When the Bombs Stopped: The Legacy of War in Rural Cambodia (Princeton Studies in International History and Politics, 206) 0691255946, 9780691255941

Table of contents :
Cover
Contents
Acknowledgments
1. The Pen Family Farm: An Introduction
2. In Suspect Terrain: How War Changes Land
3. The Code of the Minefield: Risk, Race, and Hierarchy in Hvok Village
4. The Dark Halo: Lethal Protection from a Predatory Land Market
5. Beyond Cambodia: A Conclusion
Appendix
Notes
References
Index
A
B
C
D
E
F
G
H
I
J
K
L
M
N
O
P
Q
R
S
T
U
V
W
Y
Z

Citation preview

w h e n t h e b o m b s s to p p e d

p r i n c e to n s t u d i e s i n i n t e r n at i o n a l h i s to ry a n d p o l i t i c s Tanisha M. Fazal, G. John Ikenberry, William C. Wohlforth, and Keren Yarhi-Milo, Series Editors For a full list of titles in the series, go to https://press.princeton.edu/series/princeton -studies-in-international-history-and-politics When the Bombs Stopped: The Legacy of War in Rural Cambodia, Erin Lin The Insiders’ Game: How Elites Make War and Peace, Elizabeth N. Saunders A World Safe for Commerce: American Foreign Policy from the Revolution to the Rise of China, Dale C. Copeland The Geopolitics of Shaming: When Human Rights Pressure Works—and When It Backfires, Rochelle Terman Violent Victors: Why Bloodstained Parties Win Postwar Elections, Sarah Zukerman Daly An Unwritten Future: Realism and Uncertainty in World Politics, Jonathan Kirshner Undesirable Immigrants: Why Racism Persists in International Migration, Andrew S. Rosenberg Human Rights for Pragmatists: Social Power in Modern Times, Jack Snyder Seeking the Bomb: Strategies of Nuclear Proliferation, Vipin Narang The Spectre of War: International Communism and the Origins of World War II, Jonathan Haslam

When the Bombs Stopped t h e l e g ac y o f wa r i n ru r a l c a m b o d i a

erin lin

princeton univer sit y pre ss princeton & oxford

c 2024 by Princeton University Press Copyright  Princeton University Press is committed to the protection of copyright and the intellectual property our authors entrust to us. Copyright promotes the progress and integrity of knowledge. Thank you for supporting free speech and the global exchange of ideas by purchasing an authorized edition of this book. If you wish to reproduce or distribute any part of it in any form, please obtain permission. Requests for permission to reproduce material from this work should be sent to [email protected] Published by Princeton University Press 41 William Street, Princeton, New Jersey 08540 99 Banbury Road, Oxford OX2 6JX press.princeton.edu All Rights Reserved Library of Congress Cataloging-in-Publication Data Names: Lin, Erin, 1985– author. Title: When the bombs stopped : The legacy of war in rural Cambodia / Erin Lin. Other titles: Cambodian life in the aftermath of the American bombing Description: Princeton : Princeton University Press, [2024]. | Series: Princeton studies in international history and politics | Includes bibliographical references and index. Identifiers: LCCN 2023034340 (print) | LCCN 2023034341 (ebook) | ISBN 9780691255941 (hardback) | ISBN 9780691255958 (pbk.) | ISBN 9780691256122 (ebook) Subjects: LCSH: Unexploded ordnance—Social aspects—Cambodia. | Unexploded ordnance—Political aspects—Cambodia. | Vietnam War, 1961–1975—Influence. | Vietnam War, 1961–1975—Cambodia. Classification: LCC DS554.8 .L566 2024 (print) | LCC DS554.8 (ebook) | DDC 959.704/33596—dc23/eng/20231214 LC record available at https://lccn.loc.gov/2023034340 LC ebook record available at https://lccn.loc.gov/2023034341 British Library Cataloging-in-Publication Data is available Editorial: Bridget Flannery-McCoy and Alena Chekanov Production Editorial: Natalie Baan Cover Design: Heather Hansen Production: Lauren Reese Publicity: William Pagdatoon Copyeditor: Leah Caldwell Jacket image: Boudewijn Huysmans / Unsplash This book has been composed in Arno Pro Printed in the United States of America 10 9 8 7 6 5 4 3 2 1

contents

Acknowledgments vii 1 The Pen Family Farm: An Introduction

1

2 In Suspect Terrain: How War Changes Land

32

3 The Code of the Minefield: Risk, Race, and Hierarchy in Hvok Village

86

4 The Dark Halo: Lethal Protection from a Predatory Land Market

120

5 Beyond Cambodia: A Conclusion

163

Appendix 191 Notes 199 References 217 Index 233

v

acknowledgments

this book tells a story of rural Cambodia in the past half century. But it got its start in the Politics Department at Princeton University, the home of a rich and various set of ideas, beliefs, skills, and opinions. Although I was one of the softer-spoken students, my advisors gave generous attention to my work. I can only hope that they are all pleased with the final result. Amaney Jamal focused relentlessly on the positive and found solutions where I thought there were none. Mark Beissinger was an honest, available, and always-supportive mentor, and his recommendations of my work to former students and editors advanced my career. Deborah Yashar was the first to read the eight-page memo that became the starting point of this book. She saw the possibilities, and I thank her for the nudge. I’m additionally grateful for the guidance of faculty members Leonard Wantchekon, Kris Ramsay, David Leheny, Jake Shapiro, Lynn White, and Wangyal Shawa. At Princeton, I was part of an extraordinary cohort of graduate students. I lost direction many times, and once I had to be rescued from a bad advising situation, but I did get through, in large part because of my friends: Scott Abramson, Omar Bashir, Alex Bolton, Elise Bonner, Ryan Brutger, Sarah ElKazaz, Yanilda González, Christopher Gupta, Aram Hur, Dayna Judge, Asya Magazinnik, Tom Pavone, Bryn Rosenfeld, vii

viii a c k n o w l e d g m e n t s

Sharece Thrower, and Yang-Yang Zhou. My college mentors, Keith Darden and Rob Person, knew that everything was going to be okay, and I’m thankful they had my back. After graduate school, I moved to Columbus to begin a tenure-track job at Ohio State. The Ohio State political science faculty is composed of generous readers, fiercely loyal to the junior faculty. I owe a debt of gratitude to my mentors: Amanda Robinson, Sarah Brooks, Sara Watson, Marcus Kurtz, Alex Thompson, Chris Gelpi, Jan Pierskalla, and Philipp Rehm. The department chairs, Rick Herrmann and Greg Caldeira, looked out for me, ensuring that I had the time and space to be successful. I also received support from colleagues Jennifer Mitzen, Molly Ritchie, Inés Valdez, Nicole Yadon, Ben McKean, William Minozzi, and Michael Neblo; like my mentors, they steadfastly came to my talks and helped me improve my drafts. Outside of the department, Casey Hoy, Nick Kawa, and Kara Young fed my interdisciplinary interest in agriculture. Outside of the university, an outstanding group of researchers who worked in Cambodia helped facilitate my visits. My sincere thanks to Alison Carter, Greg Crowther, Tapley Jordanwood, Ted Paterson, Alberto Pérez Pereiro, and Frank Smith. My students Chase Harpole, Kit Wislocki, Genna Brizendine, Kyle Claassen, Connor Kolet, Lyrin McKibben, and Alayna Press were also an incredible help. Travel and time off from teaching were underwritten by generous support from the Mershon Center for International Security Studies, the Princeton Institute for International and Regional Studies, the Initiative for Food and Agricultural Transformation, and the Air Force Office of Scientific Research’s Young Investigator Program (FA9550-22-0187). Any opinions, findings, conclusions, and/or recommendations expressed in this material are those

a c k n o w l e d g m e n t s ix

of the author and do not necessarily reflect the views of the United States Air Force. In Cambodia, I spoke with many people who shared their valuable time with me; all of them allowed me to document their stories, and many of them became friends. I respectfully acknowledge these farmers, shopkeepers, deminers, educators, laborers, and landowners and my book’s debt to them. Many invited me into their homes, introduced me to their families, and taught me so many things about the land and way of life. Fieldwork is often lonely, but with them I felt the fullness of our days and the joy of being together. I wanted to write a book that paid homage to them, the survivors who grapple with the moral questions raised by American foreign policy. I realized in Cambodia that, for me, the point of political science has always been meeting new people and new worlds. Upon my return to Ohio, I decided to write a book inspired by the 1970–1980s literature by James Scott, a political scientist who listened to and learned from people in rural areas. He embraced the idea of researching and paying attention to people’s everyday experiences. My scholarly heroes, Tom Pepinsky and Dan Slater, provided advice during my pandemic book conference—and even more valuable, ideas, arguments, and the occasional joke. The three anonymous reviewers paid significant attention to the manuscript, as did my colleagues Philipp Rehm and Greg Caldeira, and I am grateful to each one of them. Their comments brought the book to a new level. Heath Sledge polished the prose to a shine. At Princeton University Press, Alena Chekanov and Bridget Flannery-McCoy enthusiastically embraced the manuscript and (almost incredibly!) were kind enough to expedite the review process so I could meet my tenure deadline.

x acknowledgments

Travel occupies a large portion of my later work on this research project. I often talk to political science classes at different universities, but occasionally I speak to high-school students, retirees, veterans, and development practitioners. I love these conversations, but I tend to log only whatever mileage my job requires. As I’ve gotten older, I find it more and more difficult to be away from home. I have friends of five years who live within a few minutes of me. They leave flowers, breads, and dog treats on my front stoop. It’s an incredibly supportive village, one that helped me recover from injury and heartache, and I’m lucky to have found such good people. I dedicate this book to my family. I was a reserved child, unusually quiet. But I always believed in myself—the way I think a person does when they are deeply loved by their family. My father, Jack, who is easygoing and kindhearted, spent a lot of his time going on adventures with me, whether to Dinosaur Valley State Park or the dollar movie theater. My mother, Yen, worked full-time as a hospital pharmacist, but was interested in creative things, a devotee of music and art. She is retired now and often stays up late to work on various projects—painting, sewing, cooking—or to learn about them online. With time, I internalized the devotion to craft that characterized her approach to art. As a child, I fought constantly with my sister Ines. We were nine years apart in age, and she bossed me around like a third parent. Now I feel unusually close to her. Since I graduated from college, she has housed, fed, and advised me at regular intervals—a textbook exemplar of sisterly love and unfailing generosity. I cherish her family, too: her husband Sanjeev, her mother-in-law Vanlila, and my wonderful nieces Sydney and Elinor. I’ve watched my nieces grow into young adults who

a c k n o w l e d g m e n t s xi

observe the world closely and feel things deeply. I look to them for inspiration any time my attention wanders while in pursuit of a solid, knotty answer. May 2023 EL Columbus, Ohio

1 The Pen Family Farm an introduction

the pen family farmhouse—wooden, tall, and aged—sits in the middle of a former battlefield in Lumphat District, Ratanakiri Province, Cambodia. The original thatch roof has been replaced with aluminum, but the house has retained much of its traditional Khmer architecture, like the hardwood frame, gabled roof, and open-air ground floor. The bedroom and living room on the second floor sit upon ten-foot pylons, allowing air to circulate during the hot season and lifting the house above floodwater during the wet season. Acres of agricultural land—some of the most fertile in the country—are visible from the kitchen and the platform table on the ground floor. When the first generation of Pens settled here sometime in the 1940s, this land was a forest basin, filled with nutrient-rich, loose, loamy soil. Over the course of twenty years, Thida and Rithy Pen cleared the majority of the trees on their ten hectares, using the wood to build the house that stands here today.1 Looking around their property from the ground floor, our view is broken by an occasional thirty-meter-tall tree, a remnant of 1

2 chapter 1

the dwarf rainforest that was here when the Pen family arrived over eighty years ago. The trees have long, smooth trunks topped by bushy branches spread against the sky, like giant broccoli crowns. They are called cheuh tiel tom in Khmer. You can cut a notch in the tree, burn the hole to stimulate sap flow, and place a bucket below to collect the sap. The liquid serves many purposes: a wood sealant, a soap ingredient, and a fire starter. For these reasons, English-speaking arborists call them “resin trees.” Thida and Rithy Pen are now retired, but they have passed their farm on to their children, including their son Leang and daughter-in-law Lom. Prematurely gray-haired, Lom takes us to their garden to see the soil, calling it by its colloquial name, dei krahom (red soil).2 The technical Cambodian name is labansiek soil, according to scientists at the International Rice Research Institute. It has a burnt, reddish-brown color. The soils formed on the sides of hills as the lava flow of ancient volcanoes created balsatic rock that has been weathered down over millennia. When you touch the grains of soil, it feels both cushioned and dense, like an expensive foam mattress. It has a clay-like texture, but also falls apart like sand when it’s dry. The soil’s microstructure is well suited for holding water, to the point that it does not need to be regularly irrigated for commercial production. Horticulturalists estimate that without fertilizer, this soil should be able to produce 1,400 to 1,600 kilograms of rice per hectare (Nesbitt 1997). Lom and Leang are experienced farmers—she can identify wild boars by their tracks and he has built a walking tractor from spare parts—and the fruit trees and vegetable garden that surround the house are thriving.3 Leang has grown up on this land, and Lom raised chickens, dogs, cows, cassava, and cashews in Prey Veng Province, Cambodia, where she was

t h e p e n fa m i ly fa r m 3

born. Together, they grow fruits and vegetables twelve months a year in the garden beds (jomkaa). Hollow-stem spinach, amaranth, wax gourd, mungbean, kabocha squash, eggplant, cucumber, winged beans, tamarind mangoes, apple bananas, plantain bananas, and Madagascar plums are among the produce that the family grows here. Despite their years of experience and hard work, Lom and Leang are more or less failing at small-scale commercial production. The couple own about three hectares (the rest of Thida and Rithy’s land is owned by Leang’s siblings), but Lom and Leang have scarcely a hectare under cultivation now, and their land is thus an odd patchwork—largely wild and overgrown but punctuated by irregularly placed and wellmaintained fields. In the cultivated areas, Lom is usually able to produce only a half-kilogram of rice per year, and even that depends on luck. (Once, she lost her entire crop in one night when a wild boar ate everything.) Why are these smart and hardworking farmers in one of the country’s most fertile areas tallying so many losses, unable to clear their share of agricultural profits? To answer this question, Lom and Leang tell me about two intertwined local events, which loom large in the family’s memory: the 1970 siting of a military base within a few kilometers from the Pen family farmhouse, and the subsequent repeated bombings of the area. The base was built by General Lon Nol, who took over the country from 1971 to 1975, seizing power from King Norodom Sihanouk on the encouragement of the US State Department. During Lon Nol’s rule, he sent the Khmer National Armed Forces, a branch of the state military, to Ratanakiri Province, where they fought Pol Pot’s guerrilla army of revolutionaries and their Vietnamese communist counterparts. Lon Nol sited the army’s Ratanakiri base on Elephant

4 chapter 1

Hill, only a few kilometers from the Pen farmhouse; it included a hospital tent, soldier barracks, and even hiding places underneath the treeline. Shortly after Lon Nol took power, the US renewed its aerial attacks on Cambodia as part of an effort to break up communist supply lines that ran from North Vietnam through Laos and Cambodia to South Vietnam. From 1970 to 1973, the US dropped bombs on Elephant Hill and throughout much of the basin where the Pens’ farm is located. As they tell me this story, Leang’s and Lom’s body language and voices betray just how immediate these fifty-year-old events still are: “There’s a bomb crater as big as our house,” Leang shouts from the house, pointing in the direction of Elephant Hill.4 Lom wants to take us to see it, but she is worried: we mustn’t stray off the path, she cautions. She is dressed casually in a bright floral shirt, loose trousers, and rubber flip flops, and weather-creases line the bottom of her eyes. “Even I still get scared. . . she admits and tilts her head to gauge my comprehension. Her land is covered with unexploded bombs, some hidden inches or even feet below the surface, and it is impossible for newcomers to detect what is lurking inches below without specialized equipment and training. I promise to be careful and follow her lead, and we make our way west on a raised clay footpath. We walk through a strangely mottled landscape. In the area close to the house, we are in brilliant sunshine, wading through emerald rice stalks in a paddy full of water and mud, working our way toward a large fallow pasture of brush and fallen trees. Only a little farther on, though, we come to a big, gravelly mound of alluvial deposits that looks like clumps of clay and steel-cut oats. Blackened tree branches stick up through the mound, encrusted with metal shards and ash.

t h e p e n fa m i ly fa r m 5

We are now at Lom’s detonation point, where she explodes or stores the stray munitions that she finds. (Professional deminers do not want to move the likely unstable bombs and usually detonate them onsite; Lom moves the bombs to a central site for detonation because she does not wish to further damage potentially arable land.) She shows me a cluster bomb that she found while tilling her soil just last week. Half of the bomb is encased in dirt, and the exposed side reveals a rusty metal casing, stone-gray with reddish-brown tints. Until I look at it closely, it looks almost identical to the rocks in the soil, except the bomb is perfectly spherical—smooth, exact manufactured curves. The Pens’ experience belies neoclassical growth models, which assert that a land’s history of political violence should play little, if any, role in contemporary production. In the past three decades, neoclassical growth theory became one of the dominant paradigms of development economics, and a bridge to other fields, notably political science, history, and anthropology. Scholars like Robert Barro and Xavier Sala-iMartin (1992) initially used these models to explore why poor US states and European countries eventually converge toward rich ones. Other economists, like Chris Blattman and Edward Miguel (2010), rooted their application of the theory less in the advanced industrial world and more in the developing world, especially countries recovering from war. The draw of this framework, in their view, is its ability to clarify how war damage to particular factors of production (i.e., physical capital, labor, human capital), technology, institutions, or prices would impact economic performance (p.38). They argue that periods of depletion are followed by periods of transitory growth until a steady state is attained—until population size and capital

6 chapter 1

stocks (inevitably) bounce back. According to these theories, postwar surges in the birthrate will offset wartime casualties and repopulate neighborhoods demolished by violence; similarly, they assert that as long as a country’s capital stock is only partially destroyed and its production function remains the same (defined here as its ability to produce maximum output, given a certain quantity of inputs and level of technology), then the economy will experience only a temporary bump in capital accumulation, for production will quickly return to prewar levels. We encounter some evidence confirming these theories in seminal pieces on interstate war. Cross-national statistical analyses from political science (Przeworski 2000) and economics (Organski and Kugler 1977) find that national economies tend to quickly converge back to steady-state growth, at least in terms of macroeconomic indicators like per capita income.5 The model’s predictions are also borne out in examinations of industrialized cases like Japan and Germany (Davis and Weinstein 2002; Brakman, Garretsen, and Schramm 2004) and semiindustrialized economies like Vietnam (Miguel and Roland 2011). But what about Cambodia? When I went to Cambodia in 2018, I wanted to see how these theories of postwar recovery were shaking out in an agrarian context, where farming was the most common occupation and local economies were dependent on rice production. I had entered Ratanakiri Province in a rented van the week before, accompanied by two American students, one who came to collect soils and another who came to practice interviewing, and two Khmer interpreters.6 We had spent the first days getting acclimated to Banlung, the provincial capital, where we walked in forests that once hid the Ho Chi Minh Trail; we met former Khmer Rouge supporters and soldiers, as well as

t h e p e n fa m i ly fa r m 7

those who fought against them and those who were imprisoned by them. Alongside these encounters with internationally known people and places, that morning at the Pens’ farm in Lumphat District may seem mundane. But it retains a certain burnish, because it exemplified both how vulnerable farmers are to unexploded bombs as they move from one of their fields to another and also the bombs’ outsized effect: as I found, an unknown number of hidden bombs create a “sphere of fear” (Perlman 2006) that dampens economic activity throughout an entire settlement. Over the course of the summer, I would crisscross the province many times, revisiting people and places in order to describe not only the undetonated and leftover weapons that remain behind in conflict zones, but also the hangover effects of those weapons on the civilian population. This was among the most dangerous of a series of journeys that I had made to rural Cambodia over the past twelve years. I made this particular journey with professional deminers, who taught me how to identify and locate bombs and introduced me to village chiefs, former soldiers, indigenous minorities (Tumpuon, Jarai, Krung, Lao), and even a Khmer Rouge commander— people who would have been difficult to access otherwise. By the end of the summer, my respondents taught me a central irony of life in postconflict Cambodia. American airstrikes were designed to avoid civilian settlements, but the remnants of these rural bombings have forced civilian farmers to regularly do one of the most dangerous and strenuous tasks that the military can assign to its professional soldiers: the identification and clearance of unexploded ordnance. Our failures in military technology—the bombs’ failures to explode on impact— have drained Cambodian fields of their safety and economic potential.

8 chapter 1

I had originally gone to Cambodia to study political folklore, but as I wended my way among the bombs, I wondered if the economic studies that have downplayed war’s lasting effects might have missed something. Top-line numbers, like gross domestic product and export ratios, could be misleading. They may not capture inequality within a country or the relevant aspects of poverty that lead to long-term growth traps, like limited access to land and slow adaptation of technology. These powerful tools for quantitative analysis may abstract the messy, complicated human reality. The Pen family’s story seemed to underscore how, by listening closely and becoming sensitive to context, an ethnographic approach can redefine the range of economic outcomes we should study: for instance, agricultural productivity variables (which are central to economic growth models, particularly in the developing world) and variables that are related to quality of life and human health, such as food insecurity and occupational safety. My ethnographic approach— observing and interacting with my respondents first—deeply informed my subsequent mass-sample surveys, which are still enormously helpful for capturing how a large population interacts with the market at a given point in time. My respondents defined what was appropriate for the study, and my econometric practices evolved to use various measures of welfare, security, and happiness. Long-term, immersive fieldwork allowed me to tell the rich and complex stories of how people navigate their postconflict environments to provide for themselves and their families from day to day. I began my fieldwork as a college senior in 2006. That year, I invested in daily Khmer lessons so I could speak with people who lived on floating houses atop Boeung Kak, a lake in the center of Phnom Penh.7 Five years later, as a graduate student, I lived in the countryside for four months, working in

t h e p e n fa m i ly fa r m 9

rice paddies and tobacco fields. I learned to coat rice with talcum powder prior to planting; talcum is a common biofertilizer that increases shoot growth and protects from fungal disease (see Tamreihao et al. 2016). I harvested, housed, and stripped tobacco remains. I harvested with the men, who worked early in the morning before peak heat and bundled the leaves with twine to bring back to the hamlet. In the afternoon, I sat with the women and children around piles of leaves, threading leaves on wooden skewers so they can hang to cure in dry houses. My hands were soon stained by black tar (a pair of disposable gloves cost five hundred riel, and a completed skewer gets you only one hundred riel). While we worked, people told stories of their routine encounters with leftover metals and occasionally munitions. The stories around the tobacco pile illustrated how, although neoclassical growth models may capture economic productivity in aggregate, they also miss critical elements of peoples’ lived experience of farming—particularly negative aspects such as risk exposure, physical injury, and social harm, as well as the historical factors that produced these dangerous environmental conditions. Over the course of the Vietnam War, the United States dropped 500,000 tons of bombs over Cambodia—more than the combined weight of every man, woman, and child in the country. What started as a secret infiltration of Laos, in which a few CIA officers would train and arm local Hmong villagers to fight the communist forces, eventually enveloped Cambodia and escalated into a nine-year war over the Ho Chi Minh Trail—a war fought primarily with bombs. It is now fifty years after the last sortie went out, and my respondents’ stories show how the half-century legacy of unexploded ordnance has sedimented the war into the layers of contemporary society. Forward-looking policies aimed at developing or modernizing

10 c h a p t e r 1

Cambodia, from economic liberalization to authoritarian consolidation, must be realized in an environment that is still haunted by the violence of the past. Drawing on original interviews, a wealth of new historical data, and extensive fieldwork at the Cambodia-Vietnam border, I show how the unintended failure of military technology creates a dark halo around postconflict land. I also show how, paradoxically, these same failures can protect the farmers even as they endanger them: the very bombs that endanger and impoverish farming communities deter predatory elites from grabbing and commodifying their land. As the stories captured in this book show, the bombing served as a critical juncture in these postconflict villages— marking the place in time where development stopped.

Why We Bomb What I learned in the field about aerial weapons undercut much of what I was taught about how political scientists and military historians currently frame their study of technology and war. For decades, the conventional view among international relations experts was that more technological innovation yielded more power and plenty. Historian Geoffrey Parker (1988), in his acclaimed work The Military Revolution, attributes the rise of European empire in the nineteenth century to the Europeans’ willingness to adopt new technologies and improve them for military use. Political scientist Kenneth Waltz (1979) argues that all great powers must exploit technological change; otherwise, they would lag behind other states and face threats to their survival. In the past several years, however, research has begun to suggest otherwise. Scholars of international security like Michael Horowitz and Stephen Biddle point out that military forces

t h e p e n f a m i l y f a r m 11

with better technology and capabilities can still lose wars (for example, the Vietnam War). Contemporary scholarship in their subfield, which is largely concerned with issues of national security, argues that it is in a great power’s best interest to invest in superior weapons, but these states do not always end up acquiring or using them. To explain this inconsistency, international security scholars have pointed to the influence of domestic politics. As Michael Horowitz (2010) notes, some militaries are better at adopting new technologies than others due to their financial capacity and organizational structure. According to Stephen Biddle (2010), military leadership plays an important role here, for leadership’s “force employment decisions”—how it chooses to organize and deploy its resources—intervene between theory and practice. Other scholars, such as Dan Reiter and Allan Stam (2010), instead highlight the regime type as the deciding factor, arguing that the foreign policy options of countries operating under democratic leadership are constrained by domestic audience costs. This new line of scholarship in international security seeks to unpack the concept of hard power to show that it is more than a budgetary byline or a count of infantry or aircraft carriers—that it is a product of domestic politics. This new tradition has helped us understand how political institutions, bureaucratic rules, and leadership ideology can intervene to impede technology acquisition and usage. But in its focus on battlefield outcomes and balance of power, it has left in the background the connections between superior military technologies and the people on whom they are used. In the past century, weapons have grown in size and sophistication and increased their deadliness and destructive range to people on the ground. One prime example is aerial weaponry, which, according to advocates, offers a more efficient and rapid

12 c h a p t e r 1

projection of force than land or naval alternatives (Schelling 2008; Pape 2014). The world’s first payload drop occurred in 1911, when Lieutenant Giulio Gavotti released four grenades, by hand, over the side of his plane during the Italian-Turkish War (De Groot 2005). A half-century later, a formation of six B-52s, dropping their bombs from thirty thousand feet, could obliterate almost everything within a box about five-eighths of a mile wide by two miles long, destroying military infrastructure faster than any battalion (Sheehan 1998). The five-hundredpound MK-82 warhead, used in the US carpet bombing of Vietnam, Cambodia, and Laos (and more recently in the 2016 Saudi bombing of Yemen), is such an effective weapon of destruction that it annihilates its victims beyond recognition. After interviewing Yemeni witnesses, New York Times journalist Jeffrey E. Stern describes the visceral experience in his 2018 article: The shell fractures into several thousand pieces, becoming a jigsaw puzzle of steel shards flying through the air at up to eight times the speed of sound. Steel moving that fast doesn’t just kill people; it rearranges them. It removes appendages from torsos; it disassembles bodies and redistributes their parts. A sphere of expanding gas coming off the bomb, meanwhile, fills a body’s hollow parts with energy, rupturing eardrums, collapsing lungs, perforating abdominal cavities and making hidden things bleed. The blast wave pushes air to such extraordinary speeds that the wind alone can cast limbs off bodies. Warheads are capable of destroying so much from a safe distance. This was the guiding insight of Robert Pape. Pape (2014) and his predecessor Thomas Schelling (2008) believed that superior military technology, like airplanes, surface-to-air

t h e p e n f a m i l y f a r m 13

missiles, and explosive warheads, would improve combat in two ways: one, faced with the prospect of devastatingly destructive airstrikes, targeted societies are more likely to fold, helping great powers win wars more decisively. Two, modern weapons would shield the flesh-and-blood soldier from the dangers associated with the battlefield. By finding a way to fight without putting soldiers at risk, coercive air power became an increasingly attractive policy tool—one with little cost in terms of domestic public opinion.8 This sentiment is echoed in Jason Lyall and Isaiah Wilson’s 2009 article, in which they describe the mechanization of militaries as largely beneficial, “increas[ing] mobility and survivability on contemporary battlefields” (p.68). Their comment underscores the implicit bias in international security: survivability is measured only in terms of our own soldiers and not carried over to adversarial populations. The problem with this approach lies in its distorted perspective, its definition of cost as maximizing the odds of the great power winning while minimizing its casualties. Clearly, the distribution of casualties and the risks are extremely unequal. Aerial weapons pose enormous costs to the people who are being bombed. Attempts to make war less brutal for the American soldier have increased its brutality for those targeted. In 1971, Vietnam War journalist Neil Sheehan characterized aerial weapons not as a military necessity but as “a political convenience, a substitute for sufficient infantrymen to hold the countryside.”9 One could argue that ground troops make missions more merciful, since an infantryman can make moral determinations of life and death with more discrimination than a bomb,10 but bombs dropped from B-52 jets flying fifty thousand feet in the air cannot differentiate between a child and a combatant. Drone operators still have this problem today.

14 c h a p t e r 1

In 2019, an American F-15 attack jet dropped two warheads on a large crowd of women and children in the Syrian town of Baghuz, thinking they were Islamic State fighters; over seventy people were killed (Philipps and Schmitt 2021). If civilians often bear the brunt of wartime violence, the question of higher responsibility hangs over the heads of state and military, as well as the domestic audiences who hold them accountable. These realities stand in contrast to Pape’s argument that superior military technology would improve combat, an argument grounded in his belief that airstrikes were less violent than alternative forms of combat: as the final sentence of his book puts it, “This approach will not only be more effective but also harm fewer civilians” (2014: p.331). According to Pape’s argument, denial bombing is more effective because it induces quicker concessions: by overflying enemy land armies and inflicting direct harm on key infrastructure, airstrikes deter opposing forces from engaging in full-scale land combat. Pape was not entirely wrong—improvements in bomb accuracy have made it easier to focus on military targets. But precision technology is not the cause of all advancements in humane war tactics in the past half century. Ever since the Vietnam War, public opinion has been more sensitive to casualties on all sides and more involved in assessing the proportionateness of response (Mueller 1994; Gelpi, Feaver, and Reifler 2009). There is now a worldwide revulsion toward firebombing, napalm, nuclear devastation, and casualties on all sides. These shifts in international norms—both in terms of public opinion and in terms of global legislation around human rights—have also influenced how military campaigns are conducted. In 1977, the Geneva Conventions were updated to ban “indiscriminate attacks against civilians, the targeting of civilian infrastructure, harm to civilians that was

t h e p e n f a m i l y f a r m 15

disproportionate to the military objective” (Filkins 2021).11 Now, when US commanders are considering a military action, defense lawyers must weigh in on whether it comports with international treaties like the Geneva Conventions that protect human rights. It forces US military leaders to justify their targets, choose them carefully, and spare lives whenever possible. If the military does not hold itself to these standards, it faces possible sanctions from Geneva Convention signatories (though this has not yet happened against the US) and condemnation in the court of public opinion, which sometimes produces real material backlash. For example, after the New York Times published investigative reports of the covered-up civilian casualties from US air operations in Afghanistan, Iraq, and Syria (Khan 2021a,b; Philipps and Schmitt 2021; Savage et al. 2022), Congress imposed restrictions on military funds until the Pentagon submitted a new civilian casualty policy.12 Defense Secretary Lloyd Austin III subsequently ordered “a standardized reporting process on civilian harm, the creation of a new military ‘center of excellence’ and the completion of a comprehensive new policy on the issue that has been in the works for nearly two years” (Schmitt, Savage, and Khan 2022). In the field of international relations, most scholars studying the role of airpower in international conflict focus heavily on its consequences on US foreign policy, warfare, and the distribution of power in the international system. They often fail to examine the equally important effects on the other half of the conflict, the ones who suffer the bombings. In one sense, humanitarian laws and modern precision weapons can be seen to save what lives can be saved in the midst of great killing. In another sense, it now seems easier than ever for great powers (and those studying them) to have a sense of deniability of the consequences. More advanced military technology now

16 c h a p t e r 1

requires more layers of mediation—analysts, satellites, planes, drones, technicians, and bases—between those inflicting damage and their targets. Between the large number of parts in the military apparatus and the growing physical distance between the target and the ones dropping the payloads, it is easy for the possessors of advanced military technology to ignore the real human costs of military action.

Consequences of Bombing Scholars of comparative politics and development economics have begun to document this overwhelming form of firepower’s extensive political and economic impacts on society. First, they note, the immediate loss in human life and property from indiscriminate aerial bombardment has increased the resolve of insurgents and swayed civilians to their side (see Kocher, Pepinsky, and Kalyvas 2011, and Dell and Querubin 2017, on Vietnam; Lyall 2009 provides contradictory evidence in Chechnya). The political antipathy toward those doing the bombing can be transmitted to the next generation, as Laia Balcells’s description (2012) of the Spanish Civil War shows: adult children share their parents’ identity as victims and seek to punish the political party associated with the bombing, a finding that likely generalizes to lasting political antipathy against other nations involved in bombing campaigns. However, according to economists, these traumatic experiences that shape political attitudes in a lasting way do not seem to have long-lasting impacts on economic development; both Hiroshima and Nagasaki experienced a population boom soon after World War II that reinvigorated the depleted workforce, and their economies quickly rebounded (Davis and Weinstein

t h e p e n f a m i l y f a r m 17

2002). Inspired by the findings on Japan, Steven Brakman, Harry Garretsen, and Marc Schramm (2004) applied a similar econometric approach to analyze the impact of bombing on postwar Germany, reaching similar conclusions: bombing leaves only a temporary impact on urban growth. Why? Perhaps the answer lies in older bombing technology’s lack of precision: according to military historian Tami Biddle (2009), air theorists have overstated the precision of long-range bombers during the two World Wars, for American and British assessment reports indicated that bombs failed to hit many of their industrial targets.13 Tony Judt’s history (2006) of Europe after 1945 demonstrates that the industrial damage was limited, allowing Germany’s economy to recover remarkably fast. He uses the example of the German automobile industry. As he points out, the Ford factory was largely undamaged, and at Volkswagen, 91 percent of the machinery survived the bombing (and the postwar looting); by 1946, he says, the Volkswagen factory was producing one of every two cars in West Germany (p.85). He writes, “By early 1947 the chief impediment to a German recovery was no longer war damage, but rather raw material and other shortages—and above all, uncertainty over the country’s political future” (p.86).14 In industrialized countries such as Japan and Germany, postwar economic trajectories fall in line with traditional neoclassical growth theory: when an external shock (like bombing) damages an area, the affected area will experience an accumulation of capital until it converges back to its steady state. But what can this story of remarkable post-World War II economic recovery in industrialized nations tell us about Southeast Asian development after the Vietnam War? For the past ten years, the scholarly narrative about postwar Southeast Asia has followed this same trajectory. In a genre-defining

18 c h a p t e r 1

article, Edward Miguel and Gerard Roland (2011) find that Vietnam War-era bombing has little significant impact on the poverty rates of South Vietnamese districts today. Their analysis of the Vietnamese district economies has added to the ongoing debate about the efficacy of bombing, drawing on expanded bodies of evidence and intricate econometric analysis to find that in the ten to twenty-five years following the Vietnam War, bombings have had no long-run impact on population density, poverty rate, consumption, literacy, or access to electricity. To reach these conclusions, they compared provinces and districts within Vietnam and regressed development indicators against historical rates of bombing. They use their regression results as evidence for long-run recovery in Vietnam. Their famous nullresults publication has launched ships, inspiring many similar studies of Laos and Cambodia (including my own). Yet the new line of inquiry launched by their groundbreaking work has produced evidence that contradicts their findings in Vietnam— bombs can actually produce long-term poverty traps—and have introduced reasons to question Miguel and Roland’s interpretation of their data (Yamada and Yamada 2021; Lin 2022; Guo 2020). This new tradition of study, which applies quantitative tools to study the legacy of war, grapples with an important inferential concern that arises when a large-N study compares the economic trajectory of bombed and nonbombed communities. One might wonder if the places where bombs were dropped were simply places that were more likely to recover quickly: richer, more educated villages might be more desirable targets, having more industry and transport infrastructure to be bombed. These more educated and organized villages might also be better at reducing damage from bombs, hiding equipment and themselves with clever strategies such as the Cu

t h e p e n f a m i l y f a r m 19

Chi tunnels, an intricate underground tunnel network developed by the Vietnamese Communists to protect villagers from bombs. This line of questioning illustrates why it is hard to do this analysis—and why different scholars who examine the same region and even use the same data might reach different conclusions. The problem is that regions “assigned” to suffer bombing were not randomly chosen, though this language (borrowed from experimental research design) implies that they were. Bombs were dropped on geographic areas with suspected communist ties. If particular social and political histories were driving contemporary behaviors around bombing and locals’ responses to bombing, then observed differences in postconflict behaviors and outcomes would not be a legacy of war; they would be a reflection of historical patterns of sociopolitical development that predate the Vietnam War. For instance, if economic outcomes are a function of a geographic location of a community (e.g., suppose that higher elevation communities are more isolated, self-sufficient economies, uninterested in commercial production), then differences between bombed communities and other communities would simply be due to the fact that bombed communities are located in areas that have these geographic characteristics that (like high elevation) are associated with economic underdevelopment. The experience of war would be irrelevant; we could explain the divergence in behavior by the fact that bombs were dropped in geographic areas with features that have been and remain today associated with less dependence of external markets. If we were to simply pool units together and run regressions, then each unit would lack a clear counterfactual comparison. It may not be useful to compare two villages with different economic trajectories that have widely divergent geographic characteristics and violence

20 c h a p t e r 1

histories, because we would not be able to tease apart the impact of the geographic factor from that of the history of violence. Fortunately, the recent works I mention here do not fall into that trap. Relying on advances in causal identification techniques, machine learning, and the other tools for handling large amounts of data that have transformed this area of literature, they each pick some form of a natural experiment that allows simple comparisons between treated and control groups. Miguel and Roland develop a strategy that exploits the geography of conflict to solve this problem of causal identification. They argue that the distance from a key war site is a source of exogenous variation in the intensity of violence.15 Since the bombing was more intense at the border between North and South Vietnam, they use the distance from the 17th parallel demilitarized zone to represent the intensity of airstrikes. This seems an appropriately random type of selection for empirical statistical analysis; because the border was arbitrarily drawn during the Cold War negotiations between US and Soviet diplomats, its placement likely did not take into account local economic conditions or US military needs. Yet, as Melissa Dell and Pablo Querubin (2017) point out, firepower is not randomly allocated: while bombs may concentrate in regions closer to North Vietnam, they are nonetheless locally targeted. The US military sought out civilian populations, particularly “places where insurgency is already on the rise, confounding simple correlations” (p.703).16 However, comparisons of district-level statistics are unable to paint a more detailed picture of bombing patterns, and these more granular comparisons can misrepresent the story. As Miguel and Roland are careful to report, they are interpreting aggregate statistics. While they found that bombed districts on average can recover as quickly as nonbombed districts, one

t h e p e n f a m i l y f a r m 21

cannot infer that these results carry over to the village or individual-level behaviors. For instance, imagine a scenario where one village received all the bombing in a district. Say that village remains underdeveloped today, while all the surrounding villages in the district, which were not bombed, have high levels of development. At the district level, it would look like bombing leads to development, but at the village- level, the opposite would be true. This ecological fallacy is why it is dangerous to assume individual behavior from group statistics. In fact, studies that measure at a more fine-grained geographic unit (at the village- and individual-level) find the opposite: bombing occasionally creates poverty traps (Yamada and Yamada 2021; Guo 2020). While these statistical evaluations have set a new standard in estimating effects and building confidence in the causal connection between violence and development outcomes, they have perhaps focused overly narrowly on causal inference, ignoring the mechanisms that link long-term socioeconomic change and war. The problem we econometricians face today is how to connect the long-term rhythms of postconflict development to the populace’s specific, local experiences of violence. Few studies link an individual’s earlylife bombing experiences to later-life economic behavior, and whether violence alters the economic risk preferences central to decision-making is an open question. If bombing does alter individual risk preferences, this could help explain why certain countries fall into cycles of conflict and poverty (Collier et al. 2003). This book picks up where prior researchers left off. Its driving theme is that people are trying to make a living in the trenches and fortifications of a postwar landscape, and it seeks to describe how they negotiate that tension and the results of the solutions they have chosen. This focus on why is novel;

22 c h a p t e r 1

previous work focuses not on why but on statistically detailed and data-driven results—a necessarily incomplete analysis that cannot attend to context and to individual stories that link history, feelings, and behavior. This book uses ethnographic observations and interviews from immersive fieldwork (the most time-honored approach to the study of rural subjects— e.g., de Tocqueville 2003; Geertz 1973; Scott 1985; Cramer 2016) to illustrate possible causal mechanisms and generate new hypotheses for testing, and integrates this qualitative analysis with statistical work, seeking to build on the strengths of both approaches. Since ethnographic observations are not scalable and findings don’t necessarily translate from one context to another, I use them to triangulate my quantitative findings. Oral narrative can fill in what regressions leave out and help us understand mechanisms that connect historical violence to rural development today. As information is gathered, we get a better sense of the overall structure of the problem. With fuller evidence gained through ethnographic observation and clearer thinking about mechanisms, I draw here a more complete picture of how bombing affects long-term economic development in various contexts in rural Cambodia, concluding that previous work has underestimated war’s long-term economic effects in certain contexts.

Why Bombs Persist Unexploded bombs—physical remnants of war—remain in the ground for decades, retaining their potential to cause lethal harm. By highlighting how old bombs left undetonated in the ground make soft, fertile soil more dangerous and— counterintuitively—less valuable to farmers, this study provides a deeper understanding of the particular mechanisms

t h e p e n f a m i l y f a r m 23

that would allow us to give substance to the empirical correlations I examine here, documenting the causal mechanisms that connect historical violence to contemporary outcomes. This empirical analysis addresses the historical legacy arguments, demonstrating why legacies of war persist and affect economic behavior and outcomes even today, fifty years later. As I show, the bombs continue to confound household agricultural management and human security despite several postwar decades of changing political regimes, ministry policy encouraging rice production, and the improvement of agricultural technologies. The following chapters draw on original econometric analysis and more than a year of ethnographic research on Cambodian agricultural practices to reorient prevailing scholarly conceptions of the economic legacy of war, offering a reinterpretation of war’s effects through the close examination of the US bombing campaign and its consequences across geography, language, and time. The book reveals a reversal of fortune that rendered Cambodia’s most fertile land unworkable. For rural Cambodian people, ranging from village chiefs of the highland tribes to urban migrants from Phnom Penh to former soldiers who fled the Khmer Rouge, unexploded ordnance produces a sphere of fear over their communities—a pervasive fear that stalls out economic growth. Moving beyond the Americacentric corpus of Vietnam War history and theories of economic recovery delimited by macroeconomic indicators, the book draws not only on immersive fieldwork in remote Cambodian villages but also declassified materials and scholarly sources spanning several languages (English, Khmer, Jarai, and Tumpuon). This comparative, multimethod approach highlights both the heterogeneity of the people involved in and affected by war and the value of consulting discrete sources of data in tandem. From oral histories of indigenous elders

24 c h a p t e r 1

to Cambodian Ministry of Interior surveys, from US military maps of payload targets to personal observation of professional deminers, these sources together shine a bright light on the enduring consequences of a presidential action taken by Richard Nixon and his advisor Henry Kissinger, who later admitted that he never regretted the decision to indiscriminately bomb a neutral country (Kissinger 1994; Grandin 2015: p.173–186). The approach I offer here—bringing ethnographic depth and specificity to bear on abstract empirical analysis—is necessary to appreciate the true breadth and persistence of bombing’s destructive nature. Scholarly debate on the efficacy of airstrikes remains limited to their immediate impact on destruction and insurgency, but these oral narratives remind us that even the most advanced military technology often fails, and these failures impact lives long beyond the intended lifespan of a carpet bomb. My respondents illuminate cultural practices that might otherwise remain obscure, since scholarship by and large focuses exclusively on American military archives and English-language sources; this work thus provides a new perspective on the history of the Vietnam War, one that better suits our complex, layered, globalized political present. This approach also pushes back against our tendency to frame interstate war in terms of deterrence, coercion, bargaining—the traditional topics in international relations. Academic concepts like these represent theories of war from the top down, as security experts try to understand the decisions made by political and military leaders. But I’m compelled to think about the inverse perspective. What does war look like from the bottom up? To fully understand what impact bombing has on its victims, we have to describe what an air campaign looks like from below. This kind of work

t h e p e n f a m i l y f a r m 25

prioritizes on-the-ground and off-the-tarmac fieldwork and asks the researcher to write about life as they find it. This kind of research frees ideas to circulate in new ways. My respondents managed to scramble the predominant scripts held by airpower experts, who focused overwhelmingly on what bombing means for “us,” not “them,” and by postwar economists, who failed to notice that not all bombs explode. These are powerful correctives. But my respondents also revealed how the problem of unexploded ordnance has been woven into the history of the nation, whose politics and ethnic ties turn out to be as layered and whorled as a handwoven Cambodian krama.17 This grounded, actor-based approach revealed an important lesson: one cannot separate this legacy of war from the social fabric in which it exists, in particular the ethnic prejudices and political inequalties in that society. My book is really about the ways in which unexploded ordnance becomes part of, and exposes, the social and political context in Cambodia. As I spent time in one multi-ethnic village, I observed how ethnic hierarchy and positionality shape the understanding of and response to the bombs in the ground. As I shadowed a team of humanitarian deminers, I was surprised to learn how demining can make less-powerful, rural Cambodians vulnerable to predation via the new land laws in the country. By attending to the intimate nuances of power and emotion, I began to see how individual lives are shaped and determined by long-term historical processes. I also came away with a grim understanding of how America’s air war looks to its Cambodian targets. Broadly speaking, my book uses a fieldwork-based approach to understand the legacies of war, showing how that legacy a) impedes growth; b) is filtered through social relations; and c) unexpectedly provides protection against predation. All of this

26 c h a p t e r 1

may be seen through an underappreciated legacy of US foreign policy. These stories about the implications of unexploded ordnance assign importance to political agency. Survivors make crucial decisions about how to frame the narratives they tell about other ethnic groups. Moreover, farmers in later time periods face another choice: to request professional demining services (though elites may steal their land afterward) or to keep bombs in their land (so elites can be deterred from taking it). Across the book’s central empirical chapters, we move from the personal to the political. The work begins by considering the war’s transformation of rural life in Cambodia: the American bombing campaign stunted the livelihoods of indigenous minorities and the Khmer ethnic majority. Focusing on Ratanakiri, the most contaminated province in Cambodia, the next chapter traces the transmission of agricultural practices— which in this area often include bomb-handling practices— from parents to children. The following chapter zooms out to examine how unexploded bombs interact with the regional predatory economy, in which elite land grabs threaten smallholder farms; it shows that, paradoxically, the same bombs that threaten local farmers’ livelihoods also serve to protect their land from being gobbled up by capitalist development. The book’s final chapter, “Beyond Cambodia”, takes a global perspective, examining how a similar legacy of unexploded ordnance from more recent bombings is affecting war refugees in Syria, Bosnia, Lebanon, Yemen, and Ukraine, hindering their return home. As in Cambodia, advanced military technology was no guarantee against failure, for despite more precise weaponry and real-time satellite monitoring, many bombs still simply do not explode on impact. The immediate, tangible damage from properly exploded bombs is terrible, but it is finite, immediate, and known: it can be repaired because it has

t h e p e n f a m i l y f a r m 27

actually happened. In contrast, the damage from unexploded bombs lingers as potential damage that cannot be repaired because it is not yet over; the pending not-yet-but-someday explosions keep people frozen, waiting and waiting for the worst to happen so they can finally move forward. As the stories in this chapter show, Cambodia’s half-century-old tragedy is being repeated right now in many other parts of the world— and it will continue to repeat as long as American manufacturers produce and sell warheads and munitions to US allies. Whether they work or fail, these advanced munitions leave permanent scars on their rural targets, often locking them into cycles of poverty and stalling out the economic recovery that we see in postwar industrialized areas.

Overview of the Book This introduction has laid out the theoretical framework of the book and contextualized it within the vast array of scholarly approaches to rural growth and development. Rather than attempting to survey all of the development literature, I reviewed studies from economics, political science, and history that help us examine recent surges in economic inequality. I argue that certain changes in prosperity can be best seen as they emerge from landmark conflict. Over the course of war, the state and other contentious actors confiscate, plunder, and extort what the dead and displaced leave behind—houses, land, and assets—as advanced military weapons reduce entire neighborhoods to rubble and disintegrated jungles, from the Tokyo firebombing during World War II to the deployment of carpet bombs and Agent Orange during the Vietnam War. As I show, this military technology leaves behind unintended political and material legacies. When undetonated and leftover

28 c h a p t e r 1

weapons remain behind in conflict zones, their physical presence haunts the civilians who live there. With each step in the process of postwar reconstruction, the book shows, local political and economic practices shift in response to the new war-sculpted environmental landscapes in which constituents must eke out their livelihoods. Chapter 2 turns to the world of Cambodian farmers, who make up 80 percent of the country’s population. After conflict, the unexploded bombs produced and intensified economic inequities, as farmers in heavily bombed areas were and still are unable to take advantage of modern agricultural machinery and innovations, such as tractors, water pumps, and new types of commercial crops that require tillage. This chapter examines both the surface conditions at the targeted sites and the number of bombs at these sites to discover how and where impact fuses fail to trigger. It shows how large numbers of unexploded bombs tend to cluster in the most fertile soil, causing farmers in these seemingly ideal areas to produce crops only at the subsistence level. Drawing on personal interviews with farmers on the Cambodia-Vietnam border, this chapter elaborates on the livelihood strategies and economic decision-making of current residents, who are often unwilling to invest in green revolution technologies that require them to be closer to the bombs. In this chapter, I describe a large-N statistical analysis of the invisible economic power of undetonated bombs, often unseen and hidden beneath the ground, across the entirety of Cambodia. Chapter 3 examines the relationship between risk and race, picking apart the tangled skein of relationship between indigenous minorities and the Khmer majority and examining how each group justifies its own bomb-handling practices and judges those of others. The remarkable changes in the

t h e p e n f a m i l y f a r m 29

agricultural and social practices within one multi-ethnic commune at the Cambodia-Vietnam border tell us much about adaptive mindset of its residents as they contend with the psychic weight of everyday struggle. A collection of oral histories across ethnicity, age, gender, and wealth betray no fantasy of resilience: Khmer brothers remember human remains hanging from the branches of the village tamarind tree after an indigenous man accidentally detonated a cluster bomb; a Khmer migrant, a former soldier, scoffs at the minorities for “knowing nothing” about bomb safety; a Tumpuon elder sets out to teach his new Khmer neighbors what places to avoid, based on his memory of the bombing. The chapter considers the fearlessness, unprocessed grief, and desire to blame carried by certain residents, particularly children and former soldiers, as they face the difficulties that still linger for the community. In chapter 4 I examine the country’s turn to authoritarianism under Cambodian Prime Minister Hun Sen, the longest-ruling leader in Asia, whose rule has coincided with the spread of land disputes and land seizures across the country. This chapter pays particular attention to the 2001 Land Law, a reform designed to redistribute land to the poorest that has instead been used by the political elite to justify their enclosure of communal land for economic development. My respondents in Ratanakiri Province reveal the odd insurance offered by land too dangerous to use: while the unexploded bombs endanger them and prevent them from using their land fully, the bombs also protect them from the Land Law, making the damaged land unattractive to commodification by speculators. If residents seek to increase their economic security by removing bombs and making the land arable, they also increase their economic risk by making the land desirable to others. I show

30 c h a p t e r 1

this process at work, tracing the process of explosives clearance in one area and its aftermath: residents in cleared areas lost their protection against the Cambodian predatory economy and were soon evicted. The conclusion synthesizes the book’s findings and implications, suggesting new directions for further inquiry. Though the effects of bombings might vary by context, there are good reasons to expect that some of these findings will apply beyond Cambodia. In the Falklands, 80 percent of the twenty-fivethousand mines are hidden in sandy beaches, which allow the mines’ positions to shift over time, making detection difficult. In the Balkans, 150,000 pieces of unexploded ordnance remain in rural landscapes such as the forests around Sarajevo and Trebevic mountainsides. In Bosnia and Herzegovina, the most heavily mined area is the fertile Doboj region, where demining operations cost an average of one thousand euros per munition and financial cutbacks have delayed removal efforts. Unfortunately, this book’s findings will likely continue to be relevant, for despite innovations in precision-guided munitions, other areas are likely to be cluster bombed; the United States still maintains a domestic arsenal of 2.2 million cluster munitions and 1.5 million abroad. While the last confirmed US airstrike involving cluster munitions took place in 2009 in Yemen, survivors of a 2016 Saudi aerial attack of Yemen found the latest generation of US-manufactured cluster bombs. If American manufacturers continue to sell cluster munitions abroad, then unexploded ordnance should be treated as an enduring political problem, deserving of continued study and attention, rather than a one-off historical phenomenon. ———

t h e p e n f a m i l y f a r m 31

It has been fifty years since the US military dropped bombs over the Pen family’s land. Technically, there is no war in Cambodia anymore, but it does not feel like a land at peace. The bomber jets flew away long ago and many of their pilots are now dead, but their weapons live on, continuing to haunt the Pen family land. Lom and Leang are still unable to safely take advantage of heavy machinery such as the community tractor shared by other farmers in the village. They cannot even use the entirety of their field. Lom blames the bombs. “We don’t know where they are, so we’re still afraid,” she says as we walk back from the detonation point.18 She prefers to stay close to the house when planting, she says; she particularly hates going to the edge of her field abutting Elephant Hill, for she fears that areas closer to the former military base will have more cluster bombs. Yet Lom and Leang feel trapped, locked into their family’s land: the house is nice, and the land itself is more fertile than they might be able to access elsewhere. “It’s more like we can’t leave,” Lom says.19 Her husband confirms that “we don’t have really have a choice, so we stay here.“20 Indeed, Leang admits, they stay inside the house as much as possible. We return to the house, eat our styrofoam clamshells of barbecued pork and rice and sit with the family for the rest of the afternoon, watching intermittent traffic (motorbikes, carts, one car) come down the dirt road.

2 In Suspect Terrain how war changes l and Where they make a desert, they call it peace. — ta c i t u s , o f t e n c i t e d b y t h e l a t e v i e t n a m c o r r e s p o n d e n t b e r n a r d f a l l .1

for twelve years or so, Vannary Leng has lived on a small farm in Ratanakiri Province, in Koun Mom District, about twenty miles west of the Pens’ property.2 Vannary’s house sits on an unpaved road that curves along the edge of a dense deciduous forest. Like the Pens’, Leng’s house has the classic Khmer features: built on stilts, made of wood, shaded by a peaked roof, surrounded by water and lush greenery. The house has neither doors nor security fences, but the east side of the ground floor has a wall—a slim slat of wood that protects the kitchen tools from the wind. This flimsy kitchen wall is papered with calendar sheets, peeling away in some places to reveal the original mahogany-hued timber, hewn from lebbeck trees from the surrounding forest. Fading into pastel are two framed portraits of the royal family: King Norodom Sihanouk, who 32

i n s u s p e c t t e r r a i n 33

died six years earlier in 2012, and Queen Norodom Monineath Sihanouk. Vannary is in her thirties, lanky and lean as a high school athlete. Although the temperature is hovering above eighty-five degrees the day I am there, she wears a teal jacket, a navy blue T-shirt with an American flag logo, and red track-suit pants. Sitting on an old wood plank, with bare feet tucked into her legs, she hollers at her neighbor who is walking down the road to ask him what he is selling this morning (the answer is white corn and mushrooms shaped like bird eggs). Her young daughter sits in her lap, flipping through a Khmer alphabet book. Vannary’s farm is small, only two hectares. Her husband, Touch, only has time to help her on the farm when he is not at work cutting timber throughout the province. For more than a decade, she has been clearing it by herself and building up its mud retaining walls. Her rice paddies are shallow rectangular basins, dammed on each side so the monsoon rains can fill them during wet season and keep the dry-season sun from scorching the plants. The dams are built from turf cut by Vannary, who turfed out her land by hand. The forest edges her land on three sides, and another farm (her mother-in-law’s) is on the fourth. In the forest alongside the property, thickets of palm (pdao) cover the space between trees; she makes rattan baskets out of the dried leaves. It is a quiet, pretty corner of land. Vannary comes from a small village in Prey Veng, an agricultural province on the east bank of the Mekong River. In English, Prey Veng translates to the Vast Forest, but today it is not so vast: over the course of the past century, families have razed trees to extend their smallholder rice farms. As the population has grown, the great diospyros forests have steadily disappeared. The farther away you get from the Mekong River—the farther you go into what was once forest land—the topsoil becomes

34 c h a p t e r 2

more clay-based, light brown, and silty in texture. There is no doubt that this soil is richer than the flinty, mottled soil on the riverbank. It holds nutrients and moisture better, and it grows better rice. Vannary told me of the difficult soil in her birthplace on the riverside, where the soil was incredibly rocky because the alluvial silt washed away each monsoon season, leaving behind stones and chunks of gravel; her family’s rice paddies had the tendency to turn very hard when dry, she said, and required constant applications of fertilizer. Touch and Vannary first met in 2005, when he traveled to Prey Veng to visit friends who lived in her village. Before long, the two fell in love and made plans to move to his hometown in Ratanakiri, where they could take care of his mother and work the available land near her house. Their dream was to convert his family’s land to a cashew orchard. When they married and settled down in this house in 2006, Vannary began the farm while Touch traveled outside the village to find more lucrative opportunities as a day laborer. She told me of her first year on the farm: before the onset of rainy season, she marked out a small patch of forest where she cut down the smaller trees, like the rattan palms, to open the site. She let the cut wood dry out and set the sun-baked brush on fire, releasing nutrients into the soil through the charcoal and ash deposits. She was very familiar with this type of slash-and-burn practice, which had been common in her hometown, but in Koun Mom, she uncovered something she had never seen before: a thin metal wire stretched above the ground. She asked her neighbors about it, and they told her it was a tripline. One side, they said, would be attached to a spring stuck to the ground, the other side to a small, spherical bomb. These tripwire bombs are easy for beginners to spot if the tripwire is intact, but other types of bombs—types without external triggers to give

i n s u s p e c t t e r r a i n 35

away their locations—are even more dangerous. Vannary’s burn cleared the ground cover, exposing many pieces of metal, some scrap, some live munitions. “Once I started burning the land, I saw unexploded bombs for the first time,” Vannary remembers.3 In hindsight, she was very lucky. She could have accidentally detonated a bomb during the burn. This was her first encounter with unexploded ordnance. She was already familiar with the effects of bombs that had exploded successfully; her hometown in Prey Veng had been bombed during the Vietnam War. While she had no direct experience with the actual bombings themselves—she heard about them from her parents, for she had not yet been born when the bombs were falling—she had grown up around bomb craters that she said looked like “man-made holes.”4 Some of these craters were used to store water or hatch fish. Others had eroded in the humid climate, the once-perfect circles turned lopsided with dirt and brush collecting on one side. All the bombs that were dropped over her family’s house, she believes, detonated upon impact. “I didn’t see any leftover bombs there, like I see them here,” she says quietly.5 Vannary pauses and looks slowly from her house to the vegetation surrounding the house, which has grown wild, giving the house an unpredictable facade at any given time. The difference between the bombs’ behavior in Prey Veng and Koun Mom, she thinks, is in the soil. In Prey Veng, the land is hard and rough as a cob. In Ratanakiri, the land feels wet, almost mushy, and is dense with blooming gardens and generous fields of rice. She observes, “Here, the soil is better. Like when I grow beans in the garden, I have this huge harvest. The same is true for our rice. We don’t even need to use fertilizer.”6 Her first days here, she congratulated herself on her luck. If someone had to be a farmer in Cambodia, there is no doubt

36 c h a p t e r 2

Ratanakiri would be the top choice by almost any measure. The land lies pretty and green throughout the year, there is a favorable population-to-land ratio, and the forests are abundant in natural resources (timber, gemstones, deer, mushrooms). But, she found, the good soil and land fertility come at a cost. “I see bombs all the time,” she says in astonishment. She points to the high grass among old lebbeck trees. “I found two bombies over there earlier this week,” she says, using the American nickname for small antipersonnel bombs.7 They were cluster munitions the size of baseballs, rust-covered and half-exposed in the mud. One was on the walking path in the forest and the other in her rice field. This time of year, when the mushrooms come out, bombs are trickier to spot; the rain and humidity that are ideal conditions for mushroom growth also help fill out the forest undergrowth, providing extra cover for anything hidden in the ground. Even on a casual walk to her mother-in-law’s house, a path she has traveled a thousand times, Vannary’s eyes are always alert for any protruding gray lumps. The unexpected cost of living on a former battlefield weighs heavily on her already strained family. Despite the productive richness of the land, she can barely produce enough rice to consume throughout the year because she cannot safely clear enough land. This problem compounds itself, for the worse her harvest is, the more her family depends on her husband to find work and wages outside of the village. This means he spends more time away from home and she has less help to manage the children and the farm. Her older children can’t help her in the fields because she doesn’t want to expose them to the danger of the bombs. The worry is constant, and finding bombies on a regular basis only reinforces her fear. I have begun with the story of Vannary’s transformation from village newcomer to local bomb guide not only because I

i n s u s p e c t t e r r a i n 37

got to know her well, but because her story illustrates so many of the facets of bombing’s consequences we will encounter in this chapter. First, it shows that bombing is actually a composite intervention. Some bombs exploded on impact and others did not, and as objects of study, locations with many leftover bombs are as important as areas that took heavy damage from successfully exploded bombs. However, most studies of airstrikes have focused almost exclusively on the location and degree of damage inflicted by a detonated bomb—that is, on a payload’s immediate destructive impact. This understandable focus on detonations has led to a particularly unfortunate outcome: the unique horror of bomb explosions has diverted attention from their similarities to the slower-acting “less lethal” forms of violence. We must therefore broaden the current framework of study, expanding from a focus on the immediate victims of airstrikes to the study of people like Vannary who continue to navigate the gray spaces between danger and safety—a task that has become increasingly common not only in postwar Cambodia but also in today’s pandemic world. Second, one of the most interesting (and most ignored) characteristics of gravity bombs is that they often do not work. These bombs can fail to detonate when they hit soft ground, leaving a potentially lethal weapon in the soil for decades after the payload drop. Fertile surfaces in particular may fail to trip the trigger’s impact fuse, which detonates the TNT and other explosive material (Moyes 2002). As I show here, unexploded ordnance adds immense physical risk to daily agricultural activities, making the most fertile land the least productive because it is so dangerous to farm. In this odd reversal of fortune, the very best land continues to be held hostage by war. This chapter shows how violent political action can change the environment over the long term, thus challenging

38 c h a p t e r 2

the conventional treatment of the physical environment as an exogenous driver of human behavior. Instead, I show how war can mediate natural processes—for example, by turning favorable geography into a liability. As I show, foreign actors can transform domestic resources through the military technology that they leave behind. Bombs thus connect the local and the global. In this modern era of airstrikes and bombardments, it is as impossible to draw a sharp line between domestic and international politics as it is to understand Cambodian economic development without understanding its natural resources and its conflict history. Finally, living with unexploded bombs is transformative. When bomb fuses fail and the bombs disappear into agricultural land, rural households are left to deal with the dangers of unexploded ordnance. Since these bombs are hidden in the ground—no one knows exactly where they are or how many of them are present—rumors of finding a single bomb can single-handedly trigger a sphere of fear that is much larger than the bomb’s immediate blast radius would be. Just as this fear turned Vannary from a confident farmer into a cautious one, it may be turning thousands of others away from their dreams of commercial agricultural production toward the reality of farming minimally, only at the subsistence level. Over the span of fifty years, these short-term adjustments in agricultural practices accumulate, leading to underproduction and poverty in the long term. These points raise key empirical questions that this chapter addresses. How exactly does bombing change the behavior of individuals in former conflict zones? Is the mechanism psychological, in that it changes the way people cognitively process the world (e.g., discounting the future and shortening their time horizons)? Or is it material, with people continuing to pursue

i n s u s p e c t t e r r a i n 39

their goals rationally in a world that imposes a huge transaction cost on daily activities? In this book, I focus on identifying and tracing the mechanisms that link a historical foreign policy to contemporary economic behavior, using a mixed methodological strategy. In the current chapter, I rely on ethnographic evidence from Vannary and other Ratanakiri residents to probe the mechanisms by which the effects of unexploded ordnance persist through time. I also quantitatively test the theory I propose here; I use a historical dataset of the US Air Force sorties flown over Cambodia from 1965 to 1973 and a causal inference design that relies on the quasi-random failure rate of bombs and pretreatment controls. These qualitative and quantitative methods work in tandem. The insights of Vannary and her fellow respondents provide us with an understanding of how and why bombing from a half-century ago continues to impact rural life today, and the statistical modeling indicates that Vannary’s experiences index broader inequities in Cambodian society.8 In this chapter, I explicitly examine the origins of Cambodia’s underdevelopment, paying special attention to how American foreign policy unintentionally changed large sections of rural land. There are two main ways in which leftover military technology enters the local economy: through unexploded ordnance hidden in the ground and through the consequent fear that spreads through the village. First, I argue that, although prior research has examined the legacy of war and rural development separately, Vannary’s experience suggests that these topics should be in conversation. Next, I present evidence of changes in economic decision-making among conflict survivors, which I find to be more rational than other scholars have claimed. Finally, I turn to the way farmers have most effectively survived: by farming minimally.

40 c h a p t e r 2

Connecting War and Agricultural Development I am nowhere near the first to try to understand either the longterm impacts of war or the economic obstacles to rural development. Many scholars have addressed the long-term impacts of war: political scientists and economists look at different ways in which war fundamentally changes humans’ interactions with governments, and on the other side of university campuses, evolutionary biologists and field scientists examine how violent weapons impact ecological systems. Many other researchers address the larger (non-war-related) economic obstacles to rural development: social scientists from political science to agricultural economics have long been interested in why rural economies often suffer from persistent inequality and low agricultural productivity, particularly in the developing world. What seems to be lacking is a scientific attempt to connect the economic changes wrought by war to the rural landscapes that it has altered, materially and enduringly.

How War Changes the Developing World Vannary’s story is striking when juxtaposed with the political economy literature on the long-term consequences of political violence, as these accounts typically overlook the role played by land and human ecology in postwar development. This literature documents that war can trigger institutional changes—electoral reforms, regime turnover, or improvements in democratization—that augment the opportunities for development. As scholars have pointed out, political procedures that stabilize the new social order begin while active war is still ongoing; international peacekeeping operations seek to

i n s u s p e c t t e r r a i n 41

alleviate fear and mistrust between conflicting actors (Fortna 2008). Stabilization through governance becomes even more paramount in the first few years following war. During this period, the promise of international aid deeply shapes internal political structures: as United Nations missions begin to play a vital role in building governing institutions and supporting actors committed to peace (Doyle and Sambanis 2006), postwar regimes begin to look outward for aid and international legitimacy, which incentivizes them to democratize, hold popular elections, and incorporate combatants into their new party system (Matanock 2017). In other words, the way states are built in the years immediately after war shape developmental pathways. Institutional changes imposed at the national level are well equipped to explain cross-national variation in economic outcomes yet are less suited for analyzing why only certain regions within postconflict countries continue to suffer from underdevelopment. Additionally, democracy-enhancing institutional reforms implemented following conflict may be withdrawn once international actors depart, raising questions about the long-term legacy of conflict on political and economic stability. An alternate line of scholarship has extensively studied the long-term impacts of wartime violence on economic development. This line of work describes how war shifts the distribution of wealth at the level of the state and the community, with valuable resources being removed through three primary methods: physical destruction, capital flight, and dispossession, as the aggressing state simply appropriates resources from individuals, communities, or the victim state. The effects of the physical destruction of resources such as forests, mines, and factories through bombing are unmissable and are a usual

42 c h a p t e r 2

target of rebuilding and reparations postwar. As many scholars have pointed out (see chapter 1), these types of lost resources are generally replaced quickly postwar, and this type of resource loss therefore has little long-term effect on economic development at either the state or community level. Resources are also lost at the state and community levels when elites carry away resources as they flee—a particularly damaging phenomenon in countries where capital is largely mobile (Collier 1999). Given that the rich rarely return, postwar development in these countries requires substantial external investment at the state level (Collier 2006). In conflicts that involve systematic and geographically concentrated mass killing, like the Third Reich’s targeted genocide during World War II, there can also be significant resource losses at the community level through victimto-perpetrator transfers of property; this may give villages near killing sites an advantage in acquiring newer and better housing, but it does not appear to otherwise impact human or economic development in the area over the long term (Charnysh and Finkel 2017). Overall, scholarship on the economic impacts of war suggests that wartime shifts in resources at the state level and the community level are short-lived, having no significant influence on local wealth or quality of life in the long run. However, this may not hold true at the household level over the long term, and few studies have addressed this question. While some studies that examine short-term effects of war on household resources and wealth, focusing on the immediate postwar years (Annan et al. 2011; Brück and Schindler 2009; Bozzoli and Brück 2009), few studies trace war’s longterm redistributive impact at the household level. How war redistributes resources at the hyperlocal level and how this redistribution contributes to household poverty thus remain important open questions.

i n s u s p e c t t e r r a i n 43

By what mechanism might war affect long-term household economic health? A burgeoning literature points to violent trauma as having lingering effects on intergenerational preferences. Studies found that bombing and mass deportation, respectively, reduced long-term political support for the perpetrator in Spain and Ukraine (Balcells 2012; Rozenas, Schutte and Zhukov 2017). Another study found that these effects persist across generations: the death of a family member during the deportation of Crimean Tatars increased the strength of ethnic identification and hostile attitudes toward perpetrators in the minds of both victims and their children (Lupu and Peisakhin 2017). (Perpetrators have also displayed this type of intergenerational resentment; historical experience with the Black Plague pogroms, in which Jews were blamed for spreading the disease, are highly predictive of anti-Jewish discrimination prior to World War II, according to Voigtländer and Voth 2012.) While this body of work examines this phenomenon across multiple forms of violence, each case study arrives at a similar explanation: family socialization, in which attitudes are passed down from parent to child. This explanation relies on two assumptions (which may be flawed): first, that parents want to make children more like themselves, and second, that parents have the resources, individual ability, and institutional capacity to control their children’s beliefs (Bisin and Verdier 2001). Most prior work, then, has focused on short-term economic outcomes of war. Recently, however, scholarship has examined the potential long-term impacts of war on political and economic outcomes by pioneering new methodologies: most recent scholarship in this area has focused on overcoming the substantial methodological challenges of coarse measurement and nonrandom assignment by analyzing impressively detailed quantitative datasets alongside historical knowledge of policies

44 c h a p t e r 2

that randomly ascribe treatment status. Perhaps the important question about economic legacies of war is not whether they exist, but rather where, in what manner, and why they do or do not manifest themselves. Largely missing from the current debate are substantial explorations of why certain elements of practice persist beyond the violent experience that birthed them. How do households transfer wartime profits or losses from one generation to another? Why would children choose to side with their parents’ political ideology, given an array of other party options that may cater to youth issues? By shifting the research question to describe the causal mechanism by which these legacies of war persist, we create a new, second-generation set of inquiries into the long-term economic impact of war at not only the state or community level but the household level. The bulk of the research on war and economic development focuses on the factors of production at the state level, particularly population and capital (discussed in Blattman and Miguel 2010: p.8). However, labor and capital are not the only factors of production that can be destroyed by violence; households, too, are producers, and modern warfare technologies (from Agent Orange to carpet bombing) change the physical landscape in those rural areas where households engage in small-scale agriculture. War’s impacts on nonindustrialized land and the natural environment, and on the human activities that take place there, are among the least studied areas in the research on violence. We know that these impacts exist: for example, studies have found that in several countries in Africa, civil wars have had the unintended effect of preserving populations of large animal species, for rebel fighting keeps poachers away from the grassland reserves where they would otherwise be hunting (Daskin and Pringle 2018). But these

i n s u s p e c t t e r r a i n 45

types of long-term environmental war impacts require further study and analysis to understand how they affect local economic development. For example, the extent and distribution of Agent Orange spraying in Vietnam has been documented in Nature (Stellman et al. 2003), but this dataset has not yet been linked to environmental or human health outcomes (Butler 2008).9

Determinants of Rural Development Much of the developing world remains predominantly rural with a large smallholding population; these rural economies are marked by relatively low agricultural productivity, a slow transfer of labor out of agriculture, and persistent inequality. This precisely describes the situation in Southeast Asia more broadly, and in Cambodia specifically. We can thus better understand Cambodia’s rural economy (and, eventually, the effects of war on that economy) by looking at existing research into economic development in two comparable countries, Thailand (Walker 2012) and Malaysia (Scott 1985). One of the earliest theorists of peasant behavior in Southeast Asia was James Scott, who took his family to live in the Malaysian village of Sedaka. From this vantage point, Scott observed growing rural inequality emerging in the wake of the green revolution. In Scott’s book Weapons of the Weak, we find a story of modernization and resistance in rural Malaysia. As Scott’s story has it, the peasant’s strategy of survival in the last century in Southeast Asia can be explained by the fact that farmers’ interests—both economic and cultural—were antithetical to economic development in rural areas. Malaysian farmers, he says, have historically lived with a high level of livelihood vulnerability, and a single drought or blight could

46 c h a p t e r 2

jeopardize the survival of the entire family. According to Scott, this livelihood vulnerability changed their economic behaviors: preoccupied with preventing losses that could jeopardize the survival of the entire family, these farmers operated according to safety-first, risk-minimizing principles. When the worst happened, though, farmers could depend on their communities to make up these types of losses, for a single crop failure could be compensated through the redistribution of village resources. As the rural economy modernized, argues Scott, farmers were suddenly being encouraged to run economic risks (to outlay money for machinery or improved seeds with no guarantee of stable returns) while watching their community economic support networks erode. As agricultural elites emerged, economic inequality intensified; by adopting modern farming techniques, more rice was produced with fewer workers, reducing the economic safety net available to the village as a whole. This created grievances among peasant workers, who protested the weakening of their social safety net. This, says, Scott, is the root of peasant rebellion: the defensive reaction against economic threats, whether those come in the form of economic risk (capital machinery, faster growing seed varietals, elite decisions to cut costs, or other innovations from the green revolution) or from the fraying of traditional forms of social life that had long mitigated economic risk. To describe this dynamic, Scott coined the phrase “moral economy”: this phrase captures not only the economic but the cultural and ethical dimensions of subsistence farming, which represented an entire way of life that was being displaced by modernization, mechanization, and concentration of wealth. His 1977 book Moral Economy of the Peasant describes how this works in colonial Burma and Vietnam. The existing cultural moral economy and the drive

i n s u s p e c t t e r r a i n 47

to minimize risk drove colonial peasant farmers to avoid—even to actively fight—the technological advances and commercial growth that create economic development. The parallel here is that contemporary Cambodian farmers also must cope with the legacies of major crises with rationally risk-averse behaviors. Like James Scott writes, “Living close to the subsistence margin and subject to the vagaries of weather and the claims of outsiders, the peasant household has little scope for the profit maximization calculus of traditional neoclassical economics” (1977: p.4). But this book’s argument makes an important point of departure: Cambodian smallholder farmers today have capitalist ambitions. They seek profit. Like Scott describes, they exhibit risk-averse rather than profit-maximizing behaviors. But unlike Scott’s subjects, my respondents’ motivations are more material than moral. Decades of civil war and a genocide in Cambodia have eroded the mutual assistance networks that Scott writes about. A new substance has been “woven into the tissue of peasant behavior,” to paraphrase the moral economist (1977: p.167)—but it is the lasting social and economic wreckage of American bombing. Other scholars focused less on the grassroots causes of rural poverty and more on the role of the state in rural economic development. A broad range of thinkers in the field of political science—Chalmers Johnson, Bruce Cumings, Stephen Haggard, Atul Kohli, Tuong Vu—together created what became known as the developmental state literature. Their work hinged on the role of states as agents of economic development. They point to the successful transformation of East Asian cases like South Korea, Taiwan, and Japan—historically agrarian societies that turned into “tiger” economies churning out manufactured goods—as evidence that states can play critical roles in rural economic development by prioritizing the

48 c h a p t e r 2

development of industry over other sectors like agriculture (see Johnson 1982; Haggard 1990; Kohli 2004). Recently, anthropologist Andrew Walker has also examined the relations between the developmental state and agrarian society (2008; 2009; 2012) using a different framework than that of the political scientists who preceded him: Walker focuses not on industrialization but on how the Thai state encouraged the transformation of poor smallholder farmers into a large, middle-income agrarian class. He points out that the Thai national rural development plan depended not on taxing farmers (the strategy in most other developmental states; see Bates 2005; Kohli 2004) but on developing the agricultural sector by subsidizing the peasantry. This, says Walker, offers an alternative model of the developmental state—one that supports rural growth and livelihood stability. The state focuses on sponsoring price support, land tenure, research and extension, credit sources, infrastructure, health, and education (Walker 2012: p.56). While the Thai rural economy remains relatively underproductive, farmers do not leave the sector or the rural areas en masse, as they do in other places. Instead, they extract more from state programs, attaching themselves to the largesse (and the power) of the state. Kristen Looney’s 2020 book on the rural development of Taiwan, Korea, and Japan makes a similar point. As these states industrialized over the course of the twentieth century, they also created rural mobilization campaigns that spread a mix of propaganda and material rewards to ultimately galvanize the agricultural base. These policies were less interventionist than the ones that Walker describes in Thailand, but they served an important purpose in inspiring political loyalty among smallholder farmers. Despite this recent boom in literature on agricultural development, few social scientists have directed attention to the

i n s u s p e c t t e r r a i n 49

agricultural land itself. Yet for the rural poor, who make up the majority of the developing world, land is their primary productive asset. It plays multiple roles: a source of income and employment; a form of social insurance, savings, and inheritance for future generations; a place to live, as observed by Michael Albertus, one social scientist who has begun to attend to how small plots of farmland fit into larger political economies (2015: p.2).10 Political scientist Catherine Boone explores the colonial and postcolonial system of land tenure in Africa, linking weaker agrarian institutions to localized conflict across the continent (2014). Yet we still know very little about the political behavior of people who own and work smallholder family farms, which make up 84 percent of the world’s 570 million farms (Lowder, Skoet, and Raney 2016). We know even less about the relations that farmers have with the land itself and how that impacts their political and economic decisionmaking. We can, however, posit two important things: first, there is seldom a one-to-one relationship between quantity of land and the value of the land; second, land is of variable quality, and some farmers are more constrained in how they can farm their land than others.

A New Approach Mainstream political science research, then, has mostly ignored the connection between rural economic development and land. Research in environmental studies could help bridge that gap. Scholars working in this field see humans as actors that mediate natural processes (Cronon 1996). Indeed, according to a group of climate scientists, humans are now having a greater impact on the Earth than all the planet’s natural processes combined; as these researchers argued in 2016, we have

50 c h a p t e r 2

entered a new geologic epoch, which they named the Anthropocene era. As climate scientists have documented, geographic factors like coastal terrains, weather patterns, and land arability are changing more rapidly than ever before—a change due to human activity and particularly to states’ lax regulation of greenhouse gases (Le Quéré et al. 2018). Environmental studies focuses on the dynamic relationship between humans and the land, showing how states and their constituents actively preserve, use, degrade, enhance, or ignore their geographic surroundings. This notion of humans as agents of environmental change challenges a tendency in the classic works on political geography to treat geography as a quasi-exogenous shock to human behavior. Political geography research is closely attuned to the impact of weather, terrain, and natural resources on political outcomes (Diamond 1999; Sachs and Warner 2001; Miguel, Satyanath, and Sergenti 2004; Dunning 2008) rather than to the impact of human political institutions on geography. Accounts such as Herbst (2000) highlight the “tremendous methodological advantage [of] holding the physical environment ‘constant’” (p.13), while Ross (2013) argues that resource endowments should be treated as “unaffected by a country’s economic or political features” (p.18). This traditional canon of political geography rests on the shared assumption that the natural world functions independently from the human world and that a country’s key environmental attributes are determined outside the scope of human ability (Diamond 1999; Sachs and Warner 2001). Humans can change neither their surrounding terrain (Herbst 2000; Nunn and Puga 2012) nor the original amount, quality, or location of natural resources—though once discovered these resources can be depleted (Dunning 2008; Ross 2013; see Brooks and Kurtz 2016 for a counter argument).

i n s u s p e c t t e r r a i n 51

For example, according to this tradition, geographic factors like state size and terrain ruggedness produced weak African states: “Successive sets of leaders in Africa responded to a political geography they were forced to take as a given” (Herbst 2000: p.12). Similarly, because the geologic processes that create mineral deposits precede the formation of modern states, the distribution of natural resources like oil is considered an ex ante condition, with profits as an unintended consequence. As two influential books define it, favorable geography “gives some producers access to cheaper and better-quality oil than their competitors” (Ross 2013: p.35) with “resource rents fall[ing] into the fiscal coffers of the state more or less like ‘manna from heaven’” (Dunning 2008: p.11). These arguments treat terrain, resource endowments, and resource rents as plausibly exogenous to politics, assuming that humans exist independently from environmental processes except in their ability to extract resources. I instead operationalize the environmental studies framework, exploring how humans intervene in natural processes by treating favorable geography as a variable rather than a given. In the context of this book, I focus primarily on how geography can be made more or less favorable by previous violent conflict. There are good reasons to expect that destructive weapons, like bombs, can change land—not only through the immediate destruction produced by exploded bombs but through the long-term damage of bombs that fail to detonate, which leave behind unexploded ordnance that increases the risk of using land. This is not a minor problem; undetonated bombs are far more common than you might imagine. American military bomb disposal technicians estimate that as many as one out of every five cluster munitions fail to detonate upon impact (Ismay 2017). One-third of countries in the world report a need

52 c h a p t e r 2

to clear areas of unexploded ordnance and other explosive remnants of war (Borrie 2003), but proper removal is costly. The US Department of Defense has spent $10.3 billion to clean up more than nine hundred American military sites contaminated with unexploded ordnance from weapons testing and training, and it estimates that it will take $31 billion more to complete the task (US Congressional Report 1995). Bomb failure rates are influenced by the surface conditions at the target. The majority of explosive ordnance used in aerial attacks is designed to detonate upon impact, and soft ground and dense vegetation can cushion the fall and prevent an impact fuse from sparking. Clearance professionals observe that “[s]ub-surface UXO [unexploded ordnance] is more likely to be found in soft ground. Soft ground also makes items less likely to detonate and therefore increases the proportion that remains unexploded. . . . Dense vegetation, like soft ground, also means that items of ordnance are more likely to remain unexploded” (Moyes 2002: p.9-10). Several types of soft surfaces, such as mud, snow, sand, and surface water, have been found to produce substantial numbers of duds, and these tend to penetrate the ground cover, going ten to twenty centimeters below the soft surface upon impact (McGrath and Lloyd 2000: p.26). Given that soft surfaces can be found in almost all ecosystems, unexploded ordnance is common all over the world: it is in “the mud and jungles of Southeast Asia, the soft peat of the Falklands, the sand desert of the Gulf, and farmland in the Balkans” (King 2000: p.39).11 Bombs’ tendency to fail on good soil turns the gift of favorable geography into something treacherous. In regions where the primary food is produced in paddies, soft land is the most productive (Nesbitt 1997), and bombings are thus likely to leave more unexploded ordnance on fertile land than on less

i n s u s p e c t t e r r a i n 53

fertile areas.12 In other words, the best agricultural soils become the most dangerous, and farmers face increased (and unpredictable) risk of injury or death when accessing their land. This is why the US State Department describes unexploded ordnance as especially problematic in the developing world: “[m]ost of these countries’ economies depend heavily on agriculture,” and the explosive remnants of war can “deny farmers large sections of land” (US Congressional Report 1995).

Livelihood Risk and Poverty Economic theories of risk and precautionary behavior predict that when individuals confront a situation with risk, they can take active measures to reduce the probability of a bad event, seeking to either avoid risk or prepare for potential bad outcomes. This is true for financial risk in the absence of insurance; studies find that individuals can self-protect in two ways: either by choosing a means of production that involves lower risk—and, typically, produces lower yield (Ehrlich and Becker 1972)—or by accumulating precautionary savings (Fafchamps 1992; Jalan and Ravallion 1999). It is also true for violent risk. Normal daily operations require the immediate outlay of personal safety, and victimized individuals often self-protect by opting out of situations that have high uncertainty (Callen et al. 2014); exposure to violence can actually change risk preferences, causing individuals to become more risk-averse overall (Brown et al. 2019; Jakiela and Ozier 2019; Kim and Lee 2014; see Voors et al. 2012 for contrary evidence). As these lines of work predict, farmers in Cambodia, who face both financial and violent risk in their daily work, respond by seeking to avoid risk: they develop alternative agricultural practices that lower their exposure to the violent risk of unexploded

54 c h a p t e r 2

ordnance, curtailing their full use of land and their ability to expand production through mechanization. How do these riskadjusted behaviors impact economic efficiency? A key premise of the research I describe here is that owning assets that provide a comparative advantage (like fertile land) is not sufficient to explain patterns of unit productivity. Instead, I argue that the risk environment around that asset must be taken into account. I show that different risk environments affect how people evaluate the utility of a particular economic activity— particularly the time horizon over which they perform these calculations. Many people in high-risk environments are less willing to invest in activities that impose an immediate burden on their safety in exchange for an uncertain future return— in other words, they choose immediate physical safety over potential long-term economic return. In order to self-protect, individuals farm less efficiently: they limit themselves to small amounts of land or vetted safe areas, and they do not invest in potentially dangerous farming techniques that could raise their yield. These risk-averse strategies seem to be the most viable means of securing personal safety, but they come at the cost of unit productivity and growth. The short time horizons of these calculations produce long-term economic impacts: as financially insecure farmers try to procure stable income in the short term, they avoid risky investments or accumulate excessive precautionary savings, both behaviors that prevent them from making economic decisions that could lift them out of poverty in the long term (Dercon 2002; Wood 2003). By taking measures to protect themselves from physical risks, farmers exclude themselves from development opportunities that would pay future dividends. For example, they do not adopt more intensive agricultural technologies (such as driving tractors or applying the extra fertilizer and water needed for highyield rice), both of which would require increased exposure to

i n s u s p e c t t e r r a i n 55

unexploded ordnance; they also choose not to diversify into other land-based sources of revenue such as rearing livestock or fishing, both of which would involve traversing swaths of land to chase cattle or access rivers. Because they “opt” out of development opportunities and limit their portfolio of productive assets, farmers are more likely to remain in a state of arrested economic development. My principal hypothesis, then, is the following: when households face threats of violence from their land, they minimize their risk of injury or death by farming less efficiently. Specifically, I posit that on bombed, high-fertility soils with significant unexploded ordnance, farmers respond to the danger by cultivating less of their land and investing less in agricultural capital, which reduces their overall crop production and profit margin. These households also cannot safely engage in other land-based livelihoods to supplement their income, and they are consequently forced to migrate for seasonal work or to rely on children’s labor for supplementary income; these farmers’ self-protection strategies may thus result in higher rates of poverty. In other words, assets that typically ensure a high rate of return are not the areas with the most economic activity; instead, economic activity clusters in places with lowercapacity resources that are safer to use. This theory reveals a new sphere of marginality among individuals who seem asset-rich; these individuals are unable to fully capitalize on those assets when they are at risk of violent harm, and this reduces the assets’ economic viability.

How War Changes Land It requires an explosive ordnance disposal team to tell us the technical name of the bombs that Vannary found earlier: they are BLU-42 bomblets. The US military designed the BLU-42 as

56 c h a p t e r 2

a scatterable landmine. It is painted olive drab with no markings in order to blend into the ground. When the bomblet makes contact with the ground, it releases six spring-loaded tripwires that can trigger an explosion. These cluster bombs were released by the hundreds, and the leftovers tend to be found in groups; when you find one or two bombies, you can be fairly sure that others will be close by. Cluster munitions were designed to inflict maximum damage by packing hundreds of small individual bombs into one mother pod. The concept of having a large mortar projectile break apart midair and dispense smaller lightweight submunitions is not new; Leonardo Da Vinci is credited with designing the first cluster munition during his time as a military engineer for Duke Sforza in Milan: among his drawings of lances, chariots, catapults, crossbows, and guns, he included a drawing of large, spherical mortar projectiles bursting out of a cannon and releasing small exploding pellets from behind clouds of smoke. By the time the US military designed its first cluster bombs for use in World War II, American arms manufacturers had figured out how to incorporate incendiary explosives, and the bombs started fires when they hit the ground. In 1942, these early incendiary bomblets were dropped on Tokyo in the first American bombing on Japan; in 1945, more “advanced” napalm-filled cluster bombs (BLU-1/B) were dropped on Tokyo, starting the infamous fires that decimated entire neighborhoods and killed as many as 100,000 people. After World War II, arms designers created a new type of cluster bomb, BLU-3/B, which was designed to kill or injure people rather than to damage buildings and equipment. Once ejected into the sky, the bomblet’s rounded yellow body hangs below six drag vanes, making it look like it is raining tropical fruit (within the military, this bomb was known as

i n s u s p e c t t e r r a i n 57

the “Pineapple”). When the BLU-3/B hits rock, it trips a fuse that releases 250 high-velocity steel pellets; it kills by essentially launching buckshot in all directions. Military strategists found that cluster bombs were well suited for large-scale land attacks—for instance, on areas of suspected enemy activity, where the targets were not necessarily clear and visible. The first large-scale missions using these types of antipersonnel cluster bombs took place during the Korean War, when the Air Force dispensed the bomblets over suspected North Korean supply lines (Ismay 2019). During the Vietnam War, more than two dozen different models of cluster munitions were used to disrupt communist supply routes and deter surface-to-air missile crews from attacking US warplanes in the sky. When a pilot releases a payload of cluster bombs, he is actually releasing a bomb dispenser, a long torpedo-like plastic jacket with fins, called mei in Khmer—literally, the mother. Several feet above the ground, the casing opens and ejects the payload, releasing as many as six hundred individual bomblets, each with its own explosive charge and fuse.13 As the bomblets get caught in the airstream, they disperse further, covering an area the size of two to three football fields, free-falling in a ballistic trajectory. Far from a surgical operation, cluster bombing is designed to spread the bombs over a wide berth. Although President Nixon assured the public that “we have scrupulously observed the twenty-one-mile limit [around the uninhabited jungle perimeter] on penetration of our ground combat forces into Cambodian territory,” the bombings in Vietnam, Cambodia, and Laos used this new, untested generation of bombing technology on unfamiliar terrain, and local residents reveal that the American military wrought significant destruction to civilian homes, schools, hospitals, and temples (Uk 2016; Zani 2019; Padwe 2020).

58 c h a p t e r 2

According to munitions experts, cluster munitions are exceptionally hazardous because they have such a high dud rate: different sources estimate that between 10 and 30 percent of these bombs do not explode on impact. This failing was so widespread that it became a design feature, with certain cluster bombs being used as air-delivered landmines, also known as wide-area, antipersonnel landmines (WAAPM). From 1964 to 1973, the US Air Force dropped nearly 350 million bomblets in Southeast Asia (Ismay 2019). The fate of the leftover bombs (likely between 35 million and 105 million) is unknown. While some bomblets were collected by the National Liberation Front and used as charge material for their own improvised explosive devices, and others likely became inert, many more continue to sit in the ground, pitted with rust, waiting to go off. Explosive ordnance disposal technicians see these aged bombs as highly unstable. Because the stab-sensitive detonators are sensitive to changes in pressure, weather, and humidity, the bombs can theoretically go off at any time. Post–Vietnam War, the borderland jungles of Cambodia, Vietnam, and Laos now hold one of the densest concentrations of unexploded cluster munitions in the world. Recent estimates in northeast Cambodia suggest that as many as eighty thousand air-dropped cluster munitions, containing 26 million bomblets, remain unexploded. Dud bomblets have killed or injured an estimated eleven thousand people (one-third of them children) across these three countries (Ban 2016). The United States alone has spent more than $160 million on Cambodian demining operations (Hunt 2020). These figures track with what anyone in Cambodia can observe: the bomb disposal hotline gets dozens of calls through the week. The on-call team can only handle one to two requests per day. Between the travel (the majority of roads are still dirt) and the painstaking safety

i n s u s p e c t t e r r a i n 59

protocols, each trip to identify and detonate potential ordnance takes about four to eight hours, door to door. Meanwhile, farmers keep finding bombs. Sometimes they mark the location with sticks; other times, they try to relocate them to a safer spot, farther away from houses and human activity. Occasionally, a bomb detonates by accident. A bomb accident that killed three men happened right down the road from Vannary’s house. The men were poking around a suspect object, trying to figure out if it was a cluster bomb, a piece of scrap metal, or a rock.14 Near Vannary’s village, a generation of children has grown up in the unpredictable borderland, surrounded by lurking cluster bombs. Sophal is twenty-three now and still lives in his grandmother’s house. He explains that each year bombies have a way of “floating” up to the surface of the land; when the topsoil softens during rainy season, it reveals stones and objects hidden underneath.15 Sophal’s grandmother lived through the American bombing. She had run away from her house on the road and fled to the forest. Upon her return, she found leftover bombies everywhere—in the forests, rivers, fields—and they were easy to set off.16 A cow walking on the field could accidentally detonate a bomb. Every so often, a bomb goes off by itself. Lokkru Kheang, who has taught at the elementary school for forty years, says that during certain weeks (it’s impossible to predict exactly when, for no one knows what particular conditions trigger it), he hears bombs going off on a daily basis.17 Here, explosions are firmly lodged into the human psyche. My Khmer interpreters, who lived only thirty miles away in the provincial capital of Banlung, soon felt that they were in another country. This is, I think, because people’s attitudes here are different, just as the agricultural practices are different—both shaped by the constant looming presence of the bombs. Unlike other

60 c h a p t e r 2

parts of rural Cambodia, where bombs fell and exploded, for the most part, here cluster bombs are so common that residents feel constantly in danger, and this creates a major (and often under-appreciated) behavioral dynamic that ripples through all forms of rural life. When Vannary peers over the crest of her garden, she looks down on beds of grass, the fertile soil overgrown with cruelly beautiful but useless vegetation instead of the cashew trees in orderly rows, lined with mango trees, that she once dreamed of. In other parts of the country, homesteaders were able to fulfill the dreams that were promised by the open market economy and green revolution: settle land, farm commercial crops, and reap economic reward. But here, the odds against the farmers are too high. The new risk environment created by leftover bombs reordered their priorities. These settlers cannot maximize their land usage and experiment with new crops, for they are stuck navigating the immediate physical risks posed by war that has sedimented its legacy in their soil. Their version of good farming practices has shifted dramatically: don’t go into abandoned fields, be careful in forests and overgrown areas, keep a light footprint, beware of stray objects. The bomb effects that Vannary experiences are far outside the scope of military research, which (like the academic research mentioned in chapter 1) focuses on quantifying the immediate damage caused by bombing. When US military officers examine what happens after the release of a piece of ordnance, they use the term “weapon effects”: the aggregate damage of the blast wave, metal fragmentation, fire, and radiation. On the ground, targeteers and intelligence officers conduct bomb damage assessments (also known as Combat Assessments or Battle Damage Assessments) that describe the levels of destruction wreaked on buildings and critical infrastructure,

i n s u s p e c t t e r r a i n 61

usually located in urban spaces. The reports generally tally the number of vehicles, houses, commercial properties, factories, schools, and hospitals lost, as well as the number of civilians and military personnel killed. These assessments began during the Vietnam War, when they combined information from pilot reports, gun camera film, follow-up reconnaissance photographs and reports, and human resources intelligence from behind enemy lines (Rauch 2004). By 1990, during Operation Desert Storm, the technology for measuring bombs’ destruction had advanced: highspeedsurveillancetechnologylikesatelliteimagery,surveillance drones, and improved computing and communication systems helped intelligence analysts deliver more detailed and timely damage assessments (Sopko 1999). This archive of information about bombing effectiveness continues to inform bombing tactics today. Now, when weapons specialists conduct a munitions assessment, they estimate the effectiveness of the weapon against a target then measure the actual outcomes and compare them to their estimates (Sopko 1999). These assessments recommend changes in the weapons system and tactics in follow-up strikes so the mission will be better able to achieve its immediate objectives, whether that involves taking out an enemy, destroying infrastructure, or diminishing enemy morale. These assessments probe whether or not particular weapons correlate with mission success. The military is inherently outcomeoriented, so it is not surprising that when collecting information about their airstrikes, the military is more focused on what happens when bombs detonate than in tracing the effects of failed bombs. Yet as my notes about Vannary’s village took shape, the focus solely on exploded ordnance nagged at me. I began to draw

62 c h a p t e r 2

from new ideas about how undetonated bombs can impact daily life more than actual detonations themselves. As I observed, one lone piece of unexploded ordnance can change the behavior of many individuals without killing them. Most people in Vannary’s village will not get hurt or die from a latent bomb explosion, yet, as I have shown, they all engage in bomb-avoidance strategies that carry costs: in Vannary’s hamlet, tilling the soil before planting and after harvest commonly uncovers bombs, so she avoids crops that require tilling (like potatoes, which are profitable) and plants no-till crops instead (like soybeans, which are less profitable). She uses hand tools instead of machines, so she can “keep [her] eyes peeled,” telling me, “I cut grass with a machete. Lightly. . .”18 She also limits the amount of land she works: “I only work on a small piece of land,” she says, “just planting here and there.”19 Most of her agricultural land lies fallow as, year by year, Vannary watches it turn back into forest. This phenomenon, in which the threat of violence plays an outsize role in conditioning many people’s behavior in the surrounding environment, has been examined in other academic disciplines. Evolutionary biologists use the term “fear ecology” when prey take active measures to avoid predators. Criminologists who study Rio de Janeiro’s favelas have noticed that one or two crimes unleash a sphere of fear throughout the community, causing the majority of neighborhood residents to avoid walking on certain streets or being outside at night (Perlman 2006). I am examining a similar phenomenon, but what I specifically want to get across is that these unexploded bombs play an important role in the construction, the endurance, and the diffusion of fearful behaviors not only among individuals but among entire groups of community members—and across generations.20 This ripple effect cannot be understood through

i n s u s p e c t t e r r a i n 63

Vannary’s experience alone. We must draw on my ethnographic analysis of other local residents, which documents an entire range of behavioral adjustments. My observational research of other nearby farming families has uncovered the role of fear ecology in especially fertile areas, which (as I have explained) are more likely to host unexploded bombs because of the characteristics of the fertile soil. One of the farmers I talked to is Ros Mey, a large landowner who lives in a nearby district. Ros Mey, dressed in military fatigues and a tank top, tells me he remembers the aerial attacks, particularly how “the cluster bombs didn’t really explode—just a few did. . . . They also dropped the B-52s [the MK-82 general purpose bomb, which locals nicknamed after the B-52 bombers], and a lot of them are still around.”21 After the bombing, Ros Mey became a Khmer Rouge soldier and was relocated west to Banteay Meanchey Province, where the soil was different.22 He explains,“in Banteay Meanchey, the land is not black and is dry. In the dry season, it gets cracked. . . . Over there, it was harder.” By comparison, “[m]y land here is black and doesn’t crack.”23 This same suspicion of fertile soil is expressed by Sam Oeurn, a village elder. He survived the US bombing, joining the communist revolution led by the Khmer Rouge after a year of endless air raids; he knows that bombs were dropped throughout the river basin where most of the villagers live (“when I walked alone near the river [after the bombing], I was afraid of ghosts. . . . There used to be more than four or five hundred people who lived here”).24,25 He connects the area’s soft, fertile soil to the danger of bombs. He shows me that his soil is good (“look over there; I planted that tree a year ago, and it’s grown a lot”), and then he tells me that ten years ago, a cow had walked across his field and accidentally detonated a bomb, and that he has even found a bomb close to his house—“here,

64 c h a p t e r 2

here,” he says, pointing about fifty feet away from where we are speaking.26 Other residents of Ratanakiri province also connect bomb failure to local ground conditions. Theara Thun is a sixtyyear-old farmer and construction worker who lives near the Vietnam border. He is originally from Prey Veng, a southeastern province about three hundred miles away. He was still a child during the US bombing, but he recalls the surrounding forest and the village being attacked by B-52s and “helicopters with mounted guns.”27 The village was near a military base, built by General Lon Nol, that was sited in the neighboring commune, so “the villagers were often fighting with Lon Nol’s army,” he explains. “They [the state army] were shooting at the Khmer Rouge, but the thing was—the Khmer Rouge were living among the villagers.” When he was living in his hometown in Prey Veng, he says that he “never saw bombs that were still unexploded and in the ground. Never.”28 He reflects on why, telling me how the land in Prey Veng is different from the land in Ratanakiri. “The soil in Ratanakiri is nutrient rich; it’s naturally muddy and wet. Since a lot of the land used to be hardwood forests, it’s like there’s already fertilizer in the ground. In Prey Veng, farmers have to put fertilizer in the soil.”29 When he moved to Ratanakiri four years ago with his son and daughter-in-law, he immediately found work as a day laborer helping farmers tend their land. He was shocked by the prevalence of unexploded bombs here, and both he and his son have seen people killed by cluster bombs; he recounts one such deadly explosion on a job trying to remove a large tree.30 Other respondents’ stories of working in different parts of Cambodia similarly show that land fertility varies dramatically, not only between provinces but also between neighboring villages in the same province. During the US bombing, Bun Tek

i n s u s p e c t t e r r a i n 65

lived on one of the main rivers in Ratanakiri, where he recalls the land being rocky and relatively difficult to plant. One advantage, though, was that “there were not many kinds of this bomb [pointing to a cluster bomb], but there were a lot of big fragments. They all exploded.”31 After he married, he moved near his parents-in-law, to a house in Vannary’s district. He notes that in his new district, the soil is better; “it’s black, and it has nutrients. It’s dark like charcoal.” But “they [the US] dropped bombs over here . . . so, yes, there are also a lot of cluster bombs here.”32 He, too, connects good soil with unexploded ordnance. Lom and Leang Pan, the married couple from the first chapter, also see their land as dangerous; it is the village where General Lon Nol’s military base was located. Lon Nol built “a hospital, barracks, and hiding places” near the battlefield where they fought the Khmer Rouge and Vietnamese communists.33 In turn, “the US dropped a lot of bombs in this area,” says Lom.34 They avoid certain areas “because [they’re] scared of bombs.”35 While these respondents cannot pinpoint the specific locations of bombs or measure the density of bombs in particular areas, they seem to understand that if bombing had taken place in the area, they should suspect unexploded ordnance in fertile soil. They develop an informal coding procedure to identify safe versus dangerous areas, one that combines regional estimates of bombing history with local knowledge of soil fertility. As I have mentioned, many residents minimize risk by farming less efficiently. One common strategy is to farm a fraction of total holdings. To protect themselves, Lom and Leang only plant the “five or six meters outside of the cluster bomb area.”36 Another farmer in a more remote district describes similar changes; he says, “I only work this one strip of land. If an NGO doesn’t come to take all the bombs out, I won’t move

66 c h a p t e r 2

anywhere else.”37 A village chief describes a similar pattern of behavior in his constituency. Because the villagers worry about cluster bombs, “we don’t dare work on a large scale.”38 When Sam Oeurn, the village elder, realized that he had limited space on his field to plant safely, he chose to plant a few cashew trees rather than grow cassava for commercial production. “If I planted cassava, I’d have to break apart my topsoil,” he explains.39 By planting a nut fruit tree, he only needed to dig into the earth once; then, he and his wife just collect the fruit from the trees. Concerned farmers also rely on safer but loweryield farming methods. For instance, one farmer uses hand tools instead of machines. She tills with a stick and brings a basket so she has a place to collect bombs and scrap metal.40 A retired bureaucrat from the Office of Economic Development also plows and seeds his field by hand—in large part because “my neighbors tell me to farm carefully and to stay safe.”41 In this view, the possible future returns of heavy machinery do not justify the immediate outlay of personal security. In sum, my interviews provide proof of concept that farmers find unexploded ordnance in more fertile areas and that the developmental effects of unexploded ordnance persist through time, as farmers use less efficient agricultural methods to farm more safely. But do these one-off cases represent larger inequalities in Cambodian development? It is difficult to test whether leftover bombs act as a major obstacle to economic growth across locations. People’s relative exposure to unexploded ordnance is often difficult to gauge. There are concerns about omitted variables biasing the results, especially since bomb deployment—the treatment variable— is likely driven by a particular region’s security, political, and economic conditions. This makes it difficult to tease apart the causes of local rates of underproduction, which could come

i n s u s p e c t t e r r a i n 67

from immediate bomb explosions, leftover ordnance, or prebombing factors like terrain or prewar economic capacity. To overcome these methodological challenges, I created a historical dataset of US Air Force sorties flown over Cambodia from 1965 to 1973. In the dataset, I identify the location and amount of ordnance dropped and compare this information to the agricultural output and household economic activities from the 2012 Cambodia Socioeconomic Survey while controlling for prewar economic and geographic conditions. Furthermore, I leverage the quasi-random variation in unexploded ordnance density—since impact fuse failure was “as good as random,” conditional on soil fertility—in my statistical analysis. The gold-standard approach to identifying causal effects typically relies on randomized assignment to the treatment condition, because this design relies on the fewest and weakest assumptions. I incorporated this principle by randomly assigning exposure to unexploded bombs to some farmers but not others: unexploded ordnance stems from flaws in engineering and design, and detonation was largely determined by idiosyncratic characteristics at the target’s surface, and so the naturally occurring selection process of who is exposed to unexploded ordnance mimics the random assignment that you would observe in an experimental design.42 Because of the quasi-random assignment, exposure to fear condition (unexploded ordnance) is orthogonal to the unobservable individual capacity to be a good farmer as well as to unobservable inclinations to settle on a battlefield (e.g., the extent to which someone lacks information or feels an intrinsic pull to live in a dangerous, adventurous area). As a result, I can more reasonably infer that any effect of leftover bombs on agricultural productivity is due to that treatment and not due to preexisting differences in risk tolerance, information, farming ability, or economic status.43

68 c h a p t e r 2

Data and Estimation I first collected data on US Air Force payloads. I use airstrike placement, conditional on soil fertility, as a proxy for bomb failure. This approach is not a perfect solution, as not everyone within the target zone is equally susceptible to encountering bombs, given that fields within the same drop zone will likely vary in the density and placement of ordnance. However, the insecurity associated with the risk of detonation is not the same as the realization of risk—that is, the lethality and injury rate from leftover ordnance. Gruesome events stick in the memory, leading large segments of the population to exaggerate the likelihood of their occurrence (Thaler 1983) and to engage in precautionary behavior in excess of the actual hazard (Fafchamps 1992; Jalan and Ravallion 1999). Expanding far beyond the explosion radius of each bomb, the sphere of fear transcends the immediate vicinity, formidably restricting economic activity, even for those not directly in harm’s way. The bombing data are derived from the Southeast Asia Air Combat Database, which captured daily air combat information on the Vietnam War. These data detail the coordinates of each released piece of ordnance, date of release, type of ordnance (e.g., 750-pound general purpose bomb, BLU-32 firebomb), and number of bombs/weight of the payload. The spatial distribution of the payload drops is presented in figure 2.1. I mapped the payloads onto village hamlets over the full territory of the country. I drew a circle with a five-kilometer radius around each village center. This provides a reasonable estimate of where households would locate their fields; anthropological research indicates that farmers keep their fields within close walking proximity to the hamlet so they can monitor

i n s u s p e c t t e r r a i n 69 15°N

Laos Thailand 14°N

13°N

12°N

Vietnam 11°N

N 10°N

Legend 1 US Payload Drop 300 km

102°E

103°E

104°E

105°E

106°E

107°E

figure 2.1. Payloads dropped by the US Air Force in Cambodia from 1965–1973.

for animals and thieves (Ebihara 1971), and my interviews confirmed this tendency. I use this procedure to create a variable that counts the tonnage of bombs within the close walking radius of each hamlet.44 Even though the village centroids do not follow the actual shape of individual villages, my interviews suggested that farmers select fields based on proximity to the hamlet, not the village boundaries, which fluctuate based on local interpretation and are not recorded in the national land register.

108°E

70 c h a p t e r 2

The soil fertility indicator is drawn from the Cambodian Agricultural Research and Development Institute’s 2008 soil database, which categorized the country’s land into three levels of soil productivity (low, medium, and high) based on the main limiting factors for growing rice. Since the key components of soil—minerals, organic content, water, air, and living organisms—are derived from rocks that are physically weathered into smaller pieces, it takes more than five hundred years for one inch of topsoil to form. The 2008 measures of soil fertility are thus a reasonable reflection of soil fertility in decades prior. The land fertility at the village centroid represents a best guess for the soil fertility of each household’s fields. While a more accurate indicator would use the boundaries of each field, this information is not recorded by the Ministry of Land Management (the bureau in charge of maintaining cadastral maps), so coding fertility at a more granular level is tricky without substantial local knowledge. To capture contemporary development, I draw on the 2012 Cambodia Socioeconomic Survey, a nationally representative survey of 3,840 households in 384 villages. The survey was developed to provide a clear picture of development and welfare for the smallest economic units: households, their fields, and their individual members.45 While prior studies rely on measures of individual behavior aggregated at the village or district level, drawing conclusions about microlevel behavior from macrolevel data (Miguel and Roland 2011; Dell and Querubin 2017), this survey provides the opportunity to analyze each unit of production. Furthermore, the survey’s breadth of topics allows me to measure the legacy of bombing on a wider range of outcomes: these include agricultural productivity variables that are central to economic growth models (particularly in the developing world) and variables related to quality of life and

i n s u s p e c t t e r r a i n 71

human health, such as food insecurity. For each household plot, I calculate the amount of rice produced in the past year. Rice functions both as a food crop and a cash crop. Rice is a mainstay of the Cambodian diet, providing more than half the daily calories of the poor, and surplus production is typically sold at the local market or mill. Since rice quality varies and quantity depends on the size of the paddy, I measure rice production in US dollars per square meter of paddy. I also collect data on the amount of land that was cultivated for rice compared to the actual size of the field. I use both the absolute measure (in square meters) and the relative measure (percentage of total field) as indicators of agricultural efficiency. Other agricultural behaviors operate at the household level. For instance, household members decide whether to buy modern (but heavy) tools that would increase productivity. The survey provides data on the household’s capital assets, including specific data on farm machinery; it records the number of tractors, bulldozers, threshers, semitractors, and water pumps owned by each household. I use a household’s total number of farming machines as an indicator of higher investment in agriculture. I also measure productivity in terms of surplus; household members pool rice harvest across paddies and decide how much to consume or sell, and to measure surplus, I calculate the percentage of the household’s rice crop that was sold to market. Since crop surplus has traditionally been a major source of seasonal income, I also collect the survey data on each household’s reported cash income in the past year. To examine whether households exposed to unexploded ordnance work in nonfarming sectors, including other types of land-use production (fishing, hunting, livestock, poultry) or work in other labor sectors (whether as a second job or a permanent move away from farming). First, I use the survey items on alternative land-based

72 c h a p t e r 2

livelihoods; for each household, the survey records the past year’s income from fish and from livestock and poultry. Second, I test whether unexploded bombs limit local economic activities by evaluating whether family members leave home for safer, more profitable types of work in distant labor markets. To evaluate local migratory patterns, I collect individuallevel survey responses on the number of weeks each household member has been absent from home. I also consider the effect of unexploded bombs on childbearing patterns. Given the long treatment window, parents may try to increase the family labor force by having more children; alternatively, the reduction in farming caused by unexploded bombs may provide parents more time to have and care for children, and the eventual income from grown children may offset the losses in the field. For each household, I calculate the number of family members younger than eighteen years old. Skeptics might question whether the treatment assignment is truly random, since not all parts of the country received similar levels of bombing. Even though the failure rate of bombs was exogenous to local economic characteristics, proximity to Vietnam—where all of the Ho Chi Minh trails eventually ended—did influence the general level of bombing. Because the border was bombed more, places closer to the border suffered more bombing; I therefore control for kilometers to Vietnam from the center of each village. I also control for the location of area transit points that existed before the bombings, as the decision to drop payloads was influenced by the presence of roads and waterways that were believed to be used by the Vietnamese communists. To do so, I employ a series of military topographic maps of Cambodia (1962–67), developed by the US Army Map Service for the purpose of military land navigation. The Army Corps of Engineers used aerial photographs

i n s u s p e c t t e r r a i n 73

to tag information on household locations, crop boundaries, dirt roads, and river tributaries to identify potential targets— difficult to do in a remote setting that lacked friendly support units. I used this information to construct an indicator for road density by calculating the meters of road within five kilometers of the village center. A binary variable for waterway represents whether a river, lake, or tributary is located within the five-kilometer buffer around the village centroid. I also control for the fact that bombings could have been focused on areas with high agricultural production or larger populations, which might have been seen as providing cover for the National Liberation Front; I adjust for this potentially confounding influence by controlling for prebombing village development. Specifically, I estimate the percentage of rice fields prebombing and count the number of houses within each five-kilometer village buffer. In all estimations, I add fixed effects for the five price zones identified by the Cambodia Development Resource Institute (2008) in their survey of the Cambodian rice market. Price zone fixed effects are useful for these purposes because the analysts developed them inductively, based on the four agroclimatic zones and Phnom Penh (which has rice fields in the suburbs of the capital). By including price zone effects, I adjust for the possible confounding influence of unobserved shifts in supply and demand, as well as regional variation in transport and storage costs, wages, and input costs.

Legacy Effect on Agriculture The results of the regressions demonstrate that a farmer on high-risk land (bombed, high-fertility soil) will not produce as much rice as a farmer on safer land (unbombed, high-fertility

74 c h a p t e r 2

table 2.1. Conditional impact of bombing on agriculture (OLS, unit: field) Outcome variable: Field Productivity (1) Area farmed % total field Bombing Bombing x Low Fertility Low-Fertility Soil Constant Observations Price zone fixed effects Pretreatment covariates Adjusted R2

–0.01∗∗∗ (0.003) 0.01∗∗ (0.004) 0.07 (0.05) 0.22∗∗ (0.09) 3,636   0.17

(2) Area farmed m2 –428.82∗∗∗ (41.57) 393.290∗∗∗ (58.77) –5,162.93∗∗∗ (760.58) 8,694.95∗∗∗ (1,482.31) 3,636   0.14

(3) Rice harvest USD/m2 –0.02∗∗∗ (0.005) 0.01 (0.01) –0.15† (0.09) 0.04 (0.17) 3,636   0.01

Notes: Reference category: High-Fertility Soil. Standard errors in parentheses. Bombing is transformed log(pounds + 1). † p