, of which US troops were a part, overall suffered 125,000 wounded and 60,000 killed during this period. Likewise, the Chinese and North Koreans suffered 234,000 killed (Hermes 1966: 500; Hatchimonji 2002).

Old Baldy is the quintessence of stalemated war. For two years, China and the United States suffered massive casualties, week after week, even while both governments knew that they would not win the war. Moreover, these casualties happened even though both governments knew that they wanted to end the war and knew that they should. From a strategic view, the losses make no sense—neither side thought that holding this terrain was vital to its strategic interests. This being the case, why did Chinese and American soldiers huddle under devastating exposure to 130,000 artillery rounds in a single night, as the Americans tried to push the Chinese off Old Baldy for the sixth time—long after their own governments had decided that the Korean War was bound for a stalemate?

This book is about why those men died.

While this story about Old Baldy may seem overly detailed for such an insignificant piece of land, it is instructive nonetheless. If—as most variants of the bargaining model of war assume—battle-deaths are a major driver for altering belligerents’ calculus in bargains to end war, then recounting the Old Baldy story reminds us of the human faces behind this anonymous battle-death count.

One of the biggest puzzles about war is why events like those that occurred on Old Baldy can happen. Why did those governments decide to throw away 294,000 lives fighting for strategically meaningless terrain features, as the conflict seesawed from one hill to the other? What could make this magnitude of loss acceptable? We can look at Old Baldy as a series of battles that were senseless, irrational and quite possibly inexplicable. But this book argues that what happened on Old Baldy during the Korean War was not irrational—at least in the sense that social scientists use that term—nor was it unique. What happened on Old Baldy happens in many wars, and it is explainable. In fact, there is a solid theoretical explanation for this seemingly irrational behavior. This book develops and tests that explanation.

In the case of Old Baldy, men continued to die on that hill until other men lost power, thousands of miles away, in Moscow. More generally, this book will demonstrate that the key to ending a war often comes down to changing the composition of the domestic governing coalitions in the states that are fighting it. Old Baldy may make no sense from a strategic viewpoint, but it does from the vantage point of domestic politics.

Domestic Coalition Shifts in War Termination

Throughout history, shifts in governing coalitions have critically affected war termination. For example, the execution of the Athenian democratic ruler Cleophon and the ascendancy of the pro-Spartan oligarchs in BC 404 led to Athens’ surrender to Sparta, ending the 27-year Second Peloponnesian War. Similarly, the death of Russian Empress Elizabeth in January 1762 led her Prussophile successor, Peter III, to immediately recall Russian armies who were occupying Berlin and conclude the Treaty of Saint Petersburg by May—ending the fighting between Russia and Prussia in the Seven Years’ War. During World War I, the Bolshevik Revolution in 1917 brought Vladimir Lenin and the Bolsheviks to power, who accepted the harsh terms of Brest Litovsk to make a separate peace. The next year, riots in Germany ushered in a new government that then negotiated the final war armistice, as Kaiser Wilhelm II fled to Holland. In the Chaco War, Bolivian President Salamanca was determined to continue at all costs, despite the high command’s decision to end the war. His arrest by the military and forced resignation in November 1934 was the major turning point in the war. The generals placed Vice-President José Luis Tejada Sorzano, who was known to favor peace, in power, and the war ended the next spring. During World War II, France and Italy surrendered shortly after changes in their governing coalitions, in 1940 and 1943 respectively.

This phenomenon extends to non-interstate wars, as well. A classic example is the ending of the Wars of the Roses and the installation of the Tudor Dynasty: Richard III died in the Battle of Bosworth Field and Henry VII united the warring factions to end the war. More recently, the leader of the rebel group National Union for the Total Independence of Angola (UNITA), Jonas Zavimbi, was killed in an ambush by government troops in February 2002. Less than six weeks after Zavimbi’s death, a cease-fire was signed, ending the 27-year-long Angolan civil war. Similarly, in Sierra Leone, both sides reached a settlement after Foday Sankoh, the leader of the rebel group Revolutionary United Front, was arrested. Comparable patterns can be observed in extra-systemic wars, such as Charles de Gaulle and Mikhail Gorbachev coming to power in France and the Soviet Union and ending their wars in Algeria and Afghanistan, respectively.

Scholars working on issues related to war termination have noted this phenomenon, albeit anecdotally. For example, H. E. Calahan—who argues a change in the relative power among belligerents brings peace—observes that it seems fair to conclude that a change of regime for the vanquished comes close to being a condition precedent to the making of peace (1944: 209). James Smith—who argues war ends with an international bargain—notes that it may indeed be the case that the only way to salvage something from the ruins of war is to have the leader of a particular nation resign (1995: 106). Among domestic politics scholars, Robert Rothstein concludes that because it is unlikely that the officials currently in charge can make the necessary changes in policies with which they have become identified, new personnel seem imperative (1970: 74), while Michael Handel suggests the termination of a long and stalemated war is frequently preceded by a drastic political change in leadership in the country of one of the belligerents (1978: 26).³ Bruce Bueno de Mesquita and his colleagues (1992) similarly argue many states in interstate wars experience violent regime changes before the war ends. Finally, in the landmark study of this phenomenon, Fred Iklé (1991) classifies elites in each belligerent state as hawks and doves and asserts the hawks may need to leave the government before the state can settle.

In short, the empirical record includes many examples of domestic governing coalition shifts leading to war termination, and many scholars from different theoretical perspectives have noted this tendency. However, no one has previously attempted to explain the causal mechanisms of this phenomenon in a generalizable manner or to test the argument over a broad group of cases. In this book, I introduce a new theory about shifts in domestic governing coalitions, a state’s elite foreign policy decision-making group. Moreover, I test this theory with detailed historical case studies of the Korean War and quantitative analysis of all interstate wars since 1862. While this book focuses on interstate wars as classified by Singer and Small (1994), Chapter 10 suggests ways this argument could also apply to civil wars.

DOMESTIC COALITIONS AND INTERNATIONAL BARGAINS FOR PEACE

War termination is a political bargain struck between belligerents to dispense with further combat (Bennett and Stam 1998; Filson and Werner 2002, 2004; Goemans 2000a; Pillar 1983; Powell 2002, 2004; Reiter 2003, 2009; Slantchev 2003, 2004; Smith and Stam 2004; Wagner 2000; Werner 1998, 1999; Wittman 1979). As bargaining models about war termination suggest, belligerents will settle a conflict only after they develop an overlapping bargaining space, often modeled at the point at which both sides can agree upon their relative strength. Scholars have often posited a Bayesian rational updating process by which each side updates its expectations about the war and thus changes its bottom line to create an overlapping bargaining space.⁴ Sometimes wars are short because this bargaining space develops quickly, usually because the battlefield illustrates belligerents’ relative strength in a very compelling manner. Conversely, wars are long because this bargaining space develops more slowly, for example when the battlefield becomes stalemated, belligerents cannot credibly commit, relative power is ambiguous or the issues at stake are indivisible or highly salient.

While the Bayesian model assumes the war-ending change in expectations occurs when incumbent leaders change their minds, my research suggests that in many wars, this change in expectations results from a change in the foreign policy leadership itself. Thus, my theory extends the Bayesian model by fleshing out the domestic political aspects of the bargaining model of war. I present a new theory about shifts in domestic governing coalitions and thereby refine the domestic mechanisms for explaining the international bargains that end war. As such, this analysis very consciously builds on a much wider literature within international relations about two-level games, one level being domestic and the other international (Putnam 1988; Tsebelis 1990; Snyder 1991; Mayer 1992; Fearon 1994; Schultz 1998; Smith 1998; Milner and Rosendorff 1996, 1997; Partell and Palmer 1999). It also accords with recent scholarship that examines the two-level dynamics in continuing or ending enduring interstate rivalries (Colaresi 2004, 2005; McGinnis and Williams 2001; Mor 1997; Schultz 2005). In this way, the explanation provided in this book is more complete and thus more powerful than previous attempts to generalize about and predict war termination.

Like any other social system, war termination is a given end state that can be reached through multiple causal pathways.⁵ The domestic coalition shift theory provides an alternative causal pathway to war termination beyond the standard Bayesian bargaining models—and accounts for more than a third of interstate war endings since 1862 and 41 percent since World War II. The prevalence of this alternative causal pathway suggests my theory provides greater explanatory power than other forthcoming scholarship that argues for the insignificance of domestic politics in war termination decision-making (Reiter 2009: chapter 1). For sake of comparison, Reiter’s alternative causal pathway—in which states pursue absolute victory to overcome credible commitment problems in the war termination process—accounts for only 22 percent of war endings in the post–World War II period.⁶ Thus, the domestic coalition shift theory, which includes and expands on the Bayesian model, greatly improves our understanding of war duration and war termination.

DOMESTIC COALITION SHIFTS IN OVERCOMING OBSTACLES To PEACE

The decision to go to war is a conscious policy choice. However, in order to win, leaders need to implement policies which prepare their militaries, their businessmen and their populations for the hardships and deprivations of war. They frequently need to demonize the enemy to whip up public support, mobilize their militaries and set their economies on a war footing. Enacting these policies is necessary for a state to win on the battlefield, but they make it extremely difficult to simply turn off the decision to go to war. The very policies that might make a state effective in preparing for and fighting a war are the ones that preclude an easy end to that war. Indeed, unless the resulting battlefield situation is so compelling that the outcome is obvious to all sides, leaders may find it virtually impossible to extricate their states from these sticky policies and the war itself.

This book suggests three reasons why a war, once entrenched, is difficult to end. These reasons fall into one of three categories—preference obstacles, information obstacles and entrapment obstacles.

A preference obstacle occurs when leaders do not want to end the war. This happens when leaders receive private benefits from continuing the war, or when fighting the war helps leaders stay in power.

An information obstacle occurs when leaders do not know they should end the war. In this case, the Bayesian updating process does not occur because leaders (1) receive poor quality information, (2) do not necessarily share the same information, (3) use different indicators to assess the war or (4) have personal cognitive or (5) organizational biases that prevent them from using all available information.

An entrapment obstacle occurs when leaders want to end the war but cannot. This happens when leaders face political entrapment from a domestic constituency or external allies.

These three obstacles to peace can occur in tandem, although they may not all be present at the same time or have an equally powerful effect.

The longer a war goes on, the more the dynamics of sticky policies kick in. Because of these obstacles, leaders find it more difficult to shift policies and strategies related to fighting and ending the war as quickly as the Bayesian approach would suggest. In these longer wars, belligerents become stuck and a change in expectations is needed to produce a war-terminating bargaining space. This book argues a major source of such change is a shift in belligerents’ domestic governing coalitions. In other words, policies a state enacted to wage war effectively become sticky, and the state will often require a coalition shift to get them unstuck. A coalition shift allows different political actors—with different constituencies, political interests and assessments of the war—to take power. This, in turn, may lead to a shift in the war policy, from continuing it to ending it.⁷ In other words, domestic coalition shifts can help to overcome the obstacles to peace and thus permit the war to end.

While most other empirical studies of interstate war duration and termination have focused only on wars that ended conclusively (Bennett and Stam 1996; Goemans 2000a; Reiter and Stam 2002; Reiter 2009),⁸ my model accepts the possibility of ambiguous war outcomes and theorizes explicitly about stalemates. Given the increased prevalence of stalemates in the modern era and the dearth of theory about ending these kinds of wars, my model broadens the literature’s current findings by explicitly including them. The longer the war goes on, the less likely the relative power among belligerents will be clearly illuminated on the battlefield. As a result, Bayesian updating is more difficult and belligerents are more likely to need domestic coalition shifts to overcome stickiness in expectations and achieve peace.⁹

Methods of Analysis

To test causal hypotheses about how shifts in domestic governing coalitions influence war termination, this study uses two methods of analysis: detailed qualitative case studies and quantitative analysis. Using mixed methods helps to compensate for some of the weaknesses of the respective methodologies. The detailed case studies in this book demonstrate that my theory accurately identifies the causal dynamics that determine outcomes, while the quantitative analysis demonstrates that the patterns predicted by the theory actually occur over a broad class of cases. Therefore, the two methods provide different and complementary bases for causal inference (Bennett and George 2005: 208, also chapters 2 and 10).

While quantitative analysis is useful for identifying correlations among variables to explain the duration of wars, it cannot fully address all of the causal mechanisms outlined in my argument. As Curtis Signorino (1999) and others have argued, the statistical models commonly used in international relations generally do not capture the structure of strategic interaction and interdependence very well. Because actors anticipate the decisions of other actors when they make their choices, the decisions between actors are interdependent and not independent, as many statistical estimators require. Moreover, statistical techniques can only measure actual observed events, not the unrealized potential outcomes that leaders considered but ultimately did not choose. Finally, many of my variables of interest—most notably, the three obstacles to peace—cannot be adequately proxied in a quantitative analysis.

Process-tracing in qualitative case studies is particularly well suited to avoid these pitfalls (Van Evera 1997: 64–67; Bennett and George 2005: chapter 10; King, Keohane and Verba 1994: 226–28; George and McKeown 1985: 34–41; George 1979). There are four reasons why process-tracing is particularly useful in tandem with statistical analysis. First, the process-tracing method attempts to identify the intervening causal process between the independent variables and the outcome of a given dependent variable. Process tracing requires that intervening variables be connected in particular ways, and this makes it a powerful tool for inference. In testing theories through process-tracing, all of the intervening steps in the empirical case must be as predicted by the hypotheses for the theory to pass. Indeed, process-tracing in one case can be enough to test the theory if its unique predictions are empirically validated. In other words, process tracing allows us to evaluate certain and unique predictions, which provide strong tests of a theory (Van Evera 1997: 31, 76; Bennett and George 2005: chapter 10).¹⁰

Second, process tracing is particularly important for generating and assessing evidence on causal mechanisms. Given my theory’s focus on stickiness in Bayesian updating, process-tracing is necessary for evaluating its causal mechanisms. To know whether the three obstacles to peace exist and a coalition shift is a cause of war termination, I need to (1) measure the obstacles directly, (2) assess how battlefield information (and the opposing side’s diplomatic behavior) gets interpreted by the domestic coalition, (3) show how the existing coalition’s decisions about war policy lag behind the war trends that would seem to warrant a war policy change and (4) show that coalition shifts provide the shock that brings Bayesian updating back into line with war trends. In short, process-tracing allows us to find out what decision-makers knew when, and whether and how that information was translated into a change in war policy. This minimizes the measurement error on key variables of interest.

Third, process tracing allows us to compare competing explanations simultaneously. Although two explanations may predict the same outcome, the logic of the arguments may require different paths to that outcome. In fact, process tracing encourages us to be sensitive to the possibility of equifinality—the presence of alternative causal pathways. Only case studies allow a comparison detailed enough to differentiate between two explanations and evaluate their relative importance. As Andrew Bennett and Alexander George argue, Case studies employing process tracing are particularly useful as a supplement in large-N statistical analyses, which are likely to overlook the possibility of equifinality and settle for a statement of a probabilistic finding regarding only one causal path at work (2005: 215). Because war termination is characterized by equifinality, process-tracing is indispensable for comparing alternative causal pathways in the cases.

Finally, process tracing is particularly useful for obtaining an explanation for deviant or outlier cases that are not well explained by existing theories (Bennett and George 2005: 215–16; Becker 1998: 192–96). Process tracing of deviant cases offers an opportunity to develop contingent generalizations that identify the conditions under which alternative outcomes occur. To some degree, sticky wars and stalemated battlefields are deviant cases from the perspective of existing theories of war termination: while Bayesian models accept the updating process can take time, the reasons for the lag are under-theorized. Stalemates are also understudied in the empirical research about interstate war termination, as most empirical studies have focused on wars that ended conclusively.¹¹ Therefore, process tracing within such deviant cases can help us enrich the general theory for explaining the process by which all wars end.

In sum, process tracing can complement and ameliorate some of the methodological issues with quantitative analysis. Certainly, case studies about the decision-making process require a great deal of data, which creates practical limitations on how many case studies can be completed and limits the findings’ generalizability. Luckily, this weakness is directly addressed by the quantitative analysis, which bolsters the findings’ scope.

HISTORICAL CASE STUDIES

This book examines the theory’s causal mechanisms in the three primary belligerents of the Korean War: the Soviet Union, the United States and China. I have limited the qualitative work to these primary belligerents, as they were the most important in making decisions about prosecuting and ending the war. (Chapter 3 will discuss why neither North nor South Korea qualifies as a primary belligerent.) Given the importance of the strategic interaction in understanding how an overlapping bargaining space develops in the war termination process, it was critical for me to examine belligerents in the same war. This allows me to control for other realpolitik variables and the strategic interaction among belligerents.

Beyond the justification presented in the previous paragraph, there are five reasons for this case selection. First, and most importantly, these cases are strongly associated with the independent variables of interest—the three obstacles to peace and domestic coalition shifts. Such extreme value cases can teach us the most, because the variables’ effects should stand out sharply against the background of the case more clearly (Van Evera 1997: 79–80). The variables’ effects are also unlikely to be produced by measurement error or other causes. Such cases offer particularly strong tests because the theory’s predictions are certain and unique. Each of these three cases has a different predominant obstacle to peace, which allows me to explore the dynamics of each obstacle more carefully. In addition, coalition shifts in the three cases vary in terms of shift direction (hawkish, dovish and neutral) and the reason for the shift (exogenous and endogenous shifts).¹² Moreover, because some of the belligerents experienced multiple coalition shifts, there is sufficient within-case variation. Thus, these cases provide ample evidence of and variation on the independent variables in my argument.

Second, and relatedly, because of the extreme values on the independent variables of interest, these cases provide a most likely test for the domestic coalition shift theory and a least likely test for the alternative standard bargaining models of war (King, Keohane and Verba 1994: 209–10; Eckstein 1975). Because the Korean War is a stalemated war, it meets the scope conditions for the theory, making it likely on a priori grounds that the case will support the theory. Therefore, it would be easy to discard my theory if it did not explain this case. In contrast, the Korean War is a hard case for standard bargaining models. Not only was there power symmetry between the two sides—the UN Command and the Communists—but the final armistice terms accepted in July 1953 were virtually the same as those proposed at the start of negotiations in July 1951. As the discussion of Old Baldy highlighted, why did the belligerents wait so long to make the peace given that the battlefield was stalemated throughout these two years while casualties continued to mount?

Third, the three primary belligerents provide variation on regime type, an important variable in recent scholarship about war duration and war termination (Bennett and Stam 1996, 1998; Filson and Werner 2004; Fortna 2004b; Goemans 2000a; Reiter and Stam 2002; Slantchev 2004). As Chapter 2 will explain, the domestic coalition shift theory should apply to all regime types, because all leaders answer to some constituency. However, while regime type is not important for my argument, it is a variable of interest for comparing it to alternative arguments, especially Goemans’ (2000a). Indeed, these cases are useful for testing the relative power of competing theories, because as Part II will show, the theories often make opposite predictions (Van Evera 1997: 83).

Fourth, by choosing case studies from the Korean War, I am making an empirical contribution to the literature by examining an important, yet relatively understudied war. The Korean War has been dubbed the Forgotten War for good reason. To the extent that the war receives any attention, scholars have generally focused on its dramatic first year—the war’s outbreak, the Inchon landing, the Chinese intervention and the Truman-MacArthur controversy. With the exception of Rosemary Foot’s work (1990), very little scholarship about the last two years of the war—the forgotten part of the forgotten war—exists. Moreover, very few studies have examined all of the belligerents together, taking account of the two-level dynamics that are central to my theory. For this reason, this book examines all of the major combatants in the war, even the secondary belligerents—North and South Korea. In other words, creating a satisfactory explanation of this historically significant, understudied case is in and of itself an important goal.

Finally, while the Korean War is a deviant case from the perspective of the Bayesian approach, it resembles current situations that are interesting from a policy perspective. As Stephen Van Evera argues, a theory tested in a case that resembles current policy-problem cases will more often travel to the second case, and thus, policy prescriptions deduced from the first case can more safely be applied to the second (1997: 83). Obviously, stalemated wars have policy-relevance today, because of the US’ ongoing wars in Iraq and Afghanistan. More generally, recent scholarship on war termination documents a marked increase in inter-state wars ending in stalemate since World War II (Dunnigan and Martel 1987; Fortna 2004b; Pillar 1983; Smith 1995). The Korean War also resembles current and future wars in that belligerents on both sides fought as part of international coalitions. Thus, examining this war provides traction for understanding which factors may be most relevant for explaining war termination in the future.

QUANTITATIVE ANALYSIS

While the case studies focus on the deviant case of a long stalemated war, the quantitative analysis tests the domestic coalition shift theory in all interstate wars since 1862, using two different datasets. I initially use war-level datasets—from Bennett and Stam (1996), Goemans (2000a) and the Archigos project (Goemans et al. 2009)—to establish the theory’s viability over a wider sample of wars, interstate wars from 1862 to 1990. Then, I use a new belligerent-level dataset, which I specifically constructed for this purpose, of the interstate wars from 1945 to 2006.

The Archigos dataset allows me to test the effects of changes in heads of state in all interstate wars since 1862. Given the limitations of its coding—explained in Chapter 9—the Archigos data is suggestive but not overly conclusive. While we need to move to a finer-grained measure of coalition dynamics to truly test the model’s effects—one that not only measures multiple coalition shifts within belligerent states but also measures shifts’ effects more accurately over time—the supportive results obtained from analyzing these data increase our confidence in my theory’s robustness.

Given the high degree of empirical detail necessary to code domestic governing coalition shifts, the data needed to test my argument empirically are quite extensive. Because information about domestic governing coalition composition and the timing of coalition shifts in wars after World War II is significantly more accessible, I selected this sub-sample of all interstate wars to conduct a more detailed probe of the argument. Theoretically, there should be no difference in the fundamental decision-making processes during wars before 1945, as the findings from the Archigos dataset demonstrate.

However, there are some features of the post–World War II environment that make these wars stand out as a cohesive sub-sample for a more nuanced analysis of the domestic coalition shift theory. As already noted, scholars have observed a marked increase in interstate wars ending in stalemate in the post–World War II period (Dunnigan and Martel 1987; Fortna 2004b; Pillar 1983; Smith 1995). This increase in stalemates may be driven by structural factors in the international system—the presence of nuclear weapons, the strength of the United Nations and its role in bringing interstate wars to settlement before a decisive victory and the dampening effect of bipolarity during the Cold War—as well as the global increase in democracies. Moreover, these wars occurred during the era of fast communications and round-the-clock news coverage. As a result, they are most similar in terms of strategy, doctrine and technologies to wars fought today. As the 1918 Signal Corps Manual argued, . . . The art of war has not changed with the passage of years, but it is true that the science of war has changed enormously (White 1918: 3; also Gray 2006). Thus, to examine which factors are most relevant for explaining war duration and termination in the future, it may be most useful to study recent wars, with the most theoretical and policy relevance to today.

Overview of the Book

This book is divided into three parts. The next chapter, which completes Part I, lays out the book’s theory. After reviewing the existing literature about war termination, it explains the theory and presents hypotheses to be tested in subsequent sections of the book.

Part II, divided into six chapters, traces the theory’s causal mechanisms in detail through the case studies of the Korean War. Chapter 3 briefly introduces readers to the timeline of the Korean War and explains why the Soviet Union, China and the United States were the most important belligerents in decisions about prosecuting and ending it. Chapter 3 also illustrates, with mini-case studies, North and South Korea’s insignificance in their respective side’s decisions about the war. In the process, it demonstrates how their domestic coalition shifts are unimportant for explaining the war’s ending.

The next four chapters examine the argument’s causal mechanisms in more detail in the Soviet Union, United States and China. Chapter 4 traces Soviet decision-making during the war through the lens of the preference and information obstacles. The chapter argues Soviet leader Joseph Stalin did not want to end the war because he was benefiting from it without assuming many of its costs. Indeed, while moderates opposed this war policy, they were relatively powerless to change its course as long as Joseph Stalin remained in power. When Stalin died unexpectedly in March 1953, these leaders took control of the country and quickly enacted their long-held but dormant preference to end the war.

Chapters 5 and 6 trace US decision-making during the war through the lens of the information and entrapment obstacles. These chapters explore the Truman Administration’s NSC-68 mindset, a set of assumptions about the US-Soviet relationship within which it perceived the Korean War. Fearing Soviet aggression worldwide, the Truman Administration used the Korean War to embark on a massive Cold War rearmament effort and justify its globalist agenda to Congressional and public constituencies. Likewise, believing it needed to stand firm on principles in Korea, it adopted a policy of voluntary repatriation of prisoners of war (POWs). The threat inflation that resulted from selling mobilization stoked hawkish constituencies for an all-out victory in Korea; hence, any efforts to move towards a negotiated settlement increased calls of Truman being soft on Communism. Moreover, fearing any escalation would provoke global war, the Truman Administration believed it could neither abandon the POW policy nor escalate the war to bring the Communists to terms. It was only with President Dwight Eisenhower’s inauguration in January 1953 that these obstacles were lifted and the United States could change course towards ending the war.

Chapter 7 traces Chinese decision-making in the war through the lens of the entrapment obstacle. Unlike the domestic entrapment in the US case, Chinese leaders were trapped by an external ally—Stalin. This chapter argues that as the war’s domestic costs became apparent, the Chinese government’s policy goals shifted to modernizing and creating a planned economy. A domestic governing coalition shift in the summer of 1952 led the Chinese leaders to want to end the war by compromising on POW repatriation policy. However, these leaders also needed Stalin’s support for Chinese modernization and industrialization, and Stalin was unwilling to end the war. Thus, China was trapped until Stalin’s death.

After these belligerent-level chapters, Chapter 8 brings attention to the strategic interaction between the belligerents in the war termination bargaining process. While some observers might view Stalin’s death in March 1953, and the resulting coalition shift in the Soviet Union, as solely responsible for breaking the negotiating logjam, this explanation is too simplistic. Instead, this chapter argues the timing of the war’s end depended upon the strategic interaction of all three domestic coalition shifts in the Soviet Union, the United States and China. This chapter extends the theory with two mechanisms by which domestic coalition shifts can affect the creation of an overlapping bargaining space, and it suggests these mechanisms are likely to operate in tandem when opposing sides both experience coalition shifts at about the same time. In other words, this chapter returns to the bargaining models of war and demonstrates why strategic interaction paired with analysis of domestic governing coalitions leads to the most complete explanation for why, when and how interstate war ends.

Part III tests the argument beyond the Korean War and teases out its theoretical and policy implications. Chapter 9 tests the argument with quantitative analysis, using war-level datasets of interstate wars from 1862 to 1990 and my newly constructed belligerent-level dataset of interstate wars since World War II. This chapter provides significant support for the hypotheses outlined in Chapter 2; it shows that including domestic coalition shifts in quantitative models of war duration and termination provides more explanatory leverage than existing models. This chapter also examines what effect the direction of the coalition shift (i.e. hawkish, dovish or neutral) has on war termination. The analysis in Chapter 9 enhances confidence that the patterns predicted by the theory actually hold over many cases, and it reaffirms that not all wars end by the Bayesian causal pathway that has dominated the bargaining model of war. Paired with the detailed Korean War cases, this analysis increases the generalizability of this book’s findings.

Finally, Chapter 10 summarizes these results and draws out the larger theoretical and policy implications of the argument. This chapter addresses potential ways to extend the research, such as explicitly modeling third-party intervention and extending the analysis to civil wars. Moreover, while this book examines obstacles to peace, the underlying logic is obviously much broader. War termination is only one specific manifestation of a more general phenomenon—major policy change. Indeed, the shift from continuing a war to ending it can be conceived as the most difficult policy change imaginable. Thus, the chapter concludes with general policy recommendations, based on the research findings, for leaders making decisions about prosecuting and ending war—as well as leaders approaching any major policy change.

2

Domestic Coalition Shifts in War Termination

Ending war is a tough business. Historical evidence suggests that wars, once begun, are very difficult to stop. They generally last much longer and cost much more than belligerents expect at the outset. As a result, wars frequently last well beyond the point where a rational cost-benefit calculation would dictate they end. This chapter will try to show that there is a causal link between the difficulty of ending wars and the fact that they are started and ended by politicians. Ending wars requires settling with the enemy as well as at home.

At a theoretical level, the question of war termination remains largely unanswered. While a growing literature about bargaining in war has made significant inroads in our understanding, the mechanisms that link domestic politics to the international bargains are still underspecified. Very few theories of war termination adequately encompass both levels of analysis in their explanations. In contrast, this chapter argues the longer the war, the harder it is to end it, because domestic obstacles to peace become institutionalized over time. By focusing on these obstacles, this chapter explains domestic-level impediments to the international bargains that end war.

In this chapter, I present two pathways for interstate war termination, building on Bayesian models. As Bayesian models suggest, interstate wars can only end once all belligerents develop similar expectations about the war. Bayesian models suggest that a lag in updating can delay war termination, but they do not explain how states get beyond this lag—why some states get bogged down in updating their expectations while others can update very quickly. In contrast, I look more carefully at this lag and theorize about how states get beyond it. I outline three obstacles to peace, as well as coalitional dynamics, which can lead governing coalitions to be unable to end the war—even when such a change is necessary or desirable. As a result, a governing coalition may become so immobile that ending the war is only possible by replacing the incumbent coalition entirely. Thus, while Bayesian models assume the change in expectations necessary to produce an overlapping bargaining space occurs when incumbent leaders change their minds, I argue that in many wars, such change