Social Networks and Migration in Wartime Afghanistan

Social Networks and Migration in Wartime Afghanistan Kristian Berg Harpviken

Social Networks and Migration in Wartime Afghanistan

This page intentionally left blank

Social Networks and Migration in Wartime Afghanistan Kristian Berg Harpviken Centre for the Study of Civil War (CSCW) International Peace Research Institute, Oslo (PRIO), Norway

© Kristian Berg Harpviken 2009 All rights reserved. No reproduction, copy or transmission of this publication may be made without written permission. No portion of this publication may be reproduced, copied or transmitted save with written permission or in accordance with the provisions of the Copyright, Designs and Patents Act 1988, or under the terms of any licence permitting limited copying issued by the Copyright Licensing Agency, Saffron House, 6-10 Kirby Street, London EC1N 8TS. Any person who does any unauthorized act in relation to this publication may be liable to criminal prosecution and civil claims for damages. The author has asserted his right to be identified as the author of this work in accordance with the Copyright, Designs and Patents Act 1988. First published 2009 by PALGRAVE MACMILLAN Palgrave Macmillan in the UK is an imprint of Macmillan Publishers Limited, registered in England, company number 785998, of Houndmills, Basingstoke, Hampshire RG21 6XS. Palgrave Macmillan in the US is a division of St Martin’s Press LLC, 175 Fifth Avenue, New York, NY 10010. Palgrave Macmillan is the global academic imprint of the above companies and has companies and representatives throughout the world. Palgrave® and Macmillan® are registered trademarks in the United States, the United Kingdom, Europe and other countries. ISBN-13: 978–0–230–57655–1 hardback This book is printed on paper suitable for recycling and made from fully managed and sustained forest sources. Logging, pulping and manufacturing processes are expected to conform to the environmental regulations of the country of origin. A catalogue record for this book is available from the British Library. A catalog record for this book is available from the Library of Congress. 10 18

9 17

8 16

7 15

6 14

5 13

4 12

3 11

2 10

Printed and bound in Great Britain by CPI Antony Rowe, Chippenham and Eastbourne

1 09

Contents Preface

vii

Glossary

x

Acronyms

xii

Maps of Afghanistan in the Region and the Main Fieldwork Area

xv

1 Introduction Social networks in times of crisis The study of wartime migration Wartime migration in Afghanistan Two villages of Herat Structure

1 2 4 6 9 11

2

Social Networks in Wartime Migration Social networks Flows in networks Network structures Change Conclusion

13 14 19 26 38 43

3

Escape Decisions Security and escape Material resources and escape Information and escape Conclusion

46 50 61 69 74

4

Integration at Exile Security and exile Material resources and exile Information and exile Conclusion

77 81 88 94 96

5

Return Decisions Short-distance repatriation from Pakistan Security and return v

99 103 107

vi

Contents

Material resources and return Information and return Conclusion

117 125 132

6

Reintegration at ‘Home’ Security and reintegration Material resources and reintegration Information and reintegration Conclusion

135 140 151 160 163

7

Conclusions Escape and return decisions Integration and reintegration War, migration, and the transformation of networks Studying wartime migration

167 168 173 177 181

Appendix: Researching Migration in War

184

Notes

191

Bibliography

199

Author Index

222

Subject Index

223

Preface This book was brought to completion, thanks to the supportive and stimulating environment where I have my daily work, at the International Peace Research Institute, Oslo (PRIO), and its Centre for the Study of Civil War (CSCW). At PRIO, I have enjoyed tremendous institutional and collegial support. I consider PRIO my intellectual home in Norway: the best possible environment for pursuing my research interests. This book is the ultimate manifestation of a curiosity that was triggered by my observation, while traveling in Afghanistan in the early 1990s, that villages that appeared to be situated similarly in relation to the war had nevertheless fared very differently. For example, where one village was totally depopulated, a neighboring one remained vibrant throughout the conflict. The explanation was not to be found in simple geographical or political factors, which made me suspect there was something about collective decisionmaking that was not being taken into account in regular explanations of flight. Part of the problem is that nobody had conducted research on the ‘non-displaced’ – those who stayed on despite the threat. I was already interested in Afghan political mobilization. And, inspired by the idea of ‘refugee warrior communities’, launched by Aristide Zolberg, Astri Suhrke, and Sergio Aguayo in their 1989 book Escape from Violence, I increasingly realized how crucial the interplay between politics and displacement was – not only in the sense that politics drives migration, but also in the sense that displacement (and the lack thereof) fundamentally alters the conditions for political mobilization and may serve to motivate it. In a related manner, I was struck by our seeming inability to reconcile the image of displaced people as victims of factors beyond their control with the realization that they are also agents with their own types of resources, capable of making their own choices. Keeping all of these early inspirations in mind, then, this book explores the collective dimensions of migration decisions and integration processes. What you see here is the result of a journey that has been exciting on many levels – intellectually and geographically, as well as socially and culturally. Throughout this journey, I have engaged with and enjoyed the support of literally hundreds of remarkable individuals, each contributing in different ways. Only some will be named here, but I remain grateful to you all. What has impressed me the most has been witnessing the vii

viii

Preface

coping ability of individuals living under the most difficult conditions, suffering unimaginable losses, as well as the tremendous resources that lie in social networks and social solidarity. This work is only a station on a journey that I know I will continue to enjoy. I first of all want to thank all those people in Afghanistan who have given their time and energy to discuss with me, to respond to my questions, and to invite me to develop insight into their lives. For some, this may have been painful, but I also hope that my interest and understanding have been comforting. You have been granted anonymity and will not be named in this work, but you will know who you are. Some people have been particularly important for this project, having read or commented upon the manuscript in earlier iterations: Andrew Abbott, Richard Black, Jørgen Carling, Scott Gates, Cindy Horst, Yngve Lithman, Tormod Lunde, Shahrbanou Tadjbakhsh, and Stein Tønnesson. At various stages in the process, I have also received generous comments on background papers and draft chapters from a number of people: Bayo Adekanye, Grethe Brochmann, Chris Buckley, Ronald S. Burt, Victoria Ingrid Einagel, Fredrik Engelstad, John F. Padgett, Charles C. Ragin, John Scott, Dan Smith, Astri Suhrke, Charles Tilly, and Aristide Zolberg. All of you launched ideas that have been with me to the end, although not necessarily in a form that you will immediately recognize. A number of institutions have supported this research financially. First of all, the Department of Sociology and Human Geography at the University of Oslo granted me a doctoral scholarship, which allowed me to conduct the theoretical and empirical research that lies at the core of this book. I am also grateful to US–Norway Fulbright Foundation for Educational Exchange for a scholarship that allowed me to spend the whole of 1998 as a Visiting Researcher at the University of Chicago’s Department of Sociology. The Research Council of Norway funded my main fieldwork in Afghanistan from March to June 1999. The Nordic Institute for Asian studies provided funds for complementary fieldwork. Furthermore, I am grateful to the Norwegian Ministry of Foreign Affairs and Ockenden International, which have funded related projects that have enabled me to conduct successive return visits to Herat in 2001, 2002, 2003, and 2006. PRIO, and the CSCW, have allowed me the flexibility necessary to bring this work to completion. The Institute for the Study of International Migration (ISIM) at Georgetown University hosted me for a month in October 2008, allowing me to concentrate in the last round of thorough revisions. A number of specialized research libraries, all with individuals that did their utmost to help me identifying relevant sources and getting access to

Preface

ix

the most eccentric documents, have proven extremely important. I have spent days working in each of these libraries: Bibliotheca Afghanica in Liestal, Switzerland; the UNHCR Research and Documentation Centre in Geneva, Switzerland; the ACBAR Research and Information Centre (ARIC) in Peshawar, Pakistan; the Norwegian Refugee Council (NRC) library in Oslo, Norway; and the library of the Afghan Research and Evaluation Unit (AREU) in Kabul, Afghanistan. The fieldwork on which this book is based would not have been possible had it not been for the support of a number of individuals and organizations engaged in Afghanistan. First and foremost, I am grateful to Arne Strand, a good friend and colleague, as well as an astute observer of Afghan affairs, with whom I collaborated during the main fieldwork in Herat in 1999. I am indebted to Karin Ask, also a partner during the 1999 fieldwork, particularly for her insights on gender generally and the lives of women in the fieldwork communities specifically. Both Arne and Karin have allowed me to draw on their data for my own work. I am also extremely grateful to all those who have served as interpreters in the field or as research assistants, including Abdul Jamil, Abdullah, Dr. Adil, Mirwais Wardak, Mohammad Israel, Mohammad Suleman, all of whom did a great job under exceptionally difficult circumstances. In Afghanistan, I am particularly grateful for the facilitation of fieldwork to Barmak Pazhwak and Fazel Rabi (Christian Aid/Ecumenical Office), Daud Dildar and Abdul Khaliq (Agency for Mine Awareness in Afghanistan), Ajmal Shirzai and Eng. Baqi (Ockenden International), Eng. Ahmadi (Norwegian Project Office/Rural Rehabilitation Association for Afghanistan), as well as Haneef Atmar and Mohammed Ehsan (Norwegian Church Aid). Finally, I want to express a deep-felt thanks to family and friends, many of whom are occupied with things very different from mine, but who provide great support and inspiration. In particular, I am indebted to Anna, who can only with difficulty imagine a life in which her dad is not engaged in writing to try understanding the world, and to Kristine, who is always amazingly understanding and supportive. Kristian Berg Harpviken Oslo, 15 November 2008

Glossary Afghani

Alim Amniyat Ansar Arbab Arbaki Basij

Dari Emir Enqelab Faleke Farsi Farsiwan Hawala Hawaldar Hezb Hijra Irani gak Jamiat Jerib Jirga Jui Madrasa

Afghan currency (45,000 afghani roughly equal to 1 USD in 1999 but fluctuating greatly; devaluation of currency by January 2003, new rate is 52.10 afghani to 1 USD by 15 November 2008, and relatively stable) Learned religious leader within Sunni Islam (singular of ‘ulama’) Security (Dari) Hosts (for the displaced: the muhajerin) Village headman recognized or appointed by the government Tribal militia (largely operating in the neighborhood) Tax imposed on communities as compensation for the expenses of nominating, equipping, and maintaining fighters during war (widely used by the Taliban) Persian dialect spoken in Afghanistan Prince, nobleman, former title of the Afghan king (connotates independence) Revolution Site where employers come to recruit day labor Persian language Farsi-speaker Informal system for money transfers, mainly in the Islamic parts of the world Agent in the hawala system Party (political) Exodus or Flight, concept based on the escape of the prophet Mohammed from Mecca to Medina Persons who have adopted an Iranian lifestyle (language, clothing, customs) Society; association Land measure, equal to 0.2 hectares in the government version (but with local variations) Council, consultative assembly (equivalent to ‘shura’, used in Pashtun areas) Channel Religious school x

Glossary xi

Mahalla Maharram

Malik Masjid Muhajer Muhajerin

Mujahed Mujahedin Mullah Nimkora Pashtunwali Pilau Qala Qawm Rupee Sayed Salaffiyya Shah Shahid Shura Talib Taliban Toman Ulama Woleswal Woleswali

Neighborhood in a village or a city Accompaniment to females who are traveling; the maharram is either the husband or a male relative who stands in a sexually taboo relation to the woman Village leader (largely equivalent to arbab) Mosque Displaced person (see below) Displaced people (the term refers to the ‘hijra’, hence to those who have been displaced because of threats to their Islamic faith) Islamic resistance fighter (plural form for ‘mujahed’) A person with rudimentary Islamic training A durable split of the household (Pashtun concept) The Pashtun normative code Rice dish Place of residence, used in reference to both a closed-in family compound or a village Solidary group, based on various foci (extended family; village; tribe; ethnicity; professional group) Pakistani currency (50 rupees roughly equal to 1 USD in 1999) Religious dignitary who claims descent from the prophet Traditionalist interpretation of Islam (originally a modernist Islamic movement) King Martyr Council, consultative assembly Pupils undergoing religious education (literally: seeker of knowledge) Talib in plural Iranian currency (500 toman roughly equal to 1 USD in 1999) Learned religious leaders within Sunni Islam (plural of ‘alim’) District administrator District; also used in reference to administrative headquarter (lowest administrative unit in the Afghan government structure)

Acronyms ACBAR AI AIA AIMS AIHRC AJP AMAA AMAC AREA ARIC AREU BAAG BAFIA BIA CA/EO CAR CMI CPAU CRS DACAAR DANIDA DDR ECHO GCIM GIRI GoP HRW IAFSM ICRC IDP IEA IRC

Agency Coordinating Body for Afghan Relief Amnesty International Afghan Interim Administration Afghanistan Information Management Services Afghan Independent Human Rights Commission Afghanistan Justice Project Afghan Mine Awareness Agency Assistance to Mine-Affected Communities Agency for Rehabilitation and Energy Conservation in Afghanistan ACBAR Research and Information Centre Afghan Research and Evaluation Unit British Agencies Afghanistan Group Bureau for Aliens and Foreign Immigrants Affairs (Iranian government institution) Bureau International Afghanistan Christian Aid/Ecumenical Office Commissioner for Afghan Refugees (Pakistani government institution) Chr. Michelsen Institute Cooperation for Peace and Unity Centre for Refugee Studies (York University, Canada) Danish Committee for Aid to Afghan Refugees Danish International Development Agency Disarmament, demobilization and reintegration European Community Humanitarian Office Global Commission on International Migration Government of the Islamic Republic of Iran Government of Pakistan Human Rights Watch International Association for the Study of Forced Migration International Committee of the Red Cross Internally Displaced Person Islamic Emirate of Afghanistan (Taliban authority) International Rescue Committee xii

Acronyms

IPIS ISAF ISI KLA/UCK LTTE MADERA MFA MoP MoRR NAC NCA NDF NRC NGO NPO/RRAA NUD*IST NWFP OEF OI OSCE OSI PDPA PRA PRDU PRIO ProMIS QSR RCN RRA RRD RSC SNA SRSG SSD

xiii

Institute for International and Political Studies International Security Assistance Force Inter-Services Intelligence Kosovo Liberation Army (UÇK; Albanian: Ushtria Çlirimtare e Kosovës) Liberation Tigers of Tamil Eelam Mission d’Aide au Développement des Economies Rurales en Afghanistan Ministry of Foreign Affairs Ministry of Planning Ministry of Refugees and Repatriation Norwegian Afghanistan Committee Norwegian Church Aid National Development Framework Norwegian Refugee Council Nongovernmental organization Norwegian Project Office/Rural Rehabilitation Association for Afghanistan Non-numerical, Unstructured Data Indexing, Searching and Theorizing North-West Frontier Province (Pakistani province, bordering Afghanistan) Operation Enduring Freedom Ockenden International Organization for Security and Cooperation in Europe Open Society Institute People’s Democratic Party of Afghanistan Participatory rural appraisal Post-War Reconstruction and Development Unit (University of York, UK) International Peace Research Institute, Oslo Project Management Information System (UN Afghanistan) Qualitative Solutions and Research Research Council of Norway Rapid rural appraisal Rural Rehabilitation Department (of the IEA) Refugee Studies Centre (University of Oxford, UK) Social network analysis Special Representative of the Secretary-General Social Services Department (of the IEA)

xiv

Acronyms

TBA TISA UNAMA UNCHS UNDP UNHCR UNICEF UNMACA UNMAS UNOCA UNOCHA UNRISD UoC USCR USCRI USD WFP WHO

Traditional birth attendant Transitional Islamic State of Afghanistan UN Assistance Mission to Afghanistan United Nations Centre for Human Settlement (Habitat) United Nations Development Programme United Nations High Commissioner for Refugees United Nations Children’s Fund United Nations Mine Action Centre for Afghanistan United Nations Mine Action Service United Nations Office for the Coordination of Humanitarian Assistance Alternative acronym for UNOCA (see above) United Nations Research Institute for Social Development University of Chicago US Committee for Refugees (name change to USCRI in 2004) US Committee for Refugees and Immigrants US dollars World Food Programme World Health Organization

Maps of Afghanistan in the Region and the Main Fieldwork Area

Afghanistan and the region

xv

xvi

Maps of Afghanistan

Main fieldwork area in Enjil District, Herat Province, North-west Afghanistan

1 Introduction

What is the role of social networks when people escape situations of war? What are the effects of social networks on integration in exile? And, further down the line, how do networks effect the decision to return – or not to return – home, and what is their role in reintegration at origin? Most generally, how do old networks evolve and new ones emerge as a result of migration in war contexts? The processes most commonly referred to as ‘forced migration’ – but also denoted by such terms as ‘refugee outflux’, ‘mass movement’, ‘exodus’, or ‘streams of displaced’ – are really the combined result of difficult decisions made by numerous human beings trying to make the best possible use of their resources under conditions of extreme constraint and on the basis of scarce information. In making such decisions, people pursue various paths of action, draw on a multitude of social networks, and build new ties as opportunities arise. Hence, social collectives strongly influence how political events beyond the control of any given individual translate into concrete responses. Every individual is affiliated with numerous other individuals through social networks, and those networks are key to who can gain access to various resources, including information, various types of material resources, and security. Networks are not alone in providing access to such resources; but, in situations of war – when the state and its institutions have often broken down, or turned against its citizens – informal networks gain in significance. Networks are dynamic entities, and they are affected by changes to the larger context within which they exist. For example, when a specific group is targeted for eviction or destruction by a revolutionary state, people affiliated with the group will have to choose between denouncing it and giving up their membership, leaving the country, or standing up and fighting for their collective rights (Okamoto and Wilkes 2008). Any of these choices may involve dire consequences for the individuals concerned, and 1

2

Social Networks and Migration in Wartime Afghanistan

the concrete networks in which the idea of the collective is rooted will often be dramatically transformed. In the context of large-scale migration or displacement, networks are particularly likely to be altered – partly because of geographical separation between components of an earlier network, and partly because individuals have to forge new ties in order to adapt to a new location. Networks that existed prior to the migration are not necessarily disbanded, but may live on as essential resources for the individual despite geographical distance. Alternatively, whole networks may be displaced geographically when collectives migrate together and settle in a new location. In the remainder of this introduction, I first discuss the role of social networks in coping with crisis. Next I will briefly review existing studies of war-induced migration, emphasizing their close links to policy formulation. Following that, I give a brief overview of the past three decades of migration in Afghanistan, before introducing the locations where I conducted fieldwork. Finally, I will outline the structure of the book.

Social networks in times of crisis A key argument in this book is that studies of the role of networks in wartime migration should be inspired by the existent literatures on how people cope with disaster. The question stems from the realization that most people depend first and foremost on their own resources and their own personal relationships when seeking to cope with disaster. The immediate inclination for the researcher is therefore to look for cohesive ties, particularly those based on kinship. To resort to the help of one’s kin is a vital coping mechanism for people who lose their security or livelihood. In the 1980s, the assumption that social network density is decisive for a community’s ability to recover after a disaster took hold in disaster research (Oliver-Smith 1986, 1996; Picou, Marshall, and Gill 2004). Researchers pointed out that indigenous coping capacities had often been overlooked by professional organizations in relief and reconstruction (Anderson and Woodrow 1989). By the mid-1990s, the importance of networks for survival during times of crisis had been firmly established. Nevertheless, three central aspects of social networks remain understudied: how people’s networks help them cope, which kind of networks is most effective in various contexts, and what social networks provide. A central aim of the present study has thus been to rectify these deficiencies. Furthermore, the book seeks to overcome a general tendency to focus exclusively on the role of cohesive ties in dense primary groups, and on the benign effects of networks. Here, I will also look at other forms of ties (including bridging ties and power relationships) and explore the possible negative effects of networks.

Introduction

3

Studies of man-made disasters1 – as opposed to natural ones – suggest that they have a fragmenting effect on social networks. In line with this, it is commonly assumed that social networks break down and lose significance in times of war. Yet, this assumption is highly problematic (Duffield 2001: 125). War certainly imposes stress on social networks, which may lead to fragmentation (or even interruption) of ties, but it may also lead to the strengthening of existing relationships or the building of new ties (Wood 2008). As war is often associated with erosion of the state’s authority and destruction of civilian institutions, social networks are likely to gain in importance for people who can no longer expect any support or protection from the state. Under such uncertain circumstances, people tend to resort to their kin and friends for information, economic assistance, and security. Studies examining the impact of war on social networks suggest that the ‘breakdown of trust’ that is often associated with war (particularly civil war) strengthens the significance of ties rooted in family and kinship, and undercuts ties bridging different primary groups (Colletta and Cullen 2000; Harvey 1998). Furthermore, it is also clear that increased geographic mobility (for instance, through involuntary migration) may foster new ties that over time give access to significant new resources – in other words, foster new social networks. Both of these processes may take place simultaneously. The initial assumption by students of disaster response was that crises foster cooperation. This is referred to as the ‘therapeutic community’ hypothesis. However, studies of man-made disasters have discovered a potential for the creation of a ‘corrosive community’ characterized by distrust and a severing of ties (Erikson 1976; Freudenburg and Jones 1991). There now seems to be consensus that natural disasters tend to foster solidarity, whereas man-made disasters corrode social networks (Rosa and Freudenburg 2001). Wars are ‘man-made’ in a more fundamental sense than many of the scenarios that are usually classified as man-made disasters, and it therefore seems reasonable to assume that they also affect networks negatively. However, the issue is not necessarily that simple. Both wars and man-made disasters may have contradictory effects at one and the same time. While some ties are strengthened and become the basis for solidarity, others are weakened or dissolved. The task of the researcher is to differentiate the various ties, to find out how each of them is affected by the enduring uncertainty created by war, and to examine whether certain kinds of networks are better suited to tackling the extraordinary than others. In this book, I will focus on three kinds of resources that prove essential to migration decisions and integration processes: security, money and other material possessions, and information. I will also focus on different types of networks – distinguishing primarily between cohesive ties, bridging ties, and relationships based on power. This work underscores

4

Social Networks and Migration in Wartime Afghanistan

the dynamic interplay between particular resource flows and various network structures, and examination of how the two interconnect lies at its core. In more general terms, it is a particular objective to disentangle the apparent contradictions between the view that ‘war fosters cohesion’ and the view that ‘war crushes relationships’.

The study of wartime migration ‘Refugee studies’ – a term which over time has been replaced by ‘forced migration’, in order to also incorporate the internally displaced – emerged as a field of research only in the 1970s (Malkki 1995a). Since then, it has developed a strong self-identity, with specialized study centers, research programs, and journals. Such institution-building has served to draw increased attention to the problem of involuntary migration. At the same time, it has contributed to compartmentalization, in the sense that studies of war-induced migration take place in relative isolation from other social science studies, even migration studies (of which ‘forced migration’ should be considered a part), but also studies of disaster response, political mobilization, and collective action. One reason for this relative compartmentalization is that research on forced displacement is dominated by a policymaking agenda (Collyer 2005). Not only is it oriented toward developing implications that seem immediately policy-relevant, but its analysis is often based on assumptions and concepts that are derived from a policy discourse (Harrel-Bond 1998; Harrell-Bond and Voutira 1992). Policy formulation in this sector is critical, in that it largely decides who is granted access to scarce resources, such as asylum or humanitarian assistance (Hathaway 2007; Landau 2007). Nonetheless, the research field’s dependence on policymaking prevents it from addressing displacement dynamics more broadly (Harpviken 2006). From an analytical perspective, the largest problem with the concept of ‘forced migration’ is that it presumes a causal relationship, prior to investigation. This is why so much of the existing research focuses on the emergence of the causal factor, such as armed conflict, rather than the mechanisms by which cause (conflict) leads to effect (migration). Future research on forced migration needs to seek more inspiration from other fields of research, such as migration, risk adaptation, and collective action. The strong causal assumption inherent in the very term ‘forced migration’ is problematic, and fruitful theorization will require a conscious response to the general challenge of establishing causality in the social sciences (Abbott 1997a; Marini and Singer 1988). Perhaps most importantly, we will have to move beyond the unicausal inclination, and systematically search for understanding the multiple

Introduction

5

causes of wartime migration (or its absence). Related to this, we need to call for accounts of the mechanisms that link cause(s) and effect(s) across time and space. Through overcoming the strong causal assumption, inherent in the conceptualization of ‘forced migration’ studies, we ultimately will bring back individual action and social ties to this area of study, reaccommodating human agency (Essed, Frerks, and Schrijvers 2004; Horst 2006b; Lubkemann 2008: 15–21). Within studies of displacement, there has been a tendency to compartmentalize, with various stages of the migration cycle – escape, integration in exile, return, and reintegration – being addressed in separation. Over the past three and a half decades, the fluctuating trends in refugee policy have left highly visible marks on the research agenda (Appleyard 2001; Malkki 1995a; Zetter 1988a, 2008). In the 1970s, the last phase of massive decolonization around the globe, the main interest was in studying the causes of departure (Kunz 1973). Initially, the focus was on violence as a direct cause of flight, often in the context of decolonization and wars of liberation. Over time, interest turned increasingly toward studying ‘root causes’, first and foremost in the development sphere, inspired by the emergence of world systems dependency theory (Wallerstein 1974). A dependency perspective, however, is unfit for developing short-term policy recommendations. Starting in the 1980s, attention shifted toward managing displacement as closely as possible to its area of origin, rather than preventing displacement in the first place (Frelick 1995; Hyndman 2003). This completely changed the research focus. Throughout the 1980s, the reception and hosting of refugees (particularly from Eastern Europe) served to reconfirm the superiority of the political system in the West. Since this was the decade in which refugee studies were institutionalized, research interest increasingly turned to the challenge of integrating refugees in exile (Harrel-Bond 1998; Scudder and Colson 1982).2 Inspiration was drawn from development studies, notably from studies of how to assist in establishing a sustainable economy on the local level. The 1990s saw a defusing of Cold War tensions, and states became increasingly occupied with the threat that refugees were seen to pose to their national security (Loescher 1992; Weiner 1992). The United Nations High Commissioner for Refugees (UNHCR) declared the 1990s the decade of repatriation. Voluntary repatriation became the preferred solution, while the earlier ‘durable solutions’ – host-country integration and thirdcountry resettlement – went out of favor (Van Hear 2003a). Research interest turned to issues of repatriation (see, for example, Stein, Cuny, and Reed 1995; Zetter 1988b).3 The tightening of refugee regimes globally also stimulated interest in internal displacement as an issue for both policy and research (Cohen and Deng 1998; Vincent and Sørensen 2001).

6

Social Networks and Migration in Wartime Afghanistan

By the end of the 1990s, another shift of focus took place; this time to the reintegration of returned refugees in their country of origin (see, for example, Black and Koser 1999; Long and Oxfeld 2004; Markowitz and Stefansson 2004). This fits well with broader political concerns about so-called failed states, the devolution of the international norm of state sovereignty, the new principle of the ‘responsibility to protect’, and the general attention to post-conflict stabilization (Weiss 2004).4 In parallel, migration studies have increasingly branched out to look at the role of diasporas and transnational networks, both as a form of life (Horst 2006b; Koser 2007) and in terms of its political impacts (Adamson and Demetriou 2007; Shain and Aryasinha 2006). Although the research agenda in relation to displacement in Afghanistan mainly reflects political developments in that country itself, it has nonetheless followed the general pattern outlined above (Centlivres 1993; Marsden and Turton 2004; Turton and Marsden 2002). In the first years after 1978, researchers asked about the causes of flight; in the 1980s, they turned to issues of integration in the host countries Iran and Pakistan; in the first half of the 1990s, they shifted their attention to repatriation; and, more recently, in the post-Taliban period, the challenge of reintegrating repatriated refugees has begun to dominate the agenda, with a significant subset addressing transnational ties, both among refugees in the region and with the distant diaspora.

Wartime migration in Afghanistan The fall of the Taliban in December 2001 initiated new migration movements among the Afghan population (Kronenfeld 2008). Repatriation from the neighboring countries, Iran and Pakistan, was massive in the following two years. For the first time since war broke out in 1978, a substantial number of Afghans living in Western countries returned. Others were either hindered from returning to their place of origin or forced into displacement – particularly the Pashtuns in the north, who were blamed for having supported the Taliban (Marsden and Turton 2004). Altogether, this post-Taliban migration came on top of two and a half decades of massive migration that had dramatically altered Afghan society, with the demolition and reshaping of old networks and the emergence of new ones, including networks that span state borders in the region and beyond. Starting in 1978, armed conflict in Afghanistan led to the largest coerced movement of people in recent times. At the peak of the conflict, at the beginning of the 1990s, an estimated 7.5 million people were displaced: 3.2 million were registered as refugees in Pakistan; the government of Iran reported 2.35 million; and perhaps 2 million were

Introduction

7

internally displaced within Afghanistan (Knowles 1992). Repatriation started after the Soviet withdrawal in 1989, but only began to take place on a substantial level following the fall of Najibullah’s government in April 1992, with around 1.4 million refugees returning during the remainder of that year. However, renewed conflict between the factions of the so-called mujahedin government reduced the pace of repatriation in 1993, and in 1994 only 330,000 returned, which meant that the numbers of people being repatriated were actually lower than the numbers of those being displaced in response to new rounds of fighting, particularly in the capital Kabul (USCR 1995: 99–100). During the period when the Taliban was the country’s de facto government and controlled most of its territory (1996–2001), there was a steady flow of people from Afghanistan’s major cities, as well as a dramatic exodus from some areas where the Taliban was unable to gain lasting control, such as the Shamali plains north of Kabul and the Ghormach and Morghab districts in Badghis province. The US-led Coalition Campaign against the Taliban and Al-Qaeda in October– December 2001 initiated yet another round of displacement in certain regions, before the installment of an internationally backed transition regime in December 2001 triggered a massive repatriation movement, with an estimated 2.3 million returning in 2002 alone (UNHCR 2002, 2003b).5 All this means that at least half of the population of Afghanistan migrated at some point during a period of only two decades. Every single Afghan is well acquainted with someone who at some point was forced into migration. This has had significant effects on the entire social fabric of Afghanistan. Still, it must not be forgotten that Afghanistan’s settlement pattern before the war also to a large degree resulted from large-scale coerced movements, particularly at the end of the 19th century, when the first great modernizer, King Abdur Rahman, forcibly resettled Pashtuns from the south in the non-Pashtun north (Centlivres and Centlivres-Demont 1983). Thus, recent population movements form both continuities and ruptures with the national and regional history of migration. Pashtuns in the north, for example, faced massive resentment from other northerners after the fall of the Taliban, a movement that had mainly recruited among Pashtuns. In Afghanistan, migration patterns have been extremely variegated. As Gilles Dorronsoro (2005: 170) has pointed out, ‘there was no automatic linkage between the number of departures and the intensity of the fighting’. People have fled at different times, for different reasons, and to different destinations. Some have fled a location as others have arrived or returned. In addition to the refugees in the neighboring states, whose existence is well documented, there have also been high numbers of internally displaced persons (IDPs). Indeed, IDPs came to constitute an ever larger share of the total number of displaced as other states

8

Social Networks and Migration in Wartime Afghanistan

increasingly sought to discourage people from entering their territory (Farr 2001). For the sociologist studying migration, such a multitude of patterns represents both difficulties and advantages. One difficulty is that the sheer complexity of the phenomenon makes it difficult to generalize; an advantage is that if a theoretical framework stands the test of explaining such a complex experience, it is also likely to be useful in other contexts. While it is a truism that situations of displacement are inherently political, it needs to be emphasized that the Afghan situation is fundamentally so. Here, the main cause of displacement is violent political conflict. In addition, refugee policies pursued by the host states in the region, as well as other states, have aimed at encouraging the refugees to pursue particular courses of action in relation to the Afghan conflict (Fielden 1998). Among the displaced, on the other hand, leaders or groups have emerged, with their own political motivations and projects. They have acted within particular frameworks of refugee aid, as amply analyzed by Zolberg, Suhrke, and Aguayo when launching the term ‘refugee warrior community’ in the late 1980s, with Afghanistan serving as a master case (1986, 1989). From 1994 onward, the Taliban emerged from the Afghan refugee communities in Pakistan. Those who had been forced to migrate formed a returning army that in turn forced others to migrate, thus taking the political logic of displacement to its extreme (Harpviken 2009).6 Many scholars have shown interest in issues of migration in the Afghan context.7 Much of their research, however, has focused only on certain localities and certain time periods, with no attempt at generalization. This is both because access has been difficult for geographic, political, or security-related reasons and because the interest in promoting and funding research has reflected larger political agendas (Harpviken 1997b). There was a particularly intensive period of research – much of which was facilitated by the UNHCR – in the latter part of the 1980s and early part of the 1990s, in preparation for the massive repatriation that was expected to follow the 1989 withdrawal of Soviet forces (English 1988, 1989a, 1989b; Morton 1992a, 1992b, 1992c). The bulk of the first wave of research on Afghan displacement was conducted among refugees in Pakistan. Their ability to adapt to the situation of exile was the main focus, while some studies also addressed the causes and motivations of flight, or the potential for repatriation. In response to continued war during the reign of the so-called mujahedin government that was set up in 1992, international interest declined rapidly, as did the interest in research (but see Monsutti 2004b, 2005). Following the terror attacks on New York and Washington on 11 September 2001, and the pursuant US-led intervention, there was an upsurge of research interest in Afghanistan. A second wave of migration research, first and foremost from the Afghan Research and Evaluation Unit (AREU) in Kabul, followed (Abbasi-Shavazi et al. 2005a,

Introduction

9

2005b; Collective for Social Science Research 2005, 2006a, 2006b, 2006c; Stigter 2005a, 2005b; Stigter and Monsutti 2005; Turton and Marsden 2002; Saito 2008a, 2008b; Saito and Hunte 2007).8 This second wave of research has focused mainly on the living conditions and future expectations of Afghans in exile in the neighboring countries, and has been particularly articulate about the enduring transnational character of Afghan communities. The book at hand incorporates and critically discusses previous research on Afghan migration. Empirically, the book also adds significantly to the empirics of Afghan wartime migration, rooted in fieldwork conducted during the Taliban regime of the late 1990s, focusing fieldwork on communities of origin, and dealing primarily with migration between Afghanistan and Iran – in addition to engaging principally with networks and their transformation, regardless of whether they are transnational or locally bound.

Two villages of Herat Two villages in northwestern Afghanistan are at the core of the present work. Both villages were located on what had been the main security belt surrounding Herat city, which had seen some of the most intense and protracted war in the country. In one of them, Izhaq Suleman, the villagers’ dominant response as war escalated from 1978 onward was to ally with the Soviet-oriented regime of the People’s Democratic Party of Afghanistan (PDPA), which they supplied with militia forces. Simultaneously, the villagers cultivated contacts with the resistance. In the other village, Sara-e Nau, the majority of the inhabitants joined the armed resistance, and the village became a target for the Kabul regime’s military activities. The different choices of the two villages had deep implications for their displacement patterns, for the degree of physical destruction they underwent, for their relationships to later regimes, and for their chances of recovery. My most intensive fieldwork in both villages was conducted in 1999, when the Taliban was in power, but I also undertook four shorter visits over the following seven years, the final one in April 2006. The book is first and foremost about migration during the years of armed conflict from the 1978 coup up to 1999, when the main fieldwork was undertaken, but it does also follow some of the informants, and comment upon larger migration patterns, in the post-2001 period. Collecting solid data in a situation of war is difficult; to gather systematic knowledge – in part retrospectively and from a distance – on people’s social networks is even more so. By combining various means of gathering data, I have sought to map the various strategies pursued by individuals and households in coping with war, focusing on social networks as a resource and an area of strategic investment, a reflection

10 Social Networks and Migration in Wartime Afghanistan

of my research interest in how networks changed (for a discussion of the methodology, including a presentation of the data, see Appendix). Izhaq Suleman is a large village (an estimated 650 households in 1999). It is located just a few hundred meters from one of the major army tank bases in the country, which has been controlled by successive regimes: the PDPA in 1978–92, the mujahedin in 1992–95, the Taliban in 1995–2001, and thereafter regional strongman Ismael Khan or the Karzai government (Giustozzi 2009). Throughout most of the PDPA era, Herat and the surrounding areas were subject to intense fighting, and by 1982 Izhaq Suleman found itself in the midst of a heavily mined and guarded security belt surrounding the city. Whereas a majority of the villages in this area joined the resistance, there were two factors that made Arbab Saidu, the local leader at the outset of the war (1979), to choose an alternative path for the village of Izhaq Suleman. Firstly, a resistance strategy would lead to immediate extermination, because of the vulnerable location next to the strategically important tank base. Secondly, Arbab Saidu’s influential contacts within the PDPA administration made it possible for him to form his own militia force, siding with the government. Hence, for most of the people in Izhaq Suleman, the relative costs of staying were much lower than in neighboring villages. The village subsequently became a refuge for people displaced from other villages. Although he fought for the government, Arbab Saidu was also able to maintain contact with the mujahedin, whose families resided in Izhaq Suleman for extended periods of time. The militia also assisted the mujahedin with supplies, safe lodging, and access to Herat city. One out of the ten mahalla (quarters) of Izhaq Suleman stood out. Qala-e Muhajerin, which literally means the ‘refugee village’, was a small settlement of some 17 households in 1999. It is located only a half kilometer west of Izhaq Suleman, on the same side of the main road to Iran. In terms of administrative divisions, this settlement belongs to Izhaq Suleman, but it is a distinct community (as perceived both by the inhabitants themselves and by others), and the population lives in relative separation from the larger village. In 1999, Qala-e Muhajerin was not represented on the shura (the village council) of Izhaq Suleman.9 Qala-e Muhajerin had been established in 1992 by refugees returning from Iran, three of whom had former contacts in the village. The returnees bought plots for their houses on agricultural land that was heavily mined and therefore not cultivated. With nomadic forefathers, the Qala-e Muhajerin inhabitants specialized in keeping sheep and goats. As a new settlement established by returning refugees, Qala-e Muhajerin is an interesting case, one that I draw on for comparative purposes, most importantly in Chapter 6, on reintegration.

Introduction

11

The second main village, Sara-e Nau, lies almost directly south of Izhaq Suleman, also on Herat’s security belt. Access to Sara-e Nau and Izhaq Suleman is by different roads, but it takes only about an hour to walk from one to the other. Sara-e Nau had an estimated 130 households by 1999. In contrast to Izhaq Suleman, but similar to most other villages in the vicinity, the population sided with the mujahedin when the war escalated from 1979. The village suffered multiple attacks both by ground forces and from the air in the early stages of the war, and by 1983 the majority of the population had departed. A few families opted to stay, which in practice meant dislocating locally for extended periods. It incidentally emerged that Izhaq Suleman had been a refuge for many Sara-e Nau inhabitants during this period. From 1992, the village was gradually repopulated. Sara-e Nau proved in many ways to be an ideal contrast to Izhaq Suleman, and the two villages are systematically compared in all of the empirical chapters that follow. The main criteria in choosing the field sites were that the area was among the most heavily war-affected in the recent past, and that at least two villages with contrasting response strategies could be identified. There were other, and more practical, criteria that had to be met: security (including alternative evacuation routes) and access (via locally established door-openers). In 1999, the political and military situation was tense, and the Taliban were suspicious of westerners in general, and of researchers in particular. Gathering information on many alternative field sites would be costly, risky, and time-consuming. Choices have to be based on meager knowledge, and there is quite a bit of luck involved. Given the circumstances, I consider myself extraordinarily lucky to have been welcomed by the locals in the two villages at the outskirts of Herat.

Structure The primary ambition of this study is the development of a theoretical framework for understanding the role of social networks in migration, drawing on the multifaceted patterns of Afghan displacement over the past 30 years of war while focusing on two contrasting communities at the outskirts of Herat city in northwestern Afghanistan. At the same time, the work has close links with established concepts and approaches within general social network analysis, as well as within migration studies. Furthermore, Afghanistan has been the source of the largest occurrence of displacement in recent times, and it has served as the basis for forwarding a number of propositions about migration dynamics, despite the varied quality of the data that has been available. The secondary ambition of this study is therefore to assess the significance of this displacement.

12 Social Networks and Migration in Wartime Afghanistan

In pursuing these objectives, the findings from Afghanistan are placed in a global context, both by relating them to central ideas in research and policy, as discussed above, and by drawing on relevant case material from other places.10 The selection of case material, however, has not been driven by any ambition to systematically cover a given set of cases. Little of the available material on wartime migration has been collected with a theoretical perspective that even overlaps with the one being explored in the current study, and any comprehensive analysis of other cases within the current framework would have required extensive collection of primary data. The inclusion of references to a handful of studies from other contexts where relevant perspectives have been applied, however, allows me to situate the findings within larger debates, which often tend to be firmly rooted in specific case knowledge. In Chapter 2, I will establish a theoretical framework for analysis, based on how the general literature on displacement, disaster response, political mobilization, and war deals with the social network dimension, while also disentangling the different ways in which social networks can be structured, as well as the major resources that flow in networks. Chapters 3 to 6 place the Afghan migration dynamics – and particularly those in the villages of Enjil district – at the center. The chapters go stage by stage, looking first at the flight decision, then at integration in exile, before moving to the return decision, and eventually to the integration in Afghanistan upon return in Chapter 6. Various parts of one and the same network may be in different stages of the sequence at any given time, as when a household splits between the village of origin and an exile location. There are interesting comparisons to be made between the two decisionmaking stages (flight and return) and the two integration stages (exile and ‘home’), which challenge common assumptions in displacement theory, including the firm distinction between forced and voluntary migration, and the conception that return migration means coming ‘home’. Throughout, I draw comparisons between people who opted for various types of displacement, including those who were internally displaced. I also compare the migrants to those who elected to stay and pursued other ways to survive. The seventh and concluding chapter will weave together the most important of these threads, developing the implications of the current study for the academic analysis of displacement more broadly. Furthermore, I will revisit common assumptions about displacement in Afghanistan, and examine the extent to which those should be challenged or complemented by new insights. The questions outlined in the introduction will guide the concluding discussion, which will particularly follow up on the transformation of networks in the face of shocks such as disaster or war, seeking to distinguish between solidary and corrosive effects, including the persistence of various types of network structures.

2 Social Networks in Wartime Migration

For most people, the decision to migrate is a dramatic choice. To make such a decision in the midst of war – when your security is threatened, when there is little access to reliable information, and when control over economic resources becomes uncertain – is particularly stressful. Then, having arrived at a new location, the process of becoming integrated – finding a job, a house, and other necessities – forms another tremendous challenge, as stressful as the planning and the flight itself. Furthermore, the level of stress is not necessarily lower when a decision is made to return to one’s place of origin. And integration upon such a return may sometimes be even more difficult than it was in exile. In all of these processes, social networks play an essential role, both as causal factors for the flight itself and as resources people draw upon in tackling the challenges that subsequently arise. As a result, these networks are also transformed, taking on new shapes and new functions. Most often, individuals and households leave as part of a collective of people, and the exchange of information takes place to a great extent within the collective. The importance of a flight collective depends on what the structural constraints – such as transport facilities and security on the route – are. The way the actual flight happens – particularly whether or not it takes place on the basis of an established network – sets the stage not only for the social organization that is established in exile, but also for the way contact is maintained with the people who remain behind, as well as the prospects for eventual return. The extent to which flight is organized through established networks that survive the actual flight is therefore of crucial importance. If only part of a collective flees, this does not necessarily mean that the collective is split. Families often decide that some members should stay behind to take care of business. Some members can depart for a safe haven, while others take care of 13

14 Social Networks and Migration in Wartime Afghanistan

the collective’s resource base. Networks providing indirect social support might also be trans-local. Contacts in other areas, particularly at the destination, are a major resource. This chapter looks at insights gained within the social network analytical field, and builds on some of those insights to derive possible implications for the study of forced migration. In the following, I will build on studies of forced migration, although few of those are explicitly applying social network concepts or methodologies. I will also draw on social networkbased studies within other fields of collective behavior, particularly within migration, political mobilization, risk and disaster, and economic sociology (see also Brettell and Hollifield 2000b: 1364–1365; Cernea and McDowell 2000; Haug 2008). My ambition is not to build a full-fledged theoretical framework for the study of forced migration, since to do so would require looking far beyond the social network analytical field. Rather, the aim is to explore how the insights gained through social network analysis can be applied to forced migration. I will also seek to establish a consistent set of concepts that can be applied in future empirical work. The first part of the chapter provides a brief introduction to social network analysis (SNA). Then, I categorize the content of network flows as security, material support, and information. In the third part of the chapter, I discuss various network structures – strong ties, brokering ties, power relationships – drawing on insights from migration studies and the network literature. In the fourth section of the chapter, I examine temporal aspects of social networks, with a focus on the transformation of networks, their expected duration, and contagion dynamics. A fifth and final section sums up and offers some propositions that will be examined in the analytical chapters of this study (Chapters 3–6).

Social networks Social network analysis looks at relations between actors rather than at each individual actor and his or her attributes. This means that theoretical explanations are based on relational structures. In terms of methodology, this is reflected in a shift from attribute to relational data, from measuring each actor’s attributes to measuring aspects of their relations with each other (Scott 1991). According to Wassermann and Faust, whose 1994 book remains a primary reference, SNA builds on the following core principles (Wassermann and Faust 1994: 4): • Actors and their actions are viewed as interdependent rather than independent, autonomous units.

Social Networks in Wartime Migration

15

• Relational ties (linkages) between actors are channels for transfer or ‘flow’ of resources (either material or nonmaterial). • Network models focusing on individuals view the network structural environment as providing opportunities or constraints on individual action. • Network models conceptualize structure (social, economic, political, etc.) as lasting patterns of relations among actors. Social network analysis is informed by the structural traditions within sociology and social anthropology. Over the past two decades, with the development of new quantitative analytical techniques, it has gained a strong position within sociology and political science. Beyond adherence to the principles just listed, work branded as SNA may differ widely, partly because the types of data are different, and partly because the scholars have fundamentally different conceptions of social action. For this book, the conditions in Afghanistan and the neighboring countries prevented the collection of data that would be susceptible to quantitative analysis (see also Appendix). Insights gained through SNA are therefore used heuristically, to inform my qualitative analysis. This has the advantage of permitting an openended approach that applies a wide set of problem definitions. Increasingly, as the dialogue between field data and the established network literature moved forward, I refined my own understanding of social networks, and of the particular issues in forced migration where its application proved most fruitful. This is a potential basis for designing a well-conceived quantitatively oriented project for an area where the necessary data could be gathered. As Scott Feld has pointed out, many network studies map the social interaction between people but neglect the underlying institution that mediates the interaction, which he names foci: ‘relevant aspects of the social environment can be seen as foci around which individuals organize their social relations’ (Feld 1981: 1016). Foci have a widely differing nature: they may take the form of institutions and organizations, but may also be various types of activities and social positions. Within the migration literature, the family is the dominant focus (Murdoch 2005). A shared focus does not necessarily imply similarity of actors. The identification of foci rests on an understanding of culturally embedded practices and the way in which those are reflected in existing or emerging network structures (Emirbayer and Goodwin 1994). In the Afghan context, networks rooted in kinship are key. The household (and to some degree the extended family) is the basic social and economic unit. This necessitates an understanding of what we may refer to as relevant ‘cultural repertoires’, such as patterns of marriage and gender

16 Social Networks and Migration in Wartime Afghanistan

relations in the family. Within a patriarchal household structure, both the division of labor and the distribution of power remain relatively clear. Decisions on whether or not to migrate are taken at the household level. At the same time, households are transformed as a result of war, and by the migration that usually accompanies it. Hence, the viability of households as the primary focus should not be taken for granted. Feld develops a number of implications of his concentration on foci. Of particular interest here are his thoughts on the emergence of new foci: In order to understand network structure fully, it is important to remember that the formation of social networks and the relations to foci are interdependent. . . . [T]he more severe the restrictions on time, effort and emotion, the more individuals will experience pressures to combine their interactions with various members of their network by finding and developing new foci around which to bring more of them together. (Feld 1981: 1019–1020) Feld’s conceptualization of ‘foci’ has two central implications for the study of forced migration. First, the collectives that organize at the decisionmaking stage of the flight will normally build on some form of preexisting networks. While those networks may be loose to begin with, they may transform into more dense networks in this process. Second, the collectives formed or transformed through the migration experience assume an importance of their own. What I will refer to in the following as ‘migration collectives’ can then be expected to form new foci for network-based activities within domains such as the social, political, or economic. This means that there is a transformation of networks in two stages: First, an original network is transformed by the migration decision and the migration event as such. Then, the networks preserved or formed by the migration experience are transformed through the development of new foci after the migration has taken place. The transformation does not necessarily happen in clear-cut stages, but rather in overlapping processes, with people moving at intervals (Colson 2003). In most cases when somebody decides to leave a place, there already exists a link with a destination. His or her social network thus already includes someone at the other end. In other words, we must regard the community that is studied as a multi-local network, rather than as a geographically confined entity. Thus, to understand migration, it is essential to move from a geographical to a social definition of networks. Once we come to accept that the primary community of a group of people (in the sense of primary network)

Social Networks in Wartime Migration

17

may span several locations, we will explain geographical mobility quite differently. In the following, therefore, when the concepts of origin and destination are used, the intention is not to designate separate social groups, but only to indicate separate geographical locations. Similarly, when the concepts of decisionmaking, movement, or settlement are used, these are not necessarily tied to a logical temporal sequence. The sequence might be the other way around (as in return migration), or it might involve migration from one temporary location to the next (as in stepwise migration) (Bakewell 2004). Finally, individuals belonging to the same group can also move simultaneously in different directions and to different places, or at different times (Fuglerud 2004). There are few studies that address the role of social networks in forced migration (Brettell 2000; Hugo 2005; Shami 1993, 1996). One interesting study, which addresses the mechanisms through which a cause translates into people’s actual departure, is Allen and Hiller’s (1985) study of Vietnamese refugees. Allen and Hiller divide the process prior to migration into three phases: First, the would-be migrant starts forming an awareness of the need and the opportunity to escape, discussing his or her situation with a few acquaintances. Then, the flight perspective takes form and alternatives are explored with family and friends, normally over a considerable period of time. The observed departure of others is particularly important in advancing the process to the third stage, when the flight itself is organized. Several practical problems have to be resolved at this point, such as finding transportation and the means to pay for it, and important choices have to be made, such as who will leave and who will stay behind. Again, this takes place within collectives of close associates. Throughout this process, what one could call flight collectives are formed, groups of people who cooperate in undertaking the flight (Allen and Hiller 1985; Shami 1993). The members of a flight collective do not necessarily know one another well before they start to consider flight, but their mutual trust is based on personal recommendations. This means that the flight collective is not necessarily a social network transformed into a group, but social networks serve as an essential resource in the formation of the flight collective. The successful undertaking of flight depends on one’s successful entry into a flight collective, not, as commonly assumed, on the degree to which one is under threat. This also means that many who would not have chosen to flee as individuals come to do so because they become enrolled in a flight collective, perhaps as a result of ties to core members of the group that have a stronger commitment to escape. Works that take networks into account in studying forced migration are somewhat more common if we move the focus to integration processes.

18 Social Networks and Migration in Wartime Afghanistan

Art Hansen’s work on refugees in Zambia from the Angolan war of liberation in the early 1970s is a good representative (Hansen 1981, 1990). Hansen asked why so many refugees chose to self-settle in villages rather than in the organized camps, particularly when the economic conditions in the latter were so much better. He found that village settlement was preconditioned upon the presence of ties within the larger Luvale ethnic group; in fact, self-settlement was impossible without sponsorship from locally resident relatives. He concluded that the main explanation for the widespread self-settlement was that it permitted refugees to retain their identity through staying with their kin. Reciprocal rights and obligations then continued to be observed, in hierarchical patterns of power. Generally, self-settlement permitted more control of one’s life. An additional factor was the fear people had that, if they settled in a camp, they might be sent back to Angola by force. One might assume that the reluctance to settle in camps would be particularly strong for those with little former experience in dealing with the state or international organizations, but this was not discussed by Hansen. He speculated that the presence of welcoming kin at the flight destination is also important in people’s decision on whether or not to escape, but presented no data to underpin this contention. As discussed in the introduction, there is a need to reinsert agency into the study of war-related migration. Even when facing the gravest of threats, humans commonly base their decisions on experience and analytical skills, drawing on all available resources. My basic assumption is that structural forces are mediated by people’s social networks, and that social networks have the potential to function both as key resources and as key obstacles when people are forming a response (Brettell and Hollifield 2000a; Essed, Frerks, and Schrijvers 2004; Shanmugaratnam, Lund, and Stølen 2003). People act within structures that are beyond their direct control; and, although the choice between flight and other responses to war is taken under extreme conditions of stress, prospective escapees remain capable social agents. They have to act, and they have to bear the consequences of their actions. War migrants are usually assessed either as ‘helpless victims’ or as ‘fully responsible for their own fate’. In reality, they are never just one or the other. They are conscious and capable actors, actively engaged in social networks that define their response opportunities, but they are also subject to dangerous structural forces far beyond their direct control. Even decisions that seem to be entirely reactive in character – migration during war being a case in point – have a degree of proactiveness to them. The distinction between proactive and reactive collective action is outlined by Charles Tilly in the context of political mobilization. As Tilly points out, ‘the poor and powerless tend to begin defensively, the

Social Networks in Wartime Migration

19

rich and powerful offensively’ (Tilly 1978: 75). Tilly also emphasizes that while proactive mobilization tends to be top-down, reactive mobilization tends to be bottom-up. The proactive–reactive distinction reveals the often neglected fact that people who might not have left had the decision been theirs alone still do so when subjected to pressure from within their collective. Such pressure may come from within the household, the village, the tribal unit, or from a larger group. In research on forced migration, it is often implicitly assumed that the ‘force’ comes from outside the group. Thus, analysts tend to neglect the role of very ‘force’ that is inherent to internal power relations. From the perspective of the individual, social networks are not just part of a context that is given or that transforms solely at its own pace. Networks can also be consciously built. Allen and Hiller’s (1985) study of the formation of ‘flight collectives’ in Vietnam is an example. Even in this case, existing network resources were decisive for the building of new networks, since individuals were often only accepted as members of a new collective if they carried with them personal references. Strategic choices are made as to which relationships to invest in. Actors may also be strategic when deciding what relations to freeze or cut off entirely, not the least when associations with particular persons entail risk.

Flows in networks With reference to the literature on migration, as well as insights from SNA in other fields, I now propose to conceptualize social ties as conveyor belts for three distinct ‘flows’ that are crucial in migration decisionmaking under ‘forced’ conditions. The first is the ‘security flow’. Concern for one’s security is of course a central cause for migration in war, but security is also a resource that networks may provide. As other sources of protection break down (or transform into threats), the security value of people’s personal networks may increase tremendously. In most cases, this will require a collective of a certain size that takes us beyond the immediate solidary group, in which case various network types interact to form a collective of sufficient size to be an effective source of security. A second ‘flow’ contains material resources: property and money that may attract people to particular locations and that may be essential for the capacity to undertake a journey at all. The given network structure – cohesion, brokerage, or power (see the section on ‘Network Structures’ below) – that mediates material resources will vary. The third is the ‘information flow’. This relates to information about – and assessment of – the threat scenario at one’s current locality, as well as to information about the situation at a potential destination. Most people are

20 Social Networks and Migration in Wartime Afghanistan

likely to face an information dilemma: They must either rely on trusted sources in cohesive networks that are well known to them, but whose access to information is of a limited scope, or look for brokers with access to wider flows of information, but whom it may be more difficult to trust. The one flow that has perhaps received the widest attention in the network literature – social support – will not figure prominently here. Social support refers most often to intangible goods, such as emotional support, the offering of sympathy, and personal backing. Studies from industrialized societies indicate that such exchange tends to take place between actors of similar status (Walker, Wasserman, and Wellmann 1994), and social support seems to rely primarily on cohesive ties, typically based in near family or friendship. What is perhaps most interesting in this context is the general finding that one specific relationship is very rarely the channel for multiple types of aid: for example, emotional and material aid comes almost always from different individuals (Wellman and Wortley 1990). As we shall see, my findings indicate that the opposite is the case under conditions of severe stress and uncertainty. Then, dense networks increasingly come to be the conduits for all types of essential resources. The specification of distinct flows – as well as examination of how various types of ties and network structures are differentially suited to conveying those flows – is at the heart of network analysis. It is a prerequisite for being able to move beyond a romanticized conception of social networks as always cozy or benevolent. It also serves as a foundation for studying change in networks, the relative weight of various types of resources under shifting circumstances. In the following subsections, we will briefly examine the three flows at the core of this study – security, material resources, and information – before moving on to discuss how each of these play out within different network structures. Security When there is war and someone is assessing whether or not to escape, the key concern is often one of security: either a clear, direct threat that has already materialized or a threat that seems likely to arise in the foreseeable future. Unless the threat is direct and immediate, people will weigh the risks of staying against the risks associated with an escape. With the present global conflict pattern – dominated as it is by civil wars – the state often fails to serve as the main provider of security, either because it lacks capacity or because it has itself become a source of insecurity. During civil wars, the state is normally not an arbiter, but a party to the conflict (Bates 2008; Kjellman and Harpviken forthcoming; Milliken and Krause 2003). Hence, people either have to choose sides and seek refuge in areas

Social Networks in Wartime Migration

21

controlled by one or the other party, or they must resort to informal social networks for their protection. These can be preexisting groups or new groups with little basis in preexisting social ties. They can be neutral with regard to the armed conflict, or they can be linked to armies or political groups involved in the conflict. In general terms, one would expect the salience of social networks to increase with the degree of insecurity. This study is concerned with a narrow understanding of security: the absence of any physical threat to individuals and groups. This contrasts with the concept of ‘human security’ that has become increasingly common in the last decade (see Burgess and Owen 2004; Mack 2005; Paris 2001). This wider concept takes into consideration all threats against people’s fundamental welfare, and thus includes food security, health security, educational security, and environmental security. The new concept replaces the state with the individual as the prime referent in the security debate. In a more narrow interpretation of the idea, pursued by the governments of Canada and Norway, ‘human security’ is used in a way that aims mainly to reduce the human costs of armed conflict (Hubert 2004). My use of the term ‘security’ draws on the understanding of the Dari term amniyat, as used by my respondents in western Enjil (and in Afghanistan generally). This term is commonly used for the absence of physical threats to life and property.1 Such threats may come from the state, from other military– political organizations, from criminal groups – or from entities that are at the interface between these (for example, a criminal band operating under state cover). Although my conception of security is a more narrow one than that commonly applied in the human security debate, my concept is nevertheless one that has the individual, not the state, as the prime referent. One of my assumptions is that, in order to be effective in providing security, collectives have to be much larger than just a primary cohesive network, and in most cases they need to be rooted in a common sense of identity. Security threats, as discussed above, come in different shapes and sizes. At one end of the scale are smaller criminal gangs operating on their own (such as several groups that are active on the conflict scenes of Somalia or Colombia), and at the other are national armies or well-established highly organized rebel armies (such as the Liberation Tigers of Tamil Eelam (LTTE) in Sri Lanka or the Kosovo Liberation Army (Albanian: Ushtria Çlirimtare e Kosovës) (KLA/ UCK) in Kosovo). In between these extremes are the armies of mediumsize military–political groups. Obviously, the size of the collective that offers protection varies a great deal. Whereas relatively small collectives may fend off threats from criminals, even the largest collectives may find themselves helpless in face of the capabilities of a state or state-like organization.

22 Social Networks and Migration in Wartime Afghanistan

The insecurity that triggers a decision to escape – whether the threat is immediate or only imminent – is in most cases created by forces outside the primary collective that an individual belongs to. At the same time, however, the response strategies, which emerge within an individual’s primary networks, have major security implications. For example, if an individual is firmly enrolled in a village community that responds to external threat by full military mobilization, withdrawal may not be an option, since it could lead to a complete loss of network resources. Similarly, if a primary collective responds to a threat from an external force by offering to collaborate, it may be difficult for individuals to act otherwise without having to leave the group altogether. If most members of a primary network choose to move to a different location, this may for several reasons multiply the risks associated with staying. One of those reasons is that the group ensuring protection is being seriously reduced in size. Hence, the level of security is not only a result of the perceived external threat, but also reflects the response of the smaller collective to which one belongs. Material resources Material resources, whether in the form of money, real estate, or other goods, have a considerable impact on migration decisionmaking. Firstly, the weighing of economic conditions at one’s place of living against the expected conditions at a target locality still plays a role in a time of war, even though security is then the primary concern. Secondly, the ability to pay for the actual flight, including the costs of travel, documentation, and initial settlement, is a critical factor. Thirdly, there may be possessions at home – house, land, and business – that one wants to safeguard in expectation of a later return. Social networks play important roles for all of these factors: the ability to cope with economic hardship, the mobilization of resources to migrate, and the protection of resources at the locality which one leaves. In the network literature, it is commonly found that material resources of significance (beyond everyday mutual services, such as the lending and borrowing of household supplies) tend to come from actors who have better access to resources.2 This is in contrast to social support, which almost always comes from family members or friends with a similar economic status. Unequal control of resources, tangible resources in particular, often forms the basis for relations of dominance. For would-be refugees, the existence of ties to someone richer is significant, since potential migrants often lack the most basic resources for ensuring their survival. A lack of material resources frequently prevents flight, since both travel and documents may be costly. Recipients of material support will normally be expected to repay at a later

Social Networks in Wartime Migration

23

stage. Such exchanges have a double edge: they may be crucial in enabling flight, but may also limit the freedom of action both during and after the flight. If influenced by standard thinking on the social support role of cohesive ties, we may easily assume that flight group formation builds on actors that are largely equal (Allen and Hiller 1985). Hierarchical relations, however, can play a substantial role in group flight. Established leaders may use their control over crucial resources to ensure compliance in collective escape by people who might otherwise be inclined to stay. Existing power relations may survive the flight, even in situations where the parties have to live far from one another for long periods of time. In the context of migration – forced or not – the transferability of material resources is a key issue. As part of his transaction–cost approach to the study of firms and markets, Oliver Williamson (1981: 555–556) has introduced the term ‘site-specificity’ to denote the extent to which a piece of property (or ‘investment’) holds value only by virtue of its location – for example, because it is close to a power plant or to the agricultural areas whose products it processes.3 There is a continuum from non-site-specific to site-specific, with most types of material resources lying somewhere in-between the two categories. The least site-specific asset discussed in the present study is money, while livestock and household items are also possible to sell and convert into money, and may therefore be considered as not very site-specific. The most sitespecific assets are houses and land, which are often difficult to sell at an acceptable price under conditions of extreme insecurity. Additionally, of course, selling such assets, which often carry a family history, may carry a considerable emotional cost and cause resentment from one’s kin. Site-specific assets tend to represent a management challenge: how should owners maintain and protect their resources while they are away (Phillips 2004)? In some cases, the resources can be rented out; in other cases, it might be necessary to pay someone to take care of them. In yet other cases, the answer is a ‘split household’ strategy, well known from general migration research (Agesa and Sunwoong 2001: 203; Grasmuck and Pessar 1991; Pessar 1997). Here, while the majority of the family escapes to safety, those considered least at risk or most able to protect themselves stay behind to take care of the property. Those in exile will attempt to send money home. The cohesive network persists, despite being geographically split. Information The role of information networks – particularly transnational ties – in relation to migration has been firmly established (Castles and Miller 2003;

24 Social Networks and Migration in Wartime Afghanistan

Levitt 2001; Portes 2001). Those who have already migrated encourage further migration by providing information about opportunities at the destination site. Scholarly realization of how important this is dates back to Thomas and Znaniecki’s The Polish Peasant in Europe and America, which was first published in 1918–20 ([1918–20] 1958). In a review essay from 1976, N. P. Ritchey confirmed the importance of information about opportunities at the destination, channeled through networks of kin and friends, in relation to both stimulating new migration and directing it (Ritchey 1976; see also Marx 1990). Following up on Ritchey’s ‘information hypothesis’, Graeme J. Hugo found strong empirical evidence for the same, both in the USA and in the so-called Third World (Hugo 1981). In fact, Hugo suggested that the hypothesis was even more valid for ‘Third World’ countries, where the scarcity of other reliable information channels makes information transmitted by people one knows personally even more important. Ritchey’s information hypothesis, however, contrasts with a tendency in the network literature to see brokers, rather than relatives, as the essential providers of information in situations of distress. A major inspiration here is Mark Granovetter’s work on how people get access to information about new jobs. Granovetter (1973) showed that what he called ‘weak ties’ – relationships that tend to be more distant and characterized by infrequent encounters – were more important than strong personal ties such as kinship. Within a dense network, most people are likely to have access to the same information, while genuinely new information, which often triggers change or drastic decisions, is most likely to come from the outside – that is, from bridging ties. Furthermore, migration decisions depend not just on information about opportunities at the destination. In situations where it is not acutely necessary to depart, information about what one can expect to happen at the place of origin is at least as important. Such information is necessary for assessing one’s situation as a basis for deciding whether or not to escape, and also for deciding how to carry out the move if a decision is taken to leave. Earlier studies indicate that, for most war migrants, departure is motivated by the anticipation of threat, rather than by a threat that has already fully materialized (see, for example, Koehn 1991). On the other hand, when departures are fully forced, in the sense of being the result of an unambiguous physical threat, information is not an issue. In general terms, information is often unreliable in times of war. Rumors abound and reliable information is in short supply. There is a tendency for public information, to the extent it is available at all, to be highly unreliable. This is only in part due to general uncertainty. The main reason is often conscious manipulation by political power-holders. Hence, the individual assessments that precede a risky decision to take up flight are likely to draw heavily on how others

Social Networks in Wartime Migration

25

define the risk, but not on public information. This is why brokers become so important. They strongly influence threat assessments, particularly in those situations where people consider escape proactively, in the expectation that a fundamental threat will emerge. All flight decisions build heavily on information mediated by the brokers of specific social networks, and the less acute the threat is, the stronger is the role of such networks. In the study of social responses to risk and disaster, there is agreement on explaining action by the perception of risk, not the objective risk as such (Clarke and Short Jr. 1993; Douglas and Wildavsky 1983). This is likely to apply to decisionmaking processes in ‘forced’ situations as well. A group discussing the possibility of a joint flight will most often share new information and assess it jointly. It is here that the perceptions of others may have a crucial effect. Variation in information flows may account for some of the variation in escape patterns between populations in otherwise very similar situations. Information about the consequences of remaining or leaving is characterized by uncertainty. In most potential escape situations, there is also a scarcity of information. Potential migrants may receive various kinds of information: through public, through networks, or through observation of the actions of relevant others. The extent to which ‘actions speak louder than words’ can be expected to increase with the scarcity and unreliability of the information, as well as with the riskiness of the decision (Göbel 1998). If you witness other people flee, their action may be a more reliable piece of information than just hearing from people that they intend to do so. Hence, the process toward taking a risky decision is likely to draw heavily on how others define the risk and how they choose to respond, and much less on public information. Much of the literature on risk and disaster rejects an objectivistic approach to risk, and emphasizes the social construction of risk and risk acceptability. The practice of assessing attitudes to risk by aggregating individual responses to surveys has come under harsh criticism: responses to risk are fundamentally social and not the results of aggregated individual choices. As risk responses are formed in collective settings, ‘choice itself is sensitive to decision-making and agenda-setting procedures’ (Douglas 1985: 36). It has also been proposed that the shorter the time that is available for taking a decision, the more likely the decisionmakers are to build on institutionalized ways of problem-solving (Heimer 1985). When information is unavailable or in short supply and decisions have to be taken rapidly, existing cultural repertoires (such as the Islamic hijra – the religious narrative of Mohammad’s flight to Medina) or, more trivially, one’s own previous migration experience may be decisive for the course of action taken. In general terms, this implies that networks increase in significance

26 Social Networks and Migration in Wartime Afghanistan

as the availability of reliable information through other channels decreases, but if the time for making the decision is short, then actions are modeled on previous experience or existing cultural repertoires.

Network structures As already indicated, SNA is a bundle of various conceptualizations and techniques. In the following, I will look both at works that strictly specify the models of networks under study and at works that apply the network term more vaguely. Examples are drawn from studies of migration, political mobilization, risk and disaster, and from economic sociology. These examples are used to generate propositions for how to study forced migration. Scholars often distinguish between three types of network structure: cohesive, brokerage-based, and power-based. The archetypical social networks are cohesive, with strong mutual commitments. Such networks are often rooted in the family or other small social entities, where trust is considered largely unproblematic. Brokering ties are often used as a contrast. A broker is someone who connects actors who would not otherwise have been connected, and who derives influence from conveying information or other resources. A power structure is between actors with unequal command over key resources, where one can use his or her privileged position to dominate the other, to make the other do things he or she would not have done otherwise. As we shall see, various network structures have different functions and come into play in varying contexts. Strong ties When we talk about social networks, the intuitive association most people have is with the ties that exist between people who are close to each other, and these indeed represent the dominant network structure referred to in the social science literature. Several dimensions can be suggested as characterizing strong ties (for an overview, see Friedkin 2004). Using a commonsense approach, we might define strong ties in relation to the degree of kinship or friendship they exhibit – in Feld’s term, by the ‘focus’ of such ties.4 However, if one seeks a theoretically robust definition of a tie’s strength, it would be wrong to take the nature of the tie itself as a starting point. A more useful approach is to define the principal characteristics of a close tie in such a way that the definition applies in various social settings, and at various social levels. Wellmann and Wortley, who have done a lot of work on social support relationships, suggest a composite definition of three criteria, and regard any tie that satisfies at least two of these criteria as a strong tie. The three criteria they propose

Social Networks in Wartime Migration

27

are: (1) perception of close intimacy; (2) voluntary interaction between ego and alter (not interaction because of, for example, membership in the same organization); and (3) interaction between ego and alter in multiple social contexts. (Wellman and Wortley 1990: 564–566) At a general level, it is suggested that persons associated by a strong tie are important because they have greater motivation to help, and because they are most easily available (Granovetter 1983: 209). We may also assume that strong ties are particularly relevant in difficult times, when the risk associated with assisting somebody is greater. Only people who are really close will accept such a risk. As we shall see, studies of how people cope with situations where freedom of choice is extremely limited, as under severe poverty, have proven this to be the case. This adds to the fact that control over resources tends to be associated with command over multiple ties. When people with a scarce resource base rely on a few, but close, social ties in times of crisis, this is not only because such ties are more likely to be relied upon in times of difficulty. It is also because these are people who generally have fewer and less varied ties (Salaff, Fong, and Siu-lun 1999). Somebody with a rich resource base can normally seek help from many others (Fischer 1982). This leads to the assumption that strong ties are more important in less developed societies than in highly developed ones, since technological developments within travel and communication will increase people’s options. As mentioned, the importance of close ties for individuals coping with emergencies is well established. The extent to which social ties are effective means of coping with threats that encompass larger social entities is less clear. The literature on disaster and risk provides some evidence of the importance of informal organization for the ability of social entities to cope. Informal organization is a wide term, and my primary interest here is those social networks to which the people who are subject to disaster belong. One finding is that social networks are far more flexible than formal organizations in adapting to extraordinary circumstances, with the effect that the former often perform the bulk of relief work in the initial stage after an incident (Quarantelli 1978; Rossi 1993). However, research is relatively scarce, as a 1993 review concludes: ‘We need more work on the possibilities of network and decentralized responses to disasters, even in the presence of centralized bureaucracies’ (Clarke and Short Jr. 1993: 393–394). Now, a decade and a half later, this work has still not been carried out. Clarke and Short Jr. indicate a direction for research, however, in suggesting that social networks are likely to be essential for coping in most stateless settings where alternatives do not exist.5 It has been suggested that

28 Social Networks and Migration in Wartime Afghanistan

the extent to which people rely on either formal organizations (such as the state, aid agencies) or close networks in coping with the effects of disaster reflects cultural differences (Kreps 1984: 319). However, the finding that close networks tend to be more important in societies that do not have a highly developed state cannot be attributed to cultural differences. It is not culture that decides whether you have a strong or a weak state. It seems quite likely that all people, regardless of culture, rely less on social networks when there are professional agencies that can protect them. If we value social networks highly, this begs the uncomfortable question of whether the emergence of such professional agencies is necessarily for the better. Some researchers have claimed that not only local networks are the first resource to which people resort when responding to disaster, networks are also more effective than any other resources in limiting both short- and long-term damage (Morren 1983; Oliver-Smith 1996). If this is the case, then it seems important not to introduce formal organizations in a way that sideline local networks. In a series of publications, Massey and associates have established the relevance of social networks in migration (Massey and Espinosa 1997; Massey et al. 1987; Massey et al. 1993, 1998).6 So far, the most influential project in applying social network ideas to migration is the methodologically impressive study of Mexican labor migrants to the USA, in which close ties are seen to be crucial (Massey et al. 1987). In addition to kinship and friendship, this study also includes common place of origin ( paisanaje) and common organizational membership as foundations for ties. An overarching conclusion in this work is that while migration starts because of deteriorating socioeconomic conditions, once underway its continuance is secured through the emerging transnational networks, which supply information and other forms of assistance. The networks can be seen as a form of social capital that ‘increase the likelihood of international movement because they lower the costs and risks of movement and increase the expected net returns to migration’ (Massey et al. 1993: 448). Since the publication by Douglas Massey and associates of two influential review essays (Massey and Espinosa 1997; Massey et al. 1993), there has been practically a consensus that social networks are important in international migration, but less so in domestic migration. The premise for the consensus, however, has tended to be forgotten. In an earlier article by Edward J. Taylor (1986: 148), the general factors underlying the role of networks in international migration were suggested as: (1) the high risk incurred during migration; (2) the difficult access to labor-market information; and (3) the potential penalty for making bad forecasts. If we look closely at this list, we see that both (1) and (3) concern the level of risk, whereas (2) is about access to information. While these factors may be

Social Networks in Wartime Migration

29

more prevalent in international migration than domestic migration, there may also be forms of domestic migration – not the least during war – that have the same characteristics. Rather than the presence or absence of a state boundary, it is the extent to which there are physical and institutional obstacles to be overcome that determines the importance of networks. High risk and difficult access to information are primary characteristics of the situation that faces most potential wartime migrants. Thus, it is by no means certain that the consensus established in migration research is generally applicable. If that is not the case, then this may further reinforce the main methodological premise for the present thesis, namely, that the study of social networks forms a promising venue for continued empirical research on forced migration. The study by Massey and associates (1987) suggests a two-step model in which structural causes induce migration, whereas networks channel it and regulate its intensity once it is underway. Gurak and Caces (1992), on the other hand, have suggested that migration-inducing structural forces and network dynamics interact throughout the whole migration process. More specifically, the approach advocated through the studies of Massey and others tends to neglect the importance of networks in the decisionmaking process. The importance of close ties in decisionmaking is clearly in evidence: in many instances, the family, household, or community (with potential trans-local elements) are as relevant for the analysis of decisionmaking as is the individual (Shami 1993). However, to grasp the collective character of decisionmaking, one must not, as Massey and associates (1987) have done, focus exclusively on networks linking destination to origin, but look also at networks that exist only at one of the two places or that link one of them to third localities. Seeing households as the key unit in mediating global and domestic economic and political pressures into actual migration, Grasmuck and Pessar (1991) have studied Dominican migration to the USA.7 Like the study by Massey and associates of migration from Mexico, this study combines data collection at origin and destination, using both ethnographic and survey methods. Grasmuck and Pessar are concerned with the interplay between class and migration propensity, and they find that the educated, urban, middle, and upper-middle classes are overrepresented among the migrants. This, they conclude, is not because these people’s skills match job opportunities in the USA. Rather, it is because migration requires economic resources, and because those who are economically better-off tend to have richer and more variegated networks at their disposal. People from a lower-class or rural background are more likely to move within their own country, or to closer countries in the Caribbean. One should

30 Social Networks and Migration in Wartime Afghanistan

expect a similar pattern elsewhere, also in cases of forced migration: the greater the obstacles to successful migration, the more likely it is that the successful migrants will belong to the more resourceful social classes. In the Dominican case, it is often only a part of the household that migrates, return migration is frequent, and many households have members who regularly spend long periods in the USA to earn money. Households with migrants often adapt and make changes to their lifestyle that cannot be sustained without remittances from their family members in the USA. The result is the development of truly ‘transnational households’ (Bryceson and Vuorela 2002b; Grasmuck and Pessar 1991; Pessar 1997; Vuorela 2002). Once migration is as extensive as it is in many of the communities Grasmuck and Pessar are studying, its effects are not confined to the households that dispatch migrants, but lead to social and economic changes in whole communities, with potentially huge effects also for nonmigrant households. These feedback effects are mediated through local norms and institutions. Incentives for Dominican absentee landowners to raise cattle rather than grow crops, for example, contribute both to undercutting local unemployment and to threatening domestic food supply. Within economic sociology, one branch has been particularly concerned with the issue of immigrant entrepreneurship: how is it that some groups of immigrants successfully build up within niches of the economy, whereas others tend to struggle to achieve economic subsistence? The answer lies in networks based on kinship, friendship, or common place of origin. In Alejandro Portes’s terms, the two dominant ways in which networks promote economic accumulation are ‘bounded solidarity’ and ‘enforceable trust’ (Portes 1995; Portes and Sensenbrenner 1993). Bounded solidarity refers to the existence and potential enhancement of cultural markers – such as dress style, language, customs, and phenotype – often reinforced by the attitudes of the surrounding society. Solidarity can be reinforced by situational confrontations and might lead to the construction of an alternative account of the situation in which the group’s specific practices are integrated. Cohesive ties are strengthened by external pressures, a fact that is relevant not only in the context of exile integration, but may also be at the heart of flight collectives that can emerge in response to social exclusion in the home community. Enforceable trust, following Portes and Sensenbrenner, refers to the capacity of such immigrant groups to sanction members and is closely linked to the group’s boundedness (Portes 1995; Portes and Sensenbrenner 1993). When a group is bounded, its members will often have few alternative options, and this fosters internal cooperation. Whereas a capacity to sanction members is more costly to maintain over time

Social Networks in Wartime Migration

31

than voluntary forms of solidarity, sanctioning can be greatly helped by hostility from surrounding groups. In general terms, in those situations where escape starts becoming an option, there is often an external pressure that enhances the boundaries of the collective and restrains interaction with people outside. The level of solidarity in a community does not influence the likelihood of flight, but it does influence the degree to which flight is undertaken individually or collectively. Not only does it seem to be a tendency that people with few economic resources also have smaller and less variegated networks (consisting of strong ties exclusively), it has also been suggested that such networks tend to restrain economic mobility (Granovetter 1995). This is a recurring theme in economic sociology, where it is found that, among the poor, economic accumulation based on close ties fails, the reason being excessive demands placed on an enterprise in terms of sharing profit or granting employment (see, for example, Grasmuck and Pessar 1991: 154; Stack 1974: 104–107). Marc Granovetter (1995: 213) speculates about the implications of having few ties, and ties only within one’s own category: the heavy concentration of social energy in strong ties has the impact of fragmenting communities of the poor into encapsulated networks with poor connections between these units; individuals so encapsulated may then lose some of the advantages associated with the outreach of weak ties. This may be one more reason why poverty is self-perpetuating. The conditions under which there emerges a sensible balance between what Granovetter calls ‘coupling’ – that is, getting the best out of group solidarity – and ‘decoupling’ – that is, avoiding excessive demands that undermine accumulation – are not well understood. One of the general conclusions, though, is that conditions forced upon groups from the outside, such as discriminatory practices, can ironically prove to be beneficial for the success of the group. This is a variation of Portes’s ‘bounded solidarity’ argument. Studying political mobilization, Roger Gould has pointed out: ‘Mobilization not just depend on social ties, it also creates them’ (Gould 1991: 719).8 This important but often neglected insight can be directly transferred to forced migration. The flight collective is sometimes an organized group, such as a political opposition party that over time has come to realize that continued resistance is only possible from exile. However, as the findings of Gould and other mobilization analysts demonstrate, it might very well be that the strength of the formal

32 Social Networks and Migration in Wartime Afghanistan

organization is derived from a strong overlap with other foci, such as family, friendship, or neighborhood. In a sense, there is a sequence here, where close informal ties help establish a strong formal organization, and where the overlap (multiplexity of ties, in Gould’s terminology) underlies its strength. In the case of proactive flight, motivated by political resistance, it might be that the flight, in hindsight, will only be seen as a minor episode in the larger narrative of the political group. The key idea here is that not only does mobilization depend on preexisting ties, such ties are also vital for maintaining the momentum of the mobilization once underway. In other words, strong ties strengthen, or even emerge, as a product of the mobilization.9 Similarly, as Allen and Hiller maintain in their study of Cambodian ‘flight collectives’, the joint experience of escape is likely to lead to a strengthening of ties, occasionally even fostering strong ties where no former association existed. Bridges Strictly speaking, a bridge is a tie that links two networks that have no other ties. Bridges, also often referred to as ‘brokering ties’, are important because they enable resource flows between otherwise disconnected people.10 For practical purposes, the definition is often made less strict, since networks tend to be connected via multiple ties, most of which, however, are too loose to allow any important flow of resources. A bridge is therefore a tie that links two networks that have no other close ties. One of the factors that make a tie loose is its length – understood in terms of the number of nodes found between alter and ego. One may therefore select a maximum path length between two actors, and include only bridging ties that lie on paths shorter than that. Mark Granovetter was a pioneer in the work on bridging ties. In his study of how people get access to information about new jobs, he showed that bridging ties tend to be more important than strong ties (Granovetter 1973). This is so because, within a dense network, most people are likely to have access to the same information, while genuinely new information is most likely to come from outside of one’s immediate circle of close associates. At a macro level, Granovetter suggests, weak ties are essential for larger-scale cohesion. If people are linked only via strong ties, that means a fragmented society of strong cliques with very little capacity for concerted action above the clique level.11 To my knowledge, no study of forced migration has systematically addressed the role of brokering ties. Khalid Koser (1993, 1997), however, addresses the role of information brokerage in voluntary repatriation of refugees. Koser states that his model is inspired by migration theory more broadly, and justifies the application of network analysis by the

Social Networks in Wartime Migration

33

basic voluntariness that characterizes both ordinary migration and the repatriation of refugees. Somewhat problematically, Koser seems to assume that insights from the study of migration are irrelevant for the study of forced migration. Relatedly, Koser assumes that refugees can be distinguished from other migrants by their lack of access to information about conditions at their place of origin. Although there will be instances where information is problematic, cases where some sort of ties persists are likely to be much more frequent. Forced and voluntary migration have much more in common than Koser seems to argue. Furthermore, it is methodologically problematic to assume that certain ties are absent prior to having searched for them, whereas concluding that ties are absent after having searched is viable. Problematic assumptions aside, Koser’s work provides several interesting insights. Building on a geographical model of innovation diffusion, Koser emphasizes that information is normally brokered, and continues: ‘The premise has been that the logic of the mediator will affect the information that is supplied. There may be an intentional manipulation of information’ (Koser 1993: 175). Koser operates with four categories of information brokers: institutional sources (governments and relief organizations); media; personal sources; and kin (only the two latter constitute brokers as the term is applied in this study). In a study of Mozambican refugees in Malawi, Koser finds that most people had a variety of sources of information while in exile, although these were often not activated in the first phase, when all energy went into establishing oneself in the new place and when many people sought to avoid information about their departed homes since such news was expected to be depressing. In this case, many villages had become entirely depopulated, and as many as 36 out of Koser’s 125 respondents reported having no kin left in Mozambique. The general pattern among Koser’s respondents was that their whole social network had moved. He argues that the high reliance on kin was a particularity of this case, as neither political parties nor other institutional sources of information were present. The apparent assumption that kin is highly important only in the absence of other institutions is questionable (Collyer 2005). For reasons stated above, I assume that, when accessible, networks based on kinship and friendship are preferred sources of information even when alternative sources of information exist. Koser finds that, at the time he conducted his interviews, some three years after the escape of most respondents, most had well-developed information systems about the situation back home, and a majority had more than one source of information. Many were able to maintain contact with their kin through an elaborate process of choosing, evaluating, and cultivating brokers – for example,

34 Social Networks and Migration in Wartime Afghanistan

traders who regularly traveled across the border. However, in the final analysis, it remains unclear to what extent people had direct contact with their kin back home, and to what extent they relied on brokers. It also remains unclear what the larger implications of brokerage are. Koser concludes that the interplay between information and repatriation expectedly varies with a range of institutional factors, and he suggests a distinction between reactive (driven by factors at exile) and proactive repatriation. In the former case, information about home conditions is not important; in the latter case, it is vital. In general migration studies, it is far more common to analyze brokerage. In the Mexican study by Massey and associates, brokers are crucial for establishing the link between a given origin and destination: most migrant networks can be traced back to the fortuitous employment of some key individual. All that is necessary for a migrant network to develop is for one person to be in the right place at the right time, and obtain a position that allows him to distribute jobs and favors to others from his community. (Massey et al. 1987: 169) Despite what they say here, however, in a summary of four case studies of the emergence of Mexican ‘daughter communities’ in the USA, Massey and España (1987) pay little attention to brokerage. Their study pays much more attention to the multiple ties that exist between origin and destination, even within single families, than to the role of brokers. It seems, therefore, that while the pioneering broker is of critical importance in establishing the link in the first place, and may play a major role in providing work upon arrival, he or she loses importance once the migrants have settled down and have started to develop ties of their own. This is just an assumption, however, since the migration literature has little to say about the changing potential for brokerage over time. It seems logical that when more and more people from one origin migrate, the room for brokerage must shrink. Most people back home will then have several contacts at the destination, and vice versa. To get a clear picture of this, one would need a clear conception of brokerage, and a longitudinal study starting at a time when out-migration has not yet started or remains marginal. If our assumption is true of migration in general, there is no reason to assume that it does not apply to forced migration. One of Koser’s findings from his interviews of Mozambicans in Malawi was that over time they developed multiple sources of information about the conditions back home.

Social Networks in Wartime Migration

35

Brokerage is found to be key in the integration of Dominican immigrants in the New York economy, studied by Grasmuck and Pessar (1991: chapter 7). A classical patron–client relationship is dominant, where the employer relates to each individual’s discrete needs, with little potential for worker organizations to develop. The two authors show that a restructuring of the New York economy, with decentralization of manufacturing businesses, opened up for the kind of small-scale companies that could be run by ethnic entrepreneurs. They also point out that the people who stay in patron–client type of jobs, with low pay and few rights, are those who are most vulnerable, particularly illegal immigrants. It might be the case more generally that people at risk are more vulnerable than others to exploitative brokers. Brokering networks can have different functions. In Larissa Lomnitz’s (1977) study of a Mexican shantytown, she distinguishes two types of brokerage. The first is the pioneering migrant. Like the bridgehead migrants described by Massey et al. (1987), these are the people who departed first and who spent considerable time seeking out job and residential opportunities at the destination. The ones who followed, mainly kin from their village of origin, went directly to the shantytown and settled there. The other kind of brokerage (Lomnitz does not discuss the extent to which these are the same kind of people as in the first instance) are those ‘who specialize in mediating between the marginals and the urban industrial economy’ (Lomnitz 1977: 13). Such brokers, while financially better off than others within their group, continue to live and socialize among the people at the margins, a precondition for their business. Similarly, there is a potential role for brokers in the actual process of migrating. In an interesting study of migration from Java, Ernst Spaan (1994) has focused on the role of middlemen. In what is increasingly a commercialized business, semi-professional recruiters play an important role. Spaan finds that potential migrants often prefer the illegal option offered by village brokers they know from before, rather than official institutions. The village broker is only one end of a chain of brokers spanning from origin to destination, and the length of such chains increases the risk for the migrant. Spaan also demonstrates an interesting interplay between religious institutions and labor migration. In the Javanese case, many go to Saudi Arabia as pilgrims and then stay on as illegal migrants. Religious leaders and religious boarding schools thus tend to serve – wittingly or unwittingly – as brokers in labor migration. This suggests a networkoriented interpretation of the common cultural explanations of migration, where it might be the network that an institution offers, rather than its nominal purpose, which provides the best understanding of migration. It

36 Social Networks and Migration in Wartime Afghanistan

has been suggested that flight is a culturally accepted solution for Muslims, since Mohammad fled to Medina in the so-called hijra (Abu-Sahlieh 1996; Suhrke 1995b). The importance of religiously based networks in facilitating migration, however, remains unclear. Power Within network studies of migration – particularly studies of ‘forced’ migration – the ‘power’ of external structural forces has been overemphasized. Internal power relations are at least as important for the study of any kind of migration. There are cases where coercion from local power-holders is the key factor behind out-migration, while external structural factors only constitute background noise. This is often the case with proactive migration under conditions of war (or in the buildup to war), when escape is closely linked to political and military mobilization. There are other cases where structural causes are indeed the main cause, yet power relations – within the family or in a larger collective – may still decide who is to leave and where they shall go. Power arrangements may well become renegotiated after a flight, when a new environment may favor different sets of qualifications. A common critique of network studies guided by the cohesion concept is that they neglect power relations. Networks do not only include equal parties or close ties. Some theorists talk of horizontal versus vertical networks. The crux of the matter is that inequality operates in networks, and unless one grasps the power inherent in a relationship, the pinpointing of a tie is of little interest. With reference to SNA, David Knoke (1994: 275–278) has suggested a distinction in the power concept between domination and influence. Influence is when an actor provides information to another actor with the intention of changing the latter’s behavior, and it builds on two-way communication. By contrast, domination is when an actor controls the behavior of another actor through the application of negative or positive sanctions, a dimension Knoke sees as neglected in many network studies. The previous section on brokerage illustrated that control of information can be a foundation on which power relations are developed. Here, I would like to suggest that unequal control over material resources is particularly important as a basis for power relations. Hierarchical relations strongly influence flight situations, as potential migrants may lack the most basic resources for survival. Lack of material support is frequently a major obstacle to flight, as travel may be costly or documents expensive. Material support tends to come from actors in a different resource situation, such as those high up in a hierarchy.

Social Networks in Wartime Migration

37

Provision of resources that make escape possible is linked to an obligation on the part of the recipient to repay at a later stage. Such exchanges have a double edge: they may be crucial in enabling flight, but will also restrain the freedom of action both during and after flight. In discussing social support above, I suggested that the formation of a flight collective builds on actors who are largely equal, but that power relations are characterized by different degrees of influence rather than domination. However, domination can also play a substantial role in group flight. Established leaders may use their control of crucial resources to ensure participation in a collective escape by people who prefer to stay. Existing power relations may survive a flight, even in situations where the parties are forced to live far away from one another for a long period of time. If we apply insights from the study of collective action (Hechter 1987; Olson 1965), we will start to reveal the potential for relations of dominance in flight situations. The basic idea is that ‘free riding’ is a major obstacle to the production of a collective good, as anybody can join in its consumption, whether they contributed to its production or not. Sanctioning and control mechanisms are needed to ensure compliance. In forced migration, the free-rider problem appears to be solved in advance: those who do not join in the production will not share the benefits of flight. However, escape may in itself be a form of collective protest. Groups migrate to get away from a political system, aiming at a radically different society (Krokfors 1995: 62–63). Similar situations are found when people leave so as to build a capacity to challenge the political order at home, as reflected in the ‘refugee warriors’ concept (Lischer 2005; Zolberg, Suhrke, and Aguayo 1989). In such situations, we should expect to see processes of sanction and control operating within networks. In Grasmuck and Pessar’s study of Dominican migration to New York, the authors criticize what they see as the two dominant representations of households in the literature. The ‘moral economy’ approach emphasizes social solidarity and income pooling among members, and is almost exclusively occupied with external constraints. The ‘economic strategy’ approach emphasizes strategies for economic survival and maximization, and exaggerates the household’s freedom of action. Both approaches neglect generation- and gender-based hierarchies of power within the household. Economic change, for example, may inspire the father to adopt new inheritance practices, altering the equation for grown-up sons considering migration. Also, the migration experience is associated with greater freedom for women because of the necessity of seeking paid work; hence, women are often more hesitant than their husbands to return to their place of origin. In times when a household’s

38 Social Networks and Migration in Wartime Afghanistan

environment changes rapidly, as is of course the case when it migrates, its members have ample opportunity to renegotiate their positions within it (Grasmuck and Pessar 1991: 161; Zackariya and Shanmugaratnam 2003). This has relevance beyond the household. Any collective meeting a dramatic change in its environment, yet remaining a primary locus of identity for its members, is likely to also undergo dramatic internal change, not the least in the distribution of power.12 As Mary Douglas (1985: 59) has pointed out, one coercion technique is to stimulate or provoke a sense of risk. In the uncertainty of a flight situation, people are vulnerable and can easily be made to enter into new temporary arrangements meant at first to solve only their immediate problems, but which may not be so easy to do away with afterward. Dangers coming from the outside can both be used to secure compliance with existing authorities and to build new power relations. This is what happens when so-called political entrepreneurs use an uncertain situation to build support for a certain political purpose. The main motivation of the supporters will then often be their need for security rather than any deeper convictions.

Change War-related migration is change in two ways. The environment of the migrants changes dramatically and, in response to this, the migrants themselves change as well (Colson 2003; Stepputat and Sørensen 2001). Network analysis maps social space, and, just as in a normal map, most analysts describe social space statically, a structure at one particular point in time. Here, I will seek to map the key dynamic – or temporal – dimensions of forced migration. This includes a debate on the durability of a given state of affairs (threats, exile), inspired by Robert Merton’s concept of socially expected durations; a discussion of how, in the absence of reliable information, threat perceptions are formed and upheld in social networks, with potential contagion effects; and, finally, a reflection on the transformation of networks. Expected durations When making a decision, people normally take into account the time that they expect to live with its consequences. In contemplating escape, for example, one considers how long insecurity at the place of origin is likely to persist and how long an exile will last. Likewise, when arriving at a destination, the expected duration of the stay will influence decisions on key issues such as housing and work. To assess duration, however, is problematic, since most of the key factors are unknown and beyond

Social Networks in Wartime Migration

39

the control of the individual. Hence, expected durations are little more than informed guesses, and each individual’s expectation is strongly influenced by whatever view takes shape within a social collective. The social character of expected durations is at the heart of Merton’s work in this domain: ‘socially prescribed or culturally patterned expectations about temporal durations imbedded in social structures of various kinds’ (Merton 1984: 265–266). In his 1984 essay, Merton discusses the concept of socially expected durations that he had introduced many years earlier, and expresses his surprise that the idea had received so little attention, since it sought to capture a fundamental trait of human action. We should remain almost as surprised today, more than two decades later. However, the concept has been given some attention within the sociology of immigration, where it has been used to account for variation in integration patterns among groups of immigrants (Kelly 1995: 233–234; Roberts 1995). In his attempt to pin down expected durations going beyond individually held perceptions, Merton distinguishes three main types. Firstly, there are ‘institutionalized durations’, which are somehow codified and most often upheld by structures of authority. In the context of forced migration, examples would include obligations toward one’s state of citizenship – such as paying taxes or reporting for military duty – or the practices of the state where one seeks exile. Secondly, there are ‘collectively expected durations’, where the ‘collective’ is defined as a group that is larger than those individuals who are held together by direct ties. Shared expectations of an ethnic community under threat of persecution would fall under this category. Thirdly, there are what Merton refers to as ‘patterned temporal expectations’, rooted in cohesive relationships among family and friends, which I will refer to here as ‘interpersonally expected durations’. Within a network perspective, it is the two latter types of socially expected durations that merit the most interest. These are the two types that Merton considers most uncertain. Collectively expected durations are held within an identity group (in contrast to those based on the expectations of other groups); they are nonroutine in character, and therefore particularly uncertain. Bryan Roberts provides an extreme example: ‘If we ask those involved in this type of duration, “How long will it last?”, their likely reply would be, “As long as it takes.”’ (Roberts 1995). Here, there is an expectation of protracted, if not infinite, duration. Interpersonally expected durations are generally seen as more predictable than collective ones. Merton (1984: 281) refers to the implicitly shared understanding of gemeinschaft relations. In the context of migration, however, even close family ties may be exposed to

40 Social Networks and Migration in Wartime Afghanistan

severe stress, as individual family members cope differently with their new social setting (see, for example, Roberts 1995). When attitudes to migration often vary between husband and wife, for example, this is precisely because they perceive the effects on their daily lives – including their mutual relationship – differently (Grasmuck and Pessar 1991). The obvious importance of expected duration for the integration of migrants has been recognized not only by Roberts, but also by other analysts, although with no reference to Merton. One example is the work of Nee and Wong, who – comparing various immigrant groups in the USA – have pointed out that an enduring multi-local household strategy might prove detrimental to integration and success in the country of immigration (Nee and Wong 1985). At the other end of the chain, Grasmuck and Pessar (1991) found that migrant domination of local ownership undermined local employment and production in the Dominican Republic. This is a case where everybody stands to lose from migrants’ maintenance of a strong position in their home communities over long periods of time. Contagion and thresholds The classic treatment of thresholds is Thomas Schelling’s (1978) work on segregation processes in US city neighborhoods. Schelling assumes that collective behavior is characterized by binary choices (either–or), that actors are rational (make cost–benefit assumptions of whether or not to join), and that the cost–benefit of joining a group or a community depends in part on how many others make the same choice (Granovetter 1978). The threshold that needs to be reached in order for a certain person to join varies according to preferences. Thresholds can be from 0 percent (instigator/pioneer) to 100 percent (will not join under any circumstance); while a threshold of 20 percent means that a particular individual will join when 20 percent of a given category of people have joined. A key contribution of this model is that it allows us to explain why populations with similar aggregate preferences end up with widely different adaptation levels. Since individuals do not act until their threshold is met, distributions of thresholds can be such that very few from one category of people end up joining, while virtually everyone from another joins, even if the average threshold was the same in both populations. In accordance with the general threshold model, the curve of adoption increases steeply when the proportion of people in a population who adopt a specific course of action reaches a certain level (Oliver 1993: 284–287). In forced migration, it is often observed that the volume of out-migration fluctuates dramatically over time, a phenomenon commonly referred to as ‘vintages’. The conventional explanation of vintages – which is in line

Social Networks in Wartime Migration

41

with a conventional understanding of migrant selectivity – focuses on how specific social categories, such as a class or an ethnie, come under threat at various stages in a political conflict (Koehn 1991; Kunz 1973). A focus on information flow in networks leads to different interpretations of vintages. Peaks and troughs in vintages may be caused by the threshold effect of flows of information – or ‘rumor’ – that are mediated through a network structure, rather than by the levels of immediate threat to a specific category. However, as also held out by Schelling, adoption does not necessarily spread to the whole network: in instances where the initial discrepancy in attitudes is large, differential stands might develop (Moscovici 1985: 354), and the seemingly contagious adoption of flight may come to an end. Thresholds in adoption are more likely when actors rely heavily on observing others, as opposed to when a decision is based on substantial and multiple sources of information (Bikhchandani, Hirshleifer, and Welch 1992). While there is little reason to expect that the vintage phenomenon can be fully explained by thresholds in network models, we would expect threshold effects of the diffusion of information to be at least a partial explanation. Within the field of migration research, the idea of contagion has grown in influence in recent years, again through the work of Massey and associates (Epstein 2008; Massey et al. 1987; Massey et al. 1993). Here, the emphasis is on the evolution of transnational networks. International migration is seen as initiated by larger structural causes, but critical thresholds come into operation once it is underway. At the threshold, the network reduces the costs and risk of migration to such an extent that the probability of migration rises dramatically.13 This again leads to exponential migration patterns. If a similar mechanism can be demonstrated to have relevance in situations of war-induced migration, this would form a basis for a critique of the widespread assumption that the number of people departing from within a given community is determined by the level of external threat – an assumption that forms an implicit or explicit basis for all of the conventional ideas of ‘vintages’ in forced migration. If we can instead explain vintages through reference to the threshold mechanism, this would contribute to resolving a puzzle in research on displacement situations: the fact that out-migration from communities exposed to the same level of threat often varies from zero to all. The diffusion of information is channeled and constrained by lines of contact between informants and recipients. Furthermore, both the assessment of information and the planning of responses depend on these relations. As Ronald Burt (1987) points out, contagion can be working by two contrasting network mechanisms, either by cohesion or by structural equivalence. According to the cohesion view, it is the

42 Social Networks and Migration in Wartime Afghanistan

structure and density of relations within the group that explain the beliefs and actions of its members. By contrast, in the structural equivalence view, networks are chiefly influencing their members because the latter occupy similar positions in relation to other network positions (Lorrain and White 1971). For example, managers have similar attitudes not because they mix with the same employees, but because their structural position vis-à-vis employees is the same. Many network studies suggest that there are so-called thresholds in social action. Taking contagion as a starting point, they suggest that there are certain levels of adoption beyond which there occurs an endogenous growth in the number of people joining the action. Once a mode of action reaches a certain threshold, a network structure is likely to have feedback effects ‘pulling along’ other actors to follow the same mode of action. In practice, testing the vintage thesis against social contagion based on structural equivalence will be difficult, since the predictions that come out of the two methods tend to be the same. Expectedly, people in a decisionmaking process will act as if special vintages are at risk, and their departures will reflect an observation on their part that others of their kind (status, profession, and age) are leaving. The overall pattern will in many cases be the same. Transformation Given my expressed ambition to analyze the dynamics of networks in migration, which analytical tools from the social network tradition might I find useful? Many sociologists have struggled to come to terms with social change. In a 1997 article, Andrew Abbott claims that the old Chicago School represented vital insights, with its emphasis on the contextuality of action, both in time and in space. Abbott takes as his starting point a dichotomy between the ‘variables paradigm’ and the ‘contextualist paradigm’, and he argues that very little current theorizing, heavily rooted as it is in the ‘variables paradigm’, comes close to fulfilling the ambition of contextualizing action either in time or in space, whereas ‘one cannot understand social life without understanding the arrangements of particular social actors in particular social times and spaces’ (Abbott 1997b: 1152; Mjøset forthcoming). In Abbott’s view, the development of network analysis stands out as a response to the need for contextualization in social space. Network analysis does not, however, have any advantage over other sociological approaches when it comes to accommodating temporal contextuality. This is a problem network analysis shares with most other branches of sociology.14 It is only in the past decade that network analysts have started to develop analytical tools for dealing with change in networks. Earlier applications have typically used computer analytical tools to account for social structure

Social Networks in Wartime Migration

43

at two or more points in time, and have then proceeded to a relatively loose description of the processes linking them (Padgett and Ansell 1993). In his 1997 ‘Manifesto for a Relational Sociology’, Mustafa Emirbayer concludes that developing analytical tools for dealing with network dynamics remains a major challenge: ‘Even studies of processes-in-relations . . . too often privilege spatiality (or topographical location) over temporality and narrative unfolding’ (Emirbayer 1997: 305). In a 1996 issue of Journal of Mathematical Sociology devoted to the evolution of networks, the editors conclude that analytical tools for network dynamics are not yet well developed (Stokman and Doreian 1996). Similarly, in a special issue of the journal Social Networks on the theme of change in networks, the introduction reads: ‘One feature common to almost all this work on “personal communities” is that it has provided a picture of the ties that exist at only one time’ (Suitor, Wellman, and Morgan 1997). In fact, there is only an emerging research agenda on network dynamics, and the implicit message is that the development of new tools will in itself be a major task. Quantitative techniques that accommodate temporal change in networks and are readily applicable to larger empirical works are little more than a future promise.15 There are two ways of responding to this shortcoming in network analysis. One approach, which is currently the dominant one, is to work systematically on developing quantitative techniques for addressing change in networks. The other is to work qualitatively with data from particular contexts in order to expand our theoretical understanding of network dynamics. The former approach would invite research in settings that allow for the collection of large datasets, preferably also from already solidly documented populations, and the research questions addressed would be relatively narrow. The latter approach would invite research on societies undergoing dramatic change, where the opportunities to collect highly reliable large-n datasets are limited, but where the initial research questions can be sufficiently broad to allow refinement throughout the research process. Obviously, the work at hand is of the latter category. While such open-ended studies should result in findings that inspire the development of new quantitative techniques, their primary value lies in expanding our theoretical understanding of network transformation.

Conclusion This chapter has taken the initial research questions – about the role of networks in decisionmaking and integration, and about network transformation – as a point of departure, and tapped into the larger network literature in order to develop a conceptual framework for this

44 Social Networks and Migration in Wartime Afghanistan

study. This framework, and the propositions suggested by the present chapter, will guide the analysis in Chapters 3–6, as well as the concluding chapter of the study. While social network insights have rarely been applied to the study of forced migration, there is a rich pool of parallel mechanisms and relevant findings to draw upon within the fields of migration, economic sociology, and political mobilization. Social network analysis is here applied as a heuristic tool to support the qualitative inquiry into various mechanisms in forced migration. At the same time, there is a need to insist on clarity both in conceptualization and in operationalization – for example, by not taking the network quality of a local village or other social categories for granted. Relatedly, the widespread existence of ‘split households’ and other types of multi-local networks serves as a warning against presuming that network structures are by necessity linked to geographical proximity. The agency of forced migrants is reasserted through an emphasis on egocentric networks, as well as by the argument that there is a degree of proactiveness in almost all ‘forced migration’. This agency also comes through the way people’s network resources are intentionally built. The mediation of security – in the sense of protection against crime, along with political and military threats – has strong network dimensions that are barely touched upon in the literature. When the state is repressive or its capacities are virtually absent, people have to rely on their networks, which need to be of significant scope if they are to be effective. Material resources are commonly assumed to come from contacts of a higher status (hence, a stronger economy), but such contacts seem to be seldom available to the poor. Site-specific resources, such as housing or land, pose a particular challenge to migrants; one common response is ‘splitting’ the collective, with some staying behind to look after the property. Information, a critical resource when the credibility of public institutions breaks down, is also conveyed through networks. Importantly, information flows in situations where forecasts are based on little or no hard facts are most constructively understood as the collective formation of perceptions, rather than as the conveying of credible bits of fact. Network analysts often assume that various types of flows come through different ties. In studies of migration, and forced migration in particular, applications of network insights are considerably rougher, and such distinctions are rarely made. Close ties – as in the archetypical social network – are commonly associated with a social support role. Here, it is instead suggested that in situations of war, when trust is under threat, close ties come to serve multiple tasks, whereas other

Social Networks in Wartime Migration

45

network resources may prove of lesser value. Brokering ties, in contrast, decline in importance as trust withers, but can be massively important in directing and facilitating migration (particularly in its early stages), as well as in facilitating integration within a host society. A brokering position, when it implies influence over vital resource flows, may over time be transformed into a power position. Generally, the members of small networks dominated by powerful individuals are likely to act in concert, and power within networks may play a significant role in causing migration in war, as when local leaders decide that collective flight is the most appropriate political reaction. This is important to take into account when we consider the time dimension and the transformation of networks. For one thing, people’s expectations of how long a particular state of affairs will last are formed in social processes, with networks playing an essential role. These expected durations – for example, the expected duration of a war – will in turn affect people’s willingness to invest in existing networks and build new ones. When expectations are formed collectively, they may be contagious. And, once a particular perception takes hold, it may lead to a massive increase in migration. If so, what has been described as ‘vintages’ in forced migration might best be understood as resulting from a social process of forming and spreading perceptions. To test such a hypothesis, it is desirable to study networks over time – before, during, and after a forced migration process – to find out when they contract, expand, and transform, as well as how familiarity, assurance, and trust come into play. Since networks are mostly fluid and informal, it is hard to investigate them using quantitative techniques. The present work has not even attempted to do so, but is instead based on qualitative interviews and observation of certain communities both in Afghanistan and in exile in Pakistan, at several intervals prior to outmigration, during the exile period, and after the more recent return migration. I will return to the theoretical foundations in the concluding chapter, following the case study analysis contained in Chapters 3–6.

3 Escape Decisions

Social networks have a decisive impact on whether people decide to respond to war by escaping or through alternative strategies. As discussed above, the breakdown of the state and other formal institutions makes networks particularly important in times of war, when they become key channels of security, material resources, and information. Yet, we need to learn more about the exact roles that various types of networks play for people when they decide either to flee or to stay, as well as the ways in which old networks are transformed and new ones formed under the circumstances of war. This chapter will focus on the escape itself, while the role of social networks in exile, return, and resettlement will be left to the following chapters. Undoubtedly, the ‘root’ cause of most migration during wartime lies in a perception of insecurity (Colson 2003; Lubkemann 2008; Schmeidl 1998). The ways in which insecurity interacts with other causes was indicated already in one of the first meetings with the shura of Izhaq Suleman, when I invited views on wartime migration. The first comment, by a prominent shura member, was typical: ‘In times of war, there was one difficulty: people were afraid of death, being killed.’ As if in support of multi-causality, an elder member of the shura immediately added, ‘There were also economic problems due to the war and work was more difficult to find.’ Through subsequent interviews, I found that a significant share of the migrants, including some that fled at the peak of the fighting, would refer to economic problems as their primary motivation for leaving. More importantly, with more in-depth interviewing, it became clear that most people took their decisions with a combination of several factors in mind. My focus on the role of networks does not contradict an emphasis on war or economic deprivation as necessary causes of flight. Rather, in the realization that war itself is rarely a sufficient cause 46

Escape Decisions

47

(Birkeland 2003), the mapping of social networks is used as a tool for attempting to understand the mechanisms by which war translates into peoples’ departure. In this chapter, I will focus primarily on the role of local networks, and only secondarily on transnational ties. This is in contrast to conventional migration studies, which tend to concentrate on transnational networks, by consensus held to have a major role in both encouraging and directing migration by providing information and holding out the prospect of assistance with the process of integration in exile (Kivisto 2001; Massey et al. 1993; Portes 1995). The idea of ‘flight collectives’ – whereby a number of people plan and undertake flight together – takes us in a different direction (Allen and Hiller 1985). Enrolment in a collective has a decisive impact on whether a person decides to flee, on the choice of destination, and on whether the escape is successful. Importantly, however, the networks influencing flight decisions extend far beyond the ties between those who flee together. Split households are familiar from migration studies (Agesa and Sunwoong 2001; Bryceson and Vuorela 2002a; Vuorela 2002). Here, the collective reaches a joint decision that some members will stay while others leave, normally in the expectation that this will be a temporary arrangement. Likewise, ties to people outside the immediate family may also play important roles both in encouraging and in discouraging flight. Extensive migration between Afghanistan and its two neighbors Iran and Pakistan predated the 1978 coup (Centlivres and Centlivres-Demont 1983; Grevemeyer 1988). Afghans living in the south and east were used to a porous border with Pakistan (and, prior to 1947, British India). Many went for work, and nomads had their regular winter pastures, on the other side of the border. In relation to Iran, labor migration had picked up dramatically with the grave Afghan drought in 1971, following an agreement between Afghanistan and Iran to encourage labor migration into Iran. Most of those going there were from western and central Afghanistan, and it has been estimated that 600,000 Afghans were in Iran at the time of the 1978 coup (Rubin 1996: 3). The movement of people from western Enjil to Iran also started in the early 1970s. Among my informants in Izhaq Suleman and Sara-e Nau, however, only a handful had been to Iran prior to the 1978 revolution. Yet, the fact that migration had been going on for several years had placed it on the agenda and established certain contact points – albeit in a much more limited sense than for Afghans close to the border with Pakistan, who had been used to migration for a long time. In Enjil, following large-scale returns from Iran in the first years after 2001, the net migration stream had again turned by 2006, encouraged by what many locals described as an outright recession, caused by poor governance and deteriorating security.

48 Social Networks and Migration in Wartime Afghanistan

While internal displacement received scant attention in the 1980s, there is little doubt that both short- and long-distance internal displacement was common, particularly among people whose access to Iran or Pakistan was difficult. As short-distance displacement also meant short-term displacement, it was in many cases a conscious choice in order to maintain agricultural resources and other site-specific assets at home (BIA 1985: 15–16). Additionally, a large number of people had moved into the cities. While only some 15 percent of the total Afghan population lived in cities in 1979, almost 24 percent did 8 years later (Sliwinski 1988: 16–17).1 As repatriation slowly started in 1989, and took off in 1992, a new form of displacement was created, as many of the returnees could not go back to where they came from but settled in a different location (Kok 2002). By 2004, the government estimated that 30 percent of its population resided in the cities (Opel 2005: 28).2 There has been some research on the departure of Afghan refugees in the 1980s, focusing on household and community responses to macrolevel causes. This research has followed two broad strands (Centlivres 1993: 9–11), whose modes of explanation do not necessarily contradict each other. The first strand emphasizes war and persecution as the cause, in line with traditional conceptualizations of forced migration (see, for example, Connor 1988; Sliwinski 1989b). The other emphasizes the role of cultural or ideological factors, such as the religious imperative to escape from a regime that lacks Islamic legitimacy: the hijra (see, for example, Centlivres and Centlivres-Demont 1988b; Edwards 1986). A well-conceived example of the first strand of research is Kerry M. Connor’s (1988) work ‘The Rationales for the Movement of Afghan Refugees to Peshawar’. Connor’s article is based on interviews with 771 heads of households (nonrandomly selected; interviews conducted 1983–84). While questions were open-ended, the research design did not allow informants to report multiple reasons for departure. Connor finds that the most prominent reasons for departure were: bombing/fighting (25 percent); avoiding conscription (23 percent); imprisonment/fear of arrest (16 percent); being active in the resistance (15 percent); and anticommunism (12 percent). Connor’s findings give an interesting picture of departures to Peshawar in the early 1980s, and her empirical basis is unique. The insistence on singular reasons is unfortunate, in part because the categories of reporting are overlapping and, more importantly, because this builds on an implicit, flawed assumption of unicausality. Connor’s analysis also looks at fluctuations in refugee outflux during the first years of the war (1979–82), and identifies ‘vintages’ that the author relates mainly to particular political events and to the general intensity of

Escape Decisions

49

warfare. This part of her analysis opens up for multiple causation. Speculating on why there is a peak in flight between October 1980 and April 1981, at a time of considerable unrest, Connor (1988: 175) suggests several plausible explanations: economic deterioration, new conscription techniques, and a strong mobilization drive by in-country resistance groups. It is unclear whether the new conscription techniques play on existing network ties, and to what extent resistance groups at this stage had a capacity to organize flight, yet these are possible network factors that would support Connor’s findings. She gives scant attention to the interplay between basic causes of flight and the mechanisms for decisionmaking and execution. The correlation between war intensity and refugee movement also informs the work of Marek Sliwinski (1989b), which is based on a large-scale household survey among Afghan refugees in Pakistan, including close to 2000 families.3 Like Connor, Sliwinski establishes vintages and relates them to geographic origin. Sliwinski assumes such a strong correlation between armed conflict and forced migration that he actually uses the number of refugees in a given time period or from a given location as an indicator of the intensity of warfare in that period or place. Since he neglects the role of intermediate factors in explaining flight, Sliwinski uses displacement patterns to simply deduce the scope of Soviet military involvement. The cultural strand of analysis (Centlivres and Centlivres-Demont 1988b; Edwards 1986; Marsden 1998)4 hypothesizes that migration takes place in adherence to the religious imperative of hijra, replicating the Prophet’s flight from Mecca to Medina in protest against a ruler who lacked legitimacy in the eyes of God: many Moslems consider it a religious duty to leave what has ceased to be a land of Islam for another Moslem country. Such a hejrat protest exodus is always collective, organized and inspired by traditional notables, khan or mullah; it implies the movement of a structured group to another country, usually Pakistan, where it retains its cohesion for a certain length of time. . . . If the deciding factor is not the seriousness of the military or economic question, a particular threat can start the movement. (Guillo, Puig, and Roy 1983: 140)5 The cultural mode of explanation does not necessarily deny the role of war and insecurity as ‘root causes’, and both Connor and Sliwinski mention hijra as an additional motivation for flight. Rather than understanding the hijra as a wholly different form of motivation, I would suggest that it can be seen as a justification for one particular course of

50 Social Networks and Migration in Wartime Afghanistan

action undertaken in a situation of general insecurity or repression (see, for example, Foley 1991; Heimer 1985). This justification may influence the direction and scale of the escape, so we should perhaps see the hijra as representing a particular repertoire that gives prominence to one particular course of action among all courses possible, in which case it is a complementary rather than an alternative explanation to those emphasizing the gravity of the threat. As importantly, and in contrast to other conceptualizations of hijra, in their definition of the term Guillo, Puig, and Roy (1983) emphasize not only the religious duty to escape from an infidel regime, but also how the flight is organized collectively, under a traditional leader (Grevemeyer 1988). This suggests a network-oriented interpretation of the cultural explanation of migration: the networks and patterns of interaction that are rooted in the underlying institution are as important as its inherent cultural code for understanding migration. Leonard Kasdan (1965) has argued that cultural explanations of migration are always at risk of essentialism and fail to account for the selectivity mechanisms that exist within a population. If we emphasize culturally defined network mechanisms and the ways they come into play in migration – constraining certain courses of action while facilitating others – cultural specificity continues to play its role, while we overcome the risk of essentialism. This chapter and the three that follow have a parallel structure, reflecting my ambition to address the key questions and explore the key mechanisms presented in Chapter 2. The first section of each of chapter looks at security, distinguishing between general sources of insecurity (such as war) and specific sources (such as conscription). A second section moves on to material resources, addressing both immediate resource needs, such as money for financing flight, and more enduring control over resources, largely in the form of site-specific resources, such as houses and land. Then, I examine the subject of information, distinguishing between sources saying something about what may happen at the place of origin and sources to information about the area one may escape to. In conclusion, I revisit the main theoretical implications.

Security and escape In the villages of western Enjil, where I conducted fieldwork, security concerns were the prime motivation for departure. Security was the primary reason quoted by all who departed Sara-e Nau – a typical ‘mujahedin village’ – during the PDPA era. In the particular case of Izhaq Suleman, however, there were many who referred to economic problems

Escape Decisions

51

as the primary motivation for migration, even among those who departed during the war-intensive PDPA era. After 1992, when direct war was largely brought to a halt in the area, economic reasons became more significant (but still not dominant), and the differences between Izhaq Suleman and Sara-e Nau largely disappeared. Economic factors continued to represent a significant motivation throughout the Taliban era, not the least in the latter years when there was a severe drought. Flight collectives may play an essential role in executing the actual flight, as Allen and Hiller (1985) found among Vietnamese refugees. Flight collectives are important for the sense of security during migration. In Enjil, the overwhelming majority of those who fled had done so as part of flight collectives. These people traveled not only with their families, but in most cases as members of a group extending beyond the household to include other relatives or co-villagers. The few people who had traveled alone were young men who either left on short notice to avoid conscription or went for work in the post-1992 period, when large groups were ill-suited to bypass Iranian border controls. Allen and Hiller’s flight collectives were largely ad hoc, composed of people who did not know one another beforehand but were introduced through personal recommendations. My flight collectives, in contrast, were largely based on preexisting cohesive ties, with the extended family as the primary focus. Cohesive ties – and the robust reservoir of preexisting trust which they carried with them – played a prominent role throughout the decisionmaking process. Earlier, in Chapter 2, I suggested that, in order to be effective in providing security, collectives need to be of a considerable size, well beyond the scope of a primary cohesive network. Security threats, however, come in different shapes. On the one hand, there is a state of general and total insecurity, where the main threat is constituted by a government army or other organized armed forces, to which a response will require networks of considerable scale. On the other hand, there are more limited security threats – for example, from small criminal groups. The war that hit Herat was definitely of the former kind, and the next section will deal mainly with responses at the village level. The general state of war, however, also resulted in the specific threat of conscription, the avoidance of which requires a response at the individual level and from networks that have a smaller reach. Conscription and its response is the subject of the second section, which reveals how difficult it was to be a ‘free rider’ – to dismember oneself from the crisis – when responding to the massive war that hit western Enjil. Whether the strategy of the larger collective is compliance or resistance, the individual will at any rate be required to make a contribution.

52 Social Networks and Migration in Wartime Afghanistan

Militia or mujahedin? In western Enjil, there was a stark contrast between villages that supported the resistance (mujahedin) and those organizing militias working with the government, as illustrated in the comparison between Sara-e Nau and Izhaq Suleman.6 The mujahedin villages, Sara-e Nau included, were largely deserted in the early Karmal period – the first 3 to 4 years following the Soviet invasion. People report that the general state of war was the primary cause of their departure, although particularly dramatic events – such as an aerial bombing raid – are also said to have been instrumental in triggering departure. In Izhaq Suleman, a militia village, a significant minority refer to the economic situation as the main reason for their departure. This is the case even among those who departed in the periods under Karmal or Najib, in spite of the fact that the village economy was thriving in that period. But also in Izhaq Suleman, the majority of those who left would relate their departure to war. Many of those would add that a key issue was to escape conscription (see below), something that was an everpresent threat in a village where both the local militia commander and the government were continuously looking for new recruits. In the 1980s, the western part of Enjil became the main scene of fighting in the Herat region. The defection of several thousand soldiers from the 17th Army Division in March 1979, led by Ismael Khan, then a Senior Captain, has since become legendary. This was the first major military defeat of Babrak Karmal’s PDPA regime. It came about as a reaction to the government’s call for local army units to help enforce enrolment in its literacy program. During the mutiny, a number of PDPA leaders and Soviet advisers were killed. A massive combined air and ground attack by government forces then brought Herat city back under the PDPA’s control, with estimates of civilian losses ranging from 5000 upward (Roy 1986: 108; Urban 1990: 30–31). In the aftermath of the uprising, Ismael Khan took contact with the burgeoning resistance in Peshawar and established himself as the most prominent resistance leader in the region, with his main bases in western Enjil. The March 1979 uprising led to significant migration from Herat city and the surrounding areas, largely by people who felt that their lives were in immediate danger. Though the main lines of fighting were between the Soviet-backed PDPA government and various resistance groups, these lines were blurred by the presence of local militia groups that were loyal to the government only in principle, were in no way controlled by it, and also maintained contacts with leaders of the resistance. Already in the first half of the 1980s, such militias played a much larger role around Herat than in most other parts of the country.7 The militias expanded their role significantly from 1986 in

Escape Decisions

53

the context of the National Reconciliation policy, when the government ‘opted to fight the opposition groups with their own techniques and accelerated its investment of resources in the formation of rural militias that were often difficult to distinguish from those of the anticommunist resistance’ (Giustozzi 2003: 5). In the Herat area, the militia track was efficiently promoted by Fazl Haq Khaliqyar (Wannell 1991: 53–57), a central PDPA associate. Although they were in principle on opposite sides in the war, there was an implicit understanding between Ismael Khan’s resistance movement and the militia groups that they should avoid fighting each other. Many militia groups were previously resistance units that had responded to the government’s invitation to cooperate, while Ismael Khan himself, as we have seen, had moved in the opposite direction. Through a combination of old-time relationships and smart deals, the resistance leaders were able to move personnel and equipment through most of western Enjil, while the local militias looked the other way. The city of Herat, situated on the major transportation routes to both Iran and the Soviet Republic of Turkmenistan, was considered strategically vital by Kabul’s Soviet advisers. With help from the local militia groups, government forces were able to maintain an acceptable degree of control over the highways. In the mid-1980s, government forces established a second and a third security belt around Herat city, which cut through several villages of western Enjil, making it difficult for the mujahedin fighters and other displaced people to visit their homes. By 1989, the combined forces of the government and the ever-expanding militias caused such trouble for Ismael Khan that he moved his bases to Zindajan district further west. In the summer of 1991, the government launched a wave of attacks against the resistance, destroying the city of Zindajan (Wannell 1991: 2). With little support from Iran, and the Pakistan-based parties unable or unwilling to help, the Herati resistance had been seriously weakened by the time Najib’s government abdicated in 1992, despite the Soviet withdrawal from Afghanistan. However, the changes in Kabul now ignited a new and different political dynamic also in Herat, with the resistance assuming power and Ismael Khan emerging as the region’s almost undisputed strongman. With the exception of those villages whose leaders were able to form an effective local militia and to strike deals with both sides in the war, the western parts of Enjil became totally depopulated. Estimates suggest that only 10 percent of the prewar population remained at the end of the 1980s (UNHCR 1990: 1; Wannell 1991: 26). Those who left had three options: to take up residence in Herat city; to seek refuge in one of the militiacontrolled villages in Enjil; or to go to Iran. Not all three of these options were open to everybody, and the level of choice varied in accordance

54 Social Networks and Migration in Wartime Afghanistan

with access to resources. Each option had different implications in terms of risk – economic as well as physical – and it was extremely difficult to make a reasonable risk assessment at the time of the escape decision. The standard image of a situation with large-scale forced migration is that a breakdown of security forces people to leave (Lubkemann 2004, 2005). This is true in a sense, but some people leave while others stay, and those that leave do so in different directions, with different destinations. The choice between the various options is mediated through social networks. The course of action pursued by one’s primary collective has implications for one’s perception of security, hence one’s decision on whether or not to leave. If a majority of one’s kin or near associates flee, then this is a compelling argument for doing the same oneself. Furthermore, once a majority within a group has departed, it may be even more risky to stay behind than it would have been if the others had stayed too. The situation in the mujahedin villages of western Enjil, situated at the frontlines throughout the PDPA era, largely corresponds to the standard image. The option of staying was simply too risky to be realistic. When the question of departure was raised at the community level, such as in meetings with a shura, the flight decision was typically made in two stages. During the first stage, individuals and households that were particularly exposed, held particular resources, or saw opportunities elsewhere, departed separately, many of them in response to specific threats to themselves. At the second stage, when declining security culminated in a major event, a massive exodus of the remaining population took place. In Sara-e Nau, one informant, Karimullah, recalls a major attack in the first year of Karmal’s government, the Eid-ul Qurban bombing: Sitting in our house, we heard the sound of airplanes, helicopters, and bombardment. We ran to our garden, where we had dug out some trenches as shelter. A bomb hit 20 meters away from my wife, and a wall fell over her. I had to wait till the attack was over before I could dig her out. Luckily, she was only lightly injured. The children did well. On that day, 12 of our villagers were martyred. A few days later, Afghan and Russian forces entered the village. Knowing they were coming, we asked the mujahedin, whom we knew, to leave the village, which they did. They [the Afghan and Russian forces] searched every house of the village, but found nothing. Many had already departed; some left soon after the Eid-ul Qurban bombing; yet it took another couple of years before Sara-e Nau became (virtually) depopulated. A massive air attack in 1983, which killed a number of residents and inflicted huge damage on residential houses, was seen as

Escape Decisions

55

the event that triggered the emptying of the village. Yet, from interviews with individuals, I found that several households continued to have Sara-e Nau as their main residence, relocating only for shorter periods. This was the case for Karimullah and his household, whose decision to stay was motivated primarily by a wish to safeguard the property, and also by other factors, including a mixed relationship with the local resistance leadership. In general terms, what this means is that the self-presentation of a community’s flight tends to leave out a third stage, the continued presence in the village of a select few after all the others have departed. The close link between refugeehood and resistance has been firmly acknowledged in the Afghan case (Harpviken forthcoming-a). At first it was generally assumed that the flight itself was self-organized and that the refugees were mobilized by the resistance only while in exile. Suhrke and Klink (1987: 93–94), however, have suggested that resistance leaders brought along their client populations with the intent of waging warfare from the relative safety of exile. From the accounts of people who left for Pakistan from the areas to the immediate south of Kabul in 1980–81, I know that the resistance helped with the organization of the escape, including reception and settlement in refugee camps. In western Enjil, however, the resistance organizations do not seem to have been directly involved in arranging flights in this manner. Individual commanders, however, did lead larger flight collectives into exile. Two mid-level mujahedin commanders whom I interviewed – from Izhaq Suleman and Sara-e Nau, respectively – reported that they had left for Iran in 1980 with a large group of supporters and their families. Both of them settled in Mashad, which became a resistance center, although the freedom of operation in Iran was far more restricted than in Pakistan. This man from Izhaq Suleman, in his late 40s when I interviewed him in 1999, was bitter about the political manipulation of local commanders: We lived through two decades of war. We were unable to know what was wrong and what was right. Some people encouraged us to become mujahedin and to migrate to Iran. Interestingly, it seems that those who were associated with the mujahedin prior to flight had a relatively clear idea of their destination, and their life in exile was characterized by stable and cohesive networks. In general terms, several puzzles regarding decisions related to mobilization and escape remain open, including the question of the sequencing of escape and resistance activities: under what conditions is flight motivated by the intent to resist versus when do people who have fled war develop the consciousness in exile that inspires resistance (Harpviken 2008; Joly 2002).

56 Social Networks and Migration in Wartime Afghanistan

At a community level, Izhaq Suleman’s militia strategy constitutes the most radical alternative to resistance. Here, the established community leader at the time, Arbab Saidu, used his contact with officials within the PDPA to promote himself as a militia commander, drawing recruits from within the local village and gradually also from the surrounding area. Arbab Saidu’s contacts with the leadership at the nearby tank base remained fragile; it was his access to senior officials of the PDPA in Herat and Kabul, including Fazel Haq Khaliqyar, that proved most important. Poorly developed relationships between militia and army units operating in the same areas seemed to be the rule rather than an exception (Giustozzi 2003: 220–223). As the war intensified following the 1979 Soviet invasion, it was difficult to avoid the choice between loyalty to the government and supporting the resistance. The latter option would almost certainly provoke a full-scale attack on the village, its destruction, and the departure of most of its residents. The Izhaq Suleman dilemma – between resistance and extermination on the one hand, and loyalty to the regime on the other – was fully acknowledged in those neighboring villages that chose to resist, and which were subsequently destroyed and deserted. But Izhaq Suleman made a different choice. Arbab Saidu’s militia was rooted in a qawm, a tightly knit solidarity group, in this instance based both on tribal commonality and on geographic proximity.8 Qawmi militias, with a preexisting network of young men arming itself to become a militia, can be contrasted to militias formed by a leader recruiting its members individually (Dorronsoro and Lobato 1989: 101–102). In the case of Sara-e Nau, the situation became gradually more difficult from 1979 onward, culminating in the 1983 aerial bombardment. When the government set up its second and third security belts around Herat city, the second belt ran right through Sara-e Nau, which for years had been known as a haven of support for the mujahedin. Its inhabitants had little other choice than to flee and resist. This was not the case for Izhaq Suleman. Its inhabitants had more options, and this would later be talked about as an advantage, even by some of those who followed the more costly path of resistance. I asked a mujahed sympathizer from a neighboring village why most people in Izhaq Suleman had sided with the PDPA government: They had no choice, they were next to the tank base, with three to four thousand soldiers. On the other hand, they also received a good salary. We don’t know what sort of relationship they had with the communists before the war, but we don’t think that they had any relations with Russians. Anyhow, they were forced to work with the government since they were neighbors of the tank base.

Escape Decisions

57

The most important gain of the militia strategy was security, albeit in a limited sense. Izhaq Suleman suffered minimal material damage during the war. And, for the significant share of the population who left the village during certain periods, the continued functioning of the community lowered their costs of return. In comparison with the surrounding villages, where the bulk of the population had left by 1983, the pattern of migration was different in Izhaq Suleman, with most of the departures taking place from 1984 to 1987. Again, this can be linked to the security belt, which implied stricter control of people’s movements to and from the militia villages. In the 1984–87 period, the government also paid more attention to the local militias, trying to woo them and recruit them into the army. This raised the costs of residing in Izhaq Suleman for those who risked conscription. Generally, the main cost of the militia strategy for the villagers was the obligation to provide recruits to the militia, and to some extent also to the government army. In both Izhaq Suleman and Sara-e Nau, the position taken by the arbab at the onset of war was of great consequence. The arbab is traditionally a key position in local administration in Afghanistan, a person appointed from the local community, approved by the district authorities, who serves as a broker between the two. The arbab is tasked to represent the state in the community (both in a control and resource broker capacity), and to represent the community in relation to the state (Christensen 1980). As an officially appointed leader whose efficiency depended on community endorsement, the arbab of Izhaq Suleman had been recruited from a family with a strong local standing. Arbab Saidu was arbab in Izhaq Suleman when war broke out, and he became the militia commander. In Sara-e Nau, on the other hand, the arbab had displayed his sympathy with the mujahedin long before the Soviet invasion. As the main contact point between the community and the government, the arbab was the first to face the dilemma of whether or not to support the new PDPA government, and his actions would be interpreted as signals of where his community stood. The existence of a militia village within an area characterized by resistance and displacement gave root to an interesting complementarity between the militia and the mujahedin – between the relative safety of Izhaq Suleman and the lack thereof in the neighboring depopulated villages. The militia–mujahedin relationship was one of mutual interdependence. The militia assisted the mujahedin in accessing Herat city, provided safe havens, and helped to bring in arms and personnel. The mujahedin, in return, rarely targeted the militia or its villages. The relationship was sometimes put to the test, but it never broke down entirely. Many people associated with the mujahedin continued to live in Izhaq Suleman during the war, although this became more difficult by the mid-1980s:

58 Social Networks and Migration in Wartime Afghanistan

I was a mujahed commander during the war. I had been to Iran before. The risk of conscription made me leave soon after Karmal came to power. I fled with a large group of mujahedin. My father and mother, as well as my brothers, lived here [Izhaq Suleman] throughout the war, and took care of the property. My wife and children stayed here until 1985, and I visited them regularly. Izhaq Suleman became a refuge for many people – IDPs in today’s terminology – from the surrounding mujahedin villages. A number of households from Sara-e Nau resided in Izhaq Suleman for shorter or longer periods during the PDPA regime. While located only a short hour’s walk apart on the Herat plain, the two villages did not have close ties prior to the war, only business exchanges. Nonetheless, one out of four of the respondent households in Sara-e Nau spent time in Izhaq Suleman at some stage between 1980 and 1992. Some of those departing Sara-e Nau under dramatic circumstances simply entered Izhaq Suleman and sought out a family willing to host them. This presupposed the consent of the local militia commander. Families residing in Izhaq Suleman as internally displaced persons (IDPs) also took up residence in houses that had been left empty, or in various shacks and sheds. The combination of only loose former ties and affiliation with the mujahedin made the displaced in Izhaq Suleman vulnerable. Dependent on the host family and the local commander, often with miserable housing, these displaced persons had to work hard to maintain the support of their hosts. In the description of Karimullah, a Sara-e Nau resident who spent several periods in Izhaq Suleman: They were in a good situation; we had a bad time there. They said that your sons are with the mujahedin, you should bring them to join the military service. They forced us to work for them. All those between 16 and maybe 30 years of age were taken for military service. If you were above 30 you had to work for them, to construct homes, to cook food, even to peel potatoes and tomatoes. There were also many people from other villages in this area that came to Izhaq Suleman. It was very populated at that time, people even lived in barns. As illustrated by Karimullah’s account, the displaced in Izhaq Suleman could not rely on solid and longstanding ties to their hosts. Rooted in looser acquaintance relations, the guests were vulnerable, and hence had to subordinate themselves to the power of their local hosts. Tellingly, the sustained interaction between people from the two villages during the war did not result in any dense relationships (such as intermarriages)

Escape Decisions

59

that survived the war. In neither community do people refer to the other community as one that they visit regularly. In relation to security more generally, robust access to variegated social networks becomes more critical the higher the level of insecurity – precisely because it is in such conditions that networks are under pressure. By implication, it is only natural that people’s various access to networks plays a major role for choosing how to respond to threat. Indeed, networks constitute opportunities as well as constraints. In order to tackle a massive security threat from the simultaneous presence of government and resistance forces, villagers need ties with larger social entities. As argued above, when the state has no capacity to provide security or when it is itself a treat to people’s security, individuals and households have little choice but to lean on their social resources. Any response will have implications for one’s relationship with the contending parties, also in the form of obligations. In the case of the internally displaced in Izhaq Suleman, the cost of protection was subjugation to local rulers. To disconnect from the conflict without seeking refuge outside the country (or at least outside the larger conflict zone) was virtually impossible.9 Conscription The risk of conscription was both a real motive and a prominent rationale for escape. During the PDPA era, young men ran a strong risk of being picked up either by the army or, in the case of Izhaq Suleman and its surroundings, by the militia. As for Herat, conscription posed a problem also in the first period of Ismael Khan’s rule after 1992 and caused some people to depart for Iran. The Taliban, when they were in power, did not want to rely on Herati recruits and instead concentrated on collecting basij, a tax on the community supposedly compensating for the costs of paying and equipping fighters. Other studies of causes of flight in the early 1980s confirm the importance of conscription. In Connor’s study of refugees in Pakistan, for example, 23 percent related their departures primarily to the conscription threat (Connor 1988: 166–167). Compared to other threats, conscription is very specific. Firstly, it targets particular individuals (men of the right age group), and, secondly, it is relatively predictable (presuming an efficient enrolment regime). In western Enjil, this had the implication that those wanting to avoid conscription had few alternatives but to escape. Network resources could not provide protection from recruitment. In Izhaq Suleman, the ability of the local militia commander to enroll new recruits was his main resource: on this issue he could not afford to be lenient.

60 Social Networks and Migration in Wartime Afghanistan

All people here were armed people; they had to go to the city to keep it safe from the attacks of the mujahedin. This was the duty of Izhaq Suleman, to have 50–100 people in the city for its defence. The only recruiting ground for these people was Izhaq Suleiman. Not surprisingly, in Sara-e Nau and other mujahedin villages of Enjil, young men who had not fled the country would avoid moving into government-controlled areas. In Izhaq Suleman, younger men were at constant risk of being conscripted. Mohammad Arif was picked up by force from his house early one morning. He lost his leg when the truck transporting him struck a mine on the outskirts of the village: I had done my army service in the time of the King [before 1973], but was worried that the militia would capture me again. I was alone here with my father, I had no old son that could look after the family, and I could not escape. I went once, in 1362 [1983–84], for two months to hide from the militia. Six months after I came back, I was captured. It was then that the accident happened. As the story of Mohammad Arif illustrates, being a militia member was risky: many were killed and injured by mines and in ambushes. Since those risking conscription were also the main breadwinners of their families, they faced a difficult choice. The only way to relative safety was to escape, but then they needed to find ways to make money and send it home. Ismael Khan was unique among the Afghan mujahedin commanders in his systematic use of conscription (Giustozzi 2003: 150; Rasanayagam 2003). The family of Ehsanullah, who lived in Iran throughout the PDPA era but returned to Izhaq Suleman in 1992, also had to face the threat of conscription: We had no money when we returned, and no property here. We lived in a rented house, and it was impossible to find a job. My older son was captured during Ismael Khan’s time, and enrolled to join the fighting. My other son was terrified, and did not dare to go out of the house. Myself, I was too old to work. Gradually, our life became miserable. This highlights the severity of the conscription threat. Those who were enrolled did receive pay. And, while the remuneration was quite generous within the professional militias (Dorronsoro and Lobato 1989: 100, 107), the pay for the qawmi militia groups, such as Arbab Saidu’s unit, was low.

Escape Decisions

61

At the same time, however, it was precisely the salary and the privileges related to militia enrolment that enabled many conscripts to provide at least nominal support for their families. Mohammad Osman’s son, who had at first refused to join, had by 2002 picked up a decent job with the military in Herat, reaping the harvest of the once so unpopular engagement. Conscription is an interesting illustration of how a ‘force’ that drives migration may be located in the immediate social environment. In Izhaq Suleman, the militia enrolment was understood as compulsory for any member of the qawm. Conscription was essentially rooted in the familiarity of local networks and informal reporting, not in the bureaucratized workings of state machinery. An analysis of the link between migration and enrolment on the mujahedin side would probably have revealed similar mechanisms: the power inherent in family- and communitybased networks became the basis for recruitment, and the consequences of noncompliance could be dramatic. While the war was the larger cause of insecurity, and also the main reason for the conscription, it is necessary to know about the specific enrolment practices in order to understand how this particular kind of insecurity impacted on flight decisions. Given the targeted nature of the conscription threat, households and individuals could respond in various ways.10 The household at large might elect to stay, as long as the person(s) at risk were in safety – a ‘split household’ strategy. This was common in Izhaq Suleman, where there was sufficient security for most of the population to stay. Men in danger of being recruited often went to Iran, while the rest of the family remained. An interesting variation on the same theme appears among a number of Sara-e Nau households. Here, the larger family opted for internal displacement in nearby Izhaq Suleman, while those at risk of conscription went abroad. Karimullah and his wife, who had five sons all below the age of ten when the war started in 1978, were among the few Sara-e Nau residents who clinched on to their property throughout most of the war, with intermittent stays in the relative safety of Izhaq Suleman. The sons were sent to relatives in Iran as soon as they reached conscription age, and did not return until the mujahedin took over in 1992. Split households were common. In general, it appears that the more specific, or targeted, a security threat, the more effectively it can be met with a network strategy at the micro level, such as a ‘split household’ strategy.

Material resources and escape While the majority of my informants in western Enjil explain their departures through reference to security threats, there is a significant

62 Social Networks and Migration in Wartime Afghanistan

majority, of about one in five, that point to economic factors as their main motive. In addition, quite a few of those citing security threats added that economic concerns also affected their decision. There is considerable variation over time and between villages. None of those who departed Sara-e Nau during the PDPA period refer to economic issues as their main reason. A majority of those who left Izhaq Suleman do so, irrespective of when they departed. Furthermore, economic concerns figure prominently for those who left during the time of the Taliban, particularly in the latter years, when there was a severe drought. In Izhaq Suleman, the economic situation in the 1980s was at least as good as before the war, while it worsened in the 1990s (see also, Dorronsoro and Lobato 1989). By contrast, the destroyed village of Sara-e Nau had virtually no economy left. Despite the presence of a vibrant economy in the militia village, however, a significant share of its inhabitants (including the IDPs) did not benefit. Some of those went to Iran, where the labor market offered much better opportunities than at home. Iran’s attraction for people in western Enjil rested either on access to humanitarian support, which the Iranian government provided for registered Afghan refugees throughout the PDPA era, or on the availability of reasonably paid jobs. In the 1990s, when supplies to Afghan refugees were gradually cut down and registration of new refugees ceased, jobs became the prime source of attraction for new migrants (Fielden 1998). For some, education opportunities in Iran also counted.11 Most of the poorer people found no opportunity to raise money for traveling to Iran and were thus compelled to stay at home, where many of them were displaced locally for shorter periods of time. The network literature suggests that material resources flow along ties between unequals, thus contributing to the maintenance of power hierarchies or even the building of new ones. This rather general proposition is supported by the way in which disempowered IDPs in Izhaq Suleman during the PDPA era found themselves subordinated to local strongmen. In the context of a possible migration for the person at the receiving end, however, the will to grant economic support is compounded by uncertainty about the opportunities to receive returns for favors being made, given the widescale migration. Apart from the importance of sustaining the general livelihood of the family, two types of material resources proved particularly important for the decision to escape. One type is site-specific resources, such as land, housing, or certain types of businesses, which may restrict migration because their long-term value depends on a minimum of maintenance. A second type is convertible resources that can be used to finance the actual flight, the lack of which may restrict migration altogether

Escape Decisions

63

or give a preference for nearby localities. While the former makes it more risky to escape the better off you are, the latter affects those at the other end of the economic scale, preventing the poor from leaving. Maintaining property Real estate and other fixed property may serve to discourage migration. Material resources are particularly constraining when it is impossible to bring them to a new locality. To convert site-specific assets into mobile ones (such as money) is particularly difficult in a situation of war, when many try to sell at the same time and there are few buyers. Emotional attachments to the home serve to reinforce the hesitation toward escape. Few of the informants in western Enjil report of any land transactions. Land is the main site-specific resource, but also houses are a prime example. Land would not be destroyed, of course, and could be left behind in the expectation that it would be possible to return to cultivate it again later. The risk that one’s house would be destroyed was much greater. Many of those who decided to leave their property drew on their social networks to find someone to look after it until they returned. Both in Izhaq Suleman and in Sara-e Nau, the villagers used a splithousehold strategy, with a decision taken collectively that one or several members should leave while others stayed behind to care for the property.12 The decision on who should stay behind would to some extent reflect power relations within the collective (that is, the head of household has decisive influence), as Grassmuck and Pessar (1991) found in their seminal study of migration between the Dominican Republic and the USA. Mainly, however, the decision would be based on an assessment of the relative risk each household member would face by staying. Also of relevance, there would be a calculation of the expected duration of the security threat that motivated the departure. The consequences of split households in Izhaq Suleman and Sara-e Nau were different. Whereas the course of the war permitted property to be maintained and part of the family to stay on in Izhaq Suleman, this was not the case in Sara-e Nau, where virtually all houses were destroyed and everybody had to escape, at least for shorter periods of time. Split-household strategies in response to conscription were discussed above, exemplified by Karimullah’s household, within which all grown boys were sent to Iran, while the other members spent the PDPA era partly in their own destroyed village and partly in neighboring Izhaq Suleman. While sending the boys to Iran was a security measure, the decision to stay behind was for Karimullah and his wife motivated by the need to maintain land and other property:

64 Social Networks and Migration in Wartime Afghanistan

Myself I went to Izhaq Suleman for some time. At that time I frequently came to the village to keep an eye on the gardens, the land, and the houses. I wanted to be in my own country, not to go to Iran. If I died, I wanted to be in Afghanistan. But there was a lot of bombardment and fighting here. Karimullah would bring food supplies to the mujahedin posted in Sara e-Nau. In the long run, Karimullah’s strategy bore fruit, but at considerable cost, as his wife was seriously injured in a bombing raid on the village. When I first met Karimullah in 1999, the mid-sized family farm, as well as the compound, was well managed, partly because all five sons again lived with their parents and contributed to the joint household economy, and partly because his determination – and a good portion of luck – had helped Karimullah maintain the family property through the war. Split-household strategies are most common among those whose flight motivation is mainly economic. In Izhaq Suleman, the 1990s were characterized by economic decline, with little income generated from the fertile soil of the area. Located at the end of the Jui Nau, one of the five major irrigation channels on the Herati plain, Izhaq Suleman was gravely affected by the disruption of water-management systems, and the situation grew worse with the onset of drought in 1998. Ali Reza was in his mid-30s when I first met him in 1999. He lived in a household headed by his elder brother. The family possessed its own compound and 21 jeribs (4.2 hectares) of land in the village: I went seven times to Iran, to the same location in Teheran, where a lot of people from Izhaq Suleman live. My brother had been there twice before me. I went the first time in 1368 [1989–90]. . . and stayed for two years. Now I come back every year; my brother and I rotate, and go for about five months each. My brother came back one month ago. Now I am forced to go. The next time I met Ali Reza was in December 2002. Soon after our last encounter in 1999, after 10 years with the two rotating, both he and his brother had decided to take their families to Iran to stay: I was expelled from Iran one month ago, caught on the way to work. I left with my family for Iran three to four months after we left last time. My brother is also there with his family. My sister [and brotherin-law] is here, they are looking after our property.

Escape Decisions

65

In April 2006, I met both the brothers. Triggered by the rumor that a relative with good connections in the governor’s office tried to take over their land, they had decided both to return with their families. At the time, they managed to make a living in Herat, but both saw new trips to Iran as a part of their household’s economic strategy. This household started out with a split-household practice in response to economic decline, while maintaining the bulk of the household at the place of origin. In response to a further economic decline, they chose to bring most of the household to Iran, while maintaining a minimum presence at home so that their property could be looked after. After several years in Iran, threats to their site-specific capital at home were of a scope that their caretaker sister could not handle, and they decided to shift back in full, obviously with some accumulated capital to facilitate the transition. Nonetheless, should there be a deterioration in the local economy, it would not by unlikely that the brothers (or even their sons) go back to circular migration, while it would probably take a major setback to move the families back to Iran. For some, migration to Iran became more than a question of survival; it also offered an opportunity to build up the household economy. Some were able to strengthen their families’ economic position considerably during the war, through a split-household strategy in which all surpluses were saved and later channeled into fixed property and production in the home community. Such cases were rare, though, as they depended on a large number of working-age men within the family. Attiqullah, a man in his mid-40s, was harvesting wheat on the fields close to Sara-e Nau, alongside three of his brothers and their children, when I met him in 2001: This year the harvest is a little better, since we have irrigation water. The deep wells are owned by businessmen in the city, not by us. We are six brothers and have forty jeribs [eight hectares] to cultivate. Two brothers are in Iran to work, to provide money for the water. Twenty years earlier, this family owned just a small plot of land. During a 12-year stay in Iran (1980–92), the family had used its large labor force to the fullest, saved a considerable amount of money, and invested it in more land. For several years, distant relatives had looked after the property at home; it was only after the return in 1992 that a proper split-household strategy was applied. After the collective return, a minimum of two of the brothers would continue to work in Iran, sending cash back that was used to cover running costs and to invest in more land. Few households had a similarly ‘advantageous’ age and gender composition, but other

66 Social Networks and Migration in Wartime Afghanistan

examples confirm the massive importance of utilizing a household’s labor force in various forms of ‘split-household strategies’. Only families such as the one just described were able to improve their economic position and status at home during the period of war and displacement. Enyatullah of Izhaq Suleman, for example, was disabled while fighting for the militia. When I first met him in April 1999, he had five children, of which the oldest were a girl of 16 and a boy of 14. More than 5 years later, I met him again, together with his now 19-year-old son. They had returned from Iran just a few weeks earlier. They had departed just days after our first meeting in 1999, as a result of the economic hardship. Linking up with Enyatullah’s brother in Isfahan, both Enyatullah and the elder son were able to find decent work. When returning, they moved back into the family house, which had been maintained by a relative, and were relatively optimistic about their economic prospects, in spite of their landlessness. Despite the advantages that can be derived from a split-household strategy, there are considerable human costs. Several alternatives are open to a family leaving its property behind. One is to sublet it, or simply to get somebody in the neighborhood to keep an eye on it. The former strategy was fairly common in Izhaq Suleman during the PDPA era, and the latter was common in Sara-e Nau. In the whole western Enjil area, there seem to have been few instances of political or military power-holders taking advantage of the situation to grab land while its owners were away, although such practices have been widely reported in other regions. Dense social ties clearly helped many families hold on to their site-specific assets. Financing flight The ability to pay for travel costs, necessary documents, and initial settlement is a precondition for a ‘successful’ flight. For prospective migrants, finding these financial resources is a part of the decisionmaking process. The greater the obstacles to reaching a preferred destination, the higher these costs will be. In a study of refugees from Bosnia-Herzegovina who went to Serbia, Lukic and Nikitovic (2004) found that people from the urban middle and upper-middle classes were overrepresented, because migration demands considerable social and economic resources (see also, Grasmuck and Pessar 1991; Piore 1979). People with few means tend either to stay or to travel to a place nearby. Distance, however, is only one of the factors influencing the costs of flight. Bribes, official fees, and the costs of staying in transit may be high. It was always quite expensive to travel from the Herat region to Iran, but became much more so during periods when a ‘closed border’ policy was in place, when people relied on paying

Escape Decisions

67

smugglers or bribes to cross the border. The higher the costs involved, the more dramatic the social selection of migrants tends to be (Koser 2005, 2007). This selection, however, is due not only to differences in personal financial means, but also to differences in terms of social ties that people can depend on to mobilize financial support. The costs of migration prevent some people from departing, encourage ‘split household’ strategies, and enforce a search for credit arrangements – mainly in the form of smuggler contracts. For those who went to Iran, mobilizing the required financial resources was a challenge. Somewhat surprisingly, given the general economic status in the area, the vast majority – some three in four of the interviewees – of those leaving western Enjil during the rule of the PDPA (1978–92) financed their travel themselves. Some had savings, but most raised the money by selling either livestock or household belongings. Again, we see that, in spite of war, there must have been a functioning market, also between the villages in western Enjil and the city of Herat. The fact that most people had time to sell their possessions in order to raise money for the trip also indicates that flight was not entirely acute. People had some time to prepare for the actual departure once they had reached the decision to escape. About a quarter of the people I talked to had found it difficult to raise the money required, and for them a loan was not an option. Generally, credit was severely constrained by the onset of war. A type of ‘poverty trap’ emerged, in which uncertainty undermined the local credit market, thus further disadvantaging those who had scarce economic resources to begin with. Furthermore, this uncertainty was exacerbated by the household’s economic dependence on having members of the proper age and gender. Credit, if available at all, could be gained only within a confined group of close relatives. Inability to raise the money for making the travel to Iran seems to have been a key factor for many of those from Sara-e Nau who became IDPs in the early 1980s, such as those who stayed in Izhaq Suleman. The household of Qadeem from Sara-e Nau spent most of the PDPA era in Izhaq Suleman: We were a large family, and quite poor. My father was not able to get money for transportation or renting a house abroad. Hence, we decided to stay in the area. During the war, our mahalla [village section] had become a battlefield. Due to continuous bombardments, we lost everything we had. When we left for Ishaq Suleman, we did not have anything. When Ismael Khan took power in Herat, we came back to our village.

68 Social Networks and Migration in Wartime Afghanistan

This household lacked the larger network of relatives who might have been able to assist. It was also a household with only one breadwinner and many young children. The few belongings of the family had been totally destroyed in the bombing. The economic vulnerability of the household is spelled out in the inability to escape to Iran. Instead, they had to opt for nearby Izhaq Suleman, where the family became enmeshed in subordinate relationships. The choice of staying close to home may also have been motivated by a wish to stay near to site-specific resources such as land, gardens, or houses. It seems that, when the resource situation permitted it, households would rather pursue a ‘split household’ strategy in which one or a few household members stayed behind while the majority of the household left for the relative security of Iran. In the case of property, however, different factors pulled in opposite directions. The land reform policy and the ‘anti-feudal’ campaign pursued by the PDPA in its earliest years around 1980 implied a particular threat to the large landowners, who therefore had good reasons to stay and defend their interests. Direct threats, however, made most landowners leave. Many of them became key players in the resistance, with a strong motivation to fight the regime that had sought to liquidate them as a class. They were also able to play on old patron–client ties that seemed relatively robust in the face of the dramatic changes caused by war and displacement. With the tightening of Iranian border controls from 1992 onward, a market for human smuggling developed, with a built-in credit option. At the time of my main fieldwork in Herat in 1999, most people saw the assistance of a smuggler as a prerequisite to crossing the border safely.13 The service was expensive: smuggling cost a total of 60,000– 100,000 Iranian Toman (USD 120–200), a sum people would need 2 to 3 months to earn in Iran. Hence, the investment was large, and there was still considerable risk that something might happen along the way. The credit arrangement permitted migrants to undertake the travel without raising the money in advance, and shifted some of the risk onto the smugglers – which was positive from the migrants’ perspective. On the other hand, this arrangement normally implied a commitment to work for one particular employer designated by the smuggler. If an emigrant had no documents that could be used to obtain a work permit in Iran, these credit arrangements left the Afghan migrant heavily dependent on the agreed employer. The smugglers ranged from village-based semiprofessionals to external professionals operating from places closer to the border. Most commonly, contact was established at one of the border

Escape Decisions

69

hotels in the southwestern province of Nimroz. The payment-upondelivery arrangement serves as a form of assurance, and compensates for the lack of trust, serves as its functional equivalent (see, for example, Yamagishi and Yamagishi 1994). This was particularly important when external smugglers were used. For the local semi-professional smugglers, trust is less critical, since they could also rely on informal social control to ensure compliance. Only a minority had used the services of smugglers. Surprisingly, household resources and possible credit from relatives were the migrants’ primary means of financing flight. The only other viable credit arrangement was through the smuggler networks. Lack of means to finance the travel was sufficiently critical to prevent many from leaving (particularly in Izhaq Suleman), and encouraged others to settle for short-distance internal migration. Contrary to the common assumption that financial resources move through ties between unequals, we found that, under the great uncertainty of war, the alternative to personal capital that was employed the most was a loan from one’s kin or other close associates.

Information and escape Once war erupts, the demand for reliable information increases exponentially, while the supply of that information becomes constrained. This was certainly the case when war engulfed western Enjil from 1979 onward. Over time, people built up new channels of information, but these were still less variegated than in times of peace. For the residents of Izhaq Suleman and Sara-e Nau, there was a gradual saturation of transnational ties, with most people coming to have multiple contact points in exile, hence not being dependent on any single one. Information about the situation at the home locality was problematic, partly because existing information channels broke down or could no longer be trusted, and partly because the situation was fundamentally unpredictable. While respondents immediately start talking about their security and economic concerns in relation to a flight decision, there are few who on their own initiative start describing what sort of information they relied on and who their sources were (Williams 1993). The reason is not that access to information is unimportant, or that this is a taboo area, but that people are less conscious of its importance. When the prime sources are relatives and neighbors, access to information is even more easily taken for granted. To get data on information, therefore, the researcher must rely either on targeted questioning in open-ended interviews or on fixed questions in structured interviews. Even then,

70 Social Networks and Migration in Wartime Afghanistan

however, responses tend to focus on the more formal information channels, such as the local headman or the mujahedin group, rather than family, neighbors, and friends. The general literature on migration acknowledges the crucial role of social networks for access to information. The focus, however, has been almost exclusively on the role of ties that span locations (Carling 2008b; Lomnitz 1977; Massey et al. 1987). Information about the situation at the locality one plans to depart from is rarely problematized, despite its obvious importance. Following Mark Granovetter’s work, it is here assumed that bridging ties are preferable to strong ties as informational conduits, in large part because of their capacity to convey information that is genuinely new and original (Granovetter 1973, 1983). It has also been suggested that people who are deprived of basic means, economic or otherwise, tend to rely too heavily on cohesive ties with people in a similar situation. The uncertainty of war may, in a related manner, undermine the effectiveness of bridging ties and further increase dependence on cohesive ties. Local information What kind of information did prospective migrants in Western Enjil have access to? It must first be emphasized that we are not here talking about villages that had been secluded or isolated prior to the war, but about an area where people traveled widely and where many possessed extremely complex and variegated networks. This is well known from the anthropologist Reidar Grønhaug, who conducted fieldwork in Enjil in the early 1970s (Grønhaug 1978, 1988). The onset of war imposed significant constraints on the traditional pattern of free movement, yet many people from the mujahedin-dominated areas of western Enjil continued to travel regularly to the government-controlled Herat city. Travel was certainly more cumbersome, but business and other forms of interaction did not grind to a halt (Wannell 1991: 35). A majority of my informants referred to people in formal leadership positions when asked about their key information sources, as well as when asked, as a follow-up, whom they contacted for information in the midst of war. For Izhaq Suleman, the key source would be Arbab Saidu during the PDPA period; later, key sources would include shura representatives, particularly the person selected to represent the mahalla of residence.14 For Sara-e Nau, the local mujahedin leaders were the primary information sources during the war; later, people relied on the arbab (a former mujahedin commander) or representatives of the shura. A number of respondents added that the reason for contacting the formal representatives was that

Escape Decisions

71

the latter had access to information from the district administration (woleswali). Hence, formal state structures did not dissolve totally as a result of the war, and people continued to feel a need for information from above, despite the fact that state institutions had been transformed and become an integral part of the war. It is particularly striking that people seemed to rely so much on information from the state’s representatives during the PDPA era. It is also interesting to note that traditional types of local leaders continue to play a role as middlemen at the interface between the local community and larger organizational structures. It emerged from the open-ended interviews, however, that the actual information gained through formal channels tended to concern the events of yesterday and was of little use in judging what to expect from the future – the latter, of course, is what is most relevant for flight decisions. As it was impossible for most people to maintain cordial relations with the government and the mujahedin at the same time, it might be expected that people would choose either one or the other. Interestingly, the case was rather that most people tried, despite the difficulties, to mobilize ties so they could get at least some information from both sides. For residents of Izhaq Suleman, the easier access was to government sources via the local militia commander, but there were also opportunities to get information from the mujahedin, some of whom had family there. For Sara-e Nau, which was nominally a mujahedin village, the situation was the opposite. Here, mujahedin information was most easily available, while those with good contacts in neighboring villages (for example, Izhaq Suleman) could get access to information also from the government side. In responding to a question about who were contacted for informational updates, one young Sara-e Nau resident gave a typical answer: At the time when we lived in Izhaq Suleman we would ask Arbab Saidu, the commander of the militia. If it related to our own village, we would ask the commanders of the mujahedin for help. As the quote indicates, contact with those in formal leadership positions was used primarily for solving problems here and now, rarely to lay the foundation for assessing likely developments or emerging threats. Given the tendency for local leaders in Afghanistan – including the shura – to have a reactive rather than proactive function, this is not surprising (Harpviken, Suleman, and Taksdal 2001: 11–12). This tendency was only exacerbated by the war, since increasing uncertainty makes proactive action even more futile. Implicitly, however, information sought from local leaders about the current situation will have fed into the process of assessing likely developments.

72 Social Networks and Migration in Wartime Afghanistan

Yet, in the collection of information that would serve as a basis for a migration decision, denser networks with extended family outside the household played a more important role than contact with formal leaders. Exchange of information that could be used to gauge future developments in the home region was mixed with discussions about how to respond, and the sensitivity of such deliberations limited the choice of discussion partners. The reliability of information from one’s kin and friends is unlikely to have been any better than that of other sources. Additionally, there is the inherent risk to which Granovetter alluded in his studies of job-seeking: dense ties rarely give access to new information, but tend only to confirm what one already knows. Ultimately, however, in a situation of an unfolding war, all information is uncertain, and the hunt for information is as much a question of taking part in collectives for assessing possible threats and response options as it is about getting access to information that is genuinely different. Transnational information Transnational networks are essential to migration decisionmaking because they provide access to information about the conditions at the potential destination, as well as about the opportunities for getting there. News of good opportunities encourages new migration, and hence contributes to upholding migration streams (Massey et al. 1993: 448; Ritchey 1976). A majority of those of my respondents who had themselves escaped at some stage during the war had established contacts at the destination before they left. Those who did not have such contacts were all early leavers from the period 1979 to 1981. There had been considerable labor migration following the 1971 drought, and this was stimulated by the subsequent liberalization of travel restrictions by both the Afghan and the Iranian government. In the course of the first 3 years of war, there was a virtual saturation of information networks, so that almost everybody got access to contacts in Iran, which provided them with information (and potentially with other assistance). Hence, although the push factor of the war was the primary cause of departure, we get a sequence of departures that does not necessarily reflect the severity of armed conflict but is rather reminiscent of chain migration. A few pioneers attract others and help make sure that all those who follow have somebody they know at the destination who can support them. Persons who knew Iran before the war seem to have been particularly prone to leaving. Among those of my respondents who fled during the war, about one-fourth had been in Iran before. Almost all of them had maintained contacts in Iran. All those who had been in Iran before went back at some

Escape Decisions

73

stage in the 1980s, with the exception of one family from Sara-e Nau, who could not find the resources to go and remained in the Herat area through the whole PDPA era, most of the time staying in various localities outside of their home village. Generally, those who stayed in Afghanistan were people who had never been to Iran and had no close associates there. Lack of means of communication was a major constraint on the exchange of information between western Enjil and those residing in Iran. Throughout the 1980s and 1990s, telephone contact was possible only from the telegraph office in Herat. While it got much easier to travel to Herat after 1992, the telephone remained little utilized as a means of communication.15 Sending letters through the post was problematic all the way up to 2001, since there was no functioning distribution system in Afghanistan. By far the most frequently used means of communication was letters sent with people who served as couriers. To write letters was quite a challenge, given that a majority of the population was illiterate.16 And, it was not always easy to find a person who could be trusted as a courier. Given these constraints, written communication was infrequent and had to focus on the most important issues.17 Personal visits – either by elder men or by members of the mujahedin – became the most important form of communication. Such people could carry letters, but more importantly they conveyed oral messages and reported on their impressions The following comments come from two persons from Izhaq Suleman who left early and spent most of the PDPA era in exile: We had contact through those elders who visited the village. The government did not bother them, since they were old. They were a means of communication between us and our relatives and friends remaining in Herat. The local military commander of the government, named Arbab Saidu, had relations with both the government and the mujahedin. He supported us to get to our village. Given the constraints on communication, it is interesting to note that most households were still able to maintain contact with relatives residing in Iran. It is possible that this did not work equally well both ways, and that a share of those families who had moved collectively to Iran were unable to maintain contact with home.18 At an early stage of the war, networks between the communities in western Enjil and the possible destinations in Iran were sparse. The early leavers would serve as informational bridgeheads. Within the first 3 to

74 Social Networks and Migration in Wartime Afghanistan

4 years, most people in Sara-e Nau and Izhaq Suleman would have one or more close associates in Iran, and thus enjoyed multiple sources of information. Hence, the most important supply of information came through one’s kin, but such information was often complemented from other sources, which were less closely associated.

Conclusion In the case of Izhaq Suleman, the local arbab – a classical broker type recruited in the village while serving as the link to the government – had very good contacts within the PDPA, who trusted him and granted him considerable room for maneuver. This enabled the village population to dampen the impact of the war by setting up a militia, in support of the PDPA government. With a major army base at the village outskirts, it seemed clear that the village had to choose between loyalty and extermination. The Izhaq Suleman militia was of a qawmi kind, based on an extended network of co-residents and intermarried families. Indirectly, the militia strategy also benefited people outside the village, including local mujahedin and people displaced from neighboring villages. While a military organization based on a qawmi-type of network provides protection for the villagers as a whole, it is also in itself an internal security threat for some of them. The militia option entailed a major cost through the constant threat of conscription, and this problem continued under the rule of the mujahedin (1992–95). Not all young men wanted to enroll in the militia. Nor did all families want to support it, and they notably did not want to assist the militia in conscripting young men for the army. Under the circumstances, however, when virtually one’s whole relevant network is recruited en bloc into a militia, it seems extremely costly – in terms of network resources – to be opting out. Conscription is a highly specific threat, targeting the individual. In many instances, it also requires an immediate response. Most often, the conscription threat is tackled at the household level, with families mobilizing transnational links to send away their sons before they reach the age of conscription. A common response to the conscription threat is thus to split the household. Networks give access to material resources, and network resources are pivotal to maintaining property. Without any capital, migration may be impossible, since it involves considerable costs. At the same time, however, real estate is site-specific and serves as an obstacle to migration. In Afghanistan, the constraining power of land ownership was counterbalanced by the PDPA’s targeting of the ‘feudals’. In Izhaq Suleman and Sara-e Nau, a split-household response was common,

Escape Decisions

75

with one or more members of the family staying behind to care for the property. The literature on networks suggests a tendency for exchange of material resources to take place in relationships based on inequality, thus reinforcing existing power relations. My findings do not confirm this. In western Enjil, people had access to virtually no other material resources than their own and those they could borrow from close relatives. The poor, who were not only materially poor but also deprived in terms of social networks, did not expect to get any credit from the better-off. Several factors may explain the lack of conventional credit opportunities: general uncertainty (about value of investments, future labor), high degree of mobility (holding loan-takers accountable is difficult), and a general decline in trust. The only credit option was through engaging in contracts with smugglers, where the migrant would pay by working for an associate of the smuggler upon arrival in Iran. This gave the smuggler – and the Iranian employer – considerable power over the migrant, who would most often be considered an illegal immigrant by the Iranian authorities. The difficulties of obtaining credit illustrate the broader trend that, in times of uncertainty and upheaval, small cohesive units rely increasingly on their own economic resources for survival. The outcome is familiar, but is here played out in a particularly refined manner: those who need material assistance cannot get it; those who can get it do not need it. Dense networks linking households with their extended family play a critical role for assessing available information when migration decisions are made. When it comes to trans-local information, particularly exchange between the origin and those residing in Iran, the reliance on kin-based networks seems firm. This is also the case if we consider information used to form expectations about future developments within the locality. Again, family and friends play the key role, and even more so when the information exchanged is sensitive. At the same time, most of the people who remained in western Enjil during successive periods of war had access to formal leaders at the local level and consulted them. When people get embedded in relatively homogenous information networks, the likely effect is – in line with the Granovetter thesis – that there is little access to original, possibly also more reliable, information. Such constraints may have prevented parts of the population from migrating, simply because they lacked the relevant information. Reliance on a few small networks – as well as the absence of contacts that can provide reliable information on issues of vital importance – interacts with factors in other domains to constrain the room for informed choice, hence it also serves to exacerbate vulnerability.

76 Social Networks and Migration in Wartime Afghanistan

As war leads to an erosion of the foundation for trust, most people will have to rely on small and denser networks, privilege cohesive ties over bridging ones. If one maintains the traditional focus on the socialsupport function of networks, the privileging of cohesive ties is not necessarily disadvantageous, and it is unlikely to represent a dramatic change from the prewar situation. A focus on dense ties, however, may be more dramatic for security, economic support, and information. On the security front, the course taken by one’s larger network is decisive for the options available to the individual: If your extended family network sides with the mujahedin, it is difficult – or even unthinkable – to choose a different path. In the economic domain, credit opportunities are restricted. Information-wise, declining trust implies narrower networks with access to the same sources. The expected duration of an exile plays a major role in decisionmaking. At the early stages of a conflict, people tend to expect the absence to be short. This makes departure a less dramatic decision, and may also encourage the selection of destinations nearer to home. As the conflict extends in time, there is a gradual realization that the effects of war will extend even beyond a future settlement; those who leave understand that their absence may be long-lasting and therefore represent a long-term investment. There is, therefore, a greater will to invest heavily in migration, and to take more dramatic choices in terms of violating existing ties. A similar sequence in how one perceives the conflict is what drives stepwise migration, when people who have performed a short-distance migration of a temporary character move on to a long-distance migration of a more permanent character. In general, the same tendencies work within the domains of security, material resources, and information. War restrains the networks that people can activate to get access to critical resources, fostering increasing dependence on the household and the extended family network, while undermining bridging ties that could help connect with more variegated environments. The formation of flight collectives – which, among my Enjil respondents, were based on cohesive ties with the family as the focus – is only one way in which dense ties manifest themselves. They may be equally important in preventing migration, or in cushioning the war threat so that departure appears unnecessary. Dense ties do not depend on co-location; rather, there is a prevalence of transnational (or ‘trans-local’) households.

4 Integration at Exile

What kind of coping structures did Afghans develop while in exile? Could they build on prewar migration experiences? How were networks reconfigured among the refugees, between them and their hosts, as well as between the refugees and the people back home? Refugee studies have primarily focused on the ‘refugee condition’ in exile, and have only recently come to show interest in the continuities that bind refugees to other times and other places (Horst 2006b: 45; Koser 2007). Here, I will focus both on the ties people relied upon in exile and on how refugees coped with the challenge of maintaining and developing their ties with their home region in Afghanistan. The social networks that people maintain or build, as well as depend on, while living in exile are rarely confined to a particular exile locality. The continuities (and ruptures) with prewar collectives remain an underexplored theme, as is the flows of various content between exile and home communities even in situations involving extensive restrictions on travel and border crossing (Hugo 2005). Exile collectives take a form that reflects not only prewar networks, but also the opportunities inherent in the structures set up by host-country government bodies and international organizations (Jacobsen 2002; Michel 2002). In this regard, a comparison of the very different refugee regimes in Iran and Pakistan is informative. This difference had significant implications – both intended and unintended – for network dynamics within the refugee populations of the two countries. Refugee studies have focused on social networks either as coping mechanisms among refugees (in the form of dense ties within a primary group of refugees, most often family-based) or as a means of integration in the host society (in the form of dense or brokering ties between refugee and host population, not rarely based on emergent brokers within the refugee population) (DeVoe 1981; Jacobsen 2005). Here, emphasis will be on the 77

78 Social Networks and Migration in Wartime Afghanistan

former, but with attention paid to how relations with the host country influence these coping mechanisms. Networks are not seen to necessarily coincide with a geographic locality, but may equally well be in the form of ties between localities. Often, existent studies do not distinguish between the various contents that flow in networks, but assume that the network as such is the essence and that networks are instrumental regardless of what flows through them – security, economic support, and information (see, for example, Williams 1993). Refugees depend almost completely on the opportunities and support provided to them by their hosts; hence, the functions of various network structures in exile will also reflect the type of refugee-management structure established by the host country (Castles and Miller 2003; Khawaja 2003).1 Unlike most studies of war-related migration, which are based on data collection in exile, the primary date for this study is gathered at origin. Hence, the study does not include fieldwork in Iran among refugees from Izhaq Suleman and Sara-e Nau.2 This chapter builds on interviews made in Afghanistan with people who had stayed in Iran for extended periods of time. Most of them had returned to Afghanistan on a permanent basis, while some were on short visits. It proved difficult to obtain data of satisfactory depth on the exile condition. As a researcher, I could not directly observe the condition of exile, and people’s motivation for talking in detail and with considerable nuance about their experience in exile was much lower after their return than it probably would have been had I met them in Iran. Notably, those on a short visit were much more eager to talk about this issue. I have consulted published material, but as Pierre Centlivres (1993: 18) pointed out in his 1993 state-of-the-art paper on Afghan displacement, there has been a ‘quasi-absence of research on Afghans going to, or residing in, Iran.’ On Pakistan, however, there has been extensive research, at least until international interest declined with the fall of the Najib government in 1992.3 A series of publications from the post-Taliban period have proven extremely useful, and have been used extensively to cross-check my limited primary data (in particular: Abbasi-Shavazi et al. 2005a, 2005b, 2005c; Jamshidia and Ali Babaie 2002; Jamshidia and Anbari 2004; Monsutti 2004b; Stigter 2005a, 2005b; Stigter and Monsutti 2005). I have also benefited from several visits to Iran, where I visited refugee camps and the main settlement areas, and discussed the refugee issue with representatives of the government, UNHCR, and NGOs. All my respondents stayed with relatives while in Iran. Most of them relied on kin-based networks that extended well beyond the immediate family. Through these extended family networks, people were able to cooperate closely in coping with the multiple threats they faced in exile,

Integration at Exile

79

such as unemployment, illness, loss of residence, or even deportation. As noted in Chapter 3, a majority of the people leaving Enjil fled in groups, ranging from a couple of families to much larger flight collectives. It would be interesting to know to what extent these flight collectives stayed intact while in exile. My data do not allow me to estimate the level of interaction through the whole exile period, but it is clear that people who were enrolled in a particular network prior to and during the flight tended to go to the same place in Iran, and this makes it more than likely that the flight collectives did not collapse upon arrival. There are no indications that collectives were deliberately split by host-country institutions. Therefore, given that flight collectives were largely based on a shared origin in the same village, it would be very surprising if they were dissolved upon arrival at the destination. When Iran and Pakistan became hosts to large-scale refugee populations in the early 1980s, they could build on a long history of migration across their boundaries with Afghanistan. The two countries approached refugee management very differently, not only when it comes to cooperation with international institutions but also in relation to forms of settlement, freedom of movement, and regulation of work and business. In the case of Pakistan, Afghans were given identity passes exempting them from illegal-immigrant status and entitling them to humanitarian assistance. While assistance entitlements ended by the mid-1990s, the identity passes could still be used to establish eligibility for a UNHCR repatriation package. By 2000, new arrivals from Afghanistan were considered illegal immigrants, and the border was officially closed. Most of those who arrived in Iran between 1979 and 1992 (or resided in Iran at the time of the Soviet invasion) were recognized as ‘refugees’ and received refugee documents, referred to as ‘blue cards’.4 Those who possessed blue cards had the right to settle, and enjoyed access to free education and subsidized health services, as well as to work (Monsutti 2004b: 166). From 1992, when the mujahedin took power in Kabul, Afghans were no longer granted refugee status. By 1997, Iran stopped registering new arrivals, and launched a major registration exercise for those already residing in the country.5 Deportations became more frequent, and, together with prolonged reregistration process, stimulated considerable uncertainty among Afghans in Iran (Adelkhah and Olszewska 2007). In Pakistan, an extensive international aid apparatus was set up for the refugees, providing food aid, as well as health and education services, with generous external funding.6 The majority of refugees in Iran were not settled in camps, but settled down themselves among the general population. UNHCR (2004: 9) estimates that no more than 2.5 percent of the Afghan population in Iran ever lived in camps. Most refugees settled either in the

80 Social Networks and Migration in Wartime Afghanistan

larger cities or in the countryside along the Afghan border. Following Art Hansen’s work on Angolan refugees in Zambia and Lisa Malkki’s work on Burundian refugees in Tanzania, it is usually assumed that camp refugees are tightly controlled by the host governments, whereas self-settled refugees are less strictly controlled (Hansen 1981, 1990; Malkki 1995b). The Afghan refugee condition in Iran does not fit this pattern. Iran had effective control systems, including checkpoints along major roads (particularly in the border regions). The settlement pattern in Pakistan was very different from the one in Iran. In Pakistan, large camps were established from the first moment, primarily in areas bordering on Afghanistan (Dupree and Dupree 1987: 17–18).7 There were no fences around the camps, and residents were free to move in and out of the camp – and around the country – as they wanted (Schmeidl 2002). In general terms, therefore, the self-settled refugee population in Iran was more effectively controlled by the host state than the population in Pakistan’s camps. Scholars specializing in integration have emphasized cultural explanations for the different fates of refugees in Iran and Pakistan. The governments of both countries have referred to their duty as Islamic states rather than their international obligations as the reason for accommodating Afghan refugees (Turton 2003: 14). The religious explanation builds on the institution of hijra, where the status of the muhajir is associated with the ansar, the hosts in Medina who welcomed Muhammad and his followers after their flight from Mecca (Centlivres and Centlivres-Demont 1988a: 145). For Iran, the fact that many of the refugees arriving there belonged to Afghanistan’s Shia minority also played a role (Glazebrook and Abbasi-Shavazi 2007). Several analysts have emphasized the existence of a common normative code among Pashtuns on either side of the Afghan–Pakistani border – Pashtunwali – which prescribes hospitality to ethnic brethren (Centlivres and CentlivresDemont 1988b: 87). Common religious and ethnic identities, and shared normative frameworks, have no doubt been important both for the refugees themselves and for their hosts. Cultural affinities, however, are mediated through concrete interaction between refugees and officials, as well as other residents, in the host country; and, in such concrete interaction, economic and political interests, and not least security concerns, are often so important that they override culturally imposed obligations. The structure of this chapter is similar to the one in Chapter 3. I will first look at security in exile, and relate that to debates about the role of exilebased participation in the military struggle back home (‘refugee warriors’). The second section will concern material resources, and will examine how Afghans in Iran accessed the labor market and how economic flows tied the refugees to their home communities through remittances. Finally,

Integration at Exile

81

I will look at information flows: among the refugees, between them and their hosts, and between the refugees and people inside Afghanistan.

Security and exile Economic motives played an increasing role for flight, particularly from 1992 onward. Yet, the main motive among Afghans for migrating to Iran or Pakistan was all the time one of concern for security. To go into exile, however, does not necessarily mean to disengage from a war. Many Afghans used the relative safety of exile to get military training and to prepare for military missions inside Afghanistan. At the same time, however, the safety of exile depends on the goodwill of the host state and the host population, so the refugees needed to refrain from activities that could jeopardize their position both in Iran and in Pakistan. Refugee warriors? Afghan refugees in Pakistan were cited as a prime example of Zolberg, Suhrke, and Aguayo’s term ‘refugee warrior communities’ (Zolberg, Suhrke, and Aguayo 1986, 1989; see also Suhrke 1995a). Several studies of the same case have followed (Grare 2003; Lischer 2005: 44–72; Shahrani 1995; Terry 2002: 55–82). Most analysts of Afghan refugee mobilization emphasized the role of the international assistance community (and, in the case of Zolberg, Suhrke, and Aguayo, the legitimacy derived from the international refugee regime), as well as the role of states (host state and others) (see also Zolberg 2008: 158–172). In Pakistan, the provision of international aid was made conditional upon membership in one of the resistance parties (Amstutz 1986). M. Nazif Shahrani, looking at how religious institutions (hijra, jihad) legitimize exile-based resistance, is alone in emphasizing factors internal to the refugee population. When these researchers touch upon issues of internal social organization, all allude to ways in which whole communities shift to exile under their own leaders and with their social and political systems intact (Harpviken 2009). The ‘refugee warriors’ literature makes only scant reference to Iran; and, when it does so, it is to suggest that there was no refugee warrior community there. One reason that refugee mobilization was less intense in Iran than it was in Pakistan is that, after the start of the Iran–Iraq war in September 1980, Iran could not afford to upset the Soviets (Hunter 1987: 258–259).8 Iran therefore concentrated on engaging with Shia groups (and mainly those sharing the radical ideology of Iran’s new regime) that were only modestly engaged in the war against the PDPA and the Soviets. Frédéric Grare (2003: 89), for example, claims that ‘Iran

82 Social Networks and Migration in Wartime Afghanistan

concentrated the Afghan refugees in camps located far away from the Afghan border, and prohibited them from engaging in political activities’. He goes on to suggest that the Iranian way of dealing with the Afghan refugees may inspire refugee management in other politicized contexts. Knowledge of the political and military organization of refugees in Iran is scarce, not necessarily because Iran was not engaged in refugee mobilization, but because a strategy was consciously adopted to restrain the most conspicuous engagements in the war and to limit international interaction with the refugees.9 An indication that refugees in Iran were military active, however, is given in a report by Peter Marsden, who interviewed returnees in Farah province: It appeared that those in the refugee settlements had been encouraged by the Iranian government to send men to fight as mujahedin and had been given weapons and documentation to facilitate this. (Marsden 1997: 12) About half of the people that I interviewed reported that they had been somehow associated with the mujahedin while in Iran. Not surprisingly, there were fewer mujahedin associates among respondents with an exile background in Izhaq Suleman than was the case for Sara-e Nau.10 Among those from Sara-e Nau, about one-third had taken an active part in the armed struggle. They reported that they had spent from 6 to 10 months every year hidden somewhere in the vicinity of the village. All of these had families living in the same neighborhood in Mashad, the nearest major city to the Afghan border, a day’s journey by car from Herat. While all maintained links with various parties, they had been fighting under a local commander belonging to the Hezb-e Islami party led by Younos Khales. The group catered to people of varying ethnic and religious origin, and included both Shia and Sunni, which is quite remarkable given Khales’s fame as a staunch anti-Shiite. The members of the Sara-e Nau mujahedin went into exile at various times between 1980 and 1983. The group shifted from Hezb-e Islami (Khales) to Ismael Khan and Jamiat-e Islami in the mid-1980s, but remained intact under the same commander. They all returned in the first days after Ismael Khan took power in Herat in 1992. Habibullah was the head of a household that consisted of some 40 people at the start of the war. They were constrained by the costs of taking so many people abroad, as well as by concerns related to their housing and land. Hence, they were late leavers. Six members of his household fought with the mujahedin both prior to departure and after

Integration at Exile

83

having settled in Mashad. Habibullah reflects upon Iran’s ambiguous stance as a host country for refugee warriors: Iran did not give us weapons. When we had an injured soldier and brought him to the border, we had to contact our office in Teheran to get the documents, it could take 2–3 days. When papers were arranged, we were not allowed to accompany him, but had to leave the injured person with [our] party people in Iran. The commander of the Sara-e Nau group was among the first to leave the village in 1980, leading a collective of 12 families, all relatives. The group went straight to Khwaja Rabi in Mashad, and received the blue card after 6 months, which gave access to subsidized food, education, and health services. With the family in safety, the commander spent most of the year in Afghanistan, joined by an increasing group of relatives and fellow villagers. Most members of this group returned within weeks of the mujahedin takeover in 1992; the commander was traveling with basically the same set of people that he had left with 12 years earlier. From Izhaq Suleman, several respondents – all members of the same family line – spent the years from 1984 to 1992 together in Taybad, which is close to the border with Afghanistan. Most of them had originally fought for commander Safiullah Afzali, who was Ismael Khan’s main rival in the Herat region (although both men were associated with the same party, Jamiat-e Islami). Upon Safiullah’s death in 1986, the mujahedin from Izhaq Suleman reconciled with Ismael Khan. All the respondents from this network report a similar pattern, whereby they stayed 3 months at the front in Afghanistan, then 3 months in Iran. During the periods in Iran, they worked to gather money for the household. At the same time, however, they all report a much higher level of support in terms of food, education, and health services than those refugees who were not active in the resistance. This account is rather consistent among respondents from this particular network, and it is generally supported by data from other interviews. Although one should be careful to use this as a basis for firm conclusions, it does indicate that Iran actively supported resistance groups operating from its soil, which challenges the common assumption that Iran’s need to maintain cordial relations with the Soviet Union prevented it from actively supporting mujahedin movements (with the exception of radical Shia groups) (Amin 1982: 141; Saikal 1989: 57–59). When the Taliban had seized power, Iran continued to work with armed Afghan opposition groups. A descendant of one of the mujahedin

84 Social Networks and Migration in Wartime Afghanistan

that was resident in Taybad, Iran, in the 1980s was enrolled with Ismael Khan’s forces in Iran at the very end of 1999, soon after my first interview with him. In December 2002, he said: After you were here the last time, I went to Iran for some time. I was captured in my house, the Taliban said that you are with the mujahedin, they kept me in jail for 9–10 days, then I paid some money to be released, and I left for Iran. I came back 3–4 months before the fall of the Taliban, having spent 6 months in Mashad as a laborer. When I came back, I went to Mazar [with Ismael Khan]; we were ready to attack the Taliban. Also here in Herat, there were people who were ready to attack the Taliban at that same time. I am now working with the military. Iran continued (and became more open about) its support to armed opposition groups in the 1990s. The quote above also illustrates how the ‘exile warrior’ engagement is inherited from one generation to the next (the father was incapable of fighting owing to a war injury), and that such engagement may pay off eventually in the form of jobs for the military. The exile engagement with the mujahedin is also reflected in where people settled in Iran. The earlier people left, the more likely they were to go to Khorasan province, which is close to the Afghan border and served as a hub of the Afghan resistance. Those who were active with the mujahedin also chose Mashad, the capital of Khorasan, as their first destination. Thus, people from Sara-e Nau, a mujahedin village, often spent their exile in Khorasan, while this was not often the case for people from Izhaq Suleman, a militia village where a smaller share of those leaving, even at the early stages, joined the resistance. When we move further into the 1990s, during the reign of the mujahedin and the Taliban, refugees seem to have given higher priority to work opportunities, which were far better in main cities like Teheran and Isfahan than in Mashad. Iran served as a base for armed resistance in Afghanistan, but in a manner very different from Pakistan. In Pakistan, the government stood unanimously behind the resistance and cooperated strongly with various states and international organizations in providing political, military, and humanitarian support (Bhatia and Sedra 2008: 45; Sinno 2008: 153–154). Iran took a much more ambiguous stance, largely because its relationship with the Soviets was vulnerable and also because, unlike Pakistan, it could not draw on Saudi or US support. This did not prevent various resistance organizations from operating out of Iran, but they enjoyed far less backing and freedom of movement than those operating in Pakistan. This frustrated resistance leaders in western Afghanistan considerably, and was

Integration at Exile

85

seen as a major constraint on the effectiveness of the resistance in the area bordering on Iran (Wannell 1991). Interestingly, however, several armed organizations with an ambiguous identity – shifting between militia and mujahedin – such as the Hezbollah group operating from Jibrail in Enjil district, received considerable support in Iran. For the Iranian government, engaged as it was in an all-out war with Iraq, the main aims of its refugee policy in the 1980s seem to have been to capitalize on Afghan labor and build organizations it could work with in the future, while avoiding upsetting the Soviets and risking an additional war on its eastern front. After the end of the Iran–Iraq war in 1988, the resources were used more actively, first in the struggle for influence in the mujahedin government (1992 onward), later in active combat against the Taliban regime. The constraints placed on Afghan refugees by the Iranian government kept in check the emergence of new Afghan leaders (with some exceptions among the Shia), whereas in Pakistan a system was set up that provided all sorts of support to the resistance and the refugees, creating ideal opportunities for refugee entrepreneurs. Iran relied more on traditional leaders and their cohesive networks as a way to keep refugee involvement in the Afghan war as inconspicuous as possible.11 In Pakistan, traditional leaders were often replaced by new ones who proved more competent in liaising with Pakistani and international providers of humanitarian and military support. Everyday security Security concerns are rarely the sole reason why people move, yet they are a prime motive for wartime migration. Some individuals may even move from areas where the security situation is not so bad. On the other side of the equation is concern for security at the destination, which may be threatened by institutions of the host state, groups within the host population, or fellow migrants. The degree to which a migrant network is effective in coping with exile security challenges depends on the interface between refugee and host populations, as well as self-organization among the refugees (as opposed to the extensive presence and regulation of institutions of the state) (Hollifield 2000; Koser 2008; Stølen 2005). While the refugees in Pakistan enjoyed virtually full freedom of movement, the situation was different in Iran, where travel from one province to the other required special permission and was not permitted for holders of either blue cards or pink slips. Travel was of course particularly difficult for the many Afghans who did not carry any documents. The implementation of control measures by multiple bodies of the Iranian state was so harsh that many Afghans saw it as harassment (Monsutti 2004b: 169). From the perspective of the Afghan

86 Social Networks and Migration in Wartime Afghanistan

refugees, host-state control measures represented a security concern and were considered the major drawback of staying in Iran. My informants generally confirm the picture painted in secondary accounts: The general security condition for Afghans in Iran was excellent in the 1970s and most of the 1980s, but became increasingly problematic from the early 1990s onward. In diplomatic terms, there was a ‘refugee fatigue’ in Iran. Attitudes to Afghans changed, both within the administration and among large parts of the population. Afghans were blamed for contributing to joblessness and for being responsible for a range of criminal activities (Khosrokhavar and Roy 1996: 252–253; Monsutti 2004b: 166–170). Significant political changes in Afghanistan, such as the fall of the Taliban in 2001, contributed to the sentiment that all Afghans ought to leave.12 By 2006, returnees to Enjil were motivated by a new type of security threat. They were concerned about an imminent US attack on Iran, and some were also worried that such an attack could lead the Iranians to conscript Afghans by force. The worst security threat for Afghans in Iran was to be rounded up by the police and deported to Afghanistan. While Iranian officials claim that this happened only to undocumented immigrants, there are numerous accounts from other Afghan immigrants, confirmed by independent sources, that this happened to them too. Often, however, it was possible to pay in order to avoid deportation: If the police know that we are Afghan from the clothes or from the accent, they will chase us. Many people are caught by the police and sent back. We were also caught, but we paid to be released – 200 toman, maybe 500 toman each. People were caught primarily at construction sites and other workplaces, where they often also had their living quarters. Others were caught while traveling, or in their residences. The threat of being caught increased in 2002–03, also owing to the penalties on Iranians who employed or sublet apartments to Afghans. Once caught, the best chance for Afghans was to try to bribe the Iranian officials. If this failed, the likely route back to Afghanistan went via one of the deportation camps close to the border. The process was fraught with harassment, often including physical violence. Many also lost their savings, either because these remained with their employer (who might have reported them in the first place) or because they were charged for food and transport during the deportation process. Once somebody was caught, if immediate bribery did not work out, Afghan affiliates could do nothing to obtain

Integration at Exile

87

their release. Iranian employers could in some cases intervene on behalf of their staff, but with the new regulations of 2002, which imposed penalties on employers of illegal immigrants, this became much rarer. Afghans also sensed a change of attitude among common Iranians. Instances were reported already in 1998 and 1999, when Afghans were attacked by mobs, harassed, beaten, in some cases reportedly killed (Frelick 2001). In 1999, we interviewed a number of recent returnees, most of whom had been expelled by force. The following account is fairly typical: I came from Iran four months ago. I was pressured by the Iranians to leave after the fall of the Taliban. They took us out of the houses. It was not the action of the government, but of the people. This happened to many Afghans there, many people were killed, and houses were burnt. I lived in Waramin, Teheran. Both Afghans and Iranians live there. After the fall of Taliban, when we went to the bazaar to buy bread, they did not want to sell it to us. This was the people who lived together with us. The conditions became terrible when the Taliban fell; they had been much better before. I also lost a lot of money that remained with my employer, he refused to pay me. We were badly beaten by the Iranians. Given the Iranian style of refugee management, with tight restrictions on the opportunity to develop Afghan institutions, the migrants depended on Iranian institutions, as well as on their employers. When the host population turned against them, their situation dramatically worsened, since they were already dealing with a hostile state administration. This does not mean that the Afghans in Iran did not have network resources to draw on in other domains, such as in seeking employment, but these resources could not provide security. The deteriorating security – and the related sense of humiliation – led many refugees to go back to Afghanistan; but, since many found it difficult to secure an income in the home country, some chose to leave some family members behind in Iran to work, or to move back and forth between the two countries. Hence, we get a reversal of the split-household strategy. In the 1980s, during the periods of intense armed conflict in Afghanistan, the pattern was that the main household stayed in Iran, while a few members remained in Afghanistan to look after property. From 1992 onward, the pattern shifted, as it became more common to keep most of the household in Afghanistan, while relying on the income from a few able-bodied men who stayed in Iran for long periods of time. The split-household

88 Social Networks and Migration in Wartime Afghanistan

strategy had thus been adapted to a new security situation, with the aim of minimizing risk.

Material resources and exile The discussion on refugees, and other displaced persons, has focused on the negative effects of relying on external support for survival (Horst 2006a). Afghans in Iran – regardless of whether they departed for economic, security-related, or other reasons – could not survive on what the host state or external agencies provided, but had to make money to sustain their households. Unlike in Pakistan, and in most other host countries for forced migrants, there was never any international assistance system in Iran catering to the most essential needs. Hence, every migrant household had to secure an income on their own. During the PDPA era, the need was more modest, as those with blue cards had access to subsidized food, as well as free education and healthcare. After 1992, as such privileges were gradually cut back, the stakes rose, and the challenges faced by an archetypical ‘refugee’ were quite similar to those faced by an archetypical ‘labor migrant’. Throughout, Iran represented the additional opportunity of accumulating capital, at least for single male migrants or for households with several able-bodied men. The surplus was sent back as remittances to sustain members of the family in Afghanistan, or put aside to meet extraordinary costs, such as investments in houses or land, or to pay dowry. Making a living Afghans in Iran were allowed to take up only a handful of designated menial jobs, and the setting up of private business was prohibited (Stigter and Monsutti 2005: 8). Most Afghans were therefore unskilled workers in the informal labor market. It was easy to find such work, and it remained relatively easy throughout the 1980s (when manual labor was in high demand because of the Iran–Iraq war) and during the 1990s (with much construction work going on in Iran).13 Any able-bodied man could find work in Iran, but he might have to move away from the often impoverished rural areas close to the border, where many of the Afghans had gone initially, but where job opportunities were uneven. In this regard, Iran differed from Pakistan, where the Afghan refugees were generally permitted to take whatever work that was available, as well as to set up of their own businesses (the only restriction was that they could not buy land). Work opportunities for refugees in Pakistan, however, varied greatly

Integration at Exile

89

(Christensen 1988; Morton 1994). Among a group of Hazaras from Ghazni province in southern Afghanistan, for example, Alessandro Monsutti (2004b: 180–182) found that job opportunities often proved decisive for the choice between Iranian and Pakistani exile, given that most had solid networks in both places. The significance of jobs increased as Iranian refugee support dwindled away. This is reflected among the migrants from Izhaq Suleman and Sara-e Nau. In the 1980s, a majority went to Khorasan province (which was also the staging ground for the resistance), whereas after 1992 the majority went to large cities further away from the border, such as Teheran or Isfahan, where work opportunities were richer. The tendency was the same for single migrants and families, although it remained somewhat more common for families to settle in Khorasan and other areas close to the border. When job opportunities became more important for the choice of destination, this was not only because economic motives were more dominant among later migrants, but also because living in Iran had become more costly (the earlier refugees had even enjoyed subsidized access to food), and because the overall job market in Iran had become more differentiated. Over time, Afghan networks in Iran had also diversified, so that most prospective migrants now had contacts at multiple places (see also Stigter and Monsutti 2005). The types of jobs that legal residents in Iran could take had been restricted by law to only manual labor. In practice, the overwhelming majority of Afghans, regardless of legal status, operated in the informal labor market, where the jobs available largely coincided with those permitted by the government. The most common jobs were in construction, in digging ditches and wells, and in agriculture (Abbasi-Shavazi et al. 2005a, 2005b, 2005c). These were risky jobs, and accidents happened frequently. In western Enjil, I frequently encountered stories of Afghans who were killed, and I met many who had received permanent injuries in Iran. Afghans were at the bottom of the job hierarchy and took the jobs that Iranians hesitated to do (Stigter and Monsutti 2005: 8–9). Hence, the Afghans were often sought after by Iranian employers. Afghans’ willingness to take on risky, dirty, and physically demanding jobs at relatively low wages has remained notorious in Iran. At the same time, arguments that Afghans contribute to joblessness and to keeping wages down remain frequent and form an integral part of the still-mounting pressure for their immediate return. There are two main ways of getting work. Either one can turn up at a site where employers come to recruit day labor (a faleke) or one can join a work-team led by an Afghan or Iranian foreman. The day-labor option (which often gives access to work of some duration) remains a major

90 Social Networks and Migration in Wartime Afghanistan

avenue to employment; it seems that the work-teams have become increasingly common over the past decade and a half. The foreman is most often an Afghan who is a skilled worker, for example a mason. He subcontracts for an Iranian employer and has considerable freedom in how to organize the work and perform the task, as long as the outcome is up to standard (Monsutti 2004b: 170–176; Stigter 2005a: 8–9). The following accounts, by two men from Izhaq Suleman who both have a long and varied experience from Iran, are typical: I worked in construction, as a laborer for some masons. We were only Afghans in that project. [The principle is that] One Afghan agrees with an Iranian on a contract, then this Afghan contacts some other Afghans to work. Sometimes I work directly with an Iranian, sometimes I work for Afghans who subcontract to Iranians. The person who subcontracts is normally an engineer, a mason or something. The work-teams center around the foreman, with whom all members have some sort of relationship, normally based on family, friendship, or geographical origin. Gaining access to a team is not considered a major challenge. The work-teams are fairly unstable, as size and membership have to be adjusted to fit whatever job opportunities the foreman has identified. In general, this also implies that the teams, while building on existing social ties, do not in themselves develop into cohesive units. The setup does, however, offer opportunities to Afghans who have developed the necessary technical skill and familiarity with Iranian work-life. It might easily be thought that this is a classical broker position that gives considerable influence over the lives of team members, even with a potential for domination. In practice, the level of influence seems to have been modest, because job opportunities were generally rich, because access to alternative teams was not a problem, and because strict regulations made it difficult for Afghan entrepreneurs to set up stable businesses. Families and extended family-based networks based in Iran were not necessarily confined to one specific location. Just as some households chose to split up between Afghanistan and Iran, so did households also often choose to split up within Iran. This was particularly common among families settled in areas where the job market was failing. Splitting up, however, presupposed that there were several men in the family, so that one could stay behind with the initial group. With existing restrictions on women’s mobility within the Afghan community, long-term absence

Integration at Exile

91

on the part of the men was problematic for the rest of the family unless there was a male caretaker (Damsleth 2003; Marsden 1996a; RostamiPovey 2007). However, this could be resolved by cooperation between several households (who would normally be related). The work mobility increased the risks for all, particularly the men themselves – who were at increased risk of deportation for traveling outside the area in which they were permitted to stay. It also contributed to upholding the already fluid character of the informal labor market. Afghans in Iran were in a vulnerable position, not least due to the strict government regulations and the ways they were implemented. Afghans with no legal status were particularly vulnerable. Few Afghans in Iran gained permanent residency or citizenship, and few were able to establish themselves in stable positions of influence within an economic domain. This stands in major contrast to conditions in Pakistan, where several niches within the economy were taken over by Afghans. In the Iranian case, however, the restrictions on settlement, free movement, and types of job permitted prevented a similar development. Most Afghans had established contact directly with an Iranian when they sought a job, a house, or any other type of service. Work-teams in the informal sector might be led by Afghan foremen, but the regulations in place prevented these teams from developing into stable businesses. In the Pakistani case, a wide range of opportunities emerged for Afghan middlemen to build spheres of influence, whereas in the Iranian case relatively few such opportunities existed. Savings and remittances For most types of work, the salaries in Iran were far better than those in Afghanistan. In 1999, salaries were from two to four times the level in Afghanistan. After that, however, the demand for labor in Afghanistan increased, particularly for skilled labor in construction, and salaries went up considerably. In general terms, however, salaries in Iran continued to outweigh those in Afghanistan by at least two to one. In 2004, it was estimated that unskilled workers in Afghanistan made 100–150 afghanis a day (USD 2–3), while those in Iran made 5000 to 7000 tomans (USD 5–7). Skilled workers made 300 afghanis per day (USD 6) at home, or 8000–10,000 tomans (USD 8–10) in Iran. Single migrants often live at the work site and spend as little as possible. Interviews indicate that as much as 70–80 percent of the income may be saved (Stigter and Monsutti 2005: 10). For families, who have considerably higher living costs, the saving potential is much lower. An estimate based on my own interview material is that a normal 5–6 member family with one full-time breadwinner may be able to put aside some 20–30 percent of its income. Many claim that they are unable to save, because of high living

92 Social Networks and Migration in Wartime Afghanistan

costs, periods of unemployment, or major health expenses (Abbasi-Shavazi et al. 2005b). Until recently, the role played by remittances has been poorly documented – and also contested (Hanifi 2006; Maimbo and Passas 2005; Van Hear 2003b). In a 1997 study among repatriates in Farah province, Marsden (1997: 13) concluded that the prospect of getting remittances played at best a marginal role: the benefit of sending young people to work in Iran is as much that there would, thereby, be fewer mouths to feed as that their remittances would make an important contribution to the household economy. When I was asking my respondents in Afghanistan about the extent to which their family members abroad contributed to the household economy, the most common response was either that they did not or that they contributed only marginally.14 This was the case even in households where one or several young males had gone to Iran to work. Holding this against other information, including my own data from key informants as well as original research conducted within the exile community in Iran, it is clear that respondents underreported the importance of remittances quite severely. Underplaying the scope of remittances seems habitual, inspired by the need not to display one’s resources neither vis-a-vis affiliates outside the immediate family nor to those representing the aid system. Knowing that my findings would eventually be reported in writing, it only seemed wise for the respondents not to reveal the full extent of remittances.15 For respondents in exile, the disadvantages of a mapping of remittances would be less obvious, and it is not unlikely that many would exaggerate the volume of their contributions (see also Carling 2008a). In my key informant interviews, however, I would hear that the young men going to Iran did so mainly to work and they were saving significant amounts of money, which were either put aside or sent home. A number of recent studies confirm the importance of remittances to many Afghan households (Abbasi-Shavazi et al. 2005b; Jazayery 2002; Monsutti 2004a; Pain and Lautze 2002; Stigter and Monsutti 2005). Furthermore, the importance and volume of remittances are not something recent. The bulk of my own data are from 1999, whereas Monsutti’s main fieldwork was conducted between 1993 and 1996. Savings and remittances are not sent at regular intervals. Most migrants have taken up a loan to finance the travel, and may also spend some money on getting settled. Hence, it often takes some 2 to 3 months before saving money is possible. In most cases, remittances are not sent at

Integration at Exile

93

regular intervals, partly because income may vary, but more importantly, because it takes considerable time to minimize the risks. Demands for money from family at origin also plays a role, increasingly so in recent years, where most people have access to phone communication. The general pattern is for transfers to take place two to six times a year, but at irregular intervals (Stigter 2005a: 33; 2005b: 27). The main transfer solution was the hawala system, the informal mode of money transfers that works throughout the Islamic part of the world and well beyond (Maimbo 2003). While the hawala is an institution that may be kept up by different types of actors (Monsutti 2004a: 231–232), the general principle is the same. The client pays a sum to an agent at one location and receives a document certifying the payment, and the money can then be withdrawn by an agent at a locality far away, normally with almost no delay. Charges are lower than for formal transfers; in 2004, hawaldars charged 3 percent for transfer from Teheran to Herat (Stigter and Monsutti 2005: 10–11). In order to have the necessary trust, the choice of a hawaldar is generally based on prior acquaintance, rooted in family, locality of other types of ties. Although there have been reports of fraud (Abbasi-Shavazi et al. 2005a: 61; 2005b: 55; 2005c: 62), reliability is generally high. The hawala is a highly effective informal financial institution, operating in networks of global geographic scope, seemingly based on a particular form of enforceable trust. As such, it is in itself an interesting subject of study. Here, however, my main interest is in the role of migration and remittances in response to war. A remittance economy is in part driven by the fact that Afghans in Iran have few opportunities for permanent settlement or to build up businesses through investing locally. Afghan migrants see their stay in Iran as time-limited, and their future depends on what they can build up at home. In Pakistan, where Afghans had investment opportunities and where many assumed they would be able to stay for a long time, it seems that a larger share of Afghans established themselves with their families. The effects of the Iranian style of refugee management were to limit the migrants’ expected duration of stay, which encouraged savings and remittances, often within split households. Importantly, this discouraged adaptation and integration in Iran, both economically and otherwise, as the migrants continued to regard Afghanistan (but not necessarily their place of origin) as the only viable place to reside. Remittances are specifically targeted at close relatives, almost exclusively those who can be considered members of the (‘split’) household (see also Abbasi-Shavazi et al. 2005b: 28–29, 41–42). In Izhaq Suleman and Sara-e Nau, it was commonly stated that every household

94 Social Networks and Migration in Wartime Afghanistan

had at least one of their men working in Iran. Although this is an exaggeration, since not all households had enough male members to let one go, the common saying underlines the extreme importance of work opportunities in Iran to the household economy. The ability to tap into these opportunities, however, depended on the gender and age composition of the household. Remittance flows were restricted to the densest of ties, with immediate family as the exclusive focus.

Information and exile Access to credible and practical information is critical to adaptation at exile. Less critically, most exiles also want to maintain a sense of developments at their place of origin, to know what happens to affiliates and to have a basis for deciding whether one should start contemplating return. Information streams are commonly assumed to favor bridging types of ties, particularly when the distance is high. Local information The majority of those who departed from Izhaq Suleman and Sara-e Nau during the 1980s and 1990s knew somebody at the destination. In the first couple of years (1978–80), it was more common not to know anybody, but the majority of my interviewees state that, even during that period, they either had contacts at the destination or personal experience from migration. From the early 1980s, there was a saturation of ties to the most common destinations, and by the early 1990s most migrants seem to have had contacts at multiple destinations in Iran, which gives not only more of a choice concerning where to go, but also more independence once one has arrived. One implication is also that there was room for early leavers to become bridgeheads in particular locations and to derive considerable influence from their access to information. It did not take long, however, before this room was closed by a saturation of ties. For new migrants, the presence of associates who can help is particularly critical during the first days upon arrival. Contacts at the destination provide information about work and housing opportunities and how to relate to the authorities; in addition, they may offer temporary accommodation (Stigter 2005b: 25–26). For single men, this may serve as an introduction to a work-team, but for most it simply serves as a cushioning during the first days in a new environment. Even quite weak connections – such as the recommendation by a distant relative to somebody he knows only loosely – may be extremely valuable to the newly arrived, and something that those already settled can provide with little effort.

Integration at Exile

95

As discussed above, migration management in Iran has been characterized by strict controls on travel and work. However, the eagerness with which the authorities sought to implement their instructions varied greatly (see also Monsutti 2004a: 169). This created a deep uncertainty for migrants. Their social networks could not offer privileged access to information about the internal affairs of the Iranian state and did little to reduce such uncertainty. The function of social networks is only to deal with the consequences of official pressure – for example, by taking care of the family of somebody who is deported to Afghanistan and who may be temporarily out of the money-making business. Information about origin A significant share of the migrants from Izhaq Suleman and Sara-e Nau experienced no interruptions to their information links to their communities of origin at any point. Throughout the PDPA era, the main link was through the mujahedin, who came and went regularly, had a surprisingly wide contact network at origin, and were used as regular messengers by those in exile. Khalid Koser (1993, 1997) mapped information streams between home and exile for Mozambicans in Malawi and found that there was a significant information gap in the first period after out-migration. In western Enjil, a significant share of those who left remained engaged as mujahedin in the same area after they were settled in exile, with the effect that information streams were never entirely interrupted. After 1992, when a larger share of the migrants were single men coming for work, there was sufficient traveling back and forth for most people to have a sense of the situation at home. The majority of my male respondents with a background from exile did make visits to their home area, either because they were with the mujahedin or simply because they went back to visit family.16 Abbasi-Shavazi and associates (2005b) interviewed 50 migrants in Teheran in early 2005 and found that, from about half of the families, members had visited Afghanistan while living in Iran. In the same study, it was found that nine out of ten had relatives who had returned, and that those in exile had access to information through the latter. In general terms, information about the situation at home was not a scarce commodity, as it is in many other refugee situations where access is physically closed by long distance or a sealed border. Among those interviewed in Enjil, nine out of ten had relatives in Iran, the majority of whom had regular contact through the exchange of letters. The channels of information have changed dramatically in recent years. Throughout the PDPA era, all exchange of information was by letter or through oral messages (recorded cassettes in some cases), and

96 Social Networks and Migration in Wartime Afghanistan

relied on the physical travel of a messenger. From 1992, it was possible to make phone calls from the telegraph in Herat city; but this was timeconsuming, as the lines were long. A more radical shift has occurred after the fall of the Taliban, when cell-phone and satellite-phone systems have been developed. By 2006, anybody in Enjil who had close family in Iran would have access to a cell phone. This has been important for the frequency and intensity of information exchange (Horst 2006; Vertovec 2004). It has also made imminent contact possible, which can be important when there is an emergency such as serious illness or a deportation. It reduces the burden of long-time absences by allowing regular and real-time contact.

Conclusion The ability to use and build networks depends on the freedom of movement. It is commonly assumed in displacement research that strict host-state control of refugees rests on a camp-settlement system. Pakistan has practiced camp settlement, but Afghans there have been largely free to move around and to engage in various jobs. In Iran, however, where camp settlement was the exception rather than the rule, Afghans faced extensive restrictions concerning where they could settle, on travel within the country, and on the types of jobs they could take up. The strict control measures in Iran effectively constrained and directed the effect and strategic use of network resources, limited investment and entrepreneurship, and prevented the establishment of larger collectives composed of Afghans solely, as is commonly seen in ethnic niche economies elsewhere. The Afghans in Iran were never allowed to form any niche economies; they were allowed only to serve as manual workers in the Iranian economy. The establishment of armed groups operating from exile is also networkdependent, as documented in the literature on Afghan ‘refugee warrior communities’ in Pakistan. Iran has also been supporting Afghan armed groups throughout the past 25 years, although the war with Iraq – and, accordingly, the need for cordial ties with the Soviets – constrained its engagement in the 1980s. With the exception of support to radical Shia groups whose involvement in active resistance was limited, Iran permitted preexisting groups to operate out of its territory and supported them modestly. This setup favored traditional leaders and smaller armed entities with little visibility in Iran. The resistance operating out of Iran was vulnerable, with uneven supplies, and this seemed to encourage the building of militia groups, as well as to foster tight relationships between militia and mujahedin. Security can be a concern also in exile; and, for

Integration at Exile

97

Afghans in Iran, police controls were always a concern, and increasingly so over time. In later years, there was a mounting threat of deportation and several instances where local mobs attacked Afghans. In Iran, the Afghans had scant social resources to play upon in responding to these problems, although it did happen that Iranian employers intervened on their behalf. The chief strategy was to shift the bulk of the household back to Afghanistan, while only the income-earners of the family remained in Iran. Widespread Afghan networks were extremely helpful for finding jobs in Iran. Even the loosest of ties seemed to be effective for the identification of job opportunities. In Iran, Afghans were restricted to certain manual jobs and prohibited from setting up businesses. The result was that most Afghans worked in the informal job market, either as day laborers or as members of small work-teams led by Afghan or Iranian foremen who subcontracted to local businessmen. From the Afghan perspective, this was a labor market in constant flux, where there were always opportunities but where the major risks were to be injured in a work accident or to be rolled up for deportation in a police control. Salaries remained comparatively good, and many sent a substantial share of their income to family members in Afghanistan as remittances. It seems that an increasing share of the migrants in Iran were single Afghan men come to secure an income for their household in Afghanistan (and possibly to save up money for extraordinary expenses, such as the dowry). The split-household structure, which was partly a response to the deteriorating security conditions in Iran, was amplified by the Iranian economy, which increasingly favored single job-seekers. Access to people who were willing to share information was rich. Almost all migrants had contacts at the destination, regardless of when they left. Most had multiple contacts at multiple destinations and therefore did not depend on particular brokers in Iran. While social networks were effective in providing information on housing, jobs, and other practical issues, their reach was limited by the Iranian administrative apparatus, which was both unpredictable and difficult to approach. In contrast to other refugee settings, where it was found that there was a significant lull in information in the first period upon arrival in exile, there was always a functioning information flow between Izhaq Suleman and Sara-e Nau, on the one hand, and the exiles originating from there, on the other. In the 1980s, mujahedin went back and forth; in later years, information was brought by men and families who were either returning or visiting. More generally, the transnational flow of information was ensured by the large degree of continuity between the populations in the areas of origin inside Afghanistan and those in exile in the neighboring countries.

98 Social Networks and Migration in Wartime Afghanistan

For the overwhelming majority of Afghans, Iran was only a temporary country of residence. With mounting pressure on Afghans, no one goes to Iran today with the expectation of being able to stay, and most of those who originally thought they would settle down have had to reconsider. This expectation of temporariness is at the root of the fluidity of current Afghan migrant communities in Iran. While it limits the extent to which Afghans in Iran invest in becoming integrated in the country, it has also encouraged the development of robust migration networks with a wide reach. From the perspective of the Afghan migrant, the transient character of migrant networks reduces vulnerability in relation to the Iranian authorities. Simultaneously, it helps in keeping a clear focus on Afghanistan as the long-term place of residence, and a stay in Iran is only a step toward that. In exile, there has been considerable overlap between the networks that are instrumental for accessing material resources and those that provide information. Between exile and origin, information can also effectively go through a range of channels, whereas money and other goods, in contrast, are channeled to close associates – largely to members of the (‘split’) household. Network resources are only moderately effective in dealing with security threats in exile, which primarily emanate from the state administration. Security engagements at origin, however, are organized from exile, either in smaller preexisting groups that can act as inconspicuously as possible (Sunni groups in the PDPA era) or as radical groups closely associated with Iranian institutions (mainly Shia groups at any time after 1979). One result is that flight collectives during the 1980s to a large extent maintained their coherence in exile, where they became centers of coping as well as the foundations for armed engagement at home. After 1992, however, stable exile collectives became rarer and were replaced by rather loose and fluid networks composed mainly of single men.

5 Return Decisions

Like the decision to escape, the return decision is a migration decision, and it can be analyzed within a similar framework. Return decisions can be mapped along a continuum that ranges from reactive to proactive. Reactive returns are driven primarily by increasing pressures at the host location. An extreme case is expulsion, which gives little or no room for independent action by the displaced (Heimerl 2005; Koser 2005: 3, 22). From the late 1990s, many Afghans have been expelled from Iran, not all of whom were illegal immigrants. Various forms of involuntary repatriation have also been on the rise globally (Blitz, Sales, and Marzano 2005; Chimni 1999). The archetypical case of a proactive return is when people make their decision after an independent assessment and comparison between conditions at home and in exile. Such decisions often gradually mature in interaction both with other refugees and with close associates at the place of origin. Proactive returnees may not only respond to changing conditions but also see a chance to contribute to altering the very conditions – economic, security-related, or other – that once precipitated their flight. The conventional discourse on the return of refugees and internally displaced persons (IDPs) is founded on a number of problematic assumptions, as several leading analysts have pointed out (see, for example, Allen and Turton 1996; Bascom 2005; Cernea 1996; Hammond 1999). These assumptions stem from an ideologically and politically driven view among most refugee practitioners and researchers that a return to origin is the ideal solution (Markowitz 2004). This view has been increasingly challenged by a new drive within research, with analysts following displaced people from the return decision to their situation past the return (Allen and Turton 1996). This research shows that return is far from always successful. Nonetheless, the myths that inform the discourse on return and repatriation have so far proven very resilient (Black 2006; Fagen 2006). 99

100 Social Networks and Migration in Wartime Afghanistan

One myth is that most returns are assisted and that, without the intervention of states and relief agencies, most displaced people are unable to go home (Koser 1993, 1997). This conception led to the launch of the somewhat misleading term ‘spontaneous repatriation’, coined in reference to the presumably rare cases of refugees returning on their own initiative (Cuny and Stein 1989). In reality, most displaced people return with little or no external assistance.1 Even in cases where external assistance is provided, such as with the repatriation packages provided by UNHCR to Afghans returning from Iran and Pakistan, the assistance is unlikely to be the main factor motivating the return decision. Other factors – such as long-term economic and security prospects at the location people return to – are far more important. Hence, social networks emerge once again as key resources in their own right, mediating security and economic resources, as well as information. A second myth is that repatriation means return to a familiar environment, to ‘home’. This is evidently misleading when refugees move to a different location in their country of origin than where they originally came from, which is often the case. For them, even the term ‘return’ is problematic. Also, for those who resettle in the same place that they lived prior to their flight, what used to be home will have changed as a result of war: physically, with destroyed infrastructure and economic opportunities; socially, as networks have transformed and new people have assumed power (Janzen 2004; Stefansson 2004b). After having stayed away a long time, people also change themselves, and what used to be familiar may seem strikingly unfamiliar after years of exposure to a context that is different culturally, economically, and in terms of welfare. At the extreme, refugees may have developed a conception of their exile location as more ‘home’ than their place of origin (Bascom 2005; Preston 1999; Zetter 1988b). This is often the case for those who were children at the time of the flight and have grown up in exile. The third myth is that it is implicitly assumed that those who return are similar to – or even the same as – those who left. This assumption is normally more misconceived the longer a refugee population has stayed in exile. A case in point is the many Palestinians who continue to be defined as refugees three generations after they departed or were expelled. Similar changes occur, almost inevitably, within any exile population: the composition of the group changes; people develop new ties at the cost of old ones; and individuals change perceptions and develop new priorities as a reflection of living under different conditions (Gmelch 2004; Hammond 1999, 2004). Those who return, also in cases where they do it as part of a seemingly intact collective, will always be different from those who left.

Return Decisions

101

The literature on the role of networks in return decisionmaking is scant (Ghosh 2000; Janzen 2004; King 2000). Nonetheless, several analysts have pointed to the importance of links between exile and origin (Ammassari and Black 2001; Arowolo 2000; Black and King 2004). In his studies of repatriation of Mozambicans from Malawi, Khalid Koser (1997) concludes that return decisions are taken at the family level. Koser also suggests that, in comparison with flight decisions, return decisions tend to encompass smaller collectives. Among those of my respondents who had repatriated, four out of five reported that they had traveled in return collectives that included only their close families. Among the remaining, a majority also made explicit reference to a family decision to return, even though this issue was not being systematically asked about in the survey. In contrast to Koser’s (1997: 13) findings from Mozambique, however, I found that a majority of the returnees had consulted widely with others before deciding to return. I also found that most had traveled in groups that extended well beyond the family, yet were significantly smaller than the flight collectives. The collective dimension seems to be as prevalent during return as during flight, although the size of the traveling collective is smaller. By the end of 2004, a significant share of the Afghan refugees in Iran and Pakistan had returned to their country of origin. Repatriation peaked in two periods, both in the aftermath of regime transitions: in 1992–93, following the ascension of the so-called mujahedin government, and in 2002–03, after the installment of the Karzai administration.2 In parallel to the international returns, IDPs have been going back to their localities of origin or have moved on to new places within the country. A significant share of the returnees from Iran and Pakistan have moved to a location different from the one they left when they fled, this latter location most often being in the cities. Hence, the process of flight and return has been paralleled with massive urbanization. Return migration has been considerable also at times when new out-migration has been massive, and the Afghan war provides ample evidence of how people relate differently to the same situation. A military development or political change may remove some people’s security while enhancing the security for others. Some people see ample opportunities for return even in the midst of conflict. On their way back, they may meet their old neighbors on the way out. For the UNHCR, the major instrument in promoting repatriation has been the encashment program, which started in July 1990 under an agreement between UNHCR and the government of Pakistan. There has been considerable critique of repatriation package. Some have argued that they create a form of dependency, encouraging refugees to delay repatriation until they are compensated generously (see, for example,

102 Social Networks and Migration in Wartime Afghanistan

Koser 1993: 11; 1997: 13). Others have argued that repatriation packages may send a wrong signal that conditions back ‘home’ are now secure, and that there will be ample international support for reconstruction and re-adaptation (Turton and Marsden 2002: 27–33; see also Bialczyk 2008). By 1997, inspired by small-scale initiatives driven by the NGO sector, the UN started to experiment with what was termed the ‘group repatriation approach’ in the districts of Azra and Tezin in Logar province (UNDP 1999: 11–12; UNHCR 1998).3 The basic idea was to establish contact with communities residing in refugee camps in Pakistan who were motivated to return. There would then be a needs assessment of the situation at their place of origin, and a coordinated package of activities would be launched in response to needs identified. The group repatriation initiative was inspired by the increasing realization from the early 1990s, among practitioners and analysts alike, that return decisions are frequently taken on the level of larger collectives (see also Koser 1993; Zinzer 1991). As repatriation from Pakistan to Afghanistan picked up from the late 1980s, observers saw that households often returned as part of larger collectives. These collectives had often remained largely intact throughout the exile period (see also English 1988: 15; Knowles 1992). As for Iran, it was generally assumed that repatriation did not take place collectively, since the refugees were so dispersed (Batson 1992: 4). This assumption, however, was purely speculative, since next to nothing was known about the refugee population in Iran and few data were available on the areas in Afghanistan from which most of the refugees in Iran came. The group repatriation concept was an early attempt to give weight to the network dimension in repatriation, acknowledging the fact that collectives often return together and building on them to facilitate return. UNHCR seems to have worked on the assumption that the longer refugees stay in exile, the less likely they are to ever return (see, for example, UNHCR 2004: 1). Stein and Cuny, on the other hand, suggest that refugees become less risk averse and demonstrate more initiative over time, as the dramatic events that triggered their departure slide into the more distant past and as refugee communities in exile gradually improve their level of organization (Stein and Cuny 1991). Those analysts hypothesize that declining risk aversion may drive repatriation, particularly when the situation in the host country is experienced as increasingly precarious. It was widely believed in the early 1990s that the overwhelming majority of the Afghan refugees in the neighboring countries would quickly repatriate. Some analysts, however, thought that quite a few would opt to stay in Pakistan, given cultural similarities and a close affiliation with the host population (Kronenfeld 2008). In Iran, it was assumed that almost all perceived their stay as temporary. While many

Return Decisions

103

refugees in Iran had been able to accumulate some capital while in exile, they generally did not enjoy the same close relationship with the host population as did those in Pakistan (Centlivres 1993; Centlivres and Centlivres-Demont 1992). As of 2008, the proportion of refugees who had repatriated from Pakistan and Iran was roughly the same. Yet, it seems that most of the Afghans still in Iran expect to return at some point, whereas in Pakistan many intend to stay (Saito 2008a; Tober 2007a, 2007b). This reflects the greater tendency of Iranians to not build close relationships with the Afghans. The Pakistanis interact with and accept the refugees to a much greater extent. Local attitudes, government policies, and economic opportunities are likely to influence return decisions far more than the length of the exile.4 The overall pattern of repatriation to western Enjil was similar to that of the rest of Afghanistan, with a major peak in 1992 and substantial homeward migration also in the following year. From 1994 onward, the flow of repatriates was only small-scale – probably equaled or surpassed by the rates of out-migration – before a new peak appeared in 2002 and 2003, following the defeat of the Taliban, the installment of a new transitional authority in Kabul, and the reinstallment of former governor Ismael Khan in Herat. By 2006, emigration again exceeded immigration. Throughout the whole period, repatriation to Enjil, as well as to the larger region, was almost exclusively from Iran. Despite significant obstacles to travel, there was frequent exchange of people between the communities in western Enjil and the exile locations in Iran. The share of the population from western Enjil in exile by 1992 was extraordinarily high, estimated to represent 90 percent of the prewar population (see, for example, UNHCR 1990). While, in Pakistan, a decision for a family to return was not perceived as irreversible, the control system and associated costs in Iran were of such a scale that it was unrealistic for the whole family to be able to get back to Iran again once they had returned. This may partly explain why the repatriation from Iran seemed to pick up more slowly than that from Pakistan. Addressing return, this chapter is structured in the same manner as Chapters 3 and 4. It deviates slightly in that it starts out with a review of a comparative case study of return patterns from two localities in Pakistan by anthropologist Bernt Glatzer.

Short-distance repatriation from Pakistan From 1989 to 1992, following the withdrawal of Soviet forces but with the PDPA government still in power in Kabul, several provinces close to the Pakistani border were gradually taken over by the resistance, and

104 Social Networks and Migration in Wartime Afghanistan

the government did little to regain these areas. In 1988–89, it was widely expected that the Soviet withdrawal would invite large-scale return, but this expectation had to be modified when the Najib government proved more resilient than expected. In the ‘liberated areas’ just across the border from the refugee settlements in Pakistan’s North-West Frontier Province (NWFP), most observers thought repatriation would be immediate. In reality, most of the refugees chose to stay on the Pakistani side of the border, with some practising occasional or seasonal return. Still, there were massive differences from one area to another in terms of repatriation rates. These differences inspired Glatzer, an anthropologist working with the Danish Committee for Aid to Afghan Refugees (DACAAR) at the time, to compare a region where repatriation was massive with a region where only a few returned (Glatzer 1992). This work, which has only been presented within the ‘grey literature’ (agency reports, conference proceedings, and so on), is unique in focusing on the social dynamics of repatriation at an intermediate level of analysis. Glatzer explains the massive repatriation to the first region in terms of the maintenance of prewar social structures, reflected in a low conflict level, hence good security. He also finds that return was facilitated by the stability of prewar social organization throughout the exile. In addition, the ability to maintain ties between people at home and people in exile played a role. Glatzer’s findings were based on interviews and fieldwork among refugees from Kunar province staying in Bajaur agency of NWFP, and among refugees from Khost province who stayed in and around the city of Miram Shah.5 At the time of his fieldwork, return to Khost was well underway, whereas those originating in Kunar were holding back. In both cases, the refugees had left in the early 1980s, in response to the intense armed campaigns in the provinces bordering on Pakistan. Government and Soviet forces were driven out of Kunar province in October 1988, but several radical Islamic groups – including Hezb-e Islami (Hekmatyar) and a Saudi-oriented Salaffiya group – vied for power, with traditional authorities being marginalized (Gizabi 2006). The insecurity created by the rivalry between resistance groups was a major obstacle to return (Glatzer 1992).6 Khost was taken over by the resistance in March 1991. In this area, traditional structures had largely prevailed and the radical Islamic parties had only limited success in gaining influence (Glatzer 1992; Janata 1982). Both in Khost and in Miram Shah, there had been ongoing and intense confrontation between the resistance and the government throughout most of the 1980s, and the level of war-related destruction was high (Map 5.1). When the residents of these areas left, they went to areas that they knew in advance. Most of the Pashtuns living close to Pakistan do not

Return Decisions

Map 5.1

105

Locations in Glatzer’s Comparative Study of Repatriation

see the border as an obstacle: they are used to traveling across it; they have long-standing ties with the people living on the other side; and border control is weak or nonexistent. This was also the case with the two groups of refugees discussed here; but, while the Kunaris interacted with the local population in Bajaur in almost every domain of life (including widespread intermarriage), the hosts of the Khost population generally kept their distance. The prewar patterns of interaction carried over into the refugee situation, and the exiled Kunaris further solidified their integration with the Bajauris, whereas the Khost exiles in Miram Shah kept to themselves. This was paralleled in the degree to which the two exile groups maintained relations with those who had remained behind, with a gradual withering of ties for many of the Kunaris in Bajaur, but constant solidification of ties with home for the Khost exiles in Miram Shah. The latter frequently visited their home region, often for weeks and months at a time, and could take part in cultivating the land. With the tribal system intact, the Khosti exiles maintained their position in the communities of origin throughout the war.7 Given the frequency of return visits, for both groups of refugees, access to information was not a constraint. Nor does there seem to be any significant difference in the ability to mobilize economic resources. Both exile locations offered only modest employment opportunities, and

106 Social Networks and Migration in Wartime Afghanistan

many had to travel to the larger cities to seek employment. At home, the economic basis was severely disrupted by war, but there was also considerable reconstruction activity, including international assistance.8 Little information is available on economic resources, but there are no indications of significant differences between the two locations that could explain the widely different levels of repatriation. The level of integration at the exile location is probably a more important factor in determining the degree of repatriation than the availability of economic resources at the place of origin. The repatriation of the Kunaris was limited by a reversed version of the so-called affinity hypothesis (Ritchey 1976). For the Kunaris, who had cultivated social ties over at least a decade with the people who lived in Bajaur, the restraining effect of close ties could be as important in preventing return as they are generally thought to be in constraining out-migration. The Khost exiles, on the other hand, had remained much more closely attached to their community of origin. This made transition easy; and, given that the decision to return was taken at the level of the larger collective, the pressure from close affiliates would be promoting rather than preventing return. Security, however, emerges as a key factor in Glatzer’s studies. Not surprisingly, he finds that return was less likely if the security situation back home was bad. More interestingly, however, he also finds that people see security as something that emerges from their integration in larger collectives acting together. One of Glatzer’s interviewees in Bajaur expresses his feelings in the following way: There is neither government nor police to protect us. I look forward to returning home, but only together with all my relatives and neighbors. (Glatzer 1990: 5) In Glatzer’s analysis, the interplay between repatriation and security creates a vicious cycle: Because in Kunar there are too few people it is unsafe, because it is unsafe, more people don’t come. This circle cannot be broken by individual decisions to go back, but only by an organized mass return. (Glatzer 1990: 6) In Khost, on the other hand, the larger collective, consisting of returnees as well as those that had stayed behind, joined hands to set up a tribal militia (arbaki), in a clear demonstration of resistance to the organizational forms and legitimizations applied either by the resistance or by the government.

Return Decisions

107

All of this suggests that the security issue may generate threshold effects in repatriation to particular areas. To some areas, return is minimal, whereas to other areas it is extensive, with no apparent economic explanation for the difference. If a certain level of coherence is maintained within the collective throughout the exile period – and, even better, if coherence is maintained with the community remaining at the place of origin, as in the Khost case – then it is likely that a return decision will be taken at the collective level. Ultimately, in a situation where the state has largely broken down and there are few alternative offers of protection, it is the need for basic security that fosters the dependency of individuals and households on the course decided upon by larger collectives. Of course, this does not mean that the broader security situation is not a key factor in return decisions (it is likely that internal fighting for control between resistance parties played a major role in preventing return to Bajaur), but that, even if the security environment is seen as acceptable, people depend on their membership in larger collectives and take their decisions accordingly. The comparison of the Kunar and the Khost exiles may also contribute to our understanding of how return propensity changes over time, as a reflection of declining risk aversion. In the Khost case, with ties within the original group as well as to those residing at the location of origin, it is not unlikely that there is a growing willingness to accept a certain level of future risk when considering migration. In Kunar, where prewar ties were maintained and developed throughout exile, it seems unlikely that the intention to return would gradually increase and make prospective returnees more willing to accept risk. Stated differently, the Khost case could be characterized by increasing frustration over the obstacles to interaction with one’s primary group, whereas the Kunar case is one in which people’s primary affiliations had changed dramatically toward a composition which bound people to exile rather than motivating their return. Later reports indicate that, among the Kunaris in Bajaur, the propensity to repatriate remained small, and many claimed to be solidly integrated and wanting to remain in Pakistan (DACAAR, IRC, and MADERA 2002).

Security and return Collective assessments of security in the context of deciding whether to return or not to return are not widely addressed in the literature (but see Stefansson 2004a, 2004b). Among my informants in Enjil, networks come out as critical for people’s willingness to return, as they do in Glatzer’s findings from Khost and Kunar. Intuitively this makes sense, as most conflict and post-conflict settings are characterized by the absence of legitimate

108 Social Networks and Migration in Wartime Afghanistan

security institutions. The obvious alternative is to look to personal networks for security. As I have suggested earlier, however, networks need to be of considerable size if they are to offer credible protection, particularly if the relevant threats are in the form of organized military-cum-political groups that seek to maintain or expand their position. The sensed need for security is a key reason why early return takes place in the form of a large group of people going together, and why that first early group is likely to inspire a wave of further return as people experience that the representation of their solidary group at the location is sufficient to provide them an acceptable degree of protection. These collectives need to be of considerable size and are assumedly made up of numerous smaller but well-connected clusters, each of which is based on dense ties. In 1999, when my main fieldwork in Enjil was conducted, security was not cited as the primary reason why people did not return. Most informants referred to economic difficulties as the primary obstacle to return – perhaps not surprisingly, given the combined effect of massive drought and the economic recession brought about by the Taliban rule. When asked whether refugees in Iran were planning to return, a recent returnee in his mid-20s gave the following typical reply: They will come if they think that there is security and jobs here. The economic condition is the most important. Many people are tired of living in Iran. None of this is surprising, since there was still considerable migration to Iran – in the main motivated by economic difficulties, although some also went as a result of difficulties with the Taliban. During 1992–93, however, when repatriation peaked, issues of security played a prominent role in people’s return decisions. In the first few months after the fall of the Najib regime, the division of political and military power was unclear in most parts of the country. This was the case also in Herat, although Ismael Khan, the strongman of the region, seemed unlikely to be challenged. A quick clarification of regional leadership, however, did not prevent power struggles at lower levels in the hierarchy, resulting in grave insecurity in many localities. During the period of Taliban rule, security was relatively good throughout the region, although some – those suspected of affiliations with the mujahedin in particular – had fundamental problems in relation to the new authorities. Return collectives played a significant role. People returned home not just with their immediate families, but as part of a larger collective that included members of the extended family, former neighbors, and even

Return Decisions

109

sometimes persons and families they had got to know only while in exile. Return collectives are mostly based on cohesive ties. In line with what Koser (1997) found in his Mozambique study, there is a significant difference in size between flight collectives and return collectives. Flight collectives could include ten or more families; return collectives would commonly include only two to four households. Return travel is normally better planned, with a clear destination, and risks are perceived to be more manageable than in a situation of flight. Traveling together is less of a necessity, but for security reasons it is important – particularly for the early returnees – to know that when reaching the destination, they are not alone, but rather belong to a returning community. In the following, I will first look at how those associated with armed groups – mujahedin – approached the security dimension when planning to return. Such individuals had been proactively engaged in the armed struggle and had a direct stake in the new security setup at home. I will then move on to look at refugees who were not associated with armed groups, and hence might have to revert to a more reactive stance in relation to security, but still could find ways of activating their network resources in order to foster a sense of security. I will also briefly consider those who were expelled from Iran. These were genuinely ‘forced’ to go back to Afghanistan. My focus here will be on how prospective returnees assess their own network resources in coping with security problems back home. More generally, the main discussion is about how prospective returnees assess security during and upon return, and how they play upon their networks in order to maximize security in the return phase. Issues that relate to exile security have been discussed in Chapter 6, but are relevant here to the extent that prospective informants will hold expected security at home against the security in exile. ‘Returnee warriors’ Members of a refugee population who are associated with political or military parties will have a stronger incentive than others to return quickly if there is a regime change in the home country benefiting their party. It is in the early days of a new regime that the best prizes may be won – such as state jobs, political influence, and economic contracts and opportunities. Despite the vast debate on refugee warrior communities, little attention has been paid to the effect that rapid repatriation of militarily active refugees may have on political transitions (Harpviken 2008, 2009). One prominent voice in the recent ‘refugee warrior’ debate, Sarah Lischer (2005: 150–151), is concerned that the present-day international preference for repatriation (over integration or third-country resettlement) fosters more

110 Social Networks and Migration in Wartime Afghanistan

long-term refugee populations that are susceptible to political and military mobilization. The literature on refugee mobilization, however, pays less attention to the political risks inherent in repatriation, particularly in the context of regime change – such as in Afghanistan in 1992 and 2001 (but see Krznaric 1997). Following the US-led intervention in 2001, however, UNHCR officials expressed concern that Pashtuns in IDP camps in the Kandahar area were encouraged by local commanders in the north to come back to areas where there was still active combat (Marshall 2002).9 In Enjil, where I conducted fieldwork, those who had been active in the mujahedin parties throughout the war were the first to return after the fall of Najibullah in 1992. In fact, all respondents with such a background returned during 1992 and 1993. When asked about the primary reason, almost all of them said it was the regime change, using expressions such as the ‘victory of the mujahedin’, ‘Ismael Khan was in power’, and ‘the fall of the communist regime’. In contrast, people who had returned later mainly referred to Iranian pressure. The fact that an overwhelming majority of the 1992 and 1993 returnees claimed to have gone home because of the regime change is indicative of a strong degree of association with the mujahedin. Importantly, in 1992, many Afghans, particularly those in exile, still had confidence in the mujahedin leadership. Two to three years later, incompetence and massive infighting had done severe damage to the reputation of the mujahedin, albeit less so in Herat than in most other places. Mujahedin associates from Sara-e Nau all returned in the first few months after Najibullah’s fall, whereas in Izhaq Suleman many chose to stay another year in exile before returning. If survival and the local economy were main concerns, one would expect the pattern to be the opposite. After all, Sara-e Nau had been virtually demolished, whereas Izhaq Suleman was largely intact. From a security perspective, one can understand that some of the Izhaq Suleman mujahedin preferred to wait and see as a new power balance between the local militia and the returning refugees developed. The first mujahedin to return to the village seem to have had the double advantage of being well connected to Ismael Khan, while also being on good terms with the local militia. Others, particularly those who were more uncertain regarding their status with the militia, were less eager to be at the forefront in the early – and highly uncertain – days. With the Taliban taking over Herat in 1995, the advantage of having fought with the mujahedin was suddenly turned into a disadvantage. The Taliban kept a far closer tab on people known as mujahedin than on anybody else, even former militia or PDPA activists. Likewise, the Taliban interfered more in the local politics of Sara-e Nau and similar villages than in Izhaq Suleman. Despite the risk of persecution, most mujahedin

Return Decisions

111

supporters tried to keep a low profile, and some departed for the relative safety of Iran. Toward the end of the 1990s, as Iran–Taliban relations turned increasingly sour, Iran actively encouraged and supported the buildup of military capacity in exile. Hence, the 2001 fall of the Taliban had an effect similar to that of Najib’s 1992 abdication in encouraging a rapid repatriation of mujahedin from Iran. Much, however, was different. Firstly, many of the mujahedin based in Iran (and Pakistan) were returning as part of Operation Enduring Freedom (OEF), the campaign that ousted the Taliban (Rashid 2008: 71). Some of my local informants, for example, had been fighting in Herat or elsewhere and had been brought into Afghanistan via Turkmenistan and Uzbekistan. Secondly, the post-9/11 mujahedin returnees were more of an elite than those who returned in 1992–93. Finally, those who returned in 2002 came back to societies with a changed and functioning economic and administrative structure, and had to find their roles within it rather than define its shape. Nonetheless, by 2002, Herat’s political and administrative elite was dominated by recent returnees. For prospective returnees, the stakes were much higher in 1992–93 than in 2001–02, even for core members of the mujahedin. We have already noted that the mujahedin parties had sufficient influence to prevent return in the late 1980s, at least from Pakistan. They saw repatriation as contrary to their political interests. The ability to motivate massive repatriation in the aftermath of the regime changes of 1992 and 2001 is the opposite side of the same coin. In both cases, loyalty with mujahedin groups carried the inherent promise of being rewarded in a future political setup. After a regime change, however, the reward is most often immediate and concrete. For people who opt to stay in exile, the reward is both long-term and somewhat hypothetical. It has been difficult to establish the extent to which mujahedin parties were directly involved in the organization of repatriation from Iran or Pakistan. Some reports by humanitarian agencies establish this as a fact (see, for example, Jamal and Stigter 2002: 5; Marsden 1999: 58), but I have not found any substantial description of this in the literature. From my own empirical material, it is clear – not surprisingly – that various political parties were active in securing the quickest possible return of key cadre, including local-level commanders. Rank-and-file associates, however, seem to have been encouraged to return only indirectly, as individual commanders and other influentials have sought to ensure that as many as possible of their followers come back with them. Ultimately, the networks on which ‘returnee warriors’ are based seem remarkably consistent with those that were at work during exile.

112 Social Networks and Migration in Wartime Afghanistan

General security Most prospective returnees to western Enjil do not see themselves as being directly engaged in the security domain: they are not – and have never been – active members of the mujahedin. These people see security largely as a contextual factor beyond their control. The engagement of the mujahedin or other armed groups may for them be a part of the problem rather than the solution, although they may have to play on whatever contacts they have within such organizations in attempts to foster security for themselves. More broadly, I contend, people who are not directly associated with armed groups seek alternative strategies in order to strengthen their own sense of security, playing on immediate social ties as well as more extended networks. While such network strategies are rarely sufficient for turning a fundamentally insecure context into a secure one, they may make the difference between return and staying on in exile. The people I interviewed in Enjil in 1999 generally did not see lack of security as the prime obstacle to further returns. The people I met in 2002, 2003, and 2005, however, left a different impression. Recent returnees then described security as a prime concern, and they believed it was even more so for those who remained in Iran. One man interviewed in 2002 had two sons returning from Iran after the fall of the Taliban. He made the priorities clear: Now we have security, nothing else matters. When we have security we can eat only bread, yet we will think that it is chicken and pilau. The impression is confirmed by a recent study of Afghans in Teheran. A majority of the interviewees there cited lack of security as the most important factor preventing return, and most of the respondents expressed concern for political stability (Abbasi-Shavazi et al. 2005b: 32–33).10 Returns to Izhaq Suleman and Sara-e Nau were massive following the installation of an interim administration in December 2001. A total of 57 families and – in addition – 123 individuals returned to Izhaq Suleman in 2002 and 2003. To Sara-e Nau, a total of 16 families and 18 individuals returned within the same period (Table 5.1). Table 5.1 Returns to Izhaq Suleman and Sara-e Nau, 2002–03.11 2002

Izhaq Suleman Sara-e-Nau Total

2003

Total

Single

Family

Single

Family

Single

Family

74 11 85

20 6 26

49 7 56

37 10 47

123 18 141

57 16 73

Return Decisions

113

In both villages, the number of single returnees declined by approximately one-third from 2002 to 2003, while the number of returning families increased (close to a doubling in both cases). Most of the individuals that returned were single men who had gone to Iran primarily to work and who had departed within the past few years. If we focus here on the returning families, however, a majority of them had left in the 1980s. Some had repatriated in 1992–93, but found life difficult in Afghanistan and went back again to Iran. When repatriation patterns after the regime changes in 1992 and 2001 are compared, it appears that the response in 1992 was almost immediate (particularly for Sara-e Nau), whereas in 2001–02 it was more considered and gradual. In 1992, there was virtually no time to plan the return, since the regime change coincided with the start of the migration season. In 2002, however, there were some 3 to 4 months to get ready; nevertheless, many decided to wait an extra winter before returning. There may be several reasons for this contrast. For one thing, having a few months to reflect may have served to temper the immediate impetus to pack up and leave. Furthermore, the fact that many considered their repatriation in 1992 to have been a failure may have had an impact, not only on those who had actually tried to repatriate once before but also on other refugees. Most importantly, however, the refugees remaining in Iran by 2002 were different from those returning in the early 1990s. Their affiliation with people at home was weaker already in 1992, and had withered further during an additional decade in Iran. This has a range of implications, also in terms of economic opportunities and access to information, but perhaps most importantly for the sense of security upon return. During the 1999 fieldwork, I discussed with a prominent local mujahedin commander the problem that voluntary repatriation had almost come to a halt. He argued (perhaps overoptimistically) that all the refugees planned to return, but that their ability to do so depended on a comprehensive political settlement at the national level, which would lay the foundations for peace and security throughout the country. Glatzer’s comparison of return from Pakistan to Khost and Kunar, discussed above, concluded that returnee numbers are important in their own right, but also that people’s assessment of security at home improves with the extent to which family and other close associates have already repatriated or are in the process of doing so. The pattern of return to Izhaq Suleman and Sara-e Nau confirms Glatzer’s findings. Large-scale repatriation in itself is important in attracting more returnees, but not sufficient to make all prospective returnees feel that the security situation is satisfactory. In some cases, the problem may

114 Social Networks and Migration in Wartime Afghanistan

lie in a history of contention, as in Izhaq Suleman in 1992, when some preferred to await a clarification in militia–mujahedin relations before returning. More generally, the security derived from the sheer number of people who have already returned accounts for little. What is most important is that the prospective returnees have close ties with someone at home they can rely upon. Such ties serve to translate a generally improved security environment into a personal sense of security. The importance of networks for fostering a sense of security is important for contagion effects in repatriation (Christensen 1995: 118). In Afghanistan, for example, the regime change in 1992 stands out as a turning point for return migration, with massive return within a few weeks after the event. The drive for repatriation was strong – the mujahedin, the host states Iran and Pakistan, and the international agencies all encouraged return – yet a sense that most others within one’s network were returning seems to have been key in encouraging individual decisions to return. There were several segments of the refugee population, however, who did not return in 1992–93. Those with weak ties to the mujahedin often chose to wait and see, and later felt they had made the right choice when they observed that none of their close associates returned. Hence, the overall momentum of repatriation was lost. This shows how one can get successive waves of repatriation, with a contagion effect in each wave that motivates a significant share of the refugees to repatriate but does not reach across to other sections within the same population. In the Afghan case, a second wave started in late 2002 and continued into 2005, when it started to decline, particularly for returns from Iran. Those who still hesitate to return are likely to lack the sort of cohesive networks that contribute to a sense of security, and therefore the risks associated with return still seem unacceptably high for them. An illustration of how important the sense of security is, may be found among the refugees who escaped the civil war that erupted in Tajikistan in 1992. Local power struggles led many to flee to neighboring Afghanistan, where they settled in two different areas – the Sakhi camp near Mazar-e Sharif and several camps in Kunduz province. At the peak, an estimated 42,000 refugees from Tajikistan resided in these camps. By the time of the peace agreement between the government and the rebels in Tajikistan in June 1997, almost half of these refugees remained (USCR 1997). The information that reached the refugees about conditions in Tajikistan, however, did not encourage more people to return. There were widespread reports of harassment, crime, and even killings of returnees, as a result of general disintegration of state capacity and strong tensions at the community level (HRW 1996). Tajik opposition parties controlled the camps and discouraged repatriation by disseminating alarming news.

Return Decisions

115

In an attempt to secure an independent information channel, UNHCR facilitated the exchange of letters. Repatriation picked up only after the Organization for Security and Cooperation in Europe (OSCE) set up a radio communication system through which refugees could talk directly to people at home. Given the grave nature of the potential threats, the refugees depended on two-way communication with people they trusted in their home communities (OSI 1998). While some found that conditions were not ripe for return, the majority were able to get the security guarantees necessary to go home. Does the duration of exile make refugees more or less prone to repatriate even if their future at home looks uncertain? Whereas the UNHCR in Afghanistan has worked on the assumption that return propensity declines with the length of exile, Stein and Cuny have suggested instead that the longer an exile has lasted, the more willing are the exiles to accept risks (Stein and Cuny 1991: 1; UNHCR 2004). As discussed before, Izhaq Suleman and Sara-e Nau had different patterns of flight. Both were located in the midst of the government’s security belt around Herat city. Izhaq Suleman was under militia command and remained populated, whereas Sara-e Nau was a mujahedin site and saw long periods with no permanent population. In the case of the latter, the whole population left within a short time-span in the early 1980s, whereas in the case of Izhaq Suleman departures were much fewer (in relative terms) and they were dispersed over a much longer period of time. The majority of those who returned from exile during the 1990s did so shortly after the fall of the communist government in 1992. The wish to return seems to have been equally strong among the refugees from both Izhaq Suleman and Sara-e Nau, although a larger share of the Izhaq Suleman returnees did not return until 1993. The pattern of repatriation does not support the thesis on declining risk aversion over time, but does indicate that those who expect to make a considerable gain from returning (such as members of the mujahedin) are likely to return most quickly, despite the uncertainty involved. Also, in accordance with Glatzer’s findings, poor integration in exile and maintenance of close ties with one’s place of origin increase people’s readiness to accept risk. Just as in the departure decision, security considerations are the main factor in the decision to return. Security has a network dimension, particularly in the absence of a functioning state, in that people’s primary protection stems from their association with groups that are considerably larger than the household or the extended family. Risks are great for those who return first, particularly if they go to depopulated locations. In such situations, security will not appear satisfactory unless returnees are confident that many others will come back more or less simultaneously.

116 Social Networks and Migration in Wartime Afghanistan

Members of refugee populations, however, may be nested into different types of networks, and hence may read the security prospects upon return differently. A wave of return that attracts some may not appeal to others, and we may get successive thresholds in repatriation. Expulsion From the perspective of an Afghan wanting to stay in Iran, the most dramatic threat is that of being expelled. With reference to the larger discussion of how to understand causality in ‘forced migration’, expulsion is the clearest case of forced return migration, with the migrant being left no choice. The threat of expulsion represents a major everyday security concern for most Afghans in Iran. The ‘forced’ character of the expulsion continues throughout the deportation cycle. With the exception of the opportunity to be released upon capture by offering a bribe, the expellee really has no choice and cannot get any help from networks to prevent the repatriation. Deportations from Iran peaked in 1998 and 1999 (USCR 2001a, 2001b: 177), which coincided with the main fieldwork for this study. In the late 1990s, the majority of deportees were caught at their place of work, where they lived, or while traveling inside Iran. After the fall of the Taliban, the number of deportations fell to less than one-third of the figures in 1998 and 1999, with UNHCR reporting 26,732 in 2002 and 28,311 in 2003, all at Islam Qala (Stigter 2005b: 27). By then, a majority of the deportees were people caught in the border areas while entering Iran, partly as a result of tightened border policing. During the 1999 fieldwork, the effects of the expulsions were particularly noticeable in Izhaq Suleman, which was located alongside the main road between Herat and Islam Qala at the Iranian border. Every day, several trucks and minibuses passed by on the main road, filled with single men who carried a few belongings in a bag. The deportations were also noticeable in the villages, as involuntary returnees turned up in the bazaar and created quite a stir as they told their accounts, while receiving the sympathy of bystanders, many of whom had similar experiences. In Iran, the deportees were taken to one of two detention centers – Safed Sang or Kerman, both run by the army – and kept there for around 20 days. They reported treatment that was harsh both physically and mentally. Many also complained that they had to pay for food and transport to the border, while others claimed their money and belongings were confiscated (see also Stigter 2005b: 28). Some of the deportees had their families in Iran, but were not allowed to contact them before the expulsion. One man from Sara-e Nau, deported in early 1998, recounted:

Return Decisions

117

I was taken off the street, forced to come here. I went back to Iran with a smuggler, stayed for one year. I was then expelled from Iran without the family, and I had to go back to get them with the help of a smuggler. My first attempt failed, I was caught by the police, put in jail for two months and deported. There was little food, we were beaten and humiliated. I lost the 70,000 toman that I had paid to the smuggler. After the capture, I had to pay the Iranian government for transport from one place to another. I had an extra 50,000 toman with me; it was all lost by the time I came back to Herat. I had to work three more months here to make money for a new trip. The second attempt was successful. It took a year and a half from I was expelled till I made it back to the family. My father-in law was there, he took care of the family. I was able to send some messages from here. From the Iranian jail, it was impossible to send messages. The harsh treatment in the detention centers has created a negative image of Iran among Afghans. The extent to which expulsions prevent new migration is unclear. Most of my informants claimed not to have let expulsion discourage them from entering Iran, and the same trend is reported by other analysts (Stigter and Monsutti 2005: 7). The most dramatic consequences of deportation occur when heads of households are expelled, without being given the opportunity to contact their families. In such cases, it is essential to have members of the extended family living in the vicinity, as they can help the family cope during the months – in some cases, a year or more – which it may take for the husband to return.

Material resources and return People in Ishaq Suleman and Sara-e Nau find it easy to understand why many of their former villagers have chosen not to return. No one morally condemns those who do not return – neither those who have stayed on at origin throughout the whole war period nor those who have been in exile and have returned to resettle. In their understanding of why others choose not to return, most people refer to their own economic situation. Many of the optimists returning already in 1992 or early 1993 soon found out that economic survival in Afghanistan was more difficult than they had expected. In 1999, most households were split and relied on the income from one or several family members working in Iran. On top of all this, large parts of the country – the Herat region included – were already well into a second year of drought, with devastating effects on the

118 Social Networks and Migration in Wartime Afghanistan

local economy. All of this, of course, was well known by most refugees in Iran, who received regular news about the conditions at home. Despite the grim economic picture in Enjil, however, there was a trickle of people returning from Iran, including families who returned of their own volition and with the intention of staying. The return decision is about comparing the economic conditions in exile with those in the country of origin. Exile conditions were discussed at length in Chapter 4, and the main focus here will be on the prospective returnees’ assessment of economic conditions at home. Such an assessment does not necessarily focus only on daily income opportunities, but may also take into account site-specific assets in the form of property or even professional competence. A second economic dimension underlying the return decision is the ability to pay for the travel. For example, even if a family believes it has the resources to sustain its long-term economy at the return destination, this does not mean it has the cash to finance the move. The tension between immediate and long-term economic ability illustrates how different kinds of material resources follow different logics and flow in different types of networks. Assessing the economy at ‘home’ In the fragile situation prevailing in Afghanistan after the first return in the late 1980s, it was difficult to assess the future economic conditions in one’s home area. The expectations of the refugee households could also have changed while in exile, particularly among those whose living standards had improved. In addition, various members of the household might have developed differing expectations. When I asked the returnees to Izhaq Suleman and Sara-e Nau why others had remained in exile, most pointed to economic reasons, not lack of security. One old man in Sara-e Nau had not foreseen how difficult it would be to sustain his family after returning from Iran in 1992, and understood all too well the reluctance of others to follow his example. When asked if he knew whether remaining refugees planned to return, he replied: Apparently their life is better there than ours is here. Because they have a pretty good income in Iran, they prefer to live there. They will not find it possible to make a sufficient income here. Therefore, they prefer to live in exile rather than returning to their country. Income opportunities after a return depend heavily on the state of the local economy, as well as the resources of the family. While economic resources at the place of origin – particularly in the form of site-specific

Return Decisions

119

assets, such as land – are essential, the ability to bring resources from exile is also important. The typical image of a returnee from Pakistan is somebody who rents a truck and packs it with all that the household has acquired during its years in exile, ranging from kitchen utensils to roof beams. A similar image is unthinkable with regard to returnees from Iran, because of restrictions on the goods and money anyone can take out of the country. Basically, the only significant property one is allowed to bring with them is the UNHCR repatriation package, which is distributed at the border crossing. Though almost three-quarters of my returnees stated that they had brought household items across the border, these were only kitchen utensils, food items, and kerosene oil for cooking – not furniture, roofbeams, or window-frames. Iran’s restrictions have a long history and have been criticized repeatedly by analysts and humanitarian agencies (Centlivres and Centlivres-Demont 1992: 35; UNHCR 2004: 4). In practice, these restrictions discourage the purchase of durable household items and force prospective returnees to sell their belongings at whatever price they can get. Since refugees are allowed neither to own property nor to run a business in Iran, the possessions they are able to accumulate are modest indeed. Yet, they are seen as important by most refugees. The only real opportunities for taking money back are either to smuggle it or to use the informal Islamic banking system, the hawala, in which money paid to a businessman in Iran can be reclaimed in Herat. This is what most people do when transferring wealth to Afghanistan from Iran, whether the purpose is to support the family at home, or prepare one’s own return. The effect of the Iranian policy is to make resources that are normally mobile into site-specific items, reducing their value. A more liberal Iranian practice in allowing Afghans to bring back capital and belongings would probably have the dual effect of encouraging repatriation and strengthening the ability of those returning to cope when reestablishing their households in Afghanistan. Relatedly, competence developed during exile may also help in acquiring an income upon return (see, for example, Faist 1997b; Massey and Espinosa 1997). Not all competence is equally transferable, however. In the late 1990s, for example, when there was minimal construction activity in Herat and the surrounding areas, all skilled construction workers had problems finding work, but trained gypsum decorators seemed to be particularly disadvantaged. The result was that gypsum decorators – of whom there were quite a few from Izhaq Suleman – did not return. Many households pursue a split strategy because some of their male members have acquired a competence that is in demand in

120 Social Networks and Migration in Wartime Afghanistan

Iran but not in the home country, and because these households have access to networks that are helpful for finding work in Iran. Stepwise returns, in which one part of the household remains in exile while the other moves back to start rebuilding a life, are well known among refugees in Pakistan. In the Pakistani case, the norm has been that most of the household remains behind, while one or a few grownup males go home to rebuild their house and start cultivating the land (see, for example, English 1989b: 24). This is most common in the areas close to the border, where people often make reference to the Pashtun tradition of nimkora, which is essentially an enduring split-household strategy (Morton 1994: 33). The pattern among Afghan households returning from Iran is that most of the household returns, while one or a few male breadwinners stay behind in Iran. In Iran, the risks inherent in stepwise migration are greater than in Pakistan, as a second escape across the tightly controlled border is generally more difficult. If local conditions are not as good as hoped, households may end up in stepwise repatriation inadvertently, or they may have to live as a split family for much longer than expected. For others, however, the ability to benefit from the Iranian labor market is a prerequisite for even contemplating return (Harpviken 2003; Strand, Suhrke, and Harpviken 2004).12 The resources people expect to have access to are also important when they think about return. According to one village leader: Those inhabitants who have not returned yet are people who were living under uncertain economic conditions in the past. They were poor people. But nowadays, they have better income in Iran. This is not a statement about prewar living standards, but about the resources families will get access to at home if they return. People are more likely to return if they have land; and the more land they have, the more likely they are to return. When departing, many financed their travel by selling off animals or household items, but none of my respondents reported that they had sold their land to finance flight. When asked whether the remaining refugees were likely to return, one informant explicitly confirmed that land ownership was the critical issue: ‘Those who have land will come, those who do not will stay.’ This is a fairly accurate description of the situation from the mid-1990s onward, but there were also many landless people among those who returned immediately after the mujahedin takeover in 1992, particularly to Izhaq Suleman, where land was very unevenly distributed. The mujahedin, who had among other things been staunch opponents of the PDPA’s land reform, did not redistribute land in Herat in any way.

Return Decisions

121

The first round of repatriation (1992–93) included a number of people who had no fixed assets or known job opportunities at the place of origin. This was clearly the case in my two communities in Enjil. Peter Marsden (1999: 62) sees this as a general pattern: most of those refugees who repatriated during 1992 and 1993 did not base their decisions primarily on a prior assessment that they could survive financially upon return. A prime motive for returning in those days was simply the fall of Najib and the coming to power of the mujahedin. Those who sympathized with the latter expected that the mujahedin would provide both security and economic opportunities for their followers. In addition, the more quickly a refugee returned, the greater would the chance be to benefit from the political transition. Many came too late, or their expectations were not fulfilled for other reasons. Some had to leave Afghanistan again in order to make ends meet. Hence, there was a massive negative experience with failed returns, and this became widely known within the exile community. No wonder most people thereafter spent considerable time in preparing their return and were extremely cautious when examining their economic opportunities inside Afghanistan. The availability of site-specific resources (mainly land) has a decisive influence on decisions to return. Owners of land and other fixed assets are among the first to return. Those with little or no such property – who therefore have to rely on their education, skills, labor power, experience as traders, or accumulated mobile capital – are likely to delay their return until the government in their home country is able to safeguard the general functioning of the economy. I did not come across one single significant landowner who had chosen not to return, although many landowning families continued to complement their income by sending young men to work in Iran. Land ownership provides a sense of security for relatively stable long-term income (except during droughts) and gives work to all resident family members, but it does not provide quick substantial income. In both Izhaq Suleman and Sara-e Nau, a significant share of the land had been out of use for many years, and some of it was inaccessible owing to the massive presence of landmines. During the drought in the late 1990s, the yields were minimal in both of these villages, which were ill-placed at the dry end of long irrigation channels. After 2001, however, land was again a valuable asset. When I visited Izhaq Suleman in October 2003, I found the village full of private cars, whereas a few years earlier

122 Social Networks and Migration in Wartime Afghanistan

there had been only one or two. I was told that land in the village was now attractive for investors from Herat, and that 1 jerib (2000 square meters) of land equaled the price of a used Toyota Corolla. A shopkeeper estimated that the number of cars had passed 100, and that most of them were financed by the sale of land. In 1999, there were hardly any cases of selling land, and the prices would have been only a fraction of the prices paid in 2003. Almost 3 years later, during my 2006 visit, I found that prices had declined, yet sales of land continued. For landowners, their long-term holding on to the land during the bad times could be turned into a quick profit, but buying a car was clearly not the type of investment that would safeguard the family economy in the longer term. As mentioned, economic resources – both those needed for survival upon return and those for maintaining and expanding the households’ long-term resource base – are the key factors in return decisions. In Enjil, most people had few income opportunities, so the main opportunity for making and saving money came from the labor market in Iran. After years in exile, many Afghans found that their expectations – economic and other – had changed. This was more common among refugees in Iran than in Pakistan, since the difference between living conditions in Afghanistan and those in Iran was much greater than that between the two sides of the Afghan–Pakistani border. Although Afghans in general were economically discriminated against in Iran, and benefited only marginally from that country’s development, their expectations for a better life were boosted by their exposure to a much more developed economy (Adelkhah and Olszewska 2007). Many returnees from Iran complained about the lack of gas, electricity, and tap water in Afghanistan. AbbasiShavazi and associates, when interviewing Afghans in Teheran, were told that some returnees had been coined Irani gak by people who had not been in exile, a term denoting people with an Iranian lifestyle (Abbasi-Shavazi et al. 2005b: 31–32). The split-household strategy meant that expectations developed differently among members of the same household. In a related manner, several female returnees admitted to us that they found daily life – carrying water, collecting firewood, for example – very hard in comparison with life in Iran, but got little sympathy from their husbands. Even more dramatically divergent expectations can be found among the younger generation who had grown up in Iran. These often find it extremely difficult to adapt to the backward conditions in Afghanistan. With inspiration from the general network literature, I initially assumed that vertical ties would be important for access to economic resources. This was not confirmed by my findings in the field, neither in the expectations held by prospective returnees, nor in the networks

Return Decisions

123

reported by people who have returned. Those who did not hold resources of their own had no expectations that they would be getting access to economic resources from people who were better off. The mujahedin’s followers, however, could expect to be offered jobs in return for their loyalty (see also Marsden 1996b: 15; Stigter 2005b: 11). In a situation where no major change of the social order was to be expected, prospective returnees with no substantial land holdings realized that the critical factor for their decision would be the assistance they could get from close associates at home – mainly family. Financing the travel home An immediate challenge for a prospective returnee is to find the money required for the travel. A majority of the returnees interviewed in Enjil did not refer to this cost as a major obstacle. As compared to the costs of going to Iran – particularly for those who relied on smugglers – the costs of return were minimal. Most of the returnees were able to raise the necessary means on their own, although some had to borrow from close relatives. However, among those who remain in Iran, there are most likely many that lack the money and the network resources to finance the travel back. They also have to worry, of course, about their ability to cope once being back. The UNCHR encashment and transport programs aim to overcome the costs of return. This type of support has been in place for refugees both in Iran and in Pakistan since 1990, at least in principle, but in some periods there has been a shortage of funding, and the kind of support provided has varied over time. In December 2002, the package given in Iran to a family returning to Herat consisted of plastic sheets, plastic jerry-cans for water, washing detergent, soap, and wheat – the amount varying with the number of family members. Additionally, UNHCR provided free transport from the border to Herat, as well as additional cash support for those going to other localities. Among those who had returned to Enjil, there was little enthusiasm for the encashment program. Those with refugee cards had received support, but did not see this as a significant contribution to their repatriation and resettlement. When asked during a December 2002 interview whether UNHCR’s repatriation assistance had any impact on people’s decision to return, one prominent shura member in Sara-e Nau, who had recently returned with his family, summed up the general attitude: It does not have any impact; people were also coming during the period when the UN stopped assistance. When the security situation changed, even if there was no support, people came regardless.

124 Social Networks and Migration in Wartime Afghanistan

In a conversation 1 year later, however, the same informant had changed his opinion somewhat, arguing that repatriation support could help those who were in an extremely poor condition to make the leap. He added that those people would need additional support to reestablish themselves in their home communities. My interview data from Enjil indicate that the encashment program made little or no difference for people’s decision to return. So far, however, the poorest refugees, who would need such support the most, seem to be the ones who have chosen not to return. In the next wave of repatriation, in which the people with the least resources return home, a combination of travel support and support for reintegration may therefore be more significant. Within the wider policy debate, the repatriation package has been criticized from two sides: on the one side, for unjustifiably encouraging the return of refugees to a society not yet ready to receive them (see, for example, Turton and Marsden 2002); and, on the other side, for being a waste of resources, since the assistance does not reach the poorest and has had little impact on repatriation. The first argument is hard to sustain, as it is doubtful whether the material support given, or even the implicit encouragement signaled by the support, has made a fundamental difference in people’s decisions to return (Bialczyk 2008). To have such an effect, the support would have had to have been much larger. There is more to the second argument, though, which also has a longer history. It can be traced back to one of the first reports commissioned by UNHCR when the Soviets withdrew from Afghanistan in 1989: An organized repatriation of Afghan refugees on a large scale by UNHCR and concerned governments is ill advised. Such an endeavour could absorb enormous amounts of money and resources that could be more constructively directed to local assistance for returnees inside Afghanistan. The large majority of the refugee population will require no special assistance for repatriation. (English 1988: 1; emphasis in original) Richard English’s (1988) conclusions were based on a survey among refugees in Pakistan and are completely in accordance with what I learned from returnees to villages in the Herat region a decade and a half later. The vast majority of refugees repatriate when they themselves find that the time is ripe, weighing conditions in exile against conditions at their place of origin. For them, the ability to finance the travel is not a major constraint, as they either use accumulated household resources or obtain assistance through their social networks. For a minority, repatriation

Return Decisions

125

support may be of some importance, yet the economic challenges related to their return are also mainly related to the opportunities they face after their return. The real concern is that those who find it difficult to finance their own return will also lack the financial means and network resources to resettle successfully at home.

Information and return Return decisions are generally less reactive, less ‘forced’, than flight decisions, and therefore allow more time for the gathering of information. In contrast to escapes, most returns are also planned with the expectation of staying indefinitely at the destination. Consequently, the need for information about conditions at the return location is greater than the need for information about the destination of an escape. Trust in the sources of information is therefore also a more critical factor than in the exile phase discussed in Chapter 4. People contemplating a return are likely to lean on dense ties for such information. Furthermore, when planning their return, the information they need is highly specific – the state of their house and property, access to production facilities and markets, and so on. Such information cannot be provided by government agencies or unknown brokers. People rely on their close associates or, when possible, on personal visits. While in exile – and prior to any serious consideration of return – each individual’s access to information on conditions in the area where he or she is staying is critical to coping on a daily basis. This is particularly the case in Iran, where little assistance is to be expected from any specialized refugee-management apparatus, so most refugees have to find a job and a house themselves, as well as to contact health clinics and schools. The need to collect information about local opportunities takes priority over obtaining information from home. Once return becomes a real prospect, however, the balance in the need for information tilts. Knowledge about conditions at the place of origin now takes first priority. Given the increased need for information that is both trustworthy and specific, the sources of information that until then have been entirely satisfactory may suddenly prove deficient. Prospective returnees may therefore have to seek out new, reliable sources of information. The provision of accurate, reliable, and highly detailed information about conditions at home becomes perhaps the most critical factor in decisions to return. The information needs are more critical for those who plan to return with their whole family than for the single men who plan to go home and join their families, or the single men who plan only to return

126 Social Networks and Migration in Wartime Afghanistan

themselves and to leave the rest of their family in exile for the time being. Such single returnees will normally have some information from their own visits, and their return is not irreversible. The main focus here will be on families and refugees who have stayed a long time abroad, not so much the short-term individual migrants, for whom a return is often part of a cycle of out- and in-migration. In the following, we will briefly look at information about conditions at the exile location, before shifting attention to the channels of information that supply knowledge on conditions at home. Information about conditions in exile When people contemplate return, their information needs change. It is no longer sufficient to have access to information for carrying on daily life. People need to gauge more precisely how their living conditions are likely to develop in the future either if they stay in exile of if they go home. For refugees in Iran, the prospects in exile are a question both of how the state’s treatment of refugees is likely to develop and of what they can expect from the Iranian population at large. As discussed in Chapter 4, Afghans in Iran are generally not in a position to get access to knowledge on political currents and they lack the types of ties to Iranians that could compensate for this. Once information needs become critical – as in an acute decisionmaking situation – the Afghan refugees do not possess network resources that would enable them to find out more. This lack of effective information networks in Iran generates uncertainty. Since the beginning of the 1990s, a number of government initiatives – including reregistrations, deportation campaigns, and reduced access to welfare – have signaled the government’s opinion that it is time for the Afghans to return (Abbasi-Shavazi et al. 2005b: 16–18). Meanwhile, the treatment of the Afghans by government authorities and Iranian employers has fluctuated, often at a rhythm different from that of official government policies. This has contributed to further uncertainty. Afghans in Iranian exile often depend on a poorly informed rumor mill. The attention the exile population has paid to gathering information as part of its decisionmaking process in relation to return has varied greatly over the years. The early refugees returned spontaneously after the regime change in 1992, without taking time to weigh their opportunities in exile against their opportunities at home. The remaining refugees had more time to think and soon received negative reports from those who came back from Afghanistan after an unsuccessful return. Hence, families who stayed in exile beyond 1993 were much more focused on analyzing and weighing their possibilities in exile against those at home.

Return Decisions

127

One of their main problems was a lack of reliable information about the real conditions they could expect if staying on in exile. In Iran, the Afghan exiles mainly noticed a downward trend in terms of security. By 2002, there had been a continuing negative trend for over a decade, with no significant indications that conditions would take a turn for the better. This caused people to feel that their opportunities in exile were changing for the worse. Thus, it became even more critical to find out whether the opposite might be the case inside Afghanistan. Information about conditions at ‘home’ Information has been a topic of repatriation research and has often been found to be a critical issue, one in which social networks play a significant role (see, for example, Koser 1993, 1997). If people who contemplate return have access to information only through brokers, the trust these are able to build becomes critical. To return is a serious decision – particularly if it includes a whole household – and people want to base such a decision on reliable information. Only when a personal visit and communication with close associates at home are not possible will a prospective returnee consider relying on other brokers. My informants in Izhaq Suleman and Sara-e Nau did not consider information to have been a critical factor when they decided to repatriate. It is possible that more of them would have referred to information as a significant problem if the fieldwork had been conducted 5 years earlier, when memories of the 1992–93 repatriation were still fresh. Of the interviewed heads of household who had spent a significant period in Iran with their families, however, well over half had been visiting home during the exile period. Among the others, almost everybody received information through personal channels, and only a couple of people said that they had not had any contact with the community of origin during their exile. Thus, my informants may simply have taken the availability of information for granted. The statement of a recent expellee from Iran, whom I met in Izhaq Suleman in 2002, shows how quickly information travels: No Afghans will be staying in Iran if there is security and development here. Afghans are not happy there, they are humiliated. Those who come here, think that the situation is good, but when they see, they write back and tell the others to stay. The main issue is that there is no work, we just sit and talk, and we go back to our home. I have been here for two months, and have even not found one Afghani [Afghan currency].

128 Social Networks and Migration in Wartime Afghanistan

Information is derived in multiple ways: personal visits, letters from relatives, and now increasingly telephone calls. In general, refugees have had good access to information about conditions back home. The first wave of returnees, in 1992–93, consisted mostly of people who had used Iran as safe hinterlands from which they supported the fight of the resistance groups. The fighters went to Afghanistan regularly, and many spent most of their time there, maintaining close contact with locals, including the government-affiliated militia. In Izhaq Suleman, the mujahedin would from time to time stay overnight in the houses of people who belonged to the militia. Those who returned with the second big wave in 2002–03 had also traveled frequently back and forth, and many used Iran mainly as a source of economic survival. Some also escaped to Iran after 1995, to get away from the Taliban. The majority of the post-1992 migrants to Iran did not bring their families, and they often came back to visit their kin and assess the situation. Since the refugees mainly settled within particular neighborhoods in Iran, there was a rich and continuous flow of information back and forth. The majority of the refugees from Enjil also had access to a wealth of information about conditions at home from the earliest days of their exile. This makes Afghanistan – or at least these areas of Afghanistan – different from some other known refugee locations, such as Mozambique (Koser 1997). Koser found that the availability of information for Mozambicans in exile changed greatly over time. At first, people received little or no information, largely because the flight had happened so quickly and had been so chaotic that people lost contact with each other during the move. As more people followed, the supply of information grew, but was often contradictory and untrustworthy. Then, the sources of information dried up completely when the areas people came from were fully depopulated. After suffering for a long time from a complete lack of information, people carefully started to build new information networks, in order to follow developments back home. Most of the refugees had no family or friends who remained behind, and the information gap was filled by various actors. Koser found that, at the time of his fieldwork, some 3 years after the main escape, most had developed new, reliable information systems about the situation back home, and a majority had more than one source of information. The pattern that Koser describes is entirely different from the one I know from Enjil, where access to information was relatively rich in the early stages of the war. There was a large outflux of people soon after the Soviet invasion in 1979, but a large share of the refugees traveled regularly back and forth across the border, either to take part in the

Return Decisions

129

fighting or to visit family, and the areas they had fled from were never entirely depopulated. Already prior to the war, there had been frequent contact between communities in Enjil and certain locations in Iran, because of labor migration. It seems likely that skeletal transnational information networks existed already then. This would explain how those in exile could have relatively rich access to reliable information immediately after the first wave of flight. The Mozambicans, by contrast, had to rely – for some time – on information brokers. In Mozambique, Koser identified four types of information channels: institutional sources (governments and relief organizations); media; personal sources; and kin. He thought that the high reliance on kin was a Mozambican particularity, and explained it in terms of the absence of political parties and other institutional sources of information. Kin, however, is also the most important channel of information in the Afghan case. This makes it tempting to question the assumption that kin is central only when other channels do not exist. In all cases, the trustworthiness of the channel is seen as a critical issue, and this is a main reason why kin – even when other sources of information are available – is the preferred option. Institutional sources and the media are rarely trusted. Moreover, prospective refugees demand a kind of highly specific information that such sources are unlikely to possess, namely the precise state of affairs in one particular locality. Generally, the assumption held up in Chapter 2 – that information to potential returnees is most often mediated through brokers – has been proven wrong. Cases from Afghanistan indicate that most people get their main information from family members and other close associates. Brokering ties dominate only when access to the home location is extremely difficult, when it has been depopulated, or when polarization between various groups from the same location runs so deep that one group monopolizes all information. Furthermore, when people have stayed in exile only for a short period, the development of multiple information channels may have been prevented and the dependence on brokers may increase. What has just been said does not mean that the refugees from Afghanistan always had the information they needed. Even when there are multiple sources of information, many feel that they know too little to take the risky decision of returning with the whole family. They prefer to first gather more specific information from a trusted envoy or to go in person. One example is Gul Ahmad, a man in his early 30s, whom I met in Sara-e Nau in December 2002. Gul Ahmad had been sent home by his family to assess the situation. He had stayed for a month and a half and had used family savings made in Iran to purchase a piece of land in Herat city:

130 Social Networks and Migration in Wartime Afghanistan

I will soon have to go back to Iran, talk to my father, my brothers, then we will make a decision. . . . I have two brothers (one married, one not), two sisters, and my parents there. I am married, and have a wife and two children. The most important issue is security, then we look for economic opportunities, if we can survive here. . . . Many of my relatives have asked me to write a list of the prices of commodities to see what is the cost here. In the area we live, there are thirty-five families from Sara-e Nau, and four of them are close relatives of ours. Gul Ahmad was in regular contact with people traveling between Iran and Sara-e Nau, but he and his family wanted more specific information – about the state of their small landholding in the village and the economy in the Herat area. Gul Ahmad ended up investing the family savings in a rather attractive part of the city, effectively taking a first step in the family’s return – and upward social mobility. It is also interesting to note that many members of his extended family network had asked Gul Ahmad to serve as a scout for them, bringing back specific information on matters such as commodity prices. In itself, the trip was a considerable investment, in terms of both time (2 to 3 months) and financial outlays (travel expenses plus cost of passport/visa or help from smuggler). In the case of refugees from Khost and Kunar, studied by Glatzer, the decision to return was in almost every case based on information collected during personal visits by the head of the household. The place of origin was only a few hours away; the border was easy to pass; and most people felt – despite tension and occasional fighting – that it was safe to return for shorter visits. Hence, going on an assessment mission was considerably easier than for the Enjil refugees in Iran. The latter therefore seem to have sent a representative of a larger family group rather than just a representative for a single household. In the area of Mozambique studied by Koser, such temporary returns were not possible; so most people had to base their return decision on information acquired through kin or more distant information brokers. For the Tajikistan refugees in Afghanistan, where the problem of deep rifts in the home communities was compounded by an inability to travel between origin and exile, trust was a scarce commodity, and the opening up of channels for direct contact proved necessary to enable return. The overall pattern is that if it is difficult or impossible to make a personal assessment of conditions back home, such assessments must instead build on trust in outsiders. This makes the return decision more risky and therefore more unlikely. Within households, there may be substantial differences in the access to information. Interviewing both men and women, I found that there

Return Decisions

131

were major differences both in the general access to information and in the type of information available. Of the women interviewed, none had taken part in return visits during their stay in exile. The exchange of letters, however, was more common among the women than the men, in spite of the fact that the overwhelming majority of women are illiterate and therefore depend on others to read aloud for them and to write down what they dictate. A few women also mentioned recorded cassettes, which help them bypass the literacy barrier. Regardless, the women have less access to information from home than the men, and thus depend on what their fathers, brothers, husbands, and sons convey to them.13 When it comes to household decisionmaking, the informational dependency on men – and the husband in particular – implies that women have few resources to draw upon in forming and arguing alternative views. This informational asymmetry within the household has major implications, since we know from studies in exile that women are often far more reluctant to return than their husbands and fathers. In cases where the exile offers higher living standard or a more independent status for women, return may simply be unattractive for them.14 This is also what Grasmuck and Pessar (1991) found in a study of Dominicans working in New York. Three major international NGOs conducted a large survey in multiple refugee camps in Pakistan in 2002 and found that: women to a large extent had been left out of discussions regarding repatriation. In keeping with the traditional culture of many Afghan communities, women generally reported that they were unaware as to any discussions about repatriation that the men had held outside the house. (DACAAR, IRC, and MADERA 2002: 16) The lack of female influence on return decisions does not only relate to informational access. Women are generally allowed to express their views only within the household, though they may occasionally also discuss and develop views in contact with other women. If a collection of households return together, this may further reduce women’s influence, as several heads of households get together to discuss a joint return, at times entering into mutual commitments. The larger the size of the collectives within which prospects of return are discussed and plans made, the less the influence of women is likely to be. The strengthening of information exchange in order to stimulate further repatriation has received increasing attention from multilateral institutions, both generally (see, for example, Koser 1993) and in the

132 Social Networks and Migration in Wartime Afghanistan

Afghan context (see, for example, UNHCR 2004: 3, 22). The suggested means are both increased media coverage of the situation and, when possible, facilitation of personal visits. While information conveyed via the media (primarily radio) may motivate people to start thinking about return, this seems unlikely to satisfy the demand for reliable and specific information. To trigger the actual decision to return, particularly among refugees who have stayed away from home for a long time, the facilitation of personal visits seems much more promising. Ultimately, of course, there is no guarantee that new information will trigger a return decision, but one can be fairly certain that those who have taken decisions on the basis of reliable information are more likely to reintegrate successfully at home.

Conclusion In Chapter 2, where I outlined a theoretical framework, I suggested that the concept of flight collectives might also be applicable to return migration. My data from Izhaq Suleman and Sara-e Nau, complemented by insights from the literature, indicate that a significant share of those who return do so within groups larger than the family. Yet, those ‘return collectives’ are much smaller than the collectives of people that departed together. Most return collectives are based on cohesive ties, centered on family or common origin, while in some cases the significant ties have been developed in exile. Also in cases where there is no formation of a return collective, social ties play an essential role in the consultations preceding a decision, in accessing information, mobilizing economic resources, and coping with the general lack of security. Another finding is that ties between the exile location and ‘home’ are essential. The intensity of such ties among my informants was high, since both the mujahedin and other single men were traveling back and forth regularly. It seems likely that more of the people who still remain in exile have severed their social ties with ‘home’, but I am careful not to conclude firmly here, given that I have not conducted fieldwork in Iran or Pakistan. Nonetheless, it is safe to say that those refugees whose ties with their place of origin have been disrupted must develop new bridging ties as a precondition for return. Network resources are essential for obtaining a sense of security when contemplating return. In Enjil, just as elsewhere, those with mujahedin connections were the first to return in 1992–93. Returning mujahedin were members of protective collectives, and the prospect of getting jobs in a new administration invited swift action. Refugees who had no direct mujahedin link would depend on other, less powerful, networks for their

Return Decisions

133

personal security, and generally adapted a wait-and-see attitude. For the returnees, security was not just a contextual condition, but also something that needed to be nurtured through their own cohesive ties with larger collectivities. This collective dimension of security may contribute to creating threshold effects in return migration: once a certain share of an extended network is established in a certain area, others will feel more confident and follow. Split households were common among the Afghans, in part driven by security concerns, but predominantly (and increasingly over time) by economic ones. People who held property at the place of origin were overrepresented among the early returnees. Those who had been unable to maintain their property during the war, as was the case for many of the people from Sara-e Nau, also returned early. Property-holders persisted in showing greater propensity to return than others even during the years of drought, when there was no income to be gained from agriculture. Commonly, however, economic downturns were compensated by male members of the household going to Iran to work. The ability to tap into the Iranian labor market, and to maintain a split household, is for many a prerequisite for an economically sustainable return. The actual financing of the return was no important obstacle to return. The return travel is normally cheaper than the departure and rarely required the activation of networks. Most people were more occupied with the problem of Iranian restrictions on bringing belongings back to Afghanistan than with the problem of financing the travel. This is interesting, not least because travel support has been a chief program area for UNHCR over the past 15 years. It is worth noting, however, that returnees to other areas may have seen this differently (the poorest people in the villages I studied had never escaped, but stayed at home, although they sometimes were displaced internally); and financing may be more of an issue for some of the refugees who have yet to return. The initial assumption that information for potential returnees is most often mediated by brokers is not confirmed: the Afghanistan case rather indicates that most refugees have been able to maintain relatively rich information access through the period in exile. Nonetheless, when contemplating whether or not to return, people require extremely accurate and trustworthy information, and it is not rare for people to complement network information with personal visits. Information about the exile location is often uncertain and contradictory. With their loose connections to Iranian society, my informants found it difficult to solidify their access to information. It is interesting to note how information needs change once people start to seriously contemplate return – shifting from a preoccupation

134 Social Networks and Migration in Wartime Afghanistan

with information that is vital to daily survival in exile to a primary interest in learning about conditions at the locality to which one may return. Unlike the flight decision, which in most cases has a limited time horizon, a return decision is taken in the expectation of a permanent settlement. The return decision may, like the flight decision, be a decision that is characterized by high levels of risk, regardless of whether one stays, leaves, or – as is oftentimes the case for collectives – hedges against the various risks and splits. The strengths and weaknesses of various types of ties are clarified during the return decision phase. This is because of the serious character of the decision, which is characterized by great uncertainty but with plentiful time to weigh for and against, and once return is undertaken it is expected to be irreversible. Networks become critical for security upon return; and, while the entry point to such networks is normally cohesive ties, smaller clusters need to be nested together into larger collectives if they are to be perceived as effective. In the economic domain, people have to rely on their household or extended families. Information needs for return decisions form another bottleneck. While some find that what is conveyed by close associates suffices, others find that only personal visits are good enough. The return phase sees a continuation of the network structures at work before, during, and after flight, yet the character of the decision places networks under increasing pressure. In the security domain, cohesive networks are clustered together into larger units; for information, cohesive ties are supplanted – whenever possible – by personal visits. Ultimately, however, cohesive ties remain of vital importance.

6 Reintegration at ‘Home’

Returning to what used to be home may be just as complicated as arriving in exile. Many of the challenges are identical, the coping mechanisms similar. The fundamental challenge of reintegration is spelt out by Laura Hammond (Hammond 1999: 229): Whether a returnee comes back to his or her birthplace or settles in an entirely new environment, he/she considers return to be more of a new beginning than a return to the past. Hammond, however, may exaggerate the extent to which the repatriate is aware of this challenge. A difference between integration in exile and reintegration at home is precisely that the former is generally considered to be difficult while the latter is assumed to be relatively straightforward. This assumption is held not only by policymakers and analysts, but also by the returnees themselves, who are therefore often poorly prepared for the hindrances they have to surmount in order to reestablish a decent life in an environment they expect to be familiar. In contrast to what is often the case in exile, there is no host state and no assistance agencies to turn to. The government of the country one returns to is not likely to have much capacity to help and is unlikely to engage itself at all with reintegration at the community level. It is easy to assume that when people return to their original homes, they return to an existing resource basis, so they can make a living on their own (Black, Eastmond, and Gent 2008). This is often not the case. Even returning landowners need time to recultivate their land. Governments also find it difficult to single out returnees for special treatment, since the problems faced by the latter are not necessarily any greater than those faced by the people who never fled. If a government provides special 135

136 Social Networks and Migration in Wartime Afghanistan

treatment for returnees, then this may lead to tension between them and those who stayed (Bruce 2007; Stefansson 2004a, 2004b). On the other hand, the returnees, who have become accustomed to the services of aid agencies during their time as refugees, may have developed greater expectations of the local administration than those who stayed behind (Allen and Turton 1996: 14). The lack of assistance during the reintegration phase is also the reason for the paucity of information and research on reintegration processes (on obstacles to studies of return, see Allen and Turton 1996: 2–3). Reintegration is a recent addition to the research agenda in studies of displacement (Koser and Black 1999: 10; Preston 1999). Most studies of reintegration focus on the general preconditions for recovery in a postconflict phase, rather than on the specific situation faced by returnees. While reintegration must be studied in context, it is essential that the context be viewed from the perspective of the returnees themselves. We should not simply assume that, if a society as a whole recovers, returnees will benefit just as much as anybody else (Chimni 2003; Fagen 2003). The few existing studies of reintegration pay an inordinate amount of attention to the policies and interventions of states and international agencies, whereas in most contexts where returnees strive to reintegrate these actors play only a marginal role (Koser 2000). The realization that formal institutions mostly play a marginal role has inspired some alternative analytical approaches that emphasize how returnees fend for themselves, using whatever network resources they have (Koser and Black 1999; Van Hear 1998). Nicholas Van Hear has shown how the local population’s reception of returnees can vary just as much as the response of host populations to refugees arriving in exile (Van Hear 1998: 56). This is more clearly the case the longer the absence has lasted and the deeper the political conflict has been between those who stayed behind and those who left, and it is particularly problematic when the houses and land of those who fled have been taken over by others in their absence (Stefansson 2004b). In some cases, political disagreement with local power-holders has been the main motive for the departure. Relations among the returnees themselves may also be complicated. Additionally, relationships between the ‘home community’ as a whole – stayers and returnees alike – and their affiliates abroad often continue to be important after the bulk of the refugees have returned (Black and King 2004; Kivisto 2001). Returnees leave some behind in exile and also have often formed relationships to the local population at the place they stayed during their exile. The fundamental research challenge is to move beyond the mere contention that networks play an

Reintegration at ‘Home’ 137

essential role in reintegration, and to determine the exact roles – both integrative and disruptive – played by networks of various kinds. Micro-studies of reintegration are preoccupied with family networks (see, for example, Van Hear 1994), but the literature has not established so far what exactly family networks do in different contexts. We do not know what it is that flows in the different types of networks. The critical question is commonly considered to be whether or not returnees have family networks to rely upon when they return. The underlying premise is that when people have not sustained such networks during exile or return to a location other than where they lived prior to departure, they will lack the networks needed to facilitate reintegration. The concept ‘community alienation’ was introduced in the early 1990s to describe how individuals or collectives feel at odds with the tight communities that can develop among exiles (Stein and Cuny 1991: 20–21). Community alienation, however, may be just as prevalent during reintegration. Struggles over the attribution of responsibility for past misdeeds may tear a community apart (Allen and Turton 1996; Fagen 1996, 2003). Dramatic change may also have happened at home during the long period of absence, and those in exile have been exposed to a different society and may have altered their basic views and preferences. The most dramatic instances of community alienation are likely to occur when all these various factors are in place and interact. To the extent that reintegration researchers have focused on social networks, they often voice the concern that networks are overloaded by the needs of the returnees. There are several possible reasons for this. Firstly, when a large group of repatriates arrive who have been closely associated while in exile, they are likely to cultivate a distinct identity upon return. More importantly, however, a large coordinated return is likely to impose a large burden on the receiving society, which may be poorly equipped for the task after enduring conflict. Unfortunately, there are factors that tend to encourage simultaneous return, regardless of its disadvantages, including, as discussed in Chapter 5, the tendency to react collectively to significant events, such as a regime change, which is exacerbated by the tendency for people to copy decisions of significant others within their networks. Discussions about reintegration in Afghanistan have focused on various aspects of humanitarian assistance to the population in general (see, for example, Maley 1989: 32–34). There has been little discussion of the particular situation of repatriates and of how they can reintegrate with the population who stayed behind. Little documentation exists, in part because the main actors, including UNHCR, have been preoccupied with voluntary return packages and other short-term measures (Jamal

138 Social Networks and Migration in Wartime Afghanistan

and Stigter 2002: 2; Turton and Marsden 2002: 44). Access even to existing information may pose a problem. UNHCR, for example, compiles regular returnee monitoring reports, but does not make these public (Turton and Marsden 2002: 43–44). In both 1992–93 and 2002–03, the waves of repatriation to Afghanistan were triggered by a combination of political change at home and pressure exerted by host governments (and partly by host populations). Contagion effects among the refugees – in the sense that people copied the course of action of others – no doubt contributed significantly. There are major differences between the reintegration challenges of the first and second repatriation waves. In 1992–93, a significant share of the returning refugees owned land or other resources at their place of origin and had maintained regular contact with these, often through engagement with the resistance. However, their expectations of an improved security and economic situation were quickly disappointed as new rounds of civil war broke out. In 2002–03, a larger share of the returnees did not have access to land or other resources. They had been away longer and had weaker ties with home. Expectations of economic development and security were again disappointed, albeit less dramatically than in 1993, since the war, at least in the first few years after 2001, was of a lesser magnitude. Economic development picked up more slowly than people had expected, although Kabul and Herat saw more rapid transformation. Security was uneven, but it was particularly bad in the south and east, where the US-led Operation Enduring Freedom (OEF) continued (Suhrke 2008; Suhrke, Harpviken, and Strand 2002, 2004). By mid-2005, declining security in many parts of the country again posed a serious threat to reintegration. As in many other reintegration contexts, housing is a major problem in Afghanistan. UNHCR has therefore followed up its encashment program with a shelter support initiative. Registered returnees who hand in their refugee documents and can prove that they possess land for constructing a house may receive roof beams, doors, and windows for a two-room house, on the condition that they construct the house by their own means. While this program goes further than encashment in addressing the reintegration challenge, it still aims only at the period immediately after the return. A more fundamental concern is that the program has nothing to offer the most vulnerable: those who either do not own or are unable to prove their ownership of land. In a 2005 review of the program, it was found that only 22 percent of the registered returnees had received shelter support (Foley 2005: 33). Based on the reasonable assumption that the share of economically vulnerable and propertyless returnees gets higher the later we are in the return sequence, the absence of assistance programs

Reintegration at ‘Home’ 139

targeting this group must today be a fundamental concern (see Chr. Michelsen Institute 2005: 16; Faubert, Mojadeddi, and Sofizada 2005). Most of those who have yet not returned have stayed abroad for a long time. They have lived mainly in urban centers, with a way of living that is very different from what they can expect in Afghanistan – in economic terms, in terms of public welfare, and also in terms of popular culture (Grandi 2002). Generally, the difference between home and exile is more pressing for returnees from Iran than from Pakistan. Young people who may have lived all or most of their life in Iran find it difficult to adapt. Middle-aged women, who remember the conditions in the Afghan countryside prior to their departure, may have also become accustomed to a life in Iran with access to such amenities as electricity, gas, and running water (Rostami-Povey 2007). The contrast between those who stayed put in their areas of origin and the returnees is manifested in the different language accents that many – the young in particular – have acquired in Iran, as well as in different clothing and hairstyle. There are also political obstacles to adaptation, sometimes with grave implications for the security of both returnees and other members of the local population (Van Selm 2002). Many returning refugees have been associated with political parties, and many have taken part in the armed struggle. As Patricia Weiss Fagen has pointed out in a 1996 discussion paper on reintegration, being a ‘victim’ does not rule out the possibility that one has also been a perpetrator of violence (Fagen 1996: 7). Two types of challenge follow from the return of former belligerents. One is the challenge of reconciliation: how can individuals or groups live together after having been engaged on opposite sides in the war? Reconciliation will not be discussed here, but it is worth noting that, after almost a decade and a half of large-scale return and more than 7 years after the fall of the Taliban, virtually no measures have been undertaken to bring war criminals to justice (AIHRC 2005; AJP 2005; Niland 2004; Rashid 2008: 140–144). The effects at the local level are serious, not only because people get the impression that war criminals will not be held accountable, but also – and more damagingly – because people see that precisely these war criminals often return to powerful positions, politically as well as militarily. Hence, the second challenge is one of conflict management and prevention. Returnees may be active members of an armed group and bring conflict to their old locality, or they may come under pressure to choose sides in an ongoing local conflict. For many returnees, their association with a political-cum-military group is the only security guarantee for them and their families, and in some cases it is also the main avenue to economic survival.

140 Social Networks and Migration in Wartime Afghanistan

For western Enjil, the reintegration of those who have returned since 2002 has been easier in one sense, but more difficult in another, than for those who returned in 1992–93. Most of the first-wave repatriates came back to villages that had been totally depopulated and heavily damaged, with the exception of the handful of villages that had been working with the regime (Marsden 1994). This meant that most of the returnees faced considerable and immediate problems of reconstruction, but in the longer term, most of the returnees coped reasonably well. Western Enjil is a rich agricultural area, although the region’s output has been severely reduced by the collapse of water management related to the large channels that irrigate the Herat plain, combined with the drought at the end of the 1990s. Furthermore, being in the vicinity of Herat city implies opportunities for day-laboring, petty trading, and small businesses. By 2006, the area along the main road toward Iran, well beyond Izhaq Suleman, had developed into a contiguous business and settlement area. The returnees of the first wave filled an empty space and have largely overcome the problems they initially faced. The security situation has also been relatively good, although there were increasing problems with crime during the last years of Taliban rule, when many villages set up community-based guard systems. While the area was in a much better shape in 2002 than in 1992, and should therefore now be able to accommodate and integrate the second wave of returnees, the change for the better seems in large part to have been offset by the fact that a large share of the most recent returnees do not own any property, have little else to offer than their labor, and are not always welcome locally. Longterm reintegration challenges for the second wave of returnees, therefore, seem considerably larger than for those returning in the first wave, and are likely to be even larger for those who are yet to return. In the following, I will apply a structure familiar from the three preceding chapters, with sections on security, material resources, and information, each divided into two further subsections. I conclude by discussing the main analytical findings on reintegration.

Security and reintegration Security threats against returnees may come from a repressive regime, from various armed groups, or from criminals. As mentioned, there is security in numbers. For many of those who contemplate return, it is essential to be part of a larger solidary group that can provide basic protection. In this context, the political networks that have been built and transformed through the wars play a huge role. Upon return, preestablished relations

Reintegration at ‘Home’ 141

with local power-holders can be tremendously helpful, since these may ensure protection from crime and from opposing political groups. Security is the ultimate priority of most returnees, both refugees and Internally Displaced Persons (IDPs) (see, for example, Johnson and Leslie 2004: 16). The security situation in Afghanistan has differed dramatically from one location to the next and has fluctuated over the years. The region around Herat, which was among the areas most affected by war during 1979–92, has been one of the most secure since then, as well as during the 6 years of Taliban rule. For some individuals, such as the most prominent mujahedin leaders, security may have been a fundamental problem nonetheless, but these represent the exception. In the two Enjil communities, few mentioned any security problems when they were asked what the main challenges had been upon return. Among those interviewed during the 1999 fieldwork, only one person defined the Taliban as a challenge, but it is likely that more people would have done so if the subject had not been so politically sensitive. Two persons mentioned land mines – which had been a massive problem in the first half of the 1990s, particularly in Izhaq Suleman, but had largely been resolved through large-scale clearance well before my 1999 fieldwork.1 Generally, the respondents put economic difficulties and housing problems on top of their list of challenges. The pattern after 2001 was similar, although by now many of the returnees had much more distant ties to the power-holders – and therefore felt more vulnerable. Those returnees who had connections within the local power structure that built on mujahedin networks were most confident of their security. Reintegration of the returning warriors There are several linkages between the reintegration of returnees, on the one hand, and the demobilization and reintegration of fighters, on the other. As most observers point out, the successful reintegration of fighters is a precondition for fostering the security needed for the successful reintegration of returnees (see, for example, Faubert, Mojadeddi, and Sofizada 2005: 27). Thus, there is a strong need to invest heavily in Disarmament, Demobilization and Reintegration (DDR) programs, although this poses serious moral dilemmas: DDR programs in post-war societies often arouse debate over whether combatants . . . should be accorded special treatment over groups such as refugees and internally displaced people. (Sedra 2003: 16)

142 Social Networks and Migration in Wartime Afghanistan

A proportion of the returnees are (current or former) fighters; and, vice versa, a proportion of the fighters are found among the refugees. Nonetheless, it is common to distinguish sharply between fighters and returning refugees. The two are seen as distinct rather than overlapping groups, and the return of refugees, who are all assumed to be ‘civilians’, is seen to signify a successful peace process. Howard Adelman has examined the assumption that refugee return is a condition for – or at least a significant indicator of – a viable peace process, and he finds that that there is no relationship (Adelman 2002; see also Bhatia 2003; Chimni 2003; Eastmond and Öjendal 1999). More dramatically, however, rapid return may threaten the viability of peace if returning refugees are actually fighters or are mobilized as fighters. In that case, the returnees themselves may represent a security threat, and hence undermine a peace process.2 There is an understandable reluctance to realize that the same person may be a returning refugee and fighter – both victim and perpetrator. The ‘refugee warrior’ debate has yet to inform the analysis of reintegration. In the wake of the fall of the Najibullah government in 1992, Ismael Khan’s authority in the communities of western Enjil, as in most of Herat province, was hardly in dispute (Giustozzi 2009). In his core areas, Ismael Khan was the sovereign ruler, whereas in other areas nominally under his authority he depended on alliances with local commanders. There could be tension and conflict over resources and power at the community level, but the fact that there was no real contender for authority at the regional level constrained such local conflicts. In Sara-e Nau, there was only a marginal population that could be considered resident by 1992, virtually all of whom had been displaced for shorter periods during the 1980s. There was little controversy when, in 1992, returning mujahedin filled the position of arbab and dominated the village shura, which, in contrast to the situation in Izhaq Suleman, was established already in 1992. The main mujahedin commander from Sara-e Nau, who had been associated with various groups throughout the war but had been a confidant of Ismael Khan in recent years, occupied the position of arbab, and the shura was filled with his close associates. Those who had been internally displaced to Izhaq Suleman were perceived as ‘neutral’ rather than as supportive of the PDPA government, despite the fact that some of them had been called on for militia duty. More importantly, the IDPs did not lay claims to power, but subordinated themselves to the mujahedin structure, which was brought home wholesale from exile. Also in Izhaq Suleman, a majority of the returnees in 1992 and 1993 were mujahedin, and the rest were also in some way associated with them. In the words of one of the major commanders:

Reintegration at ‘Home’ 143

When Najib lost power, I returned with my whole family. I was the first Mujahed to enter Izhaq Suleman, one week before Ismael Khan took power in Herat. In Izhaq Suleman, however, those who had been associated with the militia – hence did not escape at any stage – maintained considerable influence in community decisionmaking. The maintenance of political influence was possibly facilitated by the killing of their most controversial figure, Arbab Saidu, in 1991. In the words of the mujahedin commander quoted above: The relation between the villagers and the government was very good. Arbab Sayed Mohammad, from this village, was with the government, but he also had good relations with the mujahedin. The contact maintained between militia and mujahedin during the period that they were officially at war with each other continued after the conflict and laid the foundations for an informal division of power in the mujahedin era. While the mujahedin got the upper hand, several of the people who had been active with the militia remained influential. The only people who lost all their influence in the aftermath of the power change were the family members of the killed local militia leader Arbab Saidu, most of whom sought exile in Iran, leaving their property in the village behind. The 1991 killing was reportedly carried out by a mujahed who had been regarded by the arbab, in the words of several informants, as being ‘like his son’. This young mujahed, who came from one of the neighboring villages, had spent many nights at the arbab’s house during the war. The arbab was killed during the night, and the mujahed, thought to be operating on the direct orders of Ismael Khan, disappeared with the arbab’s turban, probably so he could use it as evidence of a successful mission. The arbab’s oldest son took over as militia commander, but he was young and inexperienced and did not manage to gain the respect of the militia leaders in the surrounding area. He stayed in the position until the fall of Najib, and then headed for exile. Qala-e Muhajerin – which literally means the ‘refugee village’ – is a small settlement close to Izhaq Suleman. It was set up by a handful of refugees returning from Iran in 1992. Two of the three pioneers who first bought land and set up houses here had lived in Izhaq Suleman before the war, but none of the others in the 20 households that constituted Qala-e Muhajerin by 2002 had lived in the area before. The settlement was partly based on preexisting family ties, but the members originated from various places within Herat province, and ties between some of the

144 Social Networks and Migration in Wartime Afghanistan

families had been developed through marriages during the exile. The majority of the Qala-e Muhajerin residents had stayed in Iran for at least a decade, most in the cities of Taybad and Torbat-e Jam near the border. Most of the men had been active in Ismael Khan’s forces. Despite their close association with the mujahedin, as well as the local background of two of the settlement’s founders, it was perceived as a community of outsiders by most residents in Izhaq Suleman. With the coming to power of the Taliban in 1995, the advantages of those who had a mujahedin background were turned into disadvantages. The Taliban were extremely suspicious of anyone with a mujahedin past, but less suspicious of those who had been associated with the so-called communist government. The Taliban acted pragmatically, often allowing former PDPA associates – since they did not pose any organized threat – to retain their positions both within the administration, in the army, and in local power positions (such as arbab posts). Known mujahedin, by contrast, were often stripped of their positions and were kept under continuous scrutiny, since they were assumed to be plotting an armed uprising. Those who had enjoyed the largest benefits when returning in 1992 were also the hardest hit by the Taliban takeover in 1995. This shows how important political networks can be in the process of reintegration. The Taliban had reason to be suspicious. After their takeover of Herat in September 1995, Ismael Khan and many of his men again escaped to Iran, where they were once more given refuge. The host state also seemed to be more active in arming the mujahedin now than it had been in the 1980s: Iran established several training camps in the Mashad area, and some sources claim that Ismael Khan alone had five camps, with a total of some 5000 fighters (see also Griffin 2001: 80; Rasanayagam 2003: 169). In October 1996, Ismael Khan and 2000 of his fighters were flown from Iran to Mazar-e Sharif in Afghanistan’s central north, to reinforce Abdul Rashid Dostum’s forces, who continued their fight against the Taliban (Rashid 2008: 71). Among my informants, there were two who confirmed that they had been part of the forces airlifted to Mazar after having been trained and equipped in Iran.3 The Mazar operation ended badly for Ismael Khan, who in May 1997 was arrested and handed over to the Taliban by Abdul Malik, a sub-commander of Dostum who swapped sides. Ismael Khan was immediately taken to Kandahar, where he spent the next 3 years in a Taliban prison from which he escaped in early 2000 (Griffin 2001: 257; Rasanayagam 2003: 153). The residents of Sara-e Nau, who had benefited from their loyalty to the mujahedin during Ismael Khan’s first reign (1992–95), quickly found out that the situation had changed when the Taliban came to power.

Reintegration at ‘Home’ 145

Overnight, they were at a disadvantage. By 1999, the mujahedin-associated arbab had to resign, and a new one – from a family considered to have been neutral during the PDPA era – took over the post. The composition of the shura also changed. There was no permanent Taliban presence in the village, but patrols occasionally visited to ask questions or to search for specific people. A couple of persons who had been with the PDPA in the 1980s, and had later joined the Taliban in Herat, also visited regularly, and considerably influenced local decisions, although they did not hold any formal positions. On a couple of occasions when I was present, one of these men, an employee of the much-feared Taliban intelligence service in Herat, literally gate-crashed shura meetings in Sara-e Nau, creating a rather nervous atmosphere. There was none of the normally quite open debate, and the only decisions taken were those dictated by the unwelcome guest. Izhaq Suleman fared better under the Taliban, with its past as a militia rather than a mujahedin village. Here also, people known to have been associated with the mujahedin were considered a threat by the Taliban, who often held relatives, shura members, or local mullahs responsible if suspected individuals could not be found. Two of the local mullahs recounted to me the first weeks of Taliban rule: We had some other problems, after the Taliban took control here. They came to the village and requested everybody who had been a militia or a mujahedin commander to hand in their weapons. Those who did not were beaten. Many had handed in their weapons in the time of Ismael Khan, and had no weapons to hand in now. One man from the village, he was even disabled, was captured by the Taliban and beaten a lot. He was taken to a police station in Herat, where he was first beaten, and then drowned in a well. We went to Herat city and got the body from the hospital. After some discussion we decided to bring the body to the Enjil district administration, to show how people were treated. The district administrator called the governor and presented the case. We then went home to the village, where we started to prepare for Friday prayer. We had not even been able to wash our feet before people from the police in Herat came and arrested five of our men, who were brought to a police station in Herat and held for 24 hours under terrible conditions in the basement. We were then released. A majority of those who had been connected to the mujahedin were at some stage traced down and arrested by the Taliban. One man from Izhaq Suleman, during an interview in 1999, told the story of his brother, who was in Iran:

146 Social Networks and Migration in Wartime Afghanistan

Now he can not come back to visit here. He was with the mujahedin, then for some time with the Taliban, but then they [the Taliban] suspected him of hiding some arms, and he had to flee to Iran. I myself was put in prison for a month, as they tried to get information from me about my brother. Then they let me out. Nobody in Izhaq Suleman supported the Taliban, with the possible exception of a handful of internally displaced Pashtuns from the battlefronts in Badghis province to the east, who lived there temporarily in 1999. The Taliban pressured some of the village leaders – first and foremost the mullahs – to monitor what was going on and to assist in administrative tasks, such as tax collection. The latter adapted to the extent they found necessary. In many ways, this was reminiscent of how the local community had dealt with the PDPA regime in the past, although the collaboration with the Taliban was much more limited and did not include military engagement. A handful of locals strengthened their power considerably through this arrangement, serving as useful links to the troublesome rulers and invoking the Taliban’s authority when acting on their behalf – a classical broker role not unlike that commonly enacted by the arbab. The residents of Qala-e Muhajerin, known as a pure mujahedin settlement, found it particularly difficult to deal with the Taliban. Relations with Izhaq Suleman, of which Qala-e Muhajerin was administratively a part, were poor even before the Taliban came to power, but seem to have deteriorated further after 1995, when close association with well-known mujahedin, such as them, was a security risk in itself. Hence, the residents of Qala-e Muhajerin tried to keep as low a profile as possible, while paying their taxes as requested, in order to avoid Taliban attention. At the same time, some of the young men worked for or fought with the mujahedin, both in Iran and in Afghanistan, using their need to tap into the Iranian labor market as a cover story. In some cases, the Taliban found out about this and launched reprisals against their families. The 2001 regime change in Herat resembled the one in 1992, in that the force spearheading the takeover had largely been built up in exile, but the change differed in that most members of the returnee force this time had firm family roots in the area. Ismael Khan was well placed to take advantage of the post-9/11 campaign, having brought most of his exile troops to the inaccessible Ghor province earlier in the year, from where he could drive the Taliban out of northwestern Afghanistan in the wake of the US-led intervention (Harding 2001). Once Ismael Khan was back in the vicinity of Herat, claims Antonio Giustozzi (2003: 13), ‘his armed forces swelled to many thousands’.4 To Giustozzi, this illustrates the resilience of the system of patronage that underlies the power of warlords.

Reintegration at ‘Home’ 147

He believes that, during 2002, Ismael Khan commanded the largest armed force in Afghanistan. Combining Iranian support with the customs revenue from Afghanistan’s main border post, Ismael Khan was able to pay his soldiers generously, thus obtaining loyalty and discipline in return. Ismael Khan’s organization brought together an odd amalgam of various solidary groups, which were placed within a hierarchic structure. Khan’s army was both flexible and effective as long as it was well resourced, but armies bringing multiple groups together tend to be less robust than those that are based on a traditional qawmi solidarity (Jalali and Grau 2001: 51; Johnson and Leslie 2004). In Sara-e Nau, the removal of the Taliban led to a major reshuffle in the shura, with people who had stayed in the area during the war, and who had been considered ‘neutral’ by the Taliban, being replaced by mujahedin associates who had spent a long time in Iran. The arbab post was held by a man who had spent the whole war in the area, some of it in Izhaq Suleman. He had been appointed to the post by the Taliban in 1999 as a replacement for the main mujahedin commander in the village. Seeking to preempt a reversed arbab shuffle now that Ismael Khan was back in control, the incumbent arbab put forward a younger relative of his as a candidate for his replacement. This was accepted both by the local shura and by the Herat authorities. The old mujahedin network, however, continued to make itself felt, and, by late 2003, some 30 men from the village held military positions in Herat, most of them within Ismael Khan’s personal guard at the governor’s house. In Izhaq Suleman, the changes in local power relations after the fall of the Taliban were less dramatic. There was some reshuffling of shura membership, but the arbab, whose family had always balanced cleverly between the political groups, remained in his post. As in Sara-e Nau, those who were with the mujahedin were enrolled in Ismael Khan’s new security apparatus, but in Izhaq Suleman they only constituted a small share of the population. The mujahedin residents in Qala-e Muhajerin gained greatly from Ismael Khan’s reascension as the Emir of Herat. From its establishment in 1992, Qala-e Muhajerin had formally been a neighborhood – a mahalla – within Izhaq Suleman; but, after the Taliban’s fall, the settlement was invited to send its own representative to the shura. Equally important, by late 2002, about half of the households in the mahalla had found work with the military forces in Herat – not surprisingly, given that several of them had been with Ismael Khan in exile and had taken part in his anti-Taliban campaign. The villages of western Enjil reveal a tight and multifaceted relationship between military engagement and reintegration. In Herat in 2001, the

148 Social Networks and Migration in Wartime Afghanistan

forces that overthrew the Taliban were virtually indistinguishable from the first wave of refugee return. Furthermore, regime change at the national and regional levels, not surprisingly, leads to altered power relations also at the community level. More interestingly, however, militarized structures are brought back to the country from exile in the aftermath of a regime change, and the armies formerly based in exile come to constitute the new security apparatus. In Herat, these militarized structures, rooted in various local networks that were nested into a semihierarchic organization under the command of Ismael Khan, were key to ensuring stability in both the post-1992 and the post-2001 periods. A year or two after the fall of the Taliban, however, as the new government began to claim political authority and a right to Herat’s massive customs revenues, the armed units under Ismael Khan’s control became a problem, also from the perspective of local residents who were worried about the prospect of yet another round of fighting. Tension rose in western Afghanistan, as the government tried to wrest power from Ismael Khan, bringing his allies over to its side. Massive fighting broke out between Ismael Khan and Amanullah Khan in the area around Shindand air base, with the latter probably getting support from the government. In September 2004, President Karzai deposed Ismael Khan and appointed a new governor of Herat, against some local protest but without any major fighting (Constable 2004). In a social network perspective, the military and political history of the western Enjil communities illustrates the importance of power hierarchies within the security sector – the one sector where hierarchic structures play a vital role. Networks that had developed in the context of war – and to a considerable degree in exile – also proved remarkably resilient and filled a void left by a weak government. The military organizations that offered security to some, however, did not cater to the needs of others. Many find the militarized networks closed and self-protective. The later in a return sequence we are, the larger the share of returnees who are excluded from the corridors of power. One implication, of course, is that militarized networks may be great at producing security for their associates, but may at the same time cause insecurity for everybody else. Basic security To be with one of the armed contenders may give both individuals and their families a sense of security; but, as was the case for the mujahedin associates in Sara-e Nau, an association that gives security under one regime may be a massive source of insecurity under another. Serving with – or helping – an armed group may give a sense of security, but it also implies a commitment

Reintegration at ‘Home’ 149

to accept security risks. Those who elect not to engage with any armed group, however, face another kind of risk, since they have no protection. Victorious armed groups often monopolize the security domain and offer protection only to their own supporters. The best is to engage with the winning side (and, if possible, switch sides at a propitious moment), the second best not to engage with any side at all, and the worst to engage with – and stay loyal to – the losing side. During Ismael Khan’s first reign (1992–95), the majority of the returnees were mujahedin associates. Until the Taliban advanced northward in 1995, there were no credible contenders for power in the region. Yet, at the community and family level, there were real security concerns. Young men could be conscripted; there was little protection against crime; and conflicts between local clans sometimes turned violent. Among my returnee informants, the propensity for migrating back to Iran was much higher among those not associated with the mujahedin. They clearly felt more vulnerable. Those linked to the mujahedin were more successful in protecting themselves against crime: they could seek help from their comrades-inarms, and they could set up self-defense patrols in the villages. Under the Taliban, those who had not engaged with the mujahedin were at an advantage. The years 1995–99 were fairly stable, but crime rates increased considerably in the later years of the Taliban. Villages continued to organize self-defense patrols to protect against intruders, but felt uncertain about how the Taliban would view such local initiatives. There were well-founded suspicions that some of the criminal bands operating in the area pretended to represent the Taliban. In Izhaq Suleman, which was in the immediate vicinity of an artillery base, the hosting of Pashtun IDPs was a means to obtain Taliban protection. A handful of families from Murghab in Badghis, also in the northwestern parts of Afghanistan, had taken refuge in the village. They were offered housing and other assistance by the villagers. Some of the young Murghabis took up jobs with the Taliban at the artillery base across the road. People quickly realized that treating the Murghabis well was a good investment in security. In the words of one respondent: People gave them houses that were not occupied, voluntarily. People seem to be happy with them, they feel that they [the Murghabis] give them some protection. Since none of the locals worked for the Taliban, or in other ways cultivated direct contact, they found this indirect method of enhancing security useful.

150 Social Networks and Migration in Wartime Afghanistan

Certain threats are of a kind that no network resources can mitigate. The US bombing campaign in late 2001 also hit Izhaq Suleman. Several bombs, including at least two cluster bombs, hit the village, killing 12 people and injuring 14 (see also HRW 2001; Shadid 2002). There were contradictory reports as to whether or not Taliban from the army base had been hiding in the village at the time. When I visited the village just short of a year later and talked to a number of villagers, including some who had lost family members, most agreed that getting rid of the Taliban had been worth the price, since the situation under Taliban rule had been becoming increasingly difficult and other attempts at accommodating or confronting them had failed. During Ismael Khan’s second reign, international NGOs paid considerable attention to security problems in Herat, with two reports being issued by Human Rights Watch (HRW) in the latter part of 2002 (HRW 2002a, 2002b). These claimed that Ismael Khan’s services were making summary arrests of political opponents as well as committing other human rights violations. Although these claims were certainly true, I could not detect any major concerns about this among people living in the rural areas around the city. Most people thought security had improved considerably with the departure of the Taliban. Criminal offenses had also been reduced. The main worry was that the emir’s failure to comply with demands from the new government in Kabul would lead to an armed confrontation. In the run-up to the parliamentary elections of late 2004, Ismael Khan was removed from his post, but without any fighting (Walsh 2004). When security threats reach a certain scope, family networks do not provide any effective response. Networks of a larger scale are required, and are often only available through connection with an armed political group. This is why local political mobilization is often triggered by security threats, rather than a perceived need to promote a particular political program. The most well-known political networks from a period of armed conflict may continue to play prominent roles in providing security in a post-conflict phase, while they seek to transform themselves into political parties. Most of their followers, however, will continue to see them primarily as a source of security protection. Those who have no links to larger political networks find it hard to get protection. This is one reason why returnees, particularly those who have used exile as a means to avoid aligning themselves with any political faction, are particularly vulnerable. In general terms, the problem when returnees depend on social networks for security is that such security arrangements are extremely unevenly distributed, and that recent returnees – unless part of a successful fighting group – are likely to be those that fare the worst.

Reintegration at ‘Home’ 151

Material resources and reintegration Networks are needed to get economic support for resettlement – both for meeting immediate needs, such as housing and food, and for establishing a sustainable household economy. The economic landscape may have changed dramatically from how it was prior to flight. As far as local social inequality is concerned, there is a great deal of continuity: ‘those who achieved status on return had come from relatively advantaged social origins prior to departure and in exile’ (Preston 1999: 30). Even returnees with status and property, however, may find it difficult to survive in the short to medium term. Their houses may have been destroyed and their land may have been fallen out of use. The situation is not necessarily better for those who had never fled. Although they have themselves been forced to find a way to survive, they may have little or no capacity to assist the returnees. Some of the returnees in the first few years after 2001 were actually far better off than those who had stayed behind, since they had been able to accumulate capital and acquire valued competence in exile (Faubert, Mojadeddi, and Sofizada 2005; Hugo 2005; Shami 1996). This is not the case, however, in the communities I have studied. There, the majority of the recent returnees are comparatively disadvantaged. Not just security, but also economic opportunities can be enhanced – or lost – through political and military engagement, both in conflict and in post-conflict settings. New regimes may earmark jobs for those they trust, or they may use economic incentives to buy loyalty. Sometimes, an attractive posting or economic opportunity may be used by the winning party to co-opt former adversaries, or by the UN or other international actors to prevent military leaders from spoiling a peace process. This is the logic of the internationally driven DDR program for Afghanistan (UNDP 2003: 4). The idea is to offer economic opportunities to a commander in order to pull him out of his protective role vis-à-vis his followers, and to give him instead a stake in the new order (Stedman 1997, 2002; Harpviken forthcoming-b). The problem, however, is that the networks underlying a military organization are often strong and may be tied to economic opportunities on a scale that is difficult to match (particularly if narcotics, smuggling, or other forms of illegal economic activity are involved). When returnees to western Enjil were asked what had been their greatest challenges when resettling, most mentioned housing or other economic issues. In 1999, many talked about the drought that had made agriculture virtually impossible. Security concerns were hardly mentioned at all. Most people did not seem to fear renewed warfare.

152 Social Networks and Migration in Wartime Afghanistan

Also, very few were concerned by the disruption of ties with their closest family, despite the fact that most households were split between Afghanistan and Iran. Given the dominance of economic concerns in the resettlement phase, it is particularly interesting to ask what role networks play in this regard. Rebuilding an economic base Most returning families do not own a house and have to find out how to sustain a family income while at the same time building a house. The normal pattern in my Enjil villages was largely the same as that described by Peter Marsden in a 1994 brief on ‘absorptive capacity’ in Herat: As all available housing is in use returning families first seek shelter with relatives; they then begin work on their land, if they have some, before starting the reconstruction of their own homes. (Marsden 1994: 1) The challenge for returnees, therefore, is to get the sequence right. Because many need to secure an income before building a house, they may live with relatives for a long time. Those who possess economic resources (mostly in the form of land) tend to be the ones best equipped in terms of networks, while those with no economic resources are the ones who need network assistance most in order to make the transition. If such assistance is not forthcoming, they are likely to remigrate, often using a split-household strategy. Housing is the main problem for most returnees. Among my respondents, some three out of four mentioned housing as a major challenge, while one out of three referred to other economic problems. Housing was less critical for returnees to Izhaq Suleman, where the houses remained largely intact; in Sara-e Nau and other similar villages in Enjil, however, virtually everything had been destroyed, and there were no relatives to move in with. Most people had to make do with makeshift solutions while starting construction, while some lived with relatives in neighboring villages or in Herat city. Housing was also the top challenge for returnees to Qala-e Muhajerin, an entirely new settlement whose residents got little support from nearby Izhaq Suleman. Almost no one who has spent years in exile will have a house to return to. For a returnee family, it is therefore critical to have relatives who are willing to provide lodging. A man in his early 30s, head of a family from Sara-e Nau, had not maintained ties with his co-villagers during his 10 years in exile, so in early 1995:

Reintegration at ‘Home’ 153

When we returned, we did not have a house to live in. While we rebuilt the house, we had to spend four months with a relative in Shalya Khan village, who let us live in his house. As I said, we were five families who returned together. All of us settled in this one house and stayed there until the reconstruction of our houses. Those who lack network resources find the early phase of resettlement extremely difficult. Housing was a much larger problem in the first wave (1992–93), when many returned to destroyed villages, than in 2002, when there was less destruction. In the second wave, however, many no longer had any family or other close associates in their villages of origin, and therefore decided to resettle in other locations, mainly in the larger cities. To build a house is of course easier the more other economic resources a family has. Some return with sufficient money to pay for the construction of a house. They may set out building their new house before they have secured an income. The majority, however, have to secure an income first. The housing problem is one reason why people whose exile has been relatively close to the home location have fared better. Many refugees in Pakistan came from areas close to the border and could retain close contact with their home village. Some members of the family could return temporarily to build a rudimentary house, while maintaining an economic income in exile, using a temporary split-household strategy. Because of the destruction, one would expect housing to represent a much greater problem for those returning in 1992–93 than in 2002–03; however, in my two villages, there is only a slight tendency to see housing as having been more critical during the first wave than during the second. My impression during visits from 2002 and 2006 was that the pattern remained largely the same, with housing as the dominant problem. In December 2002, the arbab in one of the villages in Enjil said: Many come, but they have problems with housing here, the houses are destroyed, and it is difficult to rebuild. Those who return have to move in with other people, mostly their relatives, to live with them until they can rebuild a house. The people returning spend all the money they can bring on rebuilding their houses. An interesting finding is that returnees who were affiliated with the mujahedin were even more likely than others to see housing as the top challenge. Among those of my informants who had been affiliated with

154 Social Networks and Migration in Wartime Afghanistan

the resistance, five out of six saw housing as a major problem, whereas this was the case for only two out of six among those who had not been affiliated with the resistance. Among returnees with a mujahedin affiliation, only one out of six saw economic issues as a major challenge, while four out of six among those with no such affiliation did so. Those who were not affiliated with the resistance were either neutral or from families affiliated with the militia; apart from those from Izhaq Suleman, most of them lived as IDPs in the vicinity of their places of origin. Hence, in most cases, members of the family had either maintained a house or were capable of starting reconstruction before moving back for real. The resistance affiliates enjoyed neither of these advantages and may also, at least in the earliest wave of repatriation, have had more difficulties in finding relatives willing to house them. Not only did the vast majority of returnees emphasize the economy in general, there was also a clear focus on the importance of site-specific resources at the place of origin, particularly on the ownership of land. Agricultural land secures an income in the short term, in the first critical years upon resettlement, whether it is cultivated by the family itself or leased out to others. Interestingly, however, even when land could not contribute economically in the short term, as was the case in the difficult years of drought, the long-term economic advantage of land ownership was considered important. Among those who owned land, fewer people saw the economy as a main challenge upon return (whereas more people saw housing as the main challenge). At the time of my main fieldwork in 1999, this seemed somewhat counterintuitive, with the dramatic drought undermining agriculture. Most people, however, had returned in 1992–93, when agriculture could provide some income, despite irrigation problems. Furthermore, the families who owned land were, even after years in exile, economically more robust. When, in a different context, our respondents were asked what they considered to be the main needs of their village, four out of five referred to irrigation and flood control, in most cases listing this as the most important issue among several. The emphasis on irrigation is not surprising, since the fieldwork was conducted at a time when the area was in its second year of severe drought. Moreover, the respondents were from villages located at the end of irrigation channels, where water rarely reached. These villages had been severely affected by the breakdown in water management during the war. There is no detectable difference between landowners and others when it comes to prioritizing irrigation issues, which indicates that considerable importance is given to the local agricultural economy also among those without property. The landless emphasize local economic

Reintegration at ‘Home’ 155

interdependence and assume that village solidarity is still at work, so that reinvigorating village agriculture will be to the benefit of all. Several respondents mention the major difference between those with and those without land. In response to the open-ended question ‘How would you describe the economic situation of your village?’, several respondents emphasized the difference between landowners and the rest: ‘Here are two groups, those with land are doing well; those without land are not.’ The area was in its second year of severe drought, and agricultural production was in crisis, hence the economic robustness of landowners as compared to others becomes clearly expressed. Most of the landless returnees have little chance of acquiring land, both because of a lack of money and – at least prior to 2001, when prices were ridiculously low – because there was little land for sale. If there is any land for sale, it is often the least attractive soil. Returnees from Thailand to Cambodia’s Battambang province in 1992 found that the land allocated to them was in areas with continued armed conflict and abundant land mines (Eastmond and Öjendal 1999: 42–43). The pioneers who established Qala-e Muhajerin in 1992 used funds accumulated in Iran to purchase land from two local landowners. This land was far from attractive, situated in the midst of a minefield that stretched along the highway from Herat city. The price, on the other hand, was low. Within 5 years of the purchase, demining agencies had cleared the whole area and its value had multiplied, a source of great amusement for the residents – and irritation for the original owners. It is nonetheless interesting to note that the returnees virtually had to purchase a minefield. As with returnees to Battambang in Cambodia, unattractive land is what befalls those who come back to establish new settlements. Regardless of whether or not a family has land, paid jobs are an important source of income. Here, too, the first-wave returnees were most fortunate, since many had served the resistance and were rewarded with positions in the administration and army of Ismael Khan. Those who had worked for the outgoing communist regime – with the exception of the most prominent political and military leaders – mostly kept their jobs. Returnees who had not helped the resistance, however, did not get any public jobs. Wartime political loyalties had clear economic effects. The pattern was the same in 2001–03. Rahmatullah from Izhaq Suleman, a former mujahedin commander who had returned in 1992, confirms that the job opportunities had multiplied: My brother is again working at the tank base – ten or fifteen other people from this village are living there. In the Taliban time, there was only one person from here, he is still working there, he is a nice person.

156 Social Networks and Migration in Wartime Afghanistan

In Sara-e Nau, however, the payback for serving the mujahedin meant even more. One respondent had been loosely associated with the mujahedin during the war; he was among the last to escape and the first to return: Ismael Khan did a lot for us. From here, there are forty people now working with the government. Some are in the military; others are with the administration. Also in Qala-e Muhajerin, political networks were used to create economic opportunities. Two men had got army jobs. Several went on a daily basis to the customs house, where they were licensed to shift goods with wheelbarrows within the compound, a safe income in the hectic import climate of Herat by late 2002. Most of the Qala-e Muhajerin residents come from remote parts of Herat province. To live in the vicinity of a major city, they realized, means having access both to jobs and to markets where they can sell the bush they collect for cooking and heating. Below is an excerpt of a 2002 interview with one of the first settlers in Qala-e Muhajerin: Interviewer: Was it a good decision to settle here? Respondent: Yes, it is near to the city, if we were in the mountains, what would we have to eat? How could we go to the city for work? Interviewer: How is security here? Respondent: It is good, now it is very good. We would make the same decision today. Our sons go to the city, they work there. The satisfaction of living near Herat may also reflect changed expectations after experience of a more urban life in Iran.5 It was equally important, however, that living close to Herat made it easier to capitalize on connections with the victor, Ismael Khan. It was difficult to borrow money for resettlement, just as it had been in the flight phase (Chapter 3). Some still informed me that they had borrowed money for rebuilding their house; but, in every case, such money had been borrowed from relatives. No public credit institutions were functioning, except the drug networks that offered credit and other inputs against a promised part of the harvest (Pain 2008). However, drug production was always more restricted in Herat than in other regions of the country.6 It was also virtually impossible to borrow money from wealthy members of the local community, although this had been fairly common before the war. The one remaining option was to borrow from one’s own relatives, often from people whose economic conditions were

Reintegration at ‘Home’ 157

not much better than one’s own (see also Klijn and Pain 2007; Stigter 2005b: 6–7). Again, we may talk of a ‘poverty trap’, similar to that which made those with no resources opt for internal displacement (Chapter 3). Poor returnees, who lack a cohesive network at the place of resettlement, are finding it extremely difficult to reintegrate. A related concern, often addressed under the vignette ‘absorptive capacity’, is that even when returnees do have social network resources they can activate in the reintegration phase, there is a risk of overextending them (see, for example, Marsden 1999: 65). The coping ability gained from social networks may be overestimated. If returnees with few resources can get help only from their closest relatives, and the latter feel under an obligation to help, then there is a risk that loans will be obtained from people who need the money themselves. Either people remain committed to help their relatives, with a risk of being dragged into a web of unpaid debts, or they feel forced to violate the kin-based solidarity principles and cater for themselves. The impact of returns from exile will at any rate have a negative effect on the coping ability of the local population (see also Marsden 1999: 43). Among the returnees to Izhaq Suleman, Sara-e Nau, and Qala-e Muhajerin, only a small group were better off economically than those who had stayed behind (Faubert, Mojadeddi, and Sofizada 2005: 11; Turton and Marsden 2002: 43). The returnees that seemed to succeed best upon return were those who had maintained their property during the exile (often with the help of local relatives), who had maintained a strong network in their home area, and who had accumulated capital while in exile. People who had only accumulated capital and not cultivated land or local networks found it difficult to re-establish. This confirms that the effects of social and economic capital are mutually reinforcing. One effect of the war was that people’s ability to play on large, vertical networks had declined dramatically, leaving most people to depend on cohesive ties. This holds true for all types of economic support: help with temporary housing, access to jobs, obtaining credit. The analysis of land ownership, however, reveals that most people, including the landless, expected that an agricultural boom would reinvigorate traditional forms of local solidarity, with vertical ties between landowners and others regaining their importance as channels for the transfer of economic resources. Those who had been closely associated with the power-holders – particularly the mujahedin during Ismael Khan’s reign in Herat – were an exception, since many of them were offered jobs and other economic opportunities that helped them to reintegrate.

158 Social Networks and Migration in Wartime Afghanistan

Transnational livelihoods The survival of returnees does not depend entirely, or even mainly, on the local economy (Black and King 2004; Kivisto 2001). While in exile, people have developed new means of survival, albeit often based on opportunities that do not exist at home. This tempts many families to reintegrate only gradually, with household members who have good income opportunities remaining in – or remigrating to – exile and sending remittances to those who have returned. Paradoxically, this implies that for some to be able to return and reintegrate successfully, it may be necessary that others stay abroad. This is one of the reasons why a massive repatriation driven by a host country can have such negative consequences for people’s ability to rebuild their economies at home. The combined effect of a massive repatriation and a ban on immigration may be to discourage return among those exiles who realize that successful resettlement at home is contingent on access to the foreign labor market. While some families decide to leave an income earner behind before they return, there are others who discover only after their return that economic reintegration is impossible without access to Iranian incomes, and who then send one or two family members back. In the aftermath of the first wave of repatriation, remigration to Iran was quite common from the villages of western Enjil. Rohullah from Izhaq Suleman was the father of six children aged between 2 and 15, and had gone to Iran several times in the years between his return in 1993 and my interview with him in 2002: I am going to Iran now, I cannot survive here. My elder son is in Waramin, in Teheran, I will go to be with him. I am compelled to leave the children here, although it is not good. I have worked in the city [Herat], in construction here in the mahalla, and in collecting firewood from the hillsides. In Iran, I may earn 3000–4000 toman per day, if I am not repatriated by the government. To find work is not difficult. Like many others, Rohullah had returned to Afghanistan with the intention of staying, and engaged in whatever income-earning activities he could find. However, when the family economy could not be sustained on what he was able to make locally, he went back to Iran for periods ranging from 6 months to a year, leaving his wife and kids with his father, who lived in the same compound. Many families in western Enjil continue to use the Iranian labor market. While individual remigration to Iran has been driven mainly by economic needs, the fear of conscription has also continued to play a role, particularly

Reintegration at ‘Home’ 159

during the first reign of Ismael Khan. In this sense, those who remigrated were not very different from those who fled for the first time during the 1980s. Those who remigrated, however, had the advantage of former experience with Iran, and they had networks and contacts to draw upon. If a family still has one or several able-bodied men working in Iran, this does not mean that they think their repatriation has been unsuccessful. Indeed, the opposite may be the case. While people tend to see a split household as a temporary solution, the value added is deeply appreciated, and it may easily become a lasting solution, particularly if the family has few resources to draw on at home. Income from abroad may also help families to meet extraordinary costs, such as weddings. An elderly member of the shura of Izhaq Suleman laughed when he said: ‘Those married stay here to work in Herat; those who are not married go to Iran. This is common.’ Better-off landowning families also continue to tap into the Iranian labor market, as a means to generate cash to invest in housing or farming implements. Some of those families who grew rich in the 1990s based their fortune on a high number of young able-bodied sons, a whole army of little workers sent to Iran for cash generation. On the other hand, there were also families returning from Iran who found reintegration so difficult that they chose to remigrate collectively, despite the costs involved. For those who returned as families and then chose to go back to Iran also as families, the stakes were substantially higher than for those who returned as individuals. Whereas a man can pay the costs of travel (also when smuggling is required) with a few months’ salary, the cost of moving a whole family is much higher, hence difficult to afford. To move a family is thus a far riskier decision than to move an individual. Yet, the remigration of whole families was common in the mid-1990s, when the economic prospects in Afghanistan were grim and the Iranians were still not outright hostile to the immigrants. In the post-Taliban era, family migration to Iran is no longer an option, given the difficulties of entry (particularly for families), the end to welfare services (health, education, food subsidies), the hostile attitude of Iranians, and the increased risk of deportation. Mohammad Daud from Izhaq Suleman was in his mid-40s when I interviewed him in 1999. He had lived in Mashad and Torbat-e Jam with his family from 1985, and returned soon after the mujahedin victory: All came back together, the Afghans that I knew there, most of them came back. At least fifty families who lived there that I knew, most of them came back. Ten out of these families were from this village, but all except me returned to Iran. Those families did not have property or land here, and they could make more money in Iran. Still I have

160 Social Networks and Migration in Wartime Afghanistan

family in Iran, one brother, one nephew, six cousins, altogether ten to fifteen relatives in Iran, all in Mashad and Teheran. As this illustrates, the opportunity for family return has served as an important safety valve for those who found life in Afghanistan hard to sustain. In later years, it has been increasingly rare for whole families to remigrate to Iran. This is partly because of an expectation that the Afghan economy is improving (at least in Herat and the surrounding areas), and partly because people are now better informed prior to return – hence there is less risk that they find reintegration unexpectedly difficult. In the post-Taliban climate, it is the opportunity for single men to work in Iran that is key. As we saw in the case of short-distance returnees to Pakistan, there is a conscious application of split-household strategies to ease reintegration. In the Pakistani case, however, it was common to maintain the bulk of the household in exile while starting up agriculture and house construction. In the case of Enjil, the pattern is the opposite: here, the main household base is established at origin, and any surplus able-bodied men are sent back to Iran to generate income. These differences relate to several issues. Firstly, the Iranian border, while at a close distance, is much more difficult, hence also expensive, to cross than the Pakistani one. Secondly, families remaining in exile in Pakistan, at least in the first half of the 1990s, had access to refugee relief. Thirdly, Afghans in Pakistan did not feel alienated by the public and the administration in the same manner as do those in Iran. Fourthly, life in Iran is considerably more expensive than life in Afghanistan, whereas with Pakistan differences are small. In both cases, people from the border zones, where the migration frequency has been extremely high over the past two and a half decades, have the contacts required; what differs is the way they use them to ease reintegration. In this manner, the continued access to labor markets in neighboring countries has become an integral part of people’s reintegration strategies, and may remain so for the foreseeable future. If the borders were closed effectively, so that Afghans could not be active in the labor market of their former host states, this would seriously undermine the process of Afghanistan’s reconstruction. It would probably also hurt the economies of Iran and Pakistan (see, for example, UNHCR 2003a).

Information and reintegration I have assumed that the information role of networks is less critical in the integration phase than in the decision-making phases. It also seems likely that information is less critical in the reintegration phase

Reintegration at ‘Home’ 161

than during integration in exile, at least when people return to their community of origin and when ties have been maintained throughout the exile. If information networks are less critical in the reintegration phase than in the other three phases, it is not because information is less important, but rather because the returned migrant is likely to have multiple channels to play upon. The rebuilding of a community is also, however, about redistributing power, and it seems likely that contenders may seek to control information in order to advance their own cause. In parallel, the continued use of exile localities, as discussed above, also means that information about opportunities abroad remains important. Information in resettlement Few of my informants would emphasize access to information as a reintegration challenge, although most would indirectly touch upon it in the context of discussing other issues related to security or the economy. When asked from whom they would seek security information, almost everybody answered the village arbab, but some also mentioned ‘the village elders’ or the mahalla representative in the shura (often the same individuals). Many said that they would choose the arbab because he had direct access to the district administrator. Only one person – a former mujahedin commander – said he would go directly to the district administrator. The pattern is the same for other types of information, such as shelter support or other forms of assistance, although here the argument would be that the arbab had direct access to various aid agencies. Mahmoud returned from Iran with his family in 1992. His viewpoints are rather common: From the moment that I came back from Iran till now, the aid agencies have not assisted me or any of the neediest people of the area. Because the organizations have direct contact with the local arbab and he is trying to give all the assistance to his relatives and some of it is used by the arbab himself. . . . Some of the assistance is given to people who are rich. The arbab figure is a classical broker. The community, the district administration, and the aid agencies all see him as controlling the flow of information both ways. He may modify requirements reported by villagers so his own associates are favored. He may reserve information about aid opportunities for just a few, or may twist the content. A functioning shura, where members are more than the loyal supporters of the local arbab, can do a lot to undermine the arbab’s opportunity to manipulate

162 Social Networks and Migration in Wartime Afghanistan

information. The establishment of shuras has been encouraged by most aid agencies and by the administration in Herat (Harpviken, Suleman, and Taksdal 2001: 11–17; Marsden 1999: 66–67). In the post-Taliban era, a new program has been started to elect shuras by secret ballot and put them in charge of managing a village development fund (Boesen 2004; Leader and Atmar 2004; for a critical view, see Johnson and Leslie 2004: 189–191). A key ambition is to ensure transparency and local representativeness, as a counterweight to the arbab system. The extent to which access to information is experienced as a major constraint varies widely between the returnees. Most of my respondents had been able to maintain contact with their community of origin while in exile: some had family members who were staying behind (either in the village or in its vicinity); some came back regularly as members of the mujahedin. A minority had little or only minimal exchange with their community of origin during the exile, and stayed away for a long period of time. Whereas few in the latter category returned in the 1990s, many more have done so lately. The information difficulties they face upon return are not only due to poor networks, but also due to a lack of knowledge about how things work in present-day Afghanistan. A third category is the ‘returnees’ who settle at a location different from where they lived prior to flight. Most of the people in Qala-e Muhajerin are in the latter category; they depend on the ‘bridgehead’ returnees within their own community, who were the only ones to have a network at the location. In the Qala-e Muhajerin case, however, the lack of information has over time been partly compensated by good contacts with the province administration. Information networks are only critical to reintegration for those who are either new to a location or have not successfully maintained their networks during exile. It is then most critical in the first period after the return. The example of Qala-e Muhajerin illustrates that it may be difficult for newcomers to build robust information networks, even over the course of several years. In such situations, there is also considerable room for information brokers – and particularly for semi-institutional ones, such as the arbab – to use information control to their own benefit. Information from exile The other aspect of post-return information concerns opportunities at the former exile location (Massey et al. 1993: 448). The value of such information is likely to diminish as more and more people gain the same sort of experience, which is exactly what is likely to happen when many people return simultaneously from the same exile locations. In other

Reintegration at ‘Home’ 163

words, as the migration-relevant information held by the larger group increases, the opportunities for individual brokers to gain advantages from possessing unique information are diminishing. Everybody in Enjil has affiliates in Iran, and almost everybody has relatives there. Out of a total of 65 respondents who were asked whether they currently had relatives in Iran, only five answered no. Three of the five had never been to Iran; the other two had returned as mujahedin just after the fall of Najibullah. In communities where virtually everybody had contacts in Iran, and where a majority had themselves been to Iran for extended periods of time, it seems unlikely that anybody would have problems in finding information, or even in mobilizing contacts, in the event that they should consider migrating. Furthermore, most people had access to information about labor opportunities in Iran from multiple contacts and were not dependent on any single source. Information about migration and job opportunities in the neighboring countries is thus not a scarce good. After two and a half decades of extensive migration, such information is a major resource that is held by virtually everybody. In 1999, the Enjil communities were saturated with ties to people residing in Iran. Although many more returned after the fall of the Taliban, this does not seem to have significantly affected access to information, as most people already had multiple channels. Hence, the opportunities for brokers to derive a benefit from channeling information to prospective migrants are close to none. People’s variegated channels to information on the opportunities in Iran are one of the most significant outcomes of 25 years of migration.

Conclusion The impact upon receiving communities is massive when many migrants return within a short period of time, as was the case in western Enjil in 1992–93 and 2002–03. This is particularly so when state institutions are weak or absent, and when political-cum-military tensions remain unresolved, all of which will make ties between repatriates and hosts subject to considerable pressure, even if they remain strong. The situation in Herat, however, was exceptional in that Ismael Khan, during both of his reigns, did have a working administration, was largely able to control local tensions, and for long periods effectively shielded the region from the conflicts dominating the rest of the country. Nonetheless, in western Enjil, the war had inflicted severe damage, and the local economy – agriculture in particular – remained fragile. It is therefore not surprising that, for the returnees, network resources were critical – not only

164 Social Networks and Migration in Wartime Afghanistan

for returning to and being welcomed among one’s own, but also for obtaining security, surviving economically, and being informed. Militarized groups and networks, built or transformed in exile, played an important role in the security domain. Those associated with the winning side in the war were not only the first to return, but were often also called upon to continue their military engagement to secure the transition and offered basic protection. In a situation that was more conflictual than Herat in 1992 or 2001, the endurance of military structures built in exile fed protracted insecurity for the society at large. For those who are a part of these military networks, of course, the more insecure the general situation is, the more valuable is the protection they derive from being a member. For those who are not associated with any armed group, the main protection comes from belonging to a big family or clan. This may help to ward off criminals and perhaps small-scale incursions by armed bands, but not to counter threats from armed units in the service of the government or an organized rebel movement. In Afghanistan, the refugees who return last are those who have not been part of the mujahedin or other political networks. Although they are not perceived as threatening by others, they are far from safe because they are so vulnerable. There is security in numbers, but returnees who have not engaged with any armed groups find it very difficult to establish the sort of networks that can offer effective protection upon return. For the returnee, economic survival hinges on ties that have either maintained resources at home or are helping out during the transition. In the communities discussed here, security has not broken down totally at any point since 1992; and, when people are asked, they see the economy as their overwhelming concern. The most critical issue is housing. Returnees first have to either live with relatives or find other makeshift solutions. Land ownership remains the most important resource despite drought and agricultural recession. Those who own land have a stronger standing in their community, both in a material sense (maintaining the property) and in a social sense (maintaining networks, having a place where family members can stay while they go abroad to work). Opportunities in the labor market – particularly in the administration and the military – are closely linked to earlier military and political engagement. Those who have proved loyal are rewarded upon victory. At the geographical locality in Afghanistan, however, the economic reach of networks has protracted as a result of the war: people have little or no access to credit or economic assistance from people with a higher economic status, and this forces them to depend on support

Reintegration at ‘Home’ 165

from relatives. There is a mutually reinforcing relationship between economic resources and social networks, which means that a lack of networks constrains economic opportunities, and poverty constrains the building of networks. The split-household strategy remains important in the reintegration phase. The primary motivation for keeping some family members in exile is economic. Security considerations may also play a role, as keeping a foothold in exile makes it easier to escape should the security situation at home again deteriorate. While split households are seen generally as a temporary coping strategy, they often become quasipermanent, as reintegration proves more challenging than expected. Split-household strategies also vary. Some maintain the bulk of the family in exile while the able-bodied men carefully lay the foundations of reintegration; other families resettle while one or two incomeearners continue to tap into exile labor markets. The former is most common for returnees from Pakistan; the latter for those returning from Iran. The split households illustrate how large-scale refugee situations trigger social change, which also renders any ‘return to normal’ strategy unrealistic. Access to information through networks remains important during reintegration, particularly immediately after the return. For information about government policies or opportunities offered by aid agencies, most people depend on the village authority – the arbab – who derives considerable influence from functioning as a broker. Very few have alternative channels for accessing this kind of information. Access to information is even more critical for those who return to another place than their place of origin, such as most of the residents in Qala-e Muhajerin. For them, even accessing the arbab directly is difficult; they would rather go via one of their own villagers who have a longer history in the area. Transnational links of information, however, seem to be in relatively rich supply for almost everybody. People have many sources of information, both about opportunities in Iran and about the situation of their relatives, and there is little room for influential brokerage. For most people, the massive expansion of transnational ties, as well as knowledge of exile opportunities, is one of the most significant changes that have occurred during the war. If we assume the perspective of returnees who are solidly resettled at their place of origin, their network resources are very different from what they were before the flight. Their immediate social networks have often contracted; their relationships with local power-holders are less useful than they were in the past. The weakening of these local networks is to be

166 Social Networks and Migration in Wartime Afghanistan

compensated by new ties derived through the exile experience. Not only is there an overall transformation in most people’s network resources, their perceptions have also changed from a high degree of overlap between networks and geography to more widely dispersed networks. The length of exile is not without importance for the reintegration: time means change both at home and among the people in exile. It is commonly assumed that the longer the period of absence, the looser the networks between exile and home become. This is not necessarily the case. We often see exile communities maintaining relations with their home community that are so strong and interdependent that it is appropriate to speak of a single shared community, geographically separated but socially, politically, and economically united. Such dense interaction is greatly facilitated if travel and communications are easy. Short distance and porous borders help. Interaction may sometimes remain strong even when contextual factors work in the opposite direction, as with the western Enjil refugees who fought with the mujahedin. The more intensely home and exile communities have interacted during the period of exile – for instance through splithousehold strategies – the more likely they are to escape the general tendency for reintegration to become more and more difficult the longer the period of exile lasts.

7 Conclusions

Social networks are critically important to people who contemplate, plan, or carry out flight in war situations. In war, the environment with which people are familiar is turned upside down and institutions that have provided a degree of predictability collapse, while the amount of time for making important decisions is limited and the likely outcomes of alternative courses of action extremely difficult to foresee. Hence, social ties become the dominant means of mobilizing resources, tackling uncertainty, and carrying out an escape. The larger the obstacles to migration, the more people depend on their network resources in order to be able to carry it out. The essential role of networks comes to the foreground when we combine analysis of the causes behind forced migration with an examination of the mediating factors – those that are decisive for whether and how a threat leads to escape. The essential role of networks, however, also has dark sides – both in that people who are poorly equipped in terms of networks have few options and in that networks that are transformed or break down in the face of war may prove unreliable or may be used for repressive or bellicose purposes. The transformation of networks is difficult to pin down, but essential to understand, if we want a full picture of the social changes brought about by war. With reference to the empirical study of two communities of Enjil – and Afghan migration patterns more generally – discussed throughout the book, I shall now return to the overarching questions that I posed in the introductory chapter. I will first reexamine and refine the findings on the role of networks in migration decisions, looking at differences and similarities between the escape and the return decisions. In a subsequent section, I will compare the role of networks in integration in exile with the role upon return in a similar manner. Pulling in the larger debate about the transformations of social networks in the face 167

168 Social Networks and Migration in Wartime Afghanistan

of war and disaster, a third section will examine the impact of largescale migration on network change, drawing on a typology of patterns of network change introduced in Chapter 1. Finally, I will conclude with some general implications of this study for research on forced migration, framed in a broader context of migration and responses to war.

Escape and return decisions The actual departure of ‘forced migrants’ is rarely acute. Even in the dramatic circumstances surrounding flight in Afghanistan, such as the escape from Enjil district during the peak of war in the 1980s, most people had time to gather information, assess the situation, and plan for flight or alternative courses of action. People did not escape ‘at gunpoint’, there were no soldiers arriving in the villages with a mission to force the villagers out, and people did not escape with ‘their own troops’ in the face of an advancing enemy. Despite dramatic circumstances, people’s decisionmaking most often had a proactive character, and they had the opportunity to take certain preparatory measures. Return decisions, on the other hand, were more commonly reactive in character, taken under pressure from both the government and the population of the host country. Some returns were entirely compelled, as in the case of those expelled from Iran. To compare the decision to escape with the decision to return makes sense. In both cases, people made a difficult choice in a situation of strain and under conditions of considerable uncertainty. Political motivations are often a central element in proactive decisions – as when people escape in order to take part in resistance activities, or when they return in order to be present when the spoils of peace are to be divided under a new regime. Proactive decisions may also be motivated by other reasons – as when people considering escape see economic opportunities at the destination (such as humanitarian assistance or gainful employment) and perceive a chance to send money home, or when return is timed to benefit from new opportunities on the job market. In both the departure and the return phases, it is common for collectives of people to work together in collecting and processing information, in undertaking various decisions, and eventually in traveling together. Members of flight collectives that are formed under dramatic circumstances are likely to stick together, particularly if they end up in a place where they have little attachment. This is less likely when the displaced come from different places of origin and have not taken flight decisions together. If they later decide to return, they may travel back as a collective but dissolve once nearing home. Notwithstanding, it does

Conclusions

169

happen that displaced people with no previous association develop close relations while living in exile and that they subsequently decide to stick together when resettling in their home country. This was the case in Qala-e Muhajerin, the ‘refugee village’ of Enjil. When people return, we often see that, in order to reduce risk, they reach a collective decision that not all will return simultaneously and the process of repatriation should be carried out in steps. The resulting split often takes on a semipermanent character, since reintegration proves more difficult than expected. The original intention, however, may have been to bring about the repatriation of the whole group within a year or two. Groups of people traveling together in the return phase also tend to be smaller than groups of people fleeing together – not because the collective dimension is less important, but because the situation permits a more gradual approach. At the outset, I assumed that people have little information about what they can expect in terms of economic opportunities or security problems at the place where they live when contemplating flight, and hence that the amount of such information would be a critical factor in decisions. For the villages of Enjil, this assumption did not apply. Most people who chose to leave their place of origin in the last two and a half decades, even when war was at its most intense, had multiple independent sources of information and did not perceive access to information as critical. This is important, since it makes it much more difficult for brokers to use control of information strategically. In most cases, information sources were fairly homogenous, and the lack of variegation did limit the ability to make independent decisions. Furthermore, the assessment of the information, an integral part of the decisionmaking, took place within dense networks of family and friends. The assumption on information scarcity did hold, however, for those individuals and families who were targeted for political reasons. For them, access to relevant information was critical and in short supply. At the same time, these are also cases where information is comparatively reliable, unlike forecasts regarding the broader political and military situation in an area, where information will always be rather imprecise. I also assumed that, when at the exile location, people would have far more knowledge about what they could expect when deciding to return. Conditions in exile would be more predictable than conditions in wartorn Afghanistan. This assumption was also not fully confirmed. In the decision-to-return phase, control of information about what to expect if deciding to stay on at the exile location varied greatly. Control of information could be a key issue in Pakistan, where many refugees lived in camps, with semi-official leaders from within their own population. In such a situation, information brokers would not always find it in their

170 Social Networks and Migration in Wartime Afghanistan

best interests to encourage a quick return. They could therefore hold out the prospects of sustained or even improved conditions for exiles. In Iran, where refugees were self-settled and mainly related to Iranians for jobs or administrative issues, it was much more difficult for brokers within the refugee population to gain control of information. Signals of changes in the refugee policy would come mainly from official Iranian channels, whose uncoordinated or contradictory messages contributed to the creation of considerable uncertainty among the refugees. When people decide to flee or return, they also seek information about conditions at the ‘other’ location: the potential destination for potential emigrants, the place of origin for would-be returnees. In the first phase, people do not have many alternative sources of information. People need rather specific information; hence, a few networks and brokers play an essential role. For departure decisions, brokers have the largest opportunities in the early phase, before emigration becomes so common that everybody have multiple sources of information. The reduced influence of brokers is attributed to the fact that the first emigrants are often people who have earlier migration experience, just as Massey and his associates found in Mexican migration communities (Massey et al. 1987). When deciding to return, most people have access to multiple information sources, which is not surprising in the case under scrutiny here, given that most Afghans spent years in exile and thus have had time to develop many channels of communication (see also Koser 1997). Despite this, the need for precision on opportunities at the return location encourages many to conduct personal visits and to ‘find out for themselves’. Refugees who come from areas far away from the exile location – such as refugees from northern Afghanistan who had gone all the way to Pakistan – have larger problems with information. For such long-haulers, personal visits may be too costly to be realistic, and access to information will be even more difficult if they come from a community where only a few had fled. Networks are essential for access to material resources, but in less direct ways than I initially assumed. A majority of the respondents in Enjil raised the means to finance the travel into exile from within the household itself, often by selling tangible assets such as household items or animals. Traditionally, it had been possible to obtain credit from local associates who were better-off; this opportunity evaporated with the onset of war (and the resulting mobility), however, and many of the poorer people ended up staying or only relocating within the local neighborhood. The only exception to the credit drain was that represented by smugglers, who over time increasingly offered services against being paid back in exile while the migrant worked for an

Conclusions

171

associate of the smuggler. The migrant was rendered highly dependent in this arrangement. Site-specific assets, such as land and houses, were not sold to finance departure. Exit could easily have been discouraged by the need to hold on to one’s land, but this was countered by the fact that big landowners were specifically targeted by the PDPA regime in the early 1980s. More commonly, one – or several – family members, who were perceived not to be directly at risk, stayed behind to look after site-specific assets. Economic transactions, which otherwise often flow in relationships between unequals, were almost entirely restrained to dense relationships, with the family as the focus. Economic resources to travel home were mobilized from within the household, both by a majority of those who returned in the 1990s and by those returning in the first two post-Taliban years. Those with fixed property in Afghanistan seem to have been among the first to return. Raising money for the return is likely to be more of a problem for those who have been hesitant to return and still remain in exile, many of whom are also less well-connected via networks to their places of origin, in addition to having no property to return to. These are also likely to find conditions difficult when they return, since jobs and opportunities under new regimes are normally distributed to those perceived to have been loyal supporters and who responded quickly to the political change. Latecomers are unlikely to benefit. Networks that span boundaries play a somewhat more prominent economic role than I had expected, in contrast to the findings for the role of networks in the same locality. The Enjil data clearly indicate that remittances were significant, particularly from Iran, in spite of considerable skepticism in the existing literature about the role of remittances in Afghanistan (Marsden 1996a; Monsutti 2004b) and in spite of considerable unwillingness among informants to provide information about this subject. Families who have returned from Iran, as well as families who never left Enjil, continue to tap into the Iranian labor market, and money being made in Iran is often a prerequisite for basic survival. In many cases, people working in Iran save in order to meet extraordinary expenses in relation to marriage or investments in land. Economic considerations are a primary reason why people pursue split-household strategies in the return phase. Able-bodied men remain in Iran to contribute to the economy at home, while the bulk of the household lives in Afghanistan, in a strategy of gradual return. For refugees in Pakistan, the situation is the opposite. Here, it has been relatively common for most of the household to stay in exile, while the able-bodied men return to Afghanistan for extended periods to arrange housing and commence cultivation of the

172 Social Networks and Migration in Wartime Afghanistan

land. In both the Iranian and Pakistani cases, however, the purpose of a temporary split in the household has been to minimize risk – primarily in the economic domain, but partly also in terms of security through the maintenance of an exit option should conditions at home deteriorate. Such gradual returns, perceived only as short-term measures, often take on a semi-permanent character. Security concerns constitute the major reason for departure. It is also interesting to note the way in which social networks contribute to exacerbating or dampening threats. A case in point is Izhaq Suleiman, where the local arbab served as an interface between the village population and the PDPA regime. Being well-connected, the arbab was able to strike a deal in which the village was granted protection in exchange for contributing young men to the militia. The village’s location next door to a major army base represented an immediate risk, but also illustrated that the alternative to collaboration was likely to be extermination – a fate that eventually befell most of the neighboring villages from which a majority of people joined the resistance. Militia and mujahedin mobilization patterns were similar: both were based on preexisting loyalties rooted in extended family networks. In the militia village, conscription was an ever-present threat for young men, against which networks could not provide protection. Reprisals against Izhaq Suleman from the resistance remained limited, largely because the village maintained relations by offering it a number of favors, including the provision of indirect support to the mujahedin’s military operations. The family of active mujahedin, including some from neighboring Sara-e Nau, a classic resistance village, also had the village as a refuge. This delicate balance was maintained until the fall of Najibullah’s regime in 1992, so in that period the majority of the original village residents opted to stay rather than to depart. For Afghans in exile in Iran, the Iranian authorities represented the main security threat. These exercised strict control over the refugees and restricted their movements. Exiles could also occasionally be harassed by civilians, and there were few opportunities to stage protective arrangements. The key consideration behind the decision to return was that of balancing security problems in exile against the expected security condition at the place of origin. This appears to be the primary reason why people who had engaged actively in resistance work were among the first to return, whereas those who were politically inactive were more hesitant. Prospective returnees expected a mujahedin association to be directly convertible into security upon return. Importantly, however, it was widely believed that there is security in numbers, and this was a major reason why large return collectives prevailed in the earlier rounds of repatriation. In later rounds, groups of returnees traveling together

Conclusions

173

were smaller. The returnees saw themselves as part of a larger collective from which many had already returned and settled, subsequently conveying their assurances to the next cohort of prospective returnees. Contagion seems to have been at work in both the flight and the return phases. In the mujahedin villages of Enjil, decisions to flee were often inspired by observing the departures of others. With maturing networks in exile and information increasingly being channeled back to the locations of origin, additional migration was both encouraged and facilitated. Thomas Faist (1997a) has referred to this as ‘cumulative causation’. Dense types of networks play a marginal role both in informational access specifically and in decisionmaking generally. The fact that the departure of members of different extended networks peaked within distinct periods of time indicates that it was direct transmission (‘cohesion’) rather than general exposure (‘structural equivalence’) that was the primary mechanism at work. In return decisions, contagion is less clear, but seems to have been at work in the two major repatriations in 1992–93 and 2002–04, when whole networks moved back within rather limited time periods. At the same time, with the mature information networks linking origin with exile, negative experiences by recent returnees were immediately conveyed, resulting in ruptures in repatriation. Escape decisions are sensitive to the expected duration of stay at the target location. There is a tendency for early leavers to expect duration to be relatively short, whereas those who are concerned that the exile will last for a long time are more hesitant to depart. If instability proves to be protracted, those who stay behind have their expectations confirmed. If they should then decide to depart, they will make their decision on the basis of a more realistic assessment. When deciding to return from exile, most people expect to be going home for good. The expected time frame, however, is still an issue. As a basis for the decision, people will try to assess how long it will take to be functionally resettled at home. If this is expected to take considerable time, they may adopt a strategy of gradual resettlement, with households being split, often for several years. People who expect a particular arrangement to have a short duration are easily disappointed. In a general sense, however, the mediation of optimistic assessments of the future – through personal networks – has a strong motivating capability.

Integration and reintegration Conditions for establishing a life at a new destination after migration vary dramatically from location to location. It has been argued above that network support – from co-migrants, from people already settled at the destination, and also from those left behind – plays an essential role

174 Social Networks and Migration in Wartime Afghanistan

in the integration process, both in exile and upon return. People who are settling at a new and different location will often maintain ties with the place they left, and they may continue to play significant roles there despite their absence – as sources of information and economic support (investment, employment), or even as political-cum-military activists. There are fundamental similarities between getting integrated after flight and getting reintegrated upon return (particularly when exile has been enduring), and the two can be analyzed within a common framework. This study shows that the challenges of reintegration at home may be of a type and scope similar to – and sometimes larger than – those of exile integration. People who leave home to settle in an exile location where their network is already established may find integration easy, much easier than remaining in a place from which most of one’s social network resources have left. Many who return may do so after a long time in exile; some of those who ‘return’ have been born in exile. People may return to locations that have entirely changed, that they know only from what their parents have told them, and they may also have to establish themselves at a location that is not the one they originally came from. This was the case for many of those who settled in Afghanistan’s larger cities in the first couple of years following the fall of Najibullah in 1992 or the Taliban in 2001. Information is a critical asset for integration. In exile, people are particularly dependent on their information networks, especially if they live in an area where refugees have settled outside of any institutional framework, as they largely did in Iran. If they live in camps, the dominant pattern in Pakistan, network-mediated information is less critical, since much information is immediately available from camp officials. Generally, those who arrive first will have the greatest problems with access to information, while those arriving later in most cases will have several affiliates at the destination whom they can draw upon. In the case of Afghanistan, displacement often followed the paths of prewar migration, so that most of the refugees found that, in the places where they settled, information was in relatively rich supply. In Iran, however, the value of the information networks was constrained by the unpredictability of the policies of the host government, as well as the difficulty of building solid relations with influential Iranians. In the reintegration phase, those who return to their place of origin will normally have a variety of information sources. People opting to settle in a location that is different from their origin face bigger problems and may have to rely on local information brokers. This entails a risk of becoming overly dependent on one source for information, which then will be able

Conclusions

175

to exert considerable influence. In Afghanistan, where the state apparatus is weak, vital information – such as information on available resettlement assistance – will always depend on personal contacts. The task of building an economic foundation for survival if you enter into an assistance-based economy, such as in the camps in Pakistan during the early 1980s, is very different from integrating yourself in an economy based on self-sufficiency, as was largely the case for refugees in Iran – and also for today’s returnees to Afghanistan. Over time, the role of humanitarian assistance in Pakistan was balanced by the relative freedom of movement for refugees, including the option of setting up businesses. Afghan entrepreneurs and business networks could grow relatively strong. In Iran, by contrast, opportunities for Afghan entrepreneurship were extremely limited. Most refugees had to work for Iranian employers, either on short-term contracts or as day-laborers, all of which prevented the development of Afghan niche economies. For a majority of Afghans in Iran, access to jobs was facilitated by networks based on family or common origin. The Pakistani setting was one where both the assistance system and the general economy opened up opportunities, despite the fact that refugees initially depended less on providing for their own income than in Iran. Upon return, economic challenges constitute the major problem for most people. Obviously, those who have been able to accumulate capital in exile and bring it home fare better. Similarly, those who have been able to hold on to land or other assets at the home locality are better off. Also for them, however, the transition may be difficult, as their houses and much of the infrastructure may have been destroyed. Hence, the access to close associates who are willing to offer housing for an extended period proved essential for reactivating one’s economic base. Those who have been unable to bring back resources from exile and who have no property have faced dire problems in reintegration, but they are also far less likely to return in the first place. The convergence between strong social networks, on the one hand, and economic resources, on the other, comes out strongly in the reintegration phase. Without a capital base, you are unlikely to be able to build a diversified network; without the network, you are unable to accumulate capital. The economic importance of transnational ties for reintegration, mainly in the form of remittances, has been discussed at length. Whereas the income potential in neighboring states should not be exaggerated, particularly since the cost of traveling back and forth may be high, it is still the case that many households use labor markets outside Afghanistan as their main economic buffer. This raises a delicate policy dilemma for the

176 Social Networks and Migration in Wartime Afghanistan

neighboring states, as it suggests that successful reintegration – hence also motivation to return – depends on the ability to continue to operate in their labor markets (Harpviken 2003; Strand, Suhrke, and Harpviken 2004). Both Iran and Pakistan have responded ambiguously to this challenge: on the one hand, signaling that all refugees will eventually have to go home; on the other, recognizing that the ties that have been established represent a vital resource for economic interaction. Networks may also be important for security, particularly after a return to a state with little protective capacity. In exile, relationships with influential figures within the host population are likely to provide protection against crime or harassment, whereas strong networks within the refugee community alone give little protection. Iran permitted – even encouraged and supported – the establishment of Afghan resistance groups, but never gave them the same degree of free operation and visibility as similar groups enjoyed in Pakistan. Iranian-based groups were engaged militarily in Afghanistan, but their ability to offer protection for their members while in exile was limited. In the reintegration phase, however, where the state is likely to remain weak and where there is considerable potential for conflicts of interest to erupt, the most realistic safeguard is to enroll oneself in strong collectives, preferably with members who are well-connected politically. For such networks, in contrast to those effective for information purposes or in the economic domain, size is critical. Interestingly, as discussed above, this is a key concern to many potential repatriates already in the decisionmaking phase. Both in 1992 and in 2001, military entities based in Iran were returning more or less wholesale, filling up positions in the new administrative and political apparatus. The association with victorious military groups was particularly efficient in the Herat area, where there was always a comparatively well-functioning administration, despite shifts in power. The most systematic difference between exile integration and reintegration upon return is the expected duration of the stay. In exile, the stay is assumed to be temporary, even when in reality it becomes protracted. Upon return, the assumption is that the stay will be durable, even when survival necessitates continued reliance on resources from the outside, also by keeping some household members abroad for a long period of time. Moreover, as we know from other migration studies (such as Massey et al. 1993), once people have migrated once, they are significantly more likely to do so again; thus, even if return is expected to be permanent, unforeseen challenges in reintegration may easily lead to new plans for migration.

Conclusions

177

War, migration, and the transformation of networks In contexts of large-scale and protracted migration – Afghanistan included – a mapping of network transformation needs to take into account change in networks both at origin and in exile, as well as relations between origin and exile. This again implies a concern for changing structural conditions that affect network dynamics, both in exile and at origin, as well as factors that affect contact opportunities between the two locations. The empirical foundation of this study, with fieldwork conducted in a few localities only at certain stages of the period under scrutiny, supplemented with available literature, permits only tentative conclusions on network change. Given the scarcity of studies that empirically analyze network transformation during armed conflict, these tentative conclusions should ideally serve to inspire further research. The standard assumption – as discussed in the Introduction – is that war leads to a fragmentation of social networks (Duffield 2001; but see Wood 2008). It is suggested that wars in general – and civil wars in particular – are accompanied by a breakdown in trust and a collapse of common normative foundations that undermine existing networks, which either collapse entirely or split into subsets, each pursuing a path of its own. This is in line with disaster research, which sees the ‘corrosive community’ as the main consequence of man-made disasters (Rosa and Freudenburg 2001). A variety of the fragmentation thesis is that new networks emerge – or latent networks get activated – to replace the old ones, resulting in entirely new collectives. The new networks are likely to be an adaptation to the conditions of war, formed in an attempt to make the best of the war or possibly to seek seclusion from the dynamics of conflict. In contradiction to the fragmentation hypotheses, the main response to war is sometimes mobilization of existing ties. A vitalization of existing ties may follow vertical lines – mobilization of existing relations of domination or influence – as when traditional leaders gain significance in the face of a repressive or withering state. If we draw on disaster research, however, and its suggestion of a ‘therapeutic community’ (Erikson 1976), the expectation is mainly a vitalization of horizontal ties, where people of similar status act together on the basis of dense relationships. The cohesive ties underlying a horizontal mobilization are stable but tend to have a limited reach. All of these four avenues to network transformation – fragmentation; fragmentation and replacement; vitalization of vertical ties; vitalization of horizontal ties – represent realistic possibilities. In the following, I will seek to establish the conditions under which one or the other may enter into play, drawing on the Afghan case material.

178 Social Networks and Migration in Wartime Afghanistan

A dissolution of old ties, with no solid replacement, is difficult to find in the Afghan case. Most people have both old and new networks to draw upon. One category of particular concern, however, is the people who remained in exile after the great 2002–04 wave of repatriation. Regardless of whether they continue to live in exile or become late home-comers, they may be losers in the struggle to mobilize resourceful networks. Those who had maintained contact with home during exile tended to return quickly, both during the first major repatriation phase (1992–94) and during the second (2002–04). Many of those who remain in exile have spent 20–25 years away from home, and many have established strong social networks in their country of exile. Others may simply find themselves with little or no network resources. The remaining refugees were also poorer than those who returned early. Few owned any land and would therefore probably run into major resettlement problems if they returned. A fragmentation of ties with other Afghan exiles was more frequent among those who settled in Iran, where there was less room for Afghans to maintain large cohesive collectives, and where the room for Afghan middlemen to emerge as leaders among their own was limited. A fragmentation of ties with the home community is also likely to have been more frequent among exiles originating from the more remote areas of Afghanistan, with which it was difficult to keep in regular contact. Hence, the overrepresentation of people with poor social ties among today’s prospective returnees is a major concern in face of the intensifying repatriation drive from host states. The dissolution of old ties, and their replacement by latent or new ties, is something that should happen in situations where strong ideological movements aim at revolutionary change, as they did in Afghanistan in the 1970s and 1980s. Notably, however, the radical political movements, both communist and Islamist, had great problems in taking root throughout the country (Harpviken 1997a). In exile, the situation could be different – particularly in Pakistan, where radical Islamist parties were given considerable freedom of operation. There was an expansion in exile of the new networks that the Islamist parties drew upon, yet their ability to mobilize existing but latent ties was limited. The politicized networks of the Islamist parties did not take hold in the broader population, despite the fact that their influence – both in political terms as well as in relation to the distribution of military and humanitarian assistance – was significant. When successful, however, the new networks built in exile became a major threat to both return and reintegration, since these new ties seemed to demand a severing of traditional ties with the home population at the place of origin. This was the case for the Pakistan-based exiles from Kunar,

Conclusions

179

who ever since the province came under resistance control in 1989 had to commit to the Islamist side in the confrontation with traditionalist forces (Glatzer 1990). If we look at the Iranian side, conditions for building new networks were restrained by the fact that there was little or no role for Afghans to play in refugee management, and Afghan political parties were strictly controlled by the government. A partial exception, however, can be seen among the Afghan Shia, where new religious leaders managed to play a significant role because of their close ties with the regime in Iran. These competed with traditional Shi’ia leaders at home for political influence in Afghanistan (Harpviken 1998). Overall, the new religious– political groups were not well-positioned to gain support from a large part of Afghan society, but they retained a significant political influence at the national level by virtue of their strong international contacts. A revitalization of old vertical ties seems most likely where the response to war is not escape but active resistance. In the Afghan case, many traditional leaders, political and religious alike, became highly motivated to take the lead in resisting, since they were targeted for repression by the PDPA regime in the first years after it came to power. Many traditional leaders were killed, while others saw no other option but to escape, although many of the resources underpinning their leadership at home were not in place in exile. In Pakistan, the traditional leaders were challenged by new and younger ones, who were better at mediating with the refugee administration and the Afghan political parties. In Iran, they were hampered in playing such a leadership role since they had to fight for economic survival on equal terms with everyone else. Despite all the factors undermining the continued authority of prewar leaders during the period of exile, however, the situation in Afghanistan today is that many of the same people, or at least the same families, have returned to key positions at the local level. Those who survived were linked to the government or one of the resistance organizations, in many cases cultivating relationships with both sides. In exile, these people were partly marginalized. Upon return, their traditional resource base at home proved a great asset, and it has served to their advantage that both the Taliban and the post-Taliban regimes have sought stability by revitalizing traditional institutions at the local level. The mobilization of old ties of a horizontal type is not necessarily effective if the ambition is to gain political influence, but may be a sensible strategy if the objective is to ensure survival at the margins of a military and political conflict. It may be a natural response when an armed conflict intensifies and a breakdown of trust makes it difficult to maintain relationships beyond those most immediate. The mutual

180 Social Networks and Migration in Wartime Afghanistan

support networks that have often been found to be effective in coping with disasters are of the preexisting horizontal type. The immediate response to the crisis may have been of this kind in many Afghan communities, particularly at a stage where it still appeared realistic to stay out of the conflict. In such situations, one may come to rely on just one network as a conduit for economic resources and information, and in some cases also basic security. Such dependence on a singular network situation marks a stark contrast to the variegated networks described in Enjil of the 1970s (Grønhaug 1978). As the war escalated, it became increasingly difficult to remain disconnected from the major camps in the conflict. Where the mobilization of local horizontal ties was the initial reaction in the face of mounting military and political pressure, these small cohesive units could not grant security for long. The initial contraction strategy therefore had to be relaxed, as the smaller units found ways to link up with the warring parties. This move from contraction to conditional attachment did not necessarily undermine the cohesiveness of the horizontal networks; however, by undermining their detachment strategy, it pulled these smaller networks and their members into the mainstream of the conflict. When we look at the impact of war on networks, migration comes in as a complicating element that is rarely discussed in the literature. One suggestion that is made is simply that migration adds to the severing of ties (Harvey 1998). Insights from the larger literature on migration and transnationalism suggest that this is simplistic. Migration almost always involves an element of choice, although migration in wartime tends to be much more reactive than migration under other circumstances. Wartime migration may even reflect a conscious strategy for diversifying one’s primary networks or expanding them. Even when this is not a conscious motive, migration may contribute to expanding some network resources (while possibly severing others), either by increasing the opportunities of an existing network through pluri-localization (as with split households) or through the building of new relationships at the new location. Over time, migration potentially serves to counterbalance the effects of contraction by offering new opportunities. A strong tendency in Enjil was the choice of pluri-localization as the main coping strategy for dense groups, as part of their response to a contraction of networks in their community of origin. In the phase when the decision was taken to flee, it was common for one or more family members to stay behind to look after the property, while the others sought security and economic opportunities abroad. In the phase of return and resettlement, many families continued to rely on labor markets in exile after return,

Conclusions

181

with the bulk of the family staying permanently in Afghanistan. There are hardly any Afghans who do not have close associates in the neighboring country that could help them arrange travel or find work. New ties established in exile, however, seem to have played only a marginal role in Iran, in contrast to the experience in Pakistan, where frequent interaction with members of the host population persists, even in the aftermath of return. Most importantly, however, cohesive units may consciously split between different localities, but this does not necessarily threaten the strength of ties or inspire the development of new strong relationships. All of this indicates a considerable stability as to which networks are most effective in coping with war. From an individual perspective, of course, networks have been anything but stable, with war deaths, displacement, and conflict severing old ties. Many would find upon return that the positions they held, and the networks they belonged to before the war, have been dissolved or transformed and become unrecognizable. A significant change is also the emergence of new networks – largely in the form of military-cum-political groups and parties – that stand outside and challenge the stability and continuity represented by the old social setup, networks that do not necessarily include a large share of the population, but that may now have a disproportionate influence on the political process. These new networks are part of the new structural conditions to which the old networks have to adapt. The seeming continuity, therefore, may be deceptive. Networks that are similar to the prewar ones persist, and the basis for loyalty or authority at the local level has changed surprisingly little from prewar Afghanistan. Yet, the changed environment has affected the orientation of those traditional networks, so that they effectively contribute to drive radical change at the intermediate and macro levels.

Studying wartime migration The present work follows a critical vein of scholarship emerging over the past one and a half decades, in which the study of displacement is framed within the broader field of migration studies (Colson 2003; Lubkemann 2004, 2008; Richmond 1993; Shami 1996). As the Enjil case studies make clear, most migration decisions are to be found somewhere on the continuum between forced and voluntary. Forced migrants – like all other migrants – hold a mix of motivations at the time of departure. There are considerable similarities between ‘forced’ and ‘voluntary’ migrants: they meet a similar set of challenges; they draw on the same types of resources in tackling those challenges; and their course of action

182 Social Networks and Migration in Wartime Afghanistan

is heavily influenced by earlier migration experience. Furthermore, migration networks created during war gradually transform into resources in the postwar situation, just like the so-called migratory social capital built through other forms of migration. This study underscores scholarship that questions the validity of a sharp distinction between voluntary and forced migration. My studies in Afghanistan also tend to undermine the commonly held assumption that war is a sufficient cause in driving displacement, and it contends that factors at the macro level need to be complemented by insights about meso- and micro-level dynamics if we are to understand the dynamics of forced migration. It is also important not to confuse geographic co-location with social solidarity (see, for example, Connor 1988, 1989). Even in relatively stable situations, we find that people’s networks rarely coincide with confined geographical entities, but reach far beyond, as Reidar Grønhaug (1978, 1988) amply demonstrated in his studies from Enjil in the 1970s. Similarly, the people of even the smallest-sized settlement may relate to one another only marginally. The fundamental distinction between geographic and social proximity is amplified by massive migration experiences, such as the one in Afghanistan over the past two and a half decades. Split-household strategies, where even members of the same household live geographically apart, illustrate this well. The study of particular elements in the process of displacement, such as the decision to escape, needs to be framed in a larger context of experience and expectations held by the individual and close affiliates. An obvious example is the influence that prewar migration experience has on decisions to flee in war, with early departures coinciding strongly with migration experience. More interestingly, people almost always tend to have alternative options at any given juncture, as when choosing between escape to another country, internal displacement, or remaining on-site while seeking security from the government or another armed group. Similarly, people may pursue different integration strategies at locations of exile. All of the decisions undertaken depend on the opportunities that exist, including policies and practices of the host government and potential relief agencies, as well as the social network resources that are available. Of equal importance are people’s perceptions – particularly in terms of time – which have a direct impact on decisionmaking. Divergence in the expected duration of one’s stay at the destination – regardless of whether or not the expectation is realistic – is a central distinguishing factor between those who leave early, those who leave late, and those who stay behind. The early leavers tend to assume shorter duration.

Conclusions

183

The implication of all this is a tendency toward continuity in the displacement experience. Not only do decisions of the past directly affect opportunities and choices in the present, they also do so indirectly through their decisive effect on expectations for the future. War produces displacement as people seek to relocate to locations with more favorable prospects for survival. Such migration is involuntary in the sense that most people would have preferred not to become subject to war and unrest (though some may have contributed to it). Yet, people take choices and make the best of whatever resources and opportunities they have. Upon migration, they remain active agents who seek sensible modes of integration. All of this raises fundamental questions about the assumptions implicit in ‘forced migration’ and related terms. The implications are double-edged. They make it necessary for us to recognize that displaced people ought to be respected, dealt with, and analyzed as capable agents, just like anybody else. They also draw attention to the fact that forced migrants, while victims of armed conflict, may also be responsible for its initiation and maintenance. There is a fine line between not blaming the person displaced for his or her own misfortune and at the same time upholding the respect for, and the responsibility of, the displaced person as a capable agent. Research on displacement is always political. Why? First, displacement is a politically controversial theme in almost every country, notably when it comes to defining or limiting the rights of displaced people in relation to their host government – and, increasingly, the government in their country of origin. Second, like everybody else, displaced people act politically, a point that is not lost on political leaders looking for ways to mobilize population groups. Third, research on displacement also takes place in a politicized environment, with research agendas being tied to the objectives of actors on the international scene. The latter represents a challenge that researchers will always have to live with – if not accept – but it requires constant awareness. And then, of course, this is not a one-way street where research is defined by narrow political agendas; research may also have a deep impact on migration politics.

Appendix: Researching Migration in War This book is based on fieldwork in a country at war. The main fieldwork was conducted in spring 1999, when tension was running high between the Taliban regime and the opposition. In addition, relations between the Taliban and the international community were growing increasingly tense, and Pakistan and Saudi Arabia were the only states with diplomatic representation in Kabul. Meanwhile, the Taliban’s alliance with the Al-Qaeda network grew stronger. Fieldwork in war requires a serious risk assessment – of risk to respondents, local research staff, contacts in the field, and also the researcher (see Lee 1995: 5–10, 63–73). Is the research sufficiently important to accept the risks? Are there alternative ways of pursuing one’s research agenda? If fieldwork is the best option, in what ways can risk be minimized? Risk aversion is a constraint among researchers of armed conflict (Robben and Nordstrom 1995; Sluka 1990). The response to this problem, however, is not that researchers must be willing to accept more risk (see, for example, Barakat and Ellis 1997), but that we need to upgrade our ability to assess and manage risk, both before and during fieldwork (Van Brabant 2000; Williams et al. 1992). War makes it difficult to collect reliable data and thus constitutes a serious methodological challenge. The fundamental question is to what extent respondents and other associates in the field are able to trust the researchers. Such individuals are usually highly aware of the risks involved in sharing information with outsiders. In Afghanistan, the challenge of establishing trust is exacerbated by the fact that the country has virtually no tradition for independent research. There is also widespread skepticism toward any form of data-gathering, since this is normally a tool used by the government to expand its control over the population. It therefore becomes a major challenge to establish credible – yet truthful – explanations for what research is and what it does. Additionally, migration processes pose methodological problems of their own, even without war. One problem may be that potential informants are spread over a large area and it may be difficult to reach them, both for logistical and political reasons, despite the obvious need for multisite fieldwork (Freidberg 2001; Hannerz 2003; Marcus 1995; Wimmer and Schiller 2002). When multisite fieldwork is deemed impossible, vital data – particularly data related to those who have migrated and not maintained contact with their place of origin – are inaccessible. Furthermore, forced migration contexts encourage people to rationalize their previous acts, either in order to qualify for a certain status (such as ‘refugee’) or in order to make their past action more acceptable to others (Knudsen 1995). Again, trust between researcher and respondent is vital. Ultimately, it is only possible to restore agency to forced migrants by rooting research in their own accounts (Dyregrov, Dyregrov, and Raundalen 2000: 281–283; Ranger 1994). My approach to the fieldwork that forms a basis for this book can be described as gradualist (Atkinson and Delamont 2005). In realization of the political tension in Afghanistan in 1999, fieldwork had to include contingency plans for how to abort field activities at any moment. That things be carried out in the right order 184

Appendix 185 was extremely important, as was avoiding provoking negative reactions from anybody in a position of influence and authority. Yet, it was also imperative not to waste time, to get started on the process of collecting usable data, while avoiding methodological designs that would render data inapplicable in the event that the fieldwork had to be interrupted. Access needs to be negotiated at different levels. This presents researchers with a number of dilemmas, particularly in situations of war and under repressive regimes (Afghanistan qualifying on both counts). To cultivate relations with the authorities can make it difficult to gain the trust of common people (Lee 1995: 16–27; Williams et al. 1992). The bulk of fieldwork for this study took place at a time when the Taliban controlled Herat. The Taliban regarded the area as particularly sensitive, since they lacked popular support and were confronting a range of opposition groups, several with support from neighboring Iran. In the earliest stage, I met with Taliban officials in the capital Kabul, including successive ministers of repatriation, and discussed my research plans with them. I chose not to formally inquire about a research permit. Since I knew that there was no clear system for issuing such permits, a request would be likely to provoke suspicion and initiate a bureaucratic process that could take time and create unnecessary skepticism. Instead, I obtained a letter of recommendation from a high-level Taliban official. This letter helped in establishing contact with Taliban authorities in Herat. While the arrangement raised no expectation that we would be reporting back to national-level authorities on our findings, we did keep some key officials abreast of our activities. From the perspective of our informants, it was key that we did maintain cordial relationships with the authorities, but at the same time we acted in such a way that they could feel confident that our reporting did not place anybody at risk. It is not uncommon for researchers to work in settings where they do not know the language or do not have sufficient command of it for complex interviewing. In many such settings, there are no professional interpreters available and the only option is to recruit someone locally to act as an interpreter. The use of interpreters obviously has numerous disadvantages, as well as quite a few advantages. Yet, the methodological literature within the social sciences has little to say about how to work with interpreters (but see Laws, Harper, and Marcus 2003: 255–258). Only very rarely do texts based on fieldwork make reference to how the language issue has been tackled, and even more rarely do they discuss the pros and cons of working with local interpreters. Axel Borchgrevink, who has examined the conspicuous absence of an interpretation debate within anthropology, finds reasons both in the analysts’ need to cultivate authority and in the almost ‘sacred’ status of fieldwork within the discipline (Borchgrevink 2003: 113–115). The paradox that nobody talks about what (almost) everybody does, however, seems to apply well beyond anthropology. My own command of Dari is limited. It may be used for greetings and brief conversations and allows me to get the gist of conversations between others, but it does not allow me to conduct interviews or group meetings. In our little research group in Herat in 1999, everyone had worked frequently with interpreters in the field. None of us had come across treatment of how to deal with interpreters in the social science literature, but we all felt this deserved systematic attention (see also Strand 2003: 132–134). In parallel with identifying field sites and securing access to them through meetings with government representatives and community leaders, we started recruiting and training local interpreters. In the first round,

186 Appendix we needed three (one female, two male). Later in the fieldwork, some of these became research assistants, conducting structured interviews on their own; two more local staff were then employed. It is particularly difficult to collect reliable data in a time of war. Fear and political pressure serve as severe constraints; questions that would be considered trivial under other circumstances may be highly sensitive; systematic selection and follow-up of informants is difficult. Most fundamentally, fieldwork needs to be planned and carried out with the awareness that it may have to be aborted at any moment. This means that the researcher faces a constant dilemma between the need to collect as much data in the shortest possible time and the need to meticulously control the reliability of that data. To push ahead without carrying out the necessary exploratory work may also put the whole fieldwork (and all the people engaged in it) at risk. The way in which this dilemma is tackled has a definite effect on data quality, and this must be kept in mind during analysis. When the project was initiated in May 1997, I had already started to collect relevant publications. The main fieldwork was conducted in Herat over a period of 3 months, from March to June 1999, the first 2 months in cooperation with two other Norwegian researchers – Karin Ask and Arne Strand – who had partially overlapping research foci. I made short follow-up visits to the same area in May 2001, December 2002, September 2003, and April 2006. The collection of field data moved from a great degree of openness (open interviews with key informants and primary respondents, and observation) to fairly structured instruments (semi-structured and fully structured interviews), and back to open-ended followup interviews with former respondents during the return visits. Hence, while open-ended approaches were preferred in the initial stages in order to tune in on the field setting and develop functioning operationalizations and interview questions, the same instruments were also preferred in the later stages when checking information about specific topics with already interviewed informants. In the middle, the emphasis was on relatively structured instruments. Open interviews are the most important data uses in this book. Such interviews were dominant in the exploratory phase and formed a basis for tuning in theoretically, for establishing the working hypothesis, and for working out context-relevant operationalizations. During the return visits, open-ended interviews remained the chief mode of data-collection, but at this stage they were used to follow up with former informants, to prod more deeply into specific topics, and to test analytical suppositions (Yin 2003). In the preparatory phase, most of the interviewees were from NGOs and multilateral aid organizations. During the main period of fieldwork in 1999, almost all interviewees were local inhabitants in (or from) the chosen field sites. At later stages, particularly during the return visits, open interviews were conducted with people who had specialized knowledge, as well as with people from whom I wanted a more detailed account of their personal experience with war and migration. Open interviews were important to develop an understanding of networks and their role in migration, and to refine the formulation of some key questions about networks that I wanted to ask in a standardized formulation during all later interviews. The gathering of network information is a challenge, mainly for two reasons: its trivial character (in the eyes of the informant) and the related problem of recalling previous patterns of interaction. It is known from other contexts that people tend to take networks for granted, do not reflect

Appendix 187 upon the role they play, and do not consider them relevant to report upon (see, for example, Williams 1993). The second problem is that most respondents are interviewed after the fact – often after a considerable period of time. Some experimental work has been undertaken to assess the reliability of what people recall on networks. In a series of articles, Bernard, Killworth, and Sailer address the question of whether people’s self-reporting coincides with behavioral observations on interaction. The general conclusion from these studies is all but encouraging: ‘what people say . . . bears no useful resemblance to their behavior’ (Bernard, Killworth, and Sailer 1982: 63). This conclusion, however, has since been challenged by others, with the argument that people’s memory of relatively unimportant issues may be less reliable than their memory of important ones. The research by Bernard, Killworth, and Sailer also focuses on the recall of singular event interaction. People’s memories are far more robust when it comes to usual and regular interaction than in the case of particular events (Freeman, Romney, and Freeman 1987). Likewise, it has been demonstrated that individual group members have considerable ability to describe subgroup structure, so that aggregate perception of group structure corresponds well with behavioral data (Freeman, Freeman, and Michaelson 1989). This has the important methodological implication that key informants may do a good job in describing group structures (see, for example, Johnson 1990: 34–36). Similarly, it supports the use of complementary group interviews for mapping the structure and functions of social networks (see Bernard et al. 1984). Both of these approaches have been used in this study. Structured interviews were used only during the main fieldwork period in 1999 and were a central element in the collaboration between the three Norwegian researchers. Using our respective open interviews as a background, we worked together on question formulations and edited a pool of questions that the individual researchers could draw upon to get answers not only to their own questions, but also to those of the two others. Both the order and the emphasis of the questions varied considerably, yet each interview covered largely the same empirical ground (Massey et al. 1987: 13). In the next round, a fully structured interview form was worked out, using a selection of questions from the initial pool, which were marginally refined on the basis of experience. The sequencing of questions was also based on the experience from the open interviews. As a general principle, we asked questions about ‘here’ (the village) before ‘there’ (the exile), and about ‘now’ before ‘then’ (Patton 1990: 294–333). However, in later follow-up interviews, it proved fruitful to first revisit the past, using this as an introduction to the personal migration account. Again, a major challenge was the formulation of network questions, which was problematic owing to people’s perception of networks as ‘self-evident’. Drawing on the network literature (especially Burt 1983; Marsden 1990; McCallister and Fischer 1983), a range of formulations was tested out, refined, and eventually selected for use in the structured interviews. The fully structured interviews were conducted by the local research team and consisted of a questionnaire combining closed and open-ended questions. The questionnaire was translated into and the interviews conducted in Dari. Thirtyone such interviews were conducted – equally distributed between Izhaq Suleman and Sara-e Nau. This was less than one-third of the initial target of one hundred. Mounting security problems, particularly in the aftermath of the Shia uprising in

188 Appendix Herat in mid-May 1999 – brutally and successfully quelled by the Taliban – led us to interrupt this form of data-collection. Nonetheless, the survey interviews were useful in that they established a number of detailed migration biographies, and out of the 31 respondents, eight were interviewed again during later return visits, including two who were interviewed twice. A full overview over the interviews, including structured interviews, is given in Table A.1. Of the 125 open interviews conducted by myself, the ones carried out from 1999 onward have played the greatest role for the analysis presented here. In addition to my own interviews, I have had access to data collected by my colleagues Strand and Ask, and have used a total of 15 selected interviews that were relevant to my research focus. A number of the respondents were interviewed several times. During the main fieldwork in 1999, 26 interviews were with individuals and groups who had been interviewed previously. During the revisits, the majority of respondents had been interviewed earlier: 6 out of 7 in 2001; 8 out of 10 in 2002; 17 out of 25 in 2003; and 9 out of 12 in 2006 (in total, 40 out of 54). The shuras of Izhaq Suleman and Sara-e Nau were interviewed five and four times, respectively. Out of the total number of 210 interviews, 71 were follow-up interviews. Some of the interviews were with more than one person. In selecting the informants for all of the structured interviews, the aim was to cover a broad ground, including both people with variegated migration histories and people who had never migrated, as well as people from various walks of life. I intentionally sought a broad selection of informants by recruiting informants at different points in time – for example, by using Fridays to interview people who would otherwise be difficult to get hold of owing to work engagements in the city. Likewise, I sought to recruit people of different economic status, both by covering all mahalla (neighborhoods) within the villages to a similar extent, and by seeking to balance apparently poor households with more wealthy ones. With the exception of the ‘absent group’ (those still in exile), the informants included representatives of most major groups and life situations. I judged the collection of egocentric network data – as opposed to mapping comprehensive networks, that is, using a snowballing technique (Wasserman and Faust 1989: 33–34) – to be the best option under the circumstances. People’s confidence in the granting of anonymity, for example, would have been at risk if I had first interviewed one person about his or her relevant connections and then moved on to systematically interview each of those connections. The use of egocentric networks, though, largely deprives me of the opportunity to check Table A.1

Interviews1 1997 1999 2001 2002 2003 2006 Sums

Open interviews Open interviews (others) Structured interviews Structured interviews (others) Survey interviews (local team) Total

22

22

41 13 21 31 30 136

7 2

33

10

12

9

33

10

12

125 15 21 31 30 222

Appendix 189 the descriptions of various members within the same network against each other, which would be one central form of data triangulation. The alternative to my approach would have been to comprehensively map a cohesive type of network – for example, one where an extended family was the focus (see, for example, Monsutti 2004b). Given that my research interest, however, extended beyond the role of cohesive ties and beyond the social support role of networks, I found it preferable to establish egocentric networks. In compensation, I have sought to maximize the use of other forms of data triangulation. Ethnographic data, particularly in the form of observation, also plays an important role for this work, despite the decision – out of a concern for the security for those hosting the research – not to live in the local community (as always with Afghans, there was no scarcity of invitations). With my colleagues, and the local research team, I spent considerable time in the villages, with a view to accustoming people to our presence, developing a feel for local life, and gathering ethnographic data. This also helped in preparing the selection of informants, since it led to hundreds of casual conversations with various people and helped building an understanding of the social landscape. Ethnographic data were recorded as we went along. During interviews, we talked about and looked at the physical environment. Key informants took us around the village to tell us about the various locations, shops, and signs of war damage. These walkabouts were extremely useful, since they introduced us to the locality and made locals aware of our presence (Silverman 2001: 286). Also, when I interviewed people within their houses, I could refer to observed physical artifacts as a way of introducing particular questions. A television set, for example, could serve as an introduction to discussing life in Iran (where it had been bought), or to discussing relations with the Taliban (during whose rule televisions was banned). Through the making of such references to observable items, I could touch upon questions that might otherwise have been considered rude or politically sensitive to ask. As promised during the interviews, all respondents – with the exception of political leaders – have been anonymized. This was considered necessary in order to encourage people to talk freely and to avoid placing them unnecessarily at risk, despite the fact that many informants warmly welcomed the opportunity to tell the story of their war experience. All informants are given fictive names. Furthermore, I have altered certain biographical details, selecting items that will not affect the analysis but that make informants difficult to recognize. I have found it most convenient to implement this anonymization in one single operation throughout the text, and to do it at a late stage in the analysis, just prior to the final revision of the text. In this way, I have minimized the distortive effects of anonymization. Distortion can appear when seemingly insignificant biographical details that were modified at an early stage unknowingly become important at later stages, when the analytical focus has changed. I also found it useful to maintain throughout the analysis the intuitive associational value of real names and details. As a whole, the anonymization has forced me to provide only limited empirical detail on easily recognizable persons, but without this having substantial negative effects on the analysis. As an important supporting measure, I have relied on triangulation throughout the whole work: methodologically (by juxtaposing different methodological approaches); on data (by combining various types of data); and

190 Appendix analytically (by drawing on the work of fellow researchers, as well as opinions and explanations of local research staff and key informants) (Arksey and Knight 1999: 21). Starting in the field, concrete data, the interpretation of new findings, and modifications in the research focus were discussed with my two colleagues and with local staff. Perhaps the most important effect of working within a group of researchers was to have a forum for continuous discussion and reflection as fieldwork unfolded. To some extent, local door-openers served a similar function. More importantly, open interviews were used to corroborate earlier findings and to resolve inconsistencies. In this regard, open interviews conducted during the return visits, when I was deep into the analysis, were particularly useful. Finally, of course, the reading of various iterations of the text by competent researchers led to questions for clarification and served as a final round of quality assurance. In the analysis, I have sought to avoid the trap of anecdotalism by being comprehensive and including all of the data material. Accordingly, deviant evidence has been seen as an essential resource, and its examination has been used to confront and transform theoretical ideas. Systematic comparisons have been used – between different localities, time periods, individuals, and network structures – and have informed this work since its inception. Ultimately, however, the resulting analysis is based on data that have several flaws and weaknesses. It is difficult to collect the necessary amount of valid and reliable data in a country at war, and there were few empirical studies of forced migration available to serve as models for the research design. For these reasons, the present work is not a fully fledged empirical study with solid conclusions based on definitive evidence. On the other hand, it is also not primarily a theoretical work, claiming to present new theory while building on already established theory and earlier empirical studies. Instead, it is a work based on a constant dialogue between an inconclusive source material and a limited body of theoretical work. My hope is that precisely this kind of dialogue may have allowed me both to point out some fruitful theoretical venues for further research on forced migration and to make a contribution to the field of social network analysis.

Notes

1

Introduction

1. Examples of man-made disasters include plane crashes, breakdowns of artificial dams, or nuclear power accidents, whereas typical ‘nature-made’ disasters would be floods, hurricanes, or earthquakes. The distinction between the two is blurred, as ‘nature-made’ disasters may be triggered – and their effects worsened – by the acts of humans. 2. In her 1995 review essay, when discussing the attention to development issues among refugees in exile, Malkki provides six references, all of which were published in the period 1979–86, with four out of the six published between 1980 and 1982 (Malkki 1995a: 507). 3. In Malkki’s (1995a) review essay, three studies are seen to represent the ‘growing interest in theorizing the repatriation of refugees’ (1995a: 509). These were published in 1988, 1992, and 1994. 4. The 9/11 terrorist attacks in New York and Washington were responded to by military means, in accordance with the changing international paradigm on interventions, but they also led to a reassessment of immigration policies in the West (Nadig 2001; Schoenholtz 2003). 5. No credible estimates of the number of displaced persons during the US-led Coalition Campaign exist; most were forced to seek refuge internally, and only a portion went to organized camps for IDPs. 6. The Taliban also recruited among Pakistani citizens, particularly from fellow students of Afghans in traditionalist or Islamistic religious schools – madrasas (Harpviken 1997a). 7. For two useful reviews, see: Centlivres (1993) and Poppelwell (2007). 8. There is also an emerging literature on the Afghan diaspora and their transnational networks (Braakman and Schlenkhoff 2007; Brinkerhoff 2004) 9. On my follow-up visit in December 2003, I found that the residents of the mahalla had been permitted to select one representative to the Izhaq Suleman shura, seemingly on the initiative of the new district administration. Significantly, as a result of its previous war engagement, the population of Qala-e Muhajerin enjoyed the sympathy of Ismael Khan, the influential governor of Herat at the time. 10. Key resources in this regard have been various review essays, such as Oxfeld and Long 2004; Hein 1993; Malkki 1995a; Koser and Black 1999; Stefansson 2004a.

2

Social Networks in Wartime Migration

1. For the opposite view that common Afghan perceptions of security are almost entirely consistent with both the letter and the spirit of ‘human security’, see Feinstein International Famine Center (2004: 7). 191

192 Notes 2. The literature on patron–client relations, which is occupied with similar issues, also shares this general assumption (Eisenstadt and Roniger 1981; Knoke 1990). 3. For Williamson, site-specificity is one of three types of ‘asset specificity’, the other two being ‘physical asset specificity, as where specialized dies are required to produce a component; and human asset specificity that arises from learning by doing’ (Williamson 1981: 555). The idea of human asset specificity – that human capital is differentially valued from one labor market to the next – is a central tenet in the study of international migration. 4. Another commonly used definition is one that focuses on the frequency of interaction. On its own, this is not a robust measure, since it neglects the nature of what is being exchanged. It is intuitively recognizable that our closest ties may be with others than those with whom we interact most frequently. This fact is of utmost relevance for the study of migration, where we encounter individuals who are separated from their closest family, friends, and neighbors over long periods of time. One example is Wellman et al. 1997, in which frequency of face-to-face contact and telephone contact form two out of four indicators, along with actors’ definitions of the tie and actual provision of support. 5. One can argue that various relief organizations take on the role of the state in such cases, but in my experience such organizations are rarely present prior to the incident and their engagement is often short term. In situations where the presence of relief organizations is more enduring, as in many refugee situations, it certainly takes on a much larger significance. Even then, however, they are likely to be present only at the destination. 6. In the earlier work, the term ‘networks’ is applied, but in the 1997 article the authors have shifted to the term ‘social capital’. References to the work of Coleman (1988) and Bourdieu (1986) on the social capital concept indicate that this is a conscious move to maintain an actor-centered definition that is compatible with standard sociological methodology, as opposed to a structural definition that requires the specialized methodological repertoire of social network analysis. 7. Grasmuck and Pessar operate with a broad definition of social networks: ‘the social relations that organize and direct the circulation of labor, capital, goods, services, information, and ideologies between migrant sending and migrant receiving communities’ (Grasmuck and Pessar 1991: 13). In analyzing concrete social relations, however, their focus is on the importance of the household as a mediating mechanism and as a forum for decisionmaking, which largely overlaps with my definition of a migration collective. 8. Recent work on political mobilization has questioned the assumption that strong ties by definition contribute to mobilization (Gould 1993; Macy and Flache 1995). This work, while potentially important for forced migration studies, will not be reviewed in the current study. 9. A particularly interesting case is when the migration is a strategic move in the overall effort at political mobilization, an undisputed case of proactive migration. This is one possible avenue toward what has been coined ‘refugee-warrior communities’ (Zolberg, Suhrke, and Aguayo 1989). Another avenue is when such societies emerge through recruitment within an already exiled population. Whether the one or the other is the relevant interpretation of a given case is often a matter of considerable dispute, both politically and analytically.

Notes

193

10. A number of other concepts with overlapping definitions are used within network analysis. I prefer ‘bridges’ to ‘weak ties’ because weak ties are important only to the extent that they are bridging (Granovetter 1973: 208; 1983). I also have a preference for the term ‘bridges’ rather than ‘brokerage’ because the latter has a long history as a substantial concept in sociology and anthropology, not necessarily with the same as the meaning it has been given in social network analysis. 11. With reference to the study of the importance of strong ties among the poor, Granovetter (Granovetter 1983: 213) suggests ‘that the heavy concentration of social energy in strong ties has the impact of fragmenting communities of the poor into encapsulated networks with poor connections between these units; individuals so encapsulated may then lose some of the advantages associated with the outreach of weak ties. This may be one more reason why poverty is self-perpetuating.’ 12. This is not necessarily a consensual view. Within the broader study of responses to risk and disaster, there is both the view that societies tend to move back to a pre-disaster equilibrium and the perhaps more widely supported view that disasters are periods of significant change, when existing power structures are open to questioning and where new opportunities and constraints might lead to significant alterations (Oliver-Smith 1996: 309–310). 13. A similar mechanism lies at the core of what is often conceptualized as chain migration or circular migration, both terms describing situations where ties between origin and destination are strong. In the case of chain migration, there is a one-way stream; in the case of circular migration people migrate and return within what develops into a strong multi-local network. 14. This is the central theoretical puzzle occupying historical sociology. See, for example, Engelstad and Kalleberg (1999); Mahoney and Rueschmeyer (2003); Tilly (1984). 15. The so-called social networks community has long been primarily occupied with the development of analytical techniques through the reanalysis of existing datasets. While this is partly indicative of the cost of collecting solid network data, it is also indicative of the fact that new techniques are most easily developed in application to well-known data, a ‘laboratory approach’ in which the ingredients are already well defined.

3

Escape Decisions

1. These are percentages of the entire 1979 population, including those in exile and those killed in warfare. The figures for the percentage living in rural areas in Afghanistan are much more dramatic, with a decrease from 85 percent in 1979 to 23.2 percent in 1987. 2. The number estimates for IDPs are even less certain than those for refugees, but estimates throughout have ranged from half a million to two million (BIA 1985; Farr 2001). By early 2005, UNHCR estimates were down to 190,000 (UNHCR 2005: 2). Notably, this includes neither conventional ruralto-urban migration nor repatriates from abroad who settled somewhere else than where they originally came from. 3. For background on the survey and the methodology, see Sliwinski (1988: 2; 1989a: 56).

194 Notes 4. On the general concept of hijra and its relevance to present-day migration, see Aldeeb Abu-Sahlieh (1996), Muzaffar (2001), Shami (1996). 5. Translation adapted from BIA (BIA 1985: 14). See also Centlivres and Centlivres-Demont (1988b), Edwards (1986). 6. Peter Marsden describes a similar contrast between militia and mujahedin villages at the outskirts of Farah city, the provincial capital of Farah, to the immediate south of Herat. Also in this case, the militia village was subject to comparatively little direct war damage, despite a significant share of the locals siding with the mujahedin (Marsden 1997: 11–12, 17–19, 29–31). 7. There is only a very limited literature on the role of militias under the PDPA regime. The two most important sources are Dorronsoro and Lobato (1989) and Giustozzi (2000: 198–231). 8. Olivier Roy uses the network concept interchangeably with the idea of a qawm – defined as a ‘communal group, whose sociological basis may vary. It may be a clan (in tribal zones), a village, an ethnic group, an extended family, a professional group’ (Roy 1986: 226). Hence, Roy’s networks may not be geographically specific, which is one reason the state finds them difficult to control. For a critique of Roy’s qawm concept, see Richard Tapper (1988). 9. While it is not the focus of this chapter, it is worth pointing out that seeking refuge in a neighboring country may imply political expectations of its own, as when Afghans coming to Pakistan in the 1980s were obligated to register with one of the Afghan political parties there in order to obtain refugee status. 10. Among those who left individually for Iran during the PDPA era, a majority say they escaped to avoid conscription (a significant minority cite economic reasons). 11. On this, see also the case studies on Afghans in Iran by the Afghan Research and Evaluation Unit (AREU) (Abbasi-Shavazi et al. 2005a, 2005b, 2005c). 12. A different option is to choose a destination as close as possible to the origin, either for the whole group or for parts of the group, which may even permit regular visits to look after property. 13. The market for smuggling services fluctuated with the intensity of Iranian migration controls and largely disappeared when Iran started to grant visas to Afghans from its Herat consulate in 2002. 14. Neither Izhaq Suleman nor Sara-e Nau had an active shura (village council) during the PDPA period. 15. There are signs that this is changing in the post-Taliban era, as new cellphone systems in Afghanistan makes access far easier. 16. Among my respondents, three in ten reported an ability to read and write. 17. A couple of informants reported on exchange of cassettes, which is a creative response to the literacy problem but does require a cassette recorder, which most households do not possess. 18. Given that I have been unable to conduct systematic fieldwork in Iran, it is difficult to move much beyond speculation on this point.

4

Integration at Exile

1. The debate on resettlement of refugees in third countries offers parallel insights on the challenges of integration (Colic-Peisker and Tilbury 2003; Muggah 2003).

Notes

195

2. Given the situation for Afghans in Iran, I decided not to track down informants still residing there. I did two things to adjust for this deficiency. Firstly, I interviewed both recent returnees and close associates of people still in exile; this only partially compensates for the problem, though, since I still lack information on those exiles who have not maintained networks with home (and who, if they ever return, will probably be the last to do so). Secondly, I visited Iran both in December 2002 and in October 2003, getting feedback on work-in-progress at conferences (Harpviken 2002, 2003), as well as using the occasion to interview key informants. Monsutti (2004b: 169) reports serious problems during his fieldwork in Iran in the spring of 1996 – when conditions for Afghans there were considerably easier than at the time of my fieldwork – and many of those that he contacted refused to talk to him. 3. See, in particular: (Ahmed 1990; Centlivres 1991; Centlivres and CentlivresDemont 1987, 1988a, 1988b; Connor 1987, 1989; English 1989a; Morton 1992a, 1992b, 1992c, 1994). Even the upsurge of global interest for internally displaced persons (IDPs) in the latter half of the 1990s (see, for example, Cohen and Deng 1998) did not inspire new research on Afghan displacement (Farr 2001). 4. Often also referred to as ‘green cards’ by respondents. 5. The situation for Afghans in Iran became more difficult from 1997 onward, in large part because of new policies and even tighter controls by the authorities (Jamal and Stigter 2002). A set of new regulations, approved by the interministerial Executive Co-ordination Council for Foreign Nationals in April 2002, instituted new pressures. The so-called Regulations on Accelerating Repatriation of Afghan Nationals include 11 articles, applying to all who do not hold a valid passport with visa and residence permits (Abbasi-Shavazi et al. 2005b). The regulations introduced penalties on Iranians who employ Afghans, or who sublet their homes to them, unless they had special permission. The new regulations also denied Afghans access to administrative services, as well as the right of association. These regulations were introduced in the aftermath of the reregistration exercise which had been conducted by the Bureau for Aliens and Foreign Immigrants Affairs (BAFIA). 6. The massive aid effort was politically motivated, with Western and Islamic governments and organizations wishing to support the victims of the Soviet invasion, hence also stimulating the exile-based political and military resistance (see, for example, Baitenmann 1990; Fielden 1998). 7. By 1987, the Pakistanis attempted to establish new refugee camps in Punjab province, further away from the border, but these camps were highly unpopular for both climatic and ethno-linguistic reasons. 8. From late 1980 onward, Iran pursued contacts on both sides of the conflict. It put an end to its early role of hosting the offices of Sunni Islamic resistance organizations (including Gulbuddin Hekmatyar’s Hezb-e Islami). 9. Although it has not, to my knowledge, been thoroughly documented, there have also been widespread allegations that Iran recruited Afghans to the war with Iraq, either by force or by promising generous rewards, including citizenship (see, for example, Hug 1987) 10. Among the Qala-e Muhajerin population, the share that had been associated with the mujahedin while in exile was also high, but those were people

196 Notes

11.

12. 13.

14.

15.

16.

5

originating from various locations (Izhaq Suleman included) and who had partly met in exile through their political and military involvement. At the same time, among the Shia, Iran encouraged the emergence of new types of leaders educated in their religious schools, with close ties to leaders of the Iranian revolutions. The challenging of traditional leaders created massive armed conflict, particularly within the Hazarajat region, throughout most of the 1980s (Harpviken 1996). The 1992 fall of Najibullah ignited a similar sentiment. In the 1970s, when labor migration from Afghanistan to Iran took off, it was the booming oil industry that opened up job opportunities, and many Afghans had been severely affected by drought. Among the network-specific questions, this is the one where respondents clearly underreported their own coping ability, probably out of simple habit, but also because they may have identified foreign researchers with the assistance apparatus. Both a general suspicion of the state, which is perceived to be potentially using any data it collects for malevolent purposes, and little or no familiarity with the idea of independent research strengthen such a trend. In practice, my experience from the field is that as a foreign researcher, one is primarily associated with the international aid apparatus. Monsutti, who did fieldwork for his PhD in a different part of Afghanistan, and who had no association to aid agencies, reports of similar expectations, simply because the expatriate aid worker is a familiar figure while the researcher is not (Monsutti 2004b: 84). Eighteen females were interviewed by Karin Ask in the same area and during the same period. Out of these, 16 had been to Iran (with their families), but none of them had themselves been on return visits during their stay in exile.

Return Decisions

1. The repatriation of Mozambican refugees from Malawi following the 1992 peace agreement between Frelimo and Renamo has been considered one of the major successes of UNHCR in recent times. This case, however, also illustrates how people make their return decisions on the basis of other factors than just the peace agreement, and the extent to which ‘self-repatriation’ is the norm rather than the exception. By October 1992, when the peace agreement between the Frelimo government and the Renamo opposition forces was signed, a significant portion of the refugees had already returned (Juergensen 2000). 2. As a consequence of the 1988 Geneva agreement between the governments of Afghanistan and Pakistan, which precipitated the withdrawal of Soviet forces from Afghanistan in February 1989, the international community expected, and prepared for, massive repatriation of Afghan refugees from the neighboring countries (Rubin 1988). The Geneva accords neglected the issue of political transition in Afghanistan, and did not foresee the continuing unrest after the Soviet withdrawal. In the case of Pakistan, this was compounded by the major resistance parties, which declared that a return prior to a regime change would be an act of treason against the spirit of the jihad. One analyst referred to the Afghan refugees in Pakistan as ‘hostages in the struggle for power’ (Rizvi 1990). The resistance parties were centrally positioned among

Notes

3.

4.

5.

6.

7.

8.

9.

10. 11.

12.

197

Afghans in Pakistan and enjoyed a close relationship with Pakistan’s main military intelligence agency, Inter-Services Intelligence (ISI). Throughout the 1980s, the parties had proven their ability and will to back their messages with force, through imprisonment, abduction, and assassination of political opponents. Now, there were various reports of roadblocks and harassment to prevent repatriation (Knowles 1992). The Norwegian Project Office/Rural Rehabilitation Association for Afghanistan (NPO/RRAA) was probably the first to try out a group repatriation concept in Afghanistan with its assisted repatriation of 67 families to a village in Paktia province. The project combined community welfare, such as health and education, with income generation, along with assistance for the construction of family housing. Although it was later criticized as being too expensive, it did serve as a useful pilot project, inspiring others to try out the same idea (Heffer 1997; NPO/RRAA 1995). For more on attitudes to return among Afghans residing in Pakistan, see Collective for Social Science Research 2005, 2006a, 2006b, 2006c; Saito and Hunte 2007; Habibi and Hunte 2006. I am using ‘Khost’ here as shorthand for Mandozay and Ismailkhel in the Basin of Khost and Yaqubi, and ‘Kunar’ as shorthand for the central areas of the province, particularly in the lower Pech valley. The fact that the radical parties in general were more outspoken against repatriation before the fall of the ‘infidel’ communist government, and had a proven record of being willing to enforce their policies, may also have strengthened the hesitation to repatriate among the Kunaris. Saying that the tribal system remains intact may be a truism, but the Khost area has proven extremely resilient in the face of grave threats. Existing institutions and the role of leaders have been adapted to new challenges, but with no fundamental rupture with the past (Glatzer 1992: 165). In both cases, aid agencies were actively involved. In Kunar, the international assistance was general, rather than linked to group repatriation processes, while in Khost agencies designed their interventions in line with needs assessments that involved would-be returnees. Even in the latter case, aid does not seem to have been decisive in motivating repatriation, although it may have played a considerable role in easing resettlement and preventing new out-migration. What implications this insight had is not clear, although opportunities for delaying repatriation in the hope that conflicts could first be settled was one option that was discussed. See also Abbasi-Shavazi et al. (2005b, 2005c). These figures were collected in December 2002 and October 2003. My research assistant identified central meeting places (mosque, local shop) in each mahalla in both villages (10 mahallas in Izhaq Suleman, 4 in Sara-e Nau), and then gathered a crowd of men and asked who had returned after the fall of the Taliban (2002) and after our previous visit (2003). The list contained the name of a head of household (and his father), and the 2003 list was checked against the 2002 list to eliminate double reporting. The realization that labor migration to Iran and Pakistan is a key coping strategy for Afghans, and likely to remain so, is a core argument for the proposed

198 Notes shift from ‘refugee management’ to ‘migration management’ currently being promoted by UNHCR and others (Stigter 2005b; UNHCR 2003a). 13. The introduction of cell phones, post-2001, has changed this somewhat. While I have limited information on this, indications are that although information is generally much easier, the informational asymmetry prevails, as men in large part control the means of communication. 14. For the special cases where the Afghan men have married Iranian women, see Zahedi (2007).

6

Reintegration at ‘Home’

1. For the communities in question, land mines represent a security challenge different from the other issues discussed, in that they are first and foremost remnants of past military action, rather than instruments being used in the periods following substantial returns (post-1992 and post-2001). In other conflict areas, of course, land mines may still be used in low-intensity conflict long after the main armed action is over (Harpviken and Skåra 2003). 2. This goes beyond the question of ‘absorptive capacity’, raised above, which focuses on the capacity of the community to assist and accommodate returnees, first and foremost in the economic domain. 3. This is one issue where there could be serious underreporting, given that the main fieldwork was conducted in 1999, when the Taliban presence was felt locally, and where arrests and harassment of people suspected of being associated with Ismael Khan or other armed opposition groups were fairly common. 4. A few days after the onset of the US-led attack on 7 October 2001, a team of US Special Forces was in Uzbekistan, to be deployed with Ismael Khan (Woodward 2002: 331). 5. All the respondents in Qala-e Muhajerin had their base in two smaller cities in Mashad province: Turbat-e Jam and Kariz Tayabad. 6. In 1998 and 1999, there were a few instances of poppy cultivation in the area surrounding Herat, but this was stopped when the Taliban banned opium cultivation in August 2000. Ismael Khan maintained a strict line against drugs production during both of his reigns.

Appendix: Researching Migration in War 1. I have had access to the material collected by my two colleagues, Arne Strand and Karin Ask. Only those of their interviews that are directly relevant to my project have been included. The survey interviews in 1999 were carried out by the local research staff on their own. I have used the computer software package NUD*IST (N6) for systematization and analysis of the collected data (see also QSR 2002; Weitzmann 2000).

Bibliography Abbasi-Shavazi, Mohammad Jalal, Diana Glazebrook, Gholamreza Jamshidia, Hossein Mahmoudian, and Rasoul Sadeghi (2005a) ‘Return to Afghanistan? A Study of Afghans Living in Mashad, Islamic Republic of Iran.’ Case Study Series, October 2005. Kabul: Afghanistan Research and Evaluation Unit. ——— (2005b) ‘Return to Afghanistan? A Study of Afghans Living in Teheran.’ Case Study Series, June 2005. Kabul: Afghanistan Research and Evaluation Unit. ——— (2005c) ‘Return to Afghanistan? A Study of Afghans Living in Zahedan, Islamic Republic of Iran.’ Case Study Series, October 2005. Kabul: Afghanistan Research and Evaluation Unit. Abbott, Andrew (1997a) ‘The Causal Devolution.’ Sociological Methods and Research 27: 148–181. ——— (1997b) ‘Of Time and Space: The Contemporary Relevance of the Chicago School.’ Social Forces 75: 1149–1182. Abu-Sahlieh, Sami A. Aldeeb (1996) ‘The Islamic Conception of Migration.’ International Migration Review 30: 37–88. Adamson, Fiona B., and Madeleine Demetriou (2007) ‘Rethinking the Boundaries between “State” and “National Identity”: Incorporating Diasporas into IR Theorizing.’ European Journal of International Relations 13: 489–526. Adelkhah, Fariba, and Zuzanna Olszewska (2007) ‘The Iranian Afghans.’ Iranian Studies 40: 137–165. Adelman, Howard (2002) ‘Refugee Repatriation.’ In Ending Civil Wars: The Implementation of Peace Agreements, edited by Stephen John Stedman, Donald Rotchild, and Elizabeth M. Cousens. Boulder, CO: Lynne Rienner, pp. 273–302. Agesa, Richard U., and Kim Sunwoong (2001) ‘Rural to Urban Migration as a Household Decision.’ Review of Development Economics 5: 60–75. Ahmed, Akbar S. (1990) ‘The Impact of Afghan Refugees on Ethnicity and Politics in Baluchistan (Pakistan).’ Central Asian Survey 9: 43–56. AIHRC (2005) ‘A Call for Justice: A National Consultation on Past Human Rights Violations in Afghanistan.’ Kabul: Afghan Independent Human Rights Commission. AJP (2005) ‘Casting Shadows: War Crimes and Crimes against Humanity, 1978–2001.’ Kabul: Afghanistan Justice Project. Aldeeb Abu-Sahlieh, Sami A. (1996) ‘The Islamic Conception of Migration.’ International Migration Review 30: 37–57. Allen, Rebecca, and Harry H. Hiller (1985) ‘The Social Organization of Migration: An Analysis of the Uprooting and Flight of Vietnamese Refugees.’ International Migration 23: 434–452. Allen, Tim, and David Turton (1996) ‘Introduction: In Search of Cool Ground.’ In In Search of Cool Ground: War, Flight & Homecoming in Northeast Africa, edited by Tim Allen. Trenton, NJ: Africa World Press, pp. 1–22. Amin, Tahir (1982) Afghanistan Crisis: Implications and Options for Muslim World, Iran and Pakistan. Islamabad: Institute of Policy Studies. 199

200 Bibliography Ammassari, Shavina, and Richard Black (2001) ‘Harnessing the Potential of Migration and Return to Promote Development.’ IOM Migration Research Series. Geneva: International Organization for Migration. Amstutz, J. Bruce (1986) Afghanistan: The First Five Years of Soviet Occupation. Washington, DC: National Defense University Press. Anderson, Mary B., and Peter J. Woodrow (1989) Rising from the Ashes: Development Strategies in Times of Disaster. Boulder and San Francisco: Westview Press. Appleyard, Reginald (2001) ‘International Migration Policies: 1950–2000.’ International Migration 39: 7–20. Arksey, Hilary, and Peter Knight (1999) Interviewing for Social Scientists. London: Sage. Arowolo, Oladele O. (2000) ‘Return Migration and the Problem of Reintegration.’ International Migration 38: 59–82. Atkinson, Paul, and Sara Delamont (2005) ‘Analytic Perspectives.’ In The Sage Handbook of Qualitative Research (3rd edition), edited by Norman K. Denzin and Yvonna S. Lincoln. Thousand Oaks, CA: Sage. Baitenmann, Helga (1990) ‘NGOs and the Afghan War: The Politicisation of Humanitarian Aid.’ Third World Quarterly 12: 62–85. Bakewell, Oliver (2004) ‘Repatriation: Angolan Refugees or Migrating Villagers?’ In Refugees and the Transformation of Societies: Agency, Policies, Ethics and Politics, edited by Philomena Essed, Georg Frerks, and Joke Schrijvers. Oxford: Berghahn Books, pp. 31–41. Barakat, Sultan, and Sue Ellis (1997) ‘Researching under Fire: Issues for Consideration When Collecting Data and Information in War Circumstances, with Specific References to Relief and Reconstruction Projects.’ Disasters 20: 149–156. Bascom, Jonathan (2005) ‘The Long, “Last Step”? Reintegration of Repatriates in Eritrea.’ Journal of Refugee Studies 18: 165–180. Bates, Robert H. (2008) ‘State Failure.’ Annual Review of Political Science 11: 1–12. Batson, Wendy (1992) ‘Repatriation Report: Afghan Refugees in Pakistan – Arrivals and Departures, 1991.’ 1 March. Islamabad: UNHCR. Bernard, H. Russell, Peter D. Killworth, and Lee Sailer (1982) ‘Informant Accuracy in Social-Network Data V: An Experimental Attempt to Predict Actual Communication from Recall Data.’ Social Science Research 11: 30–66. Bernard, H. Russell, Peter Killworth, David Kronenfeld, and Lee Sailer (1984) ‘The Problem of Informant Accuracy: The Validity of Retrospective Data.’ Annual Review of Anthropology 13: 495–517. Bhatia, Michael (2003) ‘Repatriation under a Peace Process: Mandated Return in the Western Sahara.’ International Journal of Refugee Law 15: 786–822. Bhatia, Michael, and Mark Sedra (2008) Afghanistan, Arms and Conflict: Armed Groups, Disarmament and Security in a Post-War Society. London: Routledge. BIA (1985) ‘Report on the Displaced Persons in Afghanistan: International Humanitarian Inquiry Commission.’ 26 April 1985. Paris. Bialczyk, Agata (2008) ‘“Voluntary Repatriation” and the Case of Afghanistan: A Critical Examination.’ Oxford: Refugee Studies Centre, University of Oxford. Bikhchandani, Sushi, David Hirshleifer, and Ivo Welch (1992) ‘A Theory of Fads, Fashion, Custom, and Cultural Change an Informational Cascades.’ Journal of Political Economy 100: 992–1026. Birkeland, Nina M. (2003) ‘“Last Time I Fled Because of War, This Time Because of Hunger”.’ In In the Maze of Displacement: Conflict, Migration and Change,

Bibliography 201 edited by Nadarajah Shanmugaratnam, Ragnhild Lund, and Kristi Anne Stølen. Kristiansand: Høyskoleforlaget, pp. 177–200. Black, Richard (2006) ‘Return of Refugees: Retrospect and Prospect.’ In Palestinian Refugee Repatriation: Global Perspectives, edited by Michael Dumper. London: Routledge. Black, Richard, Marita Eastmond, and Saskia Gent (2008) ‘Sustainable Return in the Balkans: Beyond Property Restitution and Policy.’ International Migration 44: 5–13. Black, Richard, and Khalid Koser (Eds) (1999) The End of the Refugee Cycle: Refugee Repatriation and Reconstruction. New York: Berghahn Books. Black, Richard, and Russel King (2004) ‘Editorial Introduction: Migration, Return and Development in West Africa.’ Population, Space and Place 10: 75–83. Blitz, Brad K., Rosemary Sales, and Lisa Marzano (2005) ‘Non-voluntary Return? The Politics of Return to Afghanistan.’ Political Studies 53: 182–200. Boesen, Inger W. (2004) ‘From Subjects to Citizens: Local Participation in the National Solidarity Programme.’ Working Paper Series, August 2004. Kabul: Afghanistan Research and Evaluation Unit. Borchgrevink, Axel (2003) ‘Silencing Language: Of Anthropologists and Interpreters.’ Ethnography 4: 95–121. Bourdieu, Pierre (1986) ‘The Forms of Capital.’ In Handbook of Theory and Research for the Sociology of Education, edited by John G. Richardson. New York: Greenwood Press. Braakman, Marije, and Angela Schlenkhoff (2007) ‘Between Two Worlds: Feelings of Belonging While in Exile and the Question of Return.’ Asien 104: 9–22. Brettell, Caroline B. (2000) ‘Theorizing Migration in Anthropology: The Social Construction of Networks, Identities, Communities and Globalscapes.’ In Migration Theory: Talking across Disciplines, edited by Caroline B. Brettel and James F. Hollifield. New York: Routledge, pp. 97–135. Brettell, Caroline B., and James F. Hollifield (Eds) (2000a) Migration Theory: Talking across Disciplines. New York: Routledge. ——— (2000b) ‘Migration Theory: Talking across Disciplines (Introduction).’ In Migration Theory: Talking across Disciplines, edited by Caroline B. Brettel and James F. Hollifield. New York: Routledge, pp. 1–26. Brinkerhoff, Jennifer M. (2004) ‘Digital Diasporas and International Development: Afghan-Americans and the Reconstruction of Afghanistan.’ Public Administration and Development 24: 397–413. Bruce, John (2007) ‘Drawing a Line under the Crisis: Reconciling Returnee Land Access and Security in Post-conflict Rwanda.’ Humanitarian Policy Group Working Paper. London: Overseas Development Institute. Bryceson, Deborah, and Ulla Vuorela (2002a) ‘Transnational Families in the Twenty-First Century.’ In The Transnational Family: New European Frontiers and Global Networks, edited by Deborah Bryceson and Ulla Vuorela. Oxford and New York: Berg, pp. 3–30. ——— (Eds) (2002b) The Transnational Family: New European Frontiers and Global Networks. Oxford: Berg. Burgess, J. Peter, and Taylor Owen (Eds) (2004) ‘Special Section: What is “Human Security”?’ Security Dialogue 35: 345–371. Burt, Ronald S. (1983) ‘Network Data from Informant Interviews.’ In Applied Network Analysis: A Methodological Introduction, edited by Ronald S. Burt and Michael J. Minor. Beverly Hills: Sage, pp. 133–157.

202 Bibliography ——— (1987) ‘Social Contagion and Innovation: Cohesion versus Structural Equivalence.’ American Journal of Sociology 92: 1287–1335. Carling, Jørgen (2008a) ‘Interrogating Remittances: Core Questions for Deeper Insight and Better Policies.’ In Migration and Development: Perspectives from the South, edited by Stephen Castles and Raúl Delgado Wise. Geneva: International Organization for Migration, pp. 43–64. ——— (2008b) ‘Toward a Demography of Immigrant Communities and Their Transnational Potential.’ International Migration Review 42: 449–475. Castles, Stephen, and Mark J. Miller (2003) The Age of Migration: International Population Movements in the Modern World (Third Edition). London: Palgrave Macmillan. Centlivres, Pierre (1991) ‘Exil, relations interethniques et identitè dans la crise afghane.’ Revue du Monde Musulman et de la Mediterranè 59–60: 70–82. ——— (1993) ‘A State of the Art Review of Research on Internally Displaced, Refugees and Returnees from and in Afghanistan.’ London: The Fourth International Research and Advisory Panel Conference on Forced Migration. Centlivres, Pierre, and Micheline Centlivres-Demont (1992) Afghanistan: Case Study. Geneva: UNHCR. Centlivres, Pierre, and Micheline Centlivres-Demont (1983) ‘Frontieres et Phenomenes Migratoires en Asia Centrale: Le Cas de l` Afghanistan de 1880 a nos Jours.’ Ethnologica Helvetica 7: 83–114. ——— (1987) ‘Sociopolitical Adjustment among Afghan Refugees in Pakistan.’ Migration World 15: 15–21. ——— (1988a) ‘The Afghan Refugee in Pakistan: An Ambiguous Identity.’ Journal of Refugee Studies 1: 141–152. ——— (1988b) ‘The Afghan Refugees in Pakistan: A Nation in Exile.’ Current Sociology 36: 71–92. Cernea, Michael M. (1996) ‘Bridging the Research Divide: Studying Refugees & Development Oustees.’ In In Search of Cool Ground: War, Flight & Homecoming in Northeast Africa, edited by Tim Allen. Trenton, NJ: Africa World Press, pp. 293–317. Cernea, Michel M., and Christopher McDowell (2000) ‘Introduction: Reconstructing Resettlers’ and Refugees’ Livelihoods.’ In Risks and Reconstruction: Experiences of Resettlers and Refugees, edited by Michel M. Cernea. Washington, DC: The World Bank, pp. 1–10. Chimni, B. S. (2003) ‘Post-conflict Peacebuilding and the Return of Refugees: Concepts, Practices and Institutions.’ In Refugees and Forced Displacement, edited by Edward Newman and Joanne Van Selm. Tokyo: United Nations University Press, pp. 195–220. Chimni, B. S. (1999) ‘From Resettlement to Involuntary Repatriation: Towards a Critical History of Durable Solutions to Refugee Problems.’ Geneva: UNCHR. Chr. Michelsen Institute, Copenhagen Development Consulting and German Association of Development Consultants (2005) ‘Humanitarian and Reconstruction Assistance to Afghanistan, 2001–05: From Denmark, Ireland, the Netherlands, Sweden and the United Kingdom.’ October 2005. Copenhagen: Ministry of Foreign Affairs of Denmark. Christensen, Asger (1980) ‘Organization, Variation and Transformation in Pukhtun Society: Review Article.’ Ethnos 46: 96–108.

Bibliography 203 ——— (1988) ‘When Muslim Identity Has Different Meanings: Religion and Politics in Contemporary Afghanistan.’ In Islam: State and Society, edited by Klaus Ferdinand and Mehdi Mozaffari. London: Curzon Press, pp. 143–154. ——— (1995) Aiding Afghanistan: The Background and Prospects for Reconstruction in a Fragmented Society. Copenhagen: Nordic Institute of Asian Studies. Clarke, Lee, and James F. Short Jr. (1993) ‘Social Organization and Risk: Some Current Controversies.’ Annual Review of Sociology 19: 375–399. Cohen, Roberta, and Francis Deng (1998) Masses in Flight: The Global Crisis of Internal Displacement. Washington, DC: Brookings Institution Press. Coleman, James S. (1988) ‘Social Capital in the Creation of Human Capital.’ American Journal of Sociology 94: 95–120. Colic-Peisker, Val, and Farida Tilbury (2003) ‘“Active” and “Passive” Resettlement: The Influence of Support Services and Refugees’ own Resources on Resettlement Style.’ International Migration 41: 61–91. Collective for Social Science Research (2005) ‘Afghans in Karachi: Migration, Settlement and Social Networks.’ Kabul: Afghanistan Research and Evaluation Unit. ——— (2006a) ‘Afghans in Pakistan: Broadening the Focus.’ Kabul: Afghanistan Research and Evaluation Unit. ——— (2006b) ‘Afghans in Peshawar: Migration, Settlements and Social Networks.’ Kabul: Afghanistan Research and Evaluation Unit. ——— (2006c) ‘Afghans in Quetta: Settlements, Livelihoods, Support Networks and Cross-Border Linkages.’ Kabul: Afghanistan Research and Evaluation Unit. Colletta, Nat J., and Michelle L. Cullen (2000) Violent Conflict and the Transformation of Social Capital: Lessons from Cambodia, Rwanda, Guatemala, and Somalia. Washington, DC: The World Bank. Collyer, Michael (2005) ‘The Search for Solutions: Achievements and Challenges.’ Journal of Refugee Studies 18: 247–257. Colson, Elizabeth (2003) ‘Forced Migration and the Anthropological Response.’ Journal of Refugee Studies 16: 1–18. Connor, Kerry M. (1987) ‘An Analysis of Residential Choice among Self-Settled Afghan Refugees in Peshawar, Pakistan.’ Graduate College (Geography). Lincoln: University of Nebraska. ——— (1988) ‘Rationales for the Movement of Afghan Refugees to Peshawar.’ In Afghan Resistance: The Politics of Survival, edited by Grant M. Farr and John G. Merriam. Lahore: Vanguard books, pp. 151–190. ——— (1989) ‘Factors in the Residential Choices of Self-Settled Afghan Refugees in Peshawar, Pakistan.’ International Migration Review 23: 904–932. Constable, Pamela (2004) ‘Afghans Riot Over Dismissal of Governor in Herat.’ Washington Post, 13 September 2004. Cuny, Fredrick C., and Barry N. Stein (1989) ‘Prospects for and Promotion of Spontaneous Repatriation.’ In Refugees and International Relations, edited by Gil Loescher and Laila Monahan. Oxford: Oxford University Press. DACAAR, IRC, and MADERA (2002) ‘Expected Repatriation Patterns 2002: Results of Survey in Refugee Camps in NWFP, 18 February – 22 March 2002.’ Peshawar: DACAAR/IRC/MADERA. Damsleth, Bente (2003) ‘Coping with Disrupted Lives: A Study of Afghan Girls and Their Family Networks.’ SUM Dissertations and Thesis. Oslo: Centre for Development and the Environment, University of Oslo.

204 Bibliography DeVoe, Dorsh Marie (1981) ‘Framing Refugees as Clients.’ International Migration Review 15: 88–94. Dorronsoro, Gilles (2005) Revolution Unending: Afghanistan, 1979 to the Present. New York: Columbia University Press. Dorronsoro, Gilles, and Chantal Lobato (1989) ‘The Militia in Afghanistan.’ Central Asian Survey 8: 95–108. Douglas, Mary (1985) Risk Acceptability According to the Social Sciences. New York: Russel Sage Foundation. Douglas, Mary, and Aaron Wildavsky (1983) Risk and Culture: An Essay on the Selection of Technological and Environmental Dangers. Berkeley: University of California Press. Duffield, Mark (2001) Global Governance and the New Wars: The Merging of Development and Security. London: Zed Books. Dupree, Louis, and Nancy Dupree (1987) ‘Afghan Refugees in Pakistan.’ World Refugee Survey: 17–21. Dyregrov, Kari, Atle Dyregrov, and Magne Raundalen (2000) ‘Refugee Families’ Experience of Research Participation.’ Journal of Traumatic Stress 13: 413–426. Eastmond, Marita, and Joakim Öjendal (1999) ‘Revisiting a “Repatriation Success”: The Case of Cambodia.’ In The End of the Refugee Cycle: Refugee Repatriation and Reconstruction, edited by Richard Black and Khalid Koser. New York: Berghahn Books, pp. 38–55. Edwards, David Busby (1986) ‘Marginality and Migration: Cultural Dimensions of the Afghan Refugee Problem.’ International Migration Review 20: 313–328. Eisenstadt, Shmuel N., and Luis Roniger (1981) ‘The Study of Patron-Client Relations and Recent Developments in Sociological Theory.’ In Political Clientelism, Patronage and Development, edited by Shmuel N. Eisenstadt and Rene Lemarchand. London: Sage. Emirbayer, Mustafa (1997) ‘Manifesto for a Relational Sociology.’ American Journal of Sociology 103: 281–317. Emirbayer, Mustafa, and Jeff Goodwin (1994) ‘Network Analysis, Culture and the Problem of Agency.’ American Journal of Sociology 99: 1411–1454. Engelstad, Fredrik, and Ragnvald Kalleberg (Eds) (1999) Social Time and Social Change: Perspectives on Sociology and History. Oslo: Scandinavian University Press. English, Richard (1988) ‘Preliminary Report on Conditions Affecting the Repatriation of Afghan Refugees.’ 20 June. Geneva: UNHCR. ——— (1989a) ‘The Economic Impact of Afghan Refugee Settlement on the Tribal Areas of Northwest Pakistan.’ 25 January. Geneva: UNHCR. ——— (1989b) ‘Interim Report on Repatriation Planning for Afghan Refugees.’ 31 September. Geneva: UNHCR. Epstein, Gil S. (2008) ‘Herd and Network Effects in Migration Decision-Making.’ Journal of Ethnic and Migration Studies 34: 567–583. Erikson, Kai T. (1976) Everything in Its Path: The Destruction of Community in the Buffalo Creek Flood. New York: Simon & Schuster. Essed, Philomena, Georg Frerks, and Joke Schrijvers (2004) ‘Introduction: Refugees, Agency and Social Transformation.’ In Refugees and the Transformation of Societies: Agency, Policies, Ethics and Politics, edited by Philomena Essed, Georg Frerks, and Joke Schrijvers. Oxford: Berghahn Books, pp. 1–16. Fagen, Patricia Weiss (1996) ‘The Meaning and Modes of Reintegration.’ Geneva: World Bank/UNHCR.

Bibliography 205 ——— (2003) ‘Post-Conflict Reintegration and Reconstruction: Doing it Right Takes a While.’ In Problems of Protection: The UNHCR, Refugees and Human Rights, edited by Niklaus Steiner, Mark Gibney, and Gil Loescher. London: Routledge, pp. 197–224. ——— (2006) ‘UNHCR and Repatriation.’ In Palestinian Refugee Repatriation: Global Perspectives, edited by Michael Dumper. London: Routledge. Faist, Thomas (1997a) ‘The Crucial Meso-Level.’ In International Migration, Immobility and Development: Multidisciplinary Perspectives, edited by Grete Brochmann, Thomas Hammer, Kristof Tamas, and Thomas Faist. New York: Berg, pp. 187–217. ——— (1997b) ‘From Common Questions to Common Concepts.’ In International Migration, Immobility and Development: Multidisciplinary Perspectives, edited by Tomas Hammar, Grete Brochmann, Kristof Tamas, and Thomas Faist. Oxford: Berg, pp. 247–276. Farr, Grant M. (2001) ‘Afghanistan: Displaced in a Devastated Country.’ In Caught between Borders: Response Strategies of the Internally Displaced, edited by Marc Vincent and Birgitte Refslund Sørensen. London: Pluto Press, pp. 117–137. Faubert, Carrol, Abdullah Mojadeddi, and Abdul Hai Sofizada (2005) ‘Repatriation and Reintegration: An Appraisal of Progress in Afghanistan.’ March 2005. Kabul: UNHCR. Feinstein International Famine Center (2004) ‘Human Security and Livelihoods of Rural Afghans, 2002–2003.’ June 2004. Medford: Tufts University. Feld, Scott L. (1981) ‘The Focused Organization of Social Ties.’ American Journal of Sociology 86: 1015–1035. Fielden, Matthew B. (1998) ‘The Geopolitics of Aid: The Provision and Termination of Aid to Afghan refugees in North-West Frontier Province, Pakistan.’ Political Geography 17: 459–487. Fischer, Claude S. (1982) To Dwell among Friends: Personal Networks in Town and City. Chicago: University of Chicago Press. Foley, Pat (2005) ‘Shelter Programme Monitoring and Evaluation; Final Report.’ 28 February. Kabul: United Nations High Commissioner for Refugees. Foley, Paul (1991) ‘Heijrat: The Migration of Afghan Refugees to Pakistan, 1978–1990.’ Department of Geography. Honolulu: University of Hawaii. Freeman, Linton C., A. C. Freeman, and A. G. Michaelson (1989) ‘How Humans See Social Groups: A Test of the Sailer-Gaulin Models.’ Journal of Quantitative Anthropology 1: 229–238. Freeman, Linton C., A. K. Romney, and S. C. Freeman (1987) ‘Cognitive Structure and Informant Accuracy.’ American Anthropologist 89: 310–325. Freidberg, Susanne (2001) ‘On the Trail of the Global Green Bean: Methodological Considerations in Multi-Site Ethnography.’ Global Networks 1: 353–368. Frelick, Bill (1995) ‘Safe Haven: Safe for Whom?’ World Refugee Survey: 18–27. ——— (2001) ‘Afghans in Iran Face even Harsher Treatment.’ In Afghan Refugees Shunned and Scorned, edited by USCR. Washington, DC: U.S. Committee for Refugees, pp. 21–22. Freudenburg, William R., and Timothy R. Jones (1991) ‘Attitudes and Stress in the Presence of Technological Risk: A Test of the Supreme Court Hypothesis.’ Social Forces 69: 1143–1168. Friedkin, Noah E. (2004) ‘Social Cohesion.’ Annual Review of Sociology 30: 409–425.

206 Bibliography Fuglerud, Øyvind (2004) ‘Space and Movement in the Sri Lankan Conflict.’ In Refugees and the Transformation of Societies: Agency, Policies, Ethics and Politics, edited by Philomena Essed, Georg Frerks, and Joke Schrijvers. Oxford: Berghahn Books, pp. 42–52. Ghosh, Bimal (2000) ‘Return Migration: Reshaping Policy Approaches.’ In Return migration: Journey of Hope or Despair?, edited by Bimal Ghosh. Geneva: International Organization for Migration, pp. 181–226. Giustozzi, Antonio (2000) War, Politics and Society in Afghanistan, 1978–1992. Washington, DC: Georgetown University Press. ——— (2003) ‘Respectable Warlords? The Politics of State-Building in Post-Taleban Afghanistan.’ Working Papers, September 2003. London: Crisis States Programme, London School of Economics. ——— (2009) Empires of Mud: Wars and Warlords in Afghanistan. London: Hurst & Co. Gizabi, Akram (2006) ‘Bajaur: Tribe and Custom Continue to Protect al-Qaeda.’ Global Terrorism Analysis, 18 January 2006. Glatzer, Bernt (1990) ‘Return to Kunar? An Enquiry among Refugees from Kunar in Bajaur.’ March 1990. Peshawar: DACAAR. ——— (1992) ‘From Refugee to Resident: Effects of Aid on Repatriation.’ In From Aid during Times of War to Aid for Reconstruction and Development, edited by E. Eide and T. Skaufjord. Peshawar: Norwegian Afghanistan Committee, pp. 161–170. Glazebrook, Diana, and Mohammad Jalal Abbasi-Shavazi (2007) ‘Being Neighbours to Imam Reza: Pilgrimage Practices and Return Intentions of Hazara Afghan Refugees Living in Mashad, Iran.’ Iranian Studies 40: 187–201. Gmelch, George (2004) ‘West Indian Migrants and Their Rediscovery of Barbados.’ In Coming Home? Refugees, Migrants and Those Who Stayed Behind, edited by Ellen Oxfeld and Lynellyn D. Long. Philadelphia: University of Pennsylvania Press, pp. 206–223. Gould, Roger V. (1991) ‘Multiple Networks and Mobilization in the Paris Commune, 1871.’ American Sociological Review 56: 716–729. ——— (1993) ‘Collective Action and Network Structure.’ American Sociological Review 58: 182–196. Grandi, Fillippo (2002) ‘Hope on the Brink.’ Forced Migration Review 13: 11–13. Granovetter, Marc (1995) ‘The Economic Sociology of Firms and Entrepreneurs.’ In The Economic Sociology of Immigration: Essays on Networks, Ethnicity and Entrepreneurship, edited by Alejandro Portes. New York: Russel Sage Foundation, pp. 128–165. Granovetter, Mark (1973) ‘The Strength of Weak Ties.’ American Journal of Sociology 81: 1287–1303. ——— (1978) ‘Threshold Models of Collective Behavior.’ American Journal of Sociology 83: 1420–1443. ——— (1983) ‘The Strength of Weak Ties: A Network Theory Revisited.’ Sociological Theory 1: 201–233. Grare, Frédéric (2003) ‘The Geopolitics of Afghan Refugees in Pakistan.’ In Refugee Manipulation: War Politics and the Abuse of Human Suffering, edited by Stephen John Stedman and Fred Tanner. Washington, DC: Brookings Institution Press, pp. 57–94. Grasmuck, Sherri, and Patricia R. Pessar (1991) Between Two Islands: Dominican International Migration. Berkeley: University of California Press. Grevemeyer, Jan Heeren (1988) ‘Die Afghanen: Widerstand und Flucht.’ In Flucht und Asyl: Berichte uber Fluchtlingsgruppen. Berlin: Edition Parabolis, pp. 107–112.

Bibliography 207 Griffin, Michael (2001) Reaping the Whirlwind: The Taliban Movement in Afghanistan. London: Pluto Press. Grønhaug, Reidar (1978) ‘Scale as a Variable in Analysis: Fields in Social Organization in Herat, Northwest Afghanistan.’ In Scale and Social Organization, edited by Fredrik Barth. Oslo: Universitetsforlaget, pp. 78– 121. ——— (1988) ‘Herati People Reconstructing Their Life World: Action Groups in Large Scale Arenas in Peace and War.’ Bergen: Institute of Social Anthropology, University of Bergen. Guillo, Alain, Jean-Jose Puig, and Olivier Roy (1983) ‘La guerre en Afghanistan: Modifications des deplacements traditionnels de populations et emergence de nouveaux types de circulations.’ Ethnologica Helvetica 7: 139–153. Gurak, Douglas T., and Fe Caces (1992) ‘Migration Networks and the Shaping of Migration Systems.’ In International Migration Systems: A Global Approach, edited by Mary M. Kritz, Lin Lean Lim, and Hanna Zlotnik. Oxford: Clarendon Press, pp. 150–176. Göbel, Barbara (1998) ‘Risk, Uncertainty, and Economic Exchange in a Pastoral Community in the Andeas Highlands (Huancar, N.W. Argentina).’ In Kinship, Networks and Exchange, edited by Thomas Scweizer and Douglas R. White. Cambridge: Cambridge University Press, pp. 158–177. Habibi, Gulbadin, and Pamela Hunte (2006) ‘Afghan Returnees from NWFP, Pakistan to Nangarhar Province.’ Kabul: Afghanistan Research and Evaluation Unit. Hammond, Laura (1999) ‘Examining the Discourse of Repatriation: Towards a More Proactive Theory of Return Migration.’ In The End of the Refugee Cycle: Refugee Repatriation and Reconstruction, edited by Richard Black and Khalid Koser. Oxford: Berghahn Books, pp. 227–244. ——— (2004) ‘Tigrayan Returnees’ Notions of Home: Five Variations over a Theme.’ In Homecomings: Unsettling Paths of Return, edited by Fran Markowitz and Anders H. Stefansson. Lanham: Lexington Books, pp. 36–53. Hanifi, Shah Mahmoud (2006) ‘Material and Social Remittances to Afghanistan.’ In Converting Migration Drains into Gains Harnessing the Resources of Overseas Professionals, edited by Clay Wescott and Jennifer Brinkerhoff. Manila: Asian Development Bank, pp. 98–126. Hannerz, Ulf (2003) ‘Being there. . . and there. . . and there! Reflections on MultiSite Ethnography.’ Ethnography 4: 201–216. Hansen, Art (1981) ‘Refugee Dynamics: Angolans in Zaire 1966–1972.’ International Migration Review 15: 175–194. ——— (1990) ‘Refugee Self-Settlement versus Settlement on Government Schemes: The Long-term Consequences for Security, Integration and Economic Development of Angolan Refugees (1966–1989) in Zambia.’ Geneva: UNRISD. Harding, Luke (2001) ‘Opposition Warlords Prepare for Their Big Chance.’ Guardian, 19 September 2001. Harpviken, Kristian Berg (1996) ‘Political Mobilization among the Hazara of Afghanistan: 1978–1992.’ Oslo: Department of Sociology, University of Oslo. ——— (1997a) ‘Transcending Traditionalism: The Emergence of Non-State Military Formations in Afghanistan.’ Journal of Peace Research 34: 271–287. ——— (1997b) ‘War and Change in Afghanistan: Reflections on Research Priorities.’ In Language-Culture-Society: Contributions from the Nordic Central Asian Studies Symposium 1996, edited by M. Juntunen, S. Rosén, and B. Schlyter. London: Kegan Paul.

208 Bibliography ——— (1998) ‘The Hazara of Afghanistan: The Thorny Path towards Political Unity: 1978–1992.’ In Post-Soviet Central Asia, edited by Touraj Atabaki and John O’Kane. London: Tauris, pp. 177–203. ——— (2003) ‘Repatriating Afghans and Their Links to Iran.’ Afghan Refugees in Iran, Roundtable by the Chr. Michelsen Institute (CMI), the International Peace Research Institute, Oslo (PRIO) and the Institute for International and Political Studies (IPIS), 11 October 2003. Teheran: Unpublished manuscript. ——— (2006) ‘Conceptualizing “Forced Migration”.’ International Peace Research Institute, Oslo. ——— (2008) ‘Militant Homecoming in Afghanistan and Beyond.’ In Global Migration and Transnational Politics Working Paper, edited by Terrence Lyons and Peter Mandaville. Arlington, VA: Center for Global Studies, George Mason University. ——— (2009) ‘The Return of the Refugee Warrior: Migration and Armed Resistance in Herat.’ In Understanding Afghanistan, edited by Angela Schlenkhoff and Ceri Oeppen. London and New York: Hurst and Colombia University Press. ——— (forthcoming, ‘Understanding Warlordism: Three Biographies from Afghanistan’s Southeastern Areas.’ In Violence in the Post-Conflict State, edited by Astri Suhrke. Harpviken, Kristian Berg, and Bernt A. Skåra (2003) ‘Mine Action and Peacebuilding: Exploring the Relationship.’ Third World Quarterly 24: 809–822. Harpviken, Kristian Berg, Muhammad Suleman, and Merete Taksdal (2001) ‘Strengthening the Self-reliance of Returnee Communities: The Enjil Community Development Programme, Herat Province, Afghanistan.’ Herat: Ockenden International. Harrel-Bond, Barbare E. (1998) ‘Refugee Studies at Oxford: “Some History”.’ The Growth of Forced Migration: New Directions in the Study of Forced Migration, 24–26 March 1998. Oxford: Refugee Studies Programme, University of Oxford. Harrell-Bond, Barbara E., and E. Voutira (1992) ‘Anthropology and the Study of Refugees.’ Anthropology Today 8: 6–10. Harvey, Paul (1998) ‘Rehabilitation in Complex Political Emergencies: Is Rebuilding Civil Society the Answer?’ Disasters 22: 200–217. Hathaway, James C. (2007) ‘Forced Migration Studies: Could We Agree Just to “Date”?’ Journal of Refugee Studies 20: 349–369. Haug, Sonja (2008) ‘Migration Networks and Migration Decision-Making.’ Journal of Ethnic and Migration Studies 34: 585–605. Hechter, Michael (1987) ‘Principles of Group Solidarity.’ Berkeley: University of California Press. Heffer, Jonathan (1997) ‘Shelter and Repatriation: A Study into the Issue of Shelter in the Repatriation of Refugees to Nangarhar and Laghman, Afghanistan.’ Jalalabad: Mission d’Aide au Developpement des Economies Rurales en Afghanistan. Heimer, Carol A. (1985) ‘Social Structure, Psychology, and the Estimation of Risk.’ Annual Review of Sociology 14: 491–519. Heimerl, Daniela (2005) ‘The Return of Refugees and Internally Displaced Persons: From Coercion to Sustainability?’ International Peacekeeping 12: 377–390. Hein, Jeremy (1993) ‘Refugees, Immigrants and the State.’ Annual Review of Sociology 19: 43–59. Hollifield, James F. (2000) ‘The Politics of International Migration: How Can We “Bring the State Back In”?’ In Migration Theory: Talking across Disciplines, edited by Caroline B. Brettel and James F. Hollifield. New York: Routledge, pp. 137–186.

Bibliography 209 Horst, Cindy (2006a) ‘Refugee Livelihoods: Continuity and Transformations.’ Refugee Survey Quarterly 25: 6–22. ——— (2006b) Transnational Nomads: How Somalis Cope with Refugee Life in the Dadaab Camps of Kenya. New York: Berghahn. Horst, Heather A. (2006) ‘The Blessings and Burdens of Communication: Cell Phones in Jamaican Transnational Social Fields.’ Global Networks 6: 143–159. HRW (1996) ‘Tajik Refugees in Northern Afghanistan: Obstacles to Repatriation.’ New York: Human Rights Watch/Helsinki. ——— (2001) ‘Fatally Flawed: Cluster Bombs and Their Use by the United States in Afghanistan.’ December 2002. Washington, DC: Human Rights Watch. ——— (2002a) ‘“All Our Hopes Are Crushed”: Violence and Repression in Western Afghanistan.’ November. New York: Human Rights Watch. ——— (2002b) ‘“We Want to Live as Humans”: Repression of Women and Girls in Western Afghanistan.’ December. New York: Human Rights Watch. Hubert, Don (2004) ‘An Idea That Works in Practice.’ Security Dialogue 35: 351–352. Hug, Heiner (1987) ‘Iran’s Indoctrination of Afghan Refugees.’ Swiss Review of World Affairs XXXVll: 6–7. Hugo, Graeme (2005) ‘Migrants in Society: Diversity and Cohesion.’ September. Geneva: Global Commission on International Migration. Hugo, Graeme J. (1981) ‘Village-Community Ties, Village Norms, and Ethnic and Social Networks: A Review of Evidence from the Third World.’ In Migration Decision Making: Multidisciplinary Approaches to Microlevel Studies in Developed and Developing Countries, edited by Gordon F. De Jong and Robert W. Gardner. New York: Pergamon Press, pp. 186–224. Hunter, Shireen T. (1987) ‘The Soviet Union and the Islamic Republic of Iran.’ In Soviet-American Relations with Pakistan, Iran and Afghanistan, edited by H. Malik. Basingstoke: Macmillan, pp. 244–266. Hyndman, Jennifer (2003) ‘Preventive, Palliative, or Punitive? Safe Spaces in BosniaHerzegovina, Somalia, and Sri Lanka.’ Journal of Refugee Studies 16: 167–185. Jacobsen, Karen (2002) ‘Livelihoods in Conflict: The Pursuit of Livelihoods by Refugees and the Impact on the Human Security of Host Communities.’ International Migration 40: 95–123. ——— (2005) The Economic Life of Refugees. Bloomfield, CT: Kumarian Press. Jalali, Ali Ahmad, and Lester W. Grau (2001) Afghan Guerrilla Warfare: In the Words of the Mujahideen Fighters. London: Compendium. Jamal, Arafat, and Elca Stigter (2002) ‘Real-time Evaluation of UNHCR’s Response to the Afghanistan Emergency: Bulletin No. 3.’ 31 May 2002. Geneva: UNHCR. Jamshidia, G. R., and Y. Ali Babaie (2002) ‘Determinants of Afghan Migrants in Iran: Case Study Golshahr Mashhad.’ Journal of Social Sciences [Iran] 20: 71–90. Jamshidia, G. R., and M. Anbari (2004) ‘Social Belonging and its Impact on Repatriation of Afghan Refugees in Iran.’ Journal of Social Sciences [Iran] 23: 43–68. Janata, Alfred (1982) ‘Paktia-Paschtunen: Daheim und im Pakistanischen Exil.’ In Forschungen in und über Afghanistan: Situation der wissenschaflichen Erforschung Afghanistans and Folgen der gegenwärtigen politischen lage, edited by Siegmar-W. Breckle and Clas. M. Naumann. Bielefeld: Deutsches Orient-Institut, pp. 119–129. Janzen, John (2004) ‘Illusions of Home in the Story of a Rwandan Refugee’s Return.’ In Coming Home? Refugees, Migrants and Those Who Stayed Behind, edited by Ellen Oxfeld and Lynellyn D. Long. Philadelphia: University of Pennsylvania Press, pp. 19–33.

210 Bibliography Jazayery, Leila (2002) ‘The Migration-Development Nexus: Afghanistan Case Study.’ International Migration 40: 231–253. Johnson, Chris, and Jolyon Leslie (2004) Afghanistan: The Mirage of Peace. London: Zed. Johnson, Jeffrey C. (1990) Selecting Ethnographic Informants. Newbury Park, CA: Sage. Joly, Danièle (2002) ‘Odyssean and Rubicon Refugees: Toward a Typology of Refugees in the Land of Exile.’ International Migration 40: 3–23. Juergensen, Olaf Tataryn (2000) ‘Repatriation as Peacebuilding and Reconstruction: The Case of Northern Mozambique, 1992–1995.’ October 2000. Geneva: UNCHR. Kasdan, Leonard (1965) ‘Family Structure, Migration and the Entrepreneur.’ Comparative Studies in Society and History 7: 345–357. Kelly, M. Patricia Fernández (1995) ‘Social and Cultural Capital in the Urban Ghetto: Implications for the Economic Sociology of Immigration.’ In The Economic Sociology of Immigration: Essays on Networks, Ethnicity and Entrepreneurship, edited by Alejandro Portes. New York: Russel Sage Foundation, pp. 213–247. Khawaja, Marwan (2003) ‘Migration and the Reproduction of Poverty: The Refugee Camps in Jordan.’ International Migration 41: 27–57. Khosrokhavar, Farhad, and Olivier Roy (1996) Iran: comment sortir d’une révolution religieuse. Paris: Seuil. King, Russel (2000) ‘Generalizations from the History of Return Migration.’ In Return Migration: Journey of Hope or Despair?, edited by Bimal Ghosh. Geneva: International Organization for Migration, pp. 7–55. Kivisto, Peter (2001) ‘Theorizing Transnational Immigration: A Critical Review of Current Efforts.’ Ethnic and Racial Studies 24: 549–577. Kjellman, Kjell Erling, and Kristian Berg Harpviken (2009) ‘The Embedded State: Civil Society and the State-Centred Peacebuilding Agenda.’ In Civil Society and Peacebuilding: Concepts, Cases and Lessons, edited by Thania Paffenholz. Boulder, CO: Lynne Rienner. Klijn, Floortje, and Adam Pain (2007) ‘Finding the Money: Informal Credit Practices in Rural Afghanistan.’ Synthesis Paper Series, June. Kabul: Afghanistan Research and Evaluation Unit. Knoke, David (1990) Political Networks: The Structural Perspective. Cambridge: Cambridge University Press. ——— (1994) ‘Networks of Elite Structure and Decision Making.’ In Advances in Social Network Analysis, edited by Stanley Wasserman and Joseph Galaskiewicz. Thousand Oaks: Sage. Knowles, Michael (1992) Afghanistan: Trends and Prospects for Refugee Repatriation. Washington, DC: Refugee Policy Group. Knudsen, John Chr. (1995) ‘When Trust Is on Trial: Negotiating Refugee Narratives.’ In Mistrusting Refugees, edited by E. Valentine Daniel and John Chr. Knudsen. Berkeley: University of California Press, pp. 13–35. Koehn, Peter H. (1991) Refugees from Revolution: U.S. Policy and Third-World Migration. Boulder: CO: Westview Press. Kok, Frederik (2002) ‘Afghanistan.’ In Internally Displaced People: A Global Survey (2nd edition), edited by Stacey White. London: Earthscan, pp. 104–109. Koser, Khalid (1993) ‘Repatriation and Information: A Theoretical Model.’ In Geography and Refugees: Patterns and Processes of Change, edited by Richard Black and Vaughan Robinson. London and New York: Belhaven Press, pp. 171–184.

Bibliography 211 ——— (1997) ‘Information and Repatriation: The Case of Mozambican Refugees in Malawi.’ Journal of Refugee Studies 10: 1–18. ——— (2000) ‘Return, Readmission and Reintegration: Changing Agendas, Policy Frameworks and Operational Programmes.’ In Return Migration: Journey of Hope or Despair?, edited by Bimal Ghosh. Geneva: International Organization for Migration, pp. 57–99. ——— (2005) ‘Irregular Migration, State Security and Human Security.’ September. Geneva: Global Commission on International Migration. ——— (2007) ‘Refugees, Transnationalism and the State.’ Journal of Ethnic and Migration Studies 33: 233–254. ——— (2008) ‘Why Migrant Smuggling Pays.’ International Migration 46: 3–26. Koser, Khalid, and Richard Black (1999) ‘The End of the Refugee Cycle?’ In The End of the Refugee Cycle: Refugee Repatriation and Reconstruction, edited by Richard Black and Khalid Koser. New York: Berghahn Books, pp. 2–17. Kreps, G. A. (1984) ‘Sociological Inquiry and Disaster Research.’ Annual Review of Sociology 10: 309–330. Krokfors, Christer (1995) ‘Poverty, Environmental Stress and Culture as Factors in African Migrations.’ In The Migration Experience in Africa, edited by Jonathan Baker and Tade Akin Aida. Uppsala: The Nordic Africa Institute. Kronenfeld, Daniel A. (2008) ‘Afghan Refugees in Pakistan: Not All Refugees, Not Always in Pakistan, Not Necessarily Afghan?’ Journal of Refugee Studies 21: 43–63. Krznaric, Roman (1997) ‘Guatemalan Returnees and the Dilemma of Political Mobilization.’ Journal of Refugee Studies 10: 61–78. Kunz, Egon F. (1973) ‘The Refugee in Flight: Kinetic Models and Forms of Displacement.’ International Migration Review 7: 125–146. Landau, Loren B. (2007) ‘Can We Talk and Is Anybody Listening? Reflections on IASFM 10, “Talking across Borders: New Dialogues in Forced Migration”.’ Journal of Refugee Studies 20: 335–348. Laws, Sopie, Caroline Harper, and Rachel Marcus (2003) Research for Development: A Practical Guide. London: Sage. Leader, Nicholas, and Mohammed Haneef Atmar (2004) ‘Political Projects: Reform, Aid and the State in Afghanistan.’ In Nation-Building Unraveled? Aid, Peace and Justice in Afghanistan, edited by Antonio Donini, Norah Niland, and Karin Wermester. Bloomfield, CT: Kumarian, pp. 166–186. Lee, Raymond M. (1995) Dangerous Fieldwork. London: Sage. Levitt, Peggy (2001) The Transnational Villagers. Berkeley: University of California Press. Lischer, Sarah Kenyon (2005) Dangerous Sanctuaries: Refugee Camps, Civil Wars and the Dilemmas of Humanitarian Aid. Ithaca: Cornell University Press. Loescher, Gil (1992) ‘Refugee Movements and International Security.’ London: International Institute for Strategic Studies. Lomnitz, Larissa Adler (1977) Networks and Marginality: Life in a Mexican Shantytown. New York: Academic Press. Long, Lynellyn D., and Ellen Oxfeld (Eds) (2004) Coming Home? Refugees, Migrants and Those Who Stayed Behind. Philadelphia: University of Pennsylvania Press. Lorrain, F., and H. C. White (1971) ‘Structural Equivalence of Individuals in Social Networks.’ Journal of Mathematical Sociology 1: 49–80. Lubkemann, Stephen C. (2004) ‘Situating Migration in Wartime and Post-war Mozambique: A Critique of “Forced Migration” Research.’ In Categories and

212 Bibliography Contexts: Anthropological and Historical Studies in Critical Demography, edited by Simon Szreter, Hania Sholkamy, and A. Dharmalingam. Oxford: Oxford University Press, pp. 371–400. ——— (2005) ‘Migratory Coping in Wartime Mozambique: An Anthropology of Violence and Displacement in “Fragmented Wars”.’ Journal of Peace Research 42: 493–508. ——— (2008) Culture in Chaos: An Anthropology of the Social Condition in War. Chicago: University of Chicago Press. Lukic, Vesna, and Vladimir Nikitovic (2004) ‘Refugees from Bosnia and Herzegovina in Serbia: A Study of Refugee Selectivity.’ International Migration 42: 85–110. Mack, Andrew (Ed.) (2005) Human Security Report 2005: War and Peace in the 21st Century. Oxford: Oxford University Press. Macy, Michael W., and Andreas Flache (1995) ‘Beyond Rationality in Models of Choice.’ Annual Review of Sociology 21: 73–91. Mahoney, James, and Dietrich Rueschmeyer (Eds) (2003) Comparative Historical Analysis in the Social Sciences. Cambridge: Cambridge University Press. Maimbo, Samuel Munzele (2003) ‘The Money Exchange Dealers of Kabul: A Study of the Hawala System in Afghanistan.’ Washington, DC: The World Bank. Maimbo, Samuel Munzele, and Nikos Passas (Eds) (2005) Remittances: Development Impact and Future Prospects. Washington, DC: The World Bank. Maley, William (1989) ‘Afghan Refugees: From Diaspora to Repatriation.’ In Refugees in the Modern World, edited by Amin Saikal. Canberra: The Department of International Relations, pp. 17–44. Malkki, Lisa (1995a) ‘Refugees and Exile: From “Refugee Studies” to the National Order of Things.’ Annual Review of Anthropology 24: 495–523. ——— (1995b) Purity and Exile: Violence, Memory and National Cosmology among the Hutu Refugees in Tanzania. Chicago: University of Chicago Press. Marcus, George E. (1995) ‘Ethnography in/of the World System: The Emergence of Multi-Sited Ethnography.’ Annual Review of Anthropology 24: 95–117. Marini, Margaret Mooney, and Burton Singer (1988) ‘Causality in the Social Sciences.’ Sociological Methodology 18: 347–409. Markowitz, Fran (2004) ‘The Home(s) of Homecomings.’ In Homecomings: Unsettling Paths of Return, edited by Fran Markowitz and Anders H. Stefansson. Lanham: Lexington Books, pp. 21–33. Markowitz, Fran, and Anders H. Stefansson (Eds) (2004) Homecomings: Unsettling Paths of Return. Lanham: Lexington Books. Marsden, Peter (1994) ‘Absorptive Capacity in Districts of Herat.’ British Refugee Council. ——— (1996a) ‘Living in Exile: Report on a Study of Economic Coping Strategies among Afghan Refugees in Pakistan.’ December. London: British Agencies Afghanistan Group. ——— (1996b) ‘Report on a Study on Coping Strategies of Refugees from and Returnees to Afghanistan.’ London: British Agencies Afghanistan Group. ——— (1997) ‘Return and Reconstruction: Report on a Study of Economic Coping Strategies among Farmers in Farah Province, Afghanistan.’ London: British Agencies Afghanistan Group. ——— (1998) ‘Exile for a Cause: The Plight of Refugees.’ In Afghanistan, edited by Edward Girardet and Jonathan Walter. Geneva: Crosslines Communications, pp. 40–47.

Bibliography 213 ——— (1999) ‘Repatriation and Reconstruction: The Case of Afghanistan.’ In The End of the Refugee Cycle: Refugee Repatriation and Reconstruction, edited by Richard Black and Khalid Koser. New York: Berghahn Books, pp. 56–68. Marsden, Peter, and David Turton (2004) Preliminary Study of Assistance to Internally Displaced in Afghanistan. Copenhagen: Danida. Marsden, Peter V. (1990) ‘Network Data and Measurement.’ Annual Review of Sociology 16: 435–463. Marshall, Andrew (2002) ‘Aid Agencies Face Dilemma over Displaced Afghans.’ Reuters, 25 February. Marx, Emanuel (1990) ‘The Social World of Refugees: A Conceptual Framework.’ Journal of Refugee Studies 3: 189–203. Massey, Douglas S., and Felipe García España (1987) ‘The Social Process of International Migration.’ Science 237: 733–738. Massey, Douglas S., and Kristin E. Espinosa (1997) ‘What’s Driving Mexico-U.S. Migration? A Theoretical, Empirical and Policy Analysis.’ American Journal of Sociology 102: 939–999. Massey, Douglas, Rafael Alarcón, Jorge Durand, and Humberto Gonzales (1987) Return to Aztlan: The Social Process of International Migration from Western Mexico. Berkeley: University of California Press. Massey, Douglas S., Joaquin Arango, Graeme Hugo, Ali Kouaouci, Adela Pellegrino, and J. Edward Taylor (1993) ‘Theories of International Migration: A Review and Appraisal.’ Population and Development Review 19: 431–466. ——— (1998) Worlds in Motion: Understanding International Migration at the End of the Millennium. Oxford: Clarendon Press. McCallister, Lynne, and Claude S. Fischer (1983) ‘A Procedure for Surveying Personal Networks.’ In Applied Network Analysis: A Methodological Introduction, edited by Ronald S. Burt and Michael J. Minor. Beverly Hills: Sage, pp. 75–88. Merton, Robert K. (1984) ‘Socially Expected Durations: A Case Study of Concept Formation in Sociology.’ In Conflict and Consensus: In Honor of Lewis A. Coser, edited by Walter M. Powell and Richard Robbins. New York: Free Press, pp. 262–283. Michel, Edith F. Kauffer (2002) ‘Leadership and Social Organisation: The Integration of Guatemalan Refugees in Campeche, Mexico.’ Journal of Refugee Studies 15: 359–387. Milliken, Jennifer, and Keith Krause (2003) ‘State Failure, State Collapse and State Reconstruction: Concepts, Lessons and Strategies.’ In State Failure, Collapse and Reconstruction, edited by Jennifer Milliken. Malden, MA: Blackwell, pp. 1–21. Mjøset, Lars (2009) ‘The Contextualist Approach to Social Science Methodology.’ In The Sage Handbook of Case-Based Methods, edited by David Byrne and Charles C. Ragin. London: Sage. Monsutti, Alessandro (2004a) ‘Cooperation, Remittances and Kinship among the Hazaras.’ Iranian Studies 37: 220–240. ——— (2004b) Guerres et migrations: Réseaux sociaux et stratégies économiques des Hazaras d’Afghanistan. Neuchâtel: Institut d’ethnologie, Université de Neuchâtel. ——— (2005) War and Migration: Social Networks and Economic Strategies of the Hazaras of Afghanistan. London: Routledge. Morren, George E. B. (1983) ‘The Bushmen and the British: Problems of Identification of Drought and Response to Draught.’ In Interpretations of Calamity, edited by Kenneth Hewitt. New York: Allen & Unwin, pp. 284–297.

214 Bibliography Morton, John (1992a) ‘The Socio-Economic Status of Afghan Refugees in Pakistan: An Overview.’ May. Islamabad: UNHCR. ——— (1992b) ‘Socio-Economic Surveys of Male Afghan Refugees in the Refugee Villages of NWFP.’ May. Peshawar: UNHCR. ——— (1992c) ‘Socio-Economic Surveys of Male Afghan Refugees in the Refugee Villages of NWFP: Final Report.’ May. Peshawar: UNHCR. ——— (1994) ‘Afghan Refugees in Pakistan: Issues of Employment, Repatriation and Assistance.’ Islamabad: UNHCR. Moscovici, S. (1985) ‘Social Influence and Conformity.’ In The Handbook of Social Psychology, edited by Gardner Lindzey and Elliott Aronsen. New York: Random House. Muggah, Robert (2003) ‘A Tale of Two Solitudes: Comparing Conflict and Development-Induced Internal Displacement and Involuntary Resettlement.’ International Migration 41: 5–31. Murdoch, Steve (2005) Network North: Scottish Kin, Commercial and Covert Associations in Northern Europe, 1603–1746. Leiden: Brill. Muzaffar, Saehar A. F. (2001) ‘Practicable Ideals? A Proposal for Revitalizing the Rights of Forced Migrants in Islam.’ Journal of Refugee Studies 14: 250–275. Nadig, Aninia (2001) ‘The High Level Working Group on Asylum and Migration and its Action Plan for Afghanistan.’ Journal of Refugee Studies 14: 428–435. Nee, Victor, and Herbert Y. Wong (1985) ‘Asian American Socioeconomic Achievement: The Strength of the Family Bond.’ Sociological Perspective 28: 281–306. Niland, Norah (2004) ‘Justice Postponed: The Marginalization of Human Rights in Afghanistan.’ In Nation-Building Unraveled? Aid, Peace and Justice in Afghanistan, edited by Antonio Donini, Norah Niland, and Karin Wermester. Bloomfield, CT: Kumarian, pp. 61–82. NPO/RRAA (1995) ‘Participatory Assessment of Needs (PAN).’ Peshawar: Norwegian Project Office/Rural Rehabilitation Association for Afghanistan. Okamoto, Dina G., and Rima Wilkes (2008) ‘The Opportunities and Costs of Voice and Exit: Modelling Ethnic Group Rebellion and Emigration.’ Journal of Ethnic and Migration Studies 34: 347–369. Oliver-Smith, A. (Ed.) (1986) Natural Disasters and Cultural Response: Studies in Third World Societies. Virginia: College of William and Mary. ——— (1996) ‘Anthropological Research on Hazards and Disasters.’ Annual Review of Anthropology 25: 303–328. Oliver, Pamela E. (1993) ‘Formal Models of Collective Action.’ Annual Review of Sociology 19: 271–300. Olson, Mancur (1965) The Logic of Collective Action. Cambridge: Harvard University Press. Opel, Aftab (2005) ‘Bound for the City: A Study of Rural to Urban Labour Migration in Afghanistan.’ April 2005. Kabul: Afghanistan Research and Evaluation Unit. OSI (1998) ‘Tajikistan: Refugee Reintegration and Conflict Prevention.’ New York: Open Society Institute. Oxfeld, Ellen, and Lynellyn D. Long (2004) ‘Introduction: An Ethnography of Return.’ In Coming Home? Refugees, Migrants and Those Who Stayed Behind, edited by Ellen Oxfeld and Lynellyn D. Long. Philadelphia: University of Pennsylvania Press, pp. 1–15. Padgett, John F., and Christopher K. Ansell (1993) ‘Robust Action and the Rise of the Medici, 1400–1434.’ American Journal of Sociology 98: 1259–1319.

Bibliography 215 Pain, Adam (2008) ‘Opium Poppy and Informal Credit.’ October. Kabul: Afghanistan Research and Evaluation Unit. Pain, Adam, and Sue Lautze (2002) ‘Addressing Livelihoods in Afghanistan.’ Kabul: Afghanistan Research and Evaluation Unit. Paris, Roland (2001) ‘Human Security: Paradigm Shift of Hot Air?’ International Security 26: 87–102. Patton, Michael Quinn (1990) Qualitative Evaluation and Research Methods (2nd edition). Newbury Park: Sage. Pessar, Patricia R. (1997) ‘Introduction: New Approaches to Caribbean Emigration and Return.’ In Caribbean Circuits: New Directions in the Study of Caribbean Migration, edited by Patricia R. Pessar. New York: Center for Migration Studies. Phillips, James (2004) ‘Repatriation and Social Class in Nicaragua.’ In Coming Home? Refugees, Migrants, and Those Who Stayed Behind, edited by Lynellyn D. Long and Ellen Oxfeld. Philadelphia: University of Pennsylvania Press, pp. 150–169. Picou, J. Steven, Brent K. Marshall, and Duane A. Gill (2004) ‘Disaster, Litigation, and the Corrosive Community.’ Social Forces 82: 1497–1526. Piore, Michael J. (1979) Birds of Passage: Migrant Labor and Industrial Societies. Cambridge: Cambridge University Press. Poppelwell, Teresa (2007) ‘Afghanistan.’ Forced Migration Online Research Guide. Oxford: Refugee Studies Centre, University of Oxford. Portes, Alejandro (1995) ‘Economic Sociology and the Sociology of Immigration: A Conceptual Overview.’ In The Economic Sociology of Immigration: Essays on Networks, Ethnicity and Entrepreneurship, edited by Alejandro Portes. New York: Russel Sage Foundation, pp. 1–41. ——— (2001) ‘Introduction: The Debates and Significance of Immigrant Transnationalism.’ Global Networks 1: 181–193. Portes, Alejandro, and Julia Sensenbrenner (1993) ‘Embeddedness and Immigration: Notes on the Determinants of Economic Action.’ American Journal of Sociology 98: 1320–1350. Preston, Rosemary (1999) ‘Researching Repatriation and Reconstruction: Who Is Researching What and Why?’ In The End of the Refugee Cycle: Refugee Repatriation and Reconstruction, edited by Richard Black and Khalid Koser. New York: Berghahn Books, pp. 18–36. QSR (2002) ‘N6 (Non-numerical Unstructured Data Indexing Searching & Theorizing) Qualitative Data Analysis Program.’ Melbourne, Australia: Qualitative Solutions and Research. Quarantelli, E. L. (Ed.) (1978) Disasters: Theory and Research. London: Sage. Ranger, Terence (1994) ‘Studying Repatriation as Part of African Social History.’ In When Refugees Go Home, edited by Tim Allen and Hubert Morsink. Geneva: UN Research Institute for Social Development, pp. 279–294. Rasanayagam, Angelo (2003) Afghanistan: A Modern History. London: I. B. Tauris. Rashid, Ahmed (2008) Descent into Chaos: How the War against Islamic Extremism Is Being Lost in Pakistan, Afghanistan and Central Asia. London: Allen Lane. Richmond, Anthony H. (1993) ‘Reactive Migration: Sociological Perspectives on Refugee Movements.’ Journal of Refugee Studies 6: 7–24. Ritchey, N. P. (1976) ‘Explanations of Migration.’ In Annual Review of Sociology, edited by Alex Inkeles, James Coleman, and Neil J. Smelser. Palo Alto, CA: Annual Reviews, pp. 363–404.

216 Bibliography Rizvi, Gowher (1990) ‘The Afghan Refugees: Hostages in the Struggle for Power.’ Journal of Refugee Studies 3: 144–161. Robben, Antonius C. G. M., and Carolyn Nordstrom (1995) ‘The Anthropology and Ethnography of Violence and Sociopolitical Conflict.’ In Fieldwork under Fire: Contemporary Studies of Violence and Survival, edited by Carolyn Nordstrom and Antonius C. G. M. Robben. Berkeley: University of California Press, pp. 1–23. Roberts, Bryan R. (1995) ‘Socially Expected Durations and the Economic Adjustment of Immigrants.’ In The Economic Sociology of Immigration: Essays on Networks, Ethnicity and Entrepreneurship, edited by Alejandro Portes. New York: Russel Sage Foundation, pp. 42–86. Rosa, Eugene A., and William R. Freudenburg (2001) ‘Sociological Study of Risk.’ In International Encyclopedia of the Social and Behavioral Sciences, edited by Neil J. Smelser and Paul B. Baltes. Amsterdam: Elsevier, pp. 13356–13360. Rossi, Ino (1993) Community Reconstruction after an Earthquake: Dialectical Sociology in Action. Westport, CT/London: Praeger. Rostami-Povey, Elaheh (2007) ‘Afghan Refugees in Iran, Pakistan, the U.K. and the U.S. and Life after Return: A Comparative Gender Analysis.’ Iranian Studies 40: 241–261. Roy, Olivier (1986) Islam and Resistance in Afghanistan. Cambridge: Cambridge University Press. Rubin, Barnett R. (1996) ‘Afghanistan: The Forgotten Crisis.’ Refugee Survey Quarterly 15: 1–35. ——— (1988) ‘Prospects for Repatriation of Afghan Refugees.’ World Refugee Survey: 70–71. Saikal, Amin (1989) ‘The Regional Politics of the Afghan Crisis.’ In The Soviet Withdrawal from Afghanistan, edited by A. Saikal and W. Maley. Cambridge, New York: Cambridge University Press, pp. 52–66. Saito, Mamiko (2008a) ‘From Disappointment to Hope: Transforming Experiences of Young Afghans Returning “Home” from Pakistan and Iran.’ Briefing Paper Series, November. Kabul: Afghanistan Research and Evaluation Unit. ——— (2008b) ‘From Mohajer to Hamwatan: Afghans Return Home.’ Kabul: Afghanistan Research and Evaluation Unit. Saito, Mamiko, and Pamela Hunte (2007) ‘To Return or to Remain: The Dilemma of Second-Generation Afghans in Pakistan.’ Kabul: Afghanistan Research and Evaluation Unit. Salaff, Janet W., Eric Fong, and Wong Siu-lun (1999) ‘Using Social Networks to Exit Hong Kong.’ In Networks in the Global Village, edited by Barry Wellman. Boulder, CO: Westview, pp. 299–329. Schelling, Thomas C. (1978) Micromotives and Macrobehavior. New York: W. W. Norton. Schmeidl, Susanne (1998) ‘Causes of Forced Exodus: Five Principal Explanations in the Scholarly Literature and Six Findings from Empirical Research.’ In Whither Refugee? The Refugee Crisis: Problems and Solutions, edited by Alex P. Schmid. Leiden: Interdisciplinary Research Program on Root Causes of Human Rights Violations, pp. 5–43. ——— (2002) ‘(Human) Security Dilemmas: Long-term Implications of the Afghan Refugee Crisis.’ Third World Quarterly 23: 7–29. Schoenholtz, Andrew I. (2003) ‘Transatlantic Dialogue on Terrorism and International Migration.’ International Migration 41: 173–192.

Bibliography 217 Scott, John (1991) Social Network Analysis: A Handbook. London: Sage. Scudder, Thayer, and Elizabeth Colson (1982) ‘From Welfare to Development: A Conceptual Framework for the Analysis of Dislocated People.’ In Involuntary Migration and Resettlement: The Problems and Responses of Dislocated People, edited by Anthony Oliver-Smith and Art Hansen. Boulder, CO: Westview, pp. 267–287. Sedra, Mark (2003) ‘New Beginnings or Return to Arms? The Disarmament, Demobilization & Reintegration Process in Afghanistan.’ State Reconstruction and International Engagement in Afghanistan. Bonn: Centre for Development Research, Bonn University (ZEF) and the Crisis States Programme, London School of Economics (LSE). Shadid, Anthony (2002) ‘Victims of Circumstance.’ Middle East Report 32: 12–17. Shahrani, M. Nazif (1995) ‘Afghanistan’s Muhajirin (Muslim “Refugee–Warriors”): Politics of Mistrust and Distrust of Politics.’ In Mistrusting Refugees, edited by E. Valentine Daniel and John Chr. Knudsen. Berkley: University of California Press. Shain, Yossi, and Ravinatha P. Aryasinha (2006) ‘Spoilers or Catalysts? The Role of Diasporas in Peace Processes.’ In Challenges to Peacebuilding: Managing Spoilers during Conflict Resolution, edited by Edward Newman and Oliver Richmond. Tokyo: United Nations University Press. Shami, Seteney (1993) ‘The Social Implications of Population Displacement and Resettlement: An Overview with a Focus on the Middle East.’ International Migration Review 27: 4–32. ——— (1996) ‘Transnationalism and Refugee Studies: Rethinking Forced Migration and Identity in the Middle East.’ Journal of Refugee Studies 9: 3–25. Shanmugaratnam, Nadarajah, Ragnhild Lund, and Kristi Anne Stølen (Eds) (2003) In the Maze of Displacement: Conflict, Migration and Change. Kristiansand: Høyskoleforlaget. Silverman, David (2001). Interpreting Qualitative Data: Methods for Analysing Talk, Text and Interaction. London: Sage. Sinno, Abdelkader H. (2008) Organizations at War in Afghanistan and Beyond. Ithaca: Cornell University Press. Sliwinski, Marek (1988) ‘Afghanistan 1978– 87: War, Demography and Society.’ Central Asian Survey, Incidental Paper 6. ——— (1989a) ‘Afghanistan: The Decimation of a People.’ Orbis 33: 39–56. ——— (1989b) ‘On the Routes of “Hijrat”.’ Central Asian Survey 8: 63–93. Sluka, J. A. (1990) ‘Participant Observation in Violent Social Contexts.’ Human Organization 49: 114–126. Spaan, E. (1994) ‘Taikongs and Calos: The Role of Middlemen and Brokers in Javanese International Migration.’ International Migration Review 28: 93–113. Stack, Carol B. (1974) All Our Kin. New York: Harper & Row. Stedman, Stephen John (1997) ‘Spoiler Problems in Peace Processes.’ International Security 22: 5–53. ——— (2002) ‘Introduction.’ In Ending Civil Wars: The Implementation of Peace Agreements, edited by Stephen John Stedman, Donald Rothchild, and Elizabeth M. Cousens. Boulder, CO: Lynne Rienner, pp. 1–40. Stefansson, Anders H. (2004a) ‘Homecomings to the Future: From Diasporic Mythographies to Social Projects of Return.’ In Homecomings: Unsettling Paths of Return, edited by Fran Markowitz and Anders H. Stefansson. Lanham: Lexington Books, pp. 2–20.

218 Bibliography ——— (2004b) ‘Sarajevo Suffering: Homecoming and the Hierarchy of Homeland Hardship.’ In Homecomings: Unsettling Paths of Return, edited by Fran Markowitz and Anders H. Stefansson. Lanham: Lexington Books. Stein, Barry N., and Frederick C. Cuny (1991) ‘Repatriation under Conflict.’ World Refugee Study: 15–21. Stein, Barry N., Frederick C. Cuny, and Pat Reed (1995) Refugee Repatriation during Conflict: A New Conventional Wisdom. Dallas, TX: The Center for the Study of Societies in Conflict. Stepputat, Finn, and Ninna Nyberg Sørensen (2001) ‘The Rise and Fall of “Internally Displaced People” in the Central Peruvian Andes.’ Development and Change 32: 769–791. Stigter, Elca (2005a) ‘Transnational Networks and Migration from Faryab to Iran.’ February 2005. Kabul: Afghanistan Research and Evaluation Unit. ——— (2005b) ‘Transnational Networks and Migration from Herat to Iran.’ January 2005. Kabul: Afghanistan Research and Evaluation Unit. Stigter, Elca, and Alessandro Monsutti (2005) ‘Transnational Networks: Recognising a Regional Reality.’ April 2005. Kabul: Afghanistan Research and Evaluation Unit. Stokman, Frans N., and Patrick Doreian (1996) ‘Concluding Remarks.’ Journal of Mathematical Sociology 21: 197–199. Strand, Arne (2003) ‘Who’s Helping Who? NGO Coordination of Humanitarian Assistance.’ Department of Political Studies. York: University of York. Strand, Arne, Astri Suhrke, and Kristian Berg Harpviken (2004) ‘Afghan Refugees in Iran: From Refugee Emergency to Migration Management.’ 16 June 2004. Bergen/ Oslo: Chr. Michelsen Institute/International Peace Research Institute, Oslo. Stølen, Kristi Anne (2005) ‘Contradictory Notions of the State: Returned Refugees in Guatemala.’ In State Formation: Anthropological Perspectives, edited by Christian Krohn-Hansen and Knut G. Nustad. London: Pluto Press, pp. 142–167. Suhrke, Astri (1995a) ‘Analysing the Causes of Contemporary Refugee Flows.’ In Causes of International Migration, edited by Rob van Der Erf and Liesbeth Heering. Luxembourg: Office for Official Publications of the European Communities. ——— (1995b) ‘Refugees and Asylum in the Muslim World.’ In The Cambridge Survey of World Migration, edited by Robin Cohen. Cambridge: Cambridge University Press, pp. 457–460. ——— (2008) ‘A Contradictory Mission? NATO from Stabilization to Combat in Afghanistan.’ International Peacekeeping 15: 214–236. Suhrke, Astri, and Frank Klink (1987) ‘Contrasting Patterns of Asian Refugee Movements: The Vietnamese and Afghan Syndromes.’ In Pacific Bridges: The New Immigration from Asia and the Pacific Islands, edited by James T. Fawcett and Benjamin V. Carino. State Island, New York: Center for Migration Studies, pp. 85–102. Suhrke, Astri, Kristian Berg Harpviken, and Arne Strand (2002) ‘After Bonn: Conflictual Peacebuilding.’ Third World Quarterly 23: 875–891. ——— (2004) ‘Conflictual Peacebuilding: Afghanistan Two Years after Bonn.’ January 2002. Bergen: Chr. Michelsen Institute. Suitor, J. Jill, Barry Wellman, and David L. Morgan (1997) ‘It’s about Time: How, Why and When Networks Change.’ Social Networks 19: 1–7. Tapper, Richard (1988) ‘Minorities and the Problem of the State: Ethnicity, Development and Popular Resistance in Iran and Afghanistan.’ Third World Quarterly 10: 1027–1041.

Bibliography 219 Taylor, Edward J. (1986) ‘Differential Migration, Networks, Information and Risk.’ In Research in Human Capital and Development: Migration, Human Capital and Development, edited by Oded Stark. Greenwich: JAI Press, pp. 147–169. Terry, Fiona (2002) Condemned to Repeat: The Paradox of Humanitarian Action. Ithaca: Cornell University Press. Thomas, William I., and Florian Znaniecki ([1918–20] 1958) The Polish Peasant in Europe and America. New York: Dover Publications. Tilly, Charles (1978) ‘From Mobilization to Revolution.’ Massachusetts: AddisonWesley Publishing Company. ——— (1984) Big Structures, Large Processes, Huge Comparisons. New York: Russell Sage Foundation. Tober, Diane (2007a) ‘Introduction: Afghan Refugees and Returnees.’ Iranian Studies 40: 133–135. ——— (2007b) ‘“My Body Is Broken Like My Country”: Identity, Nation, and Repatriation among Afghan Refugees in Iran.’ Iranian Studies 40. Turton, David (2003) ‘Refugees and “Other Forced Migrants”.’ Oxford: Refugee Studies Centre. Turton, David, and Peter Marsden (2002) ‘Taking Refugees for a Ride? The Politics of Refugee Return in Afghanistan.’ Kabul: Afghanistan Research and Evaluation Unit. UNDP (1999) ‘Programme Framework for Reintegration Assistance to Returnees under the Greater Azra Initiative (Logar, Nangrahar and Paktia).’ Islamabad Papers: United Nations Development Programme. ——— (2003) ‘Partnership for Peace: Afghanistan’s New Beginnings Programme (ANBP).’ Kabul: United Nations Development Programme. UNHCR (1990) ‘Herat Province.’ 1 June 1990. Peshawar: United Nations High Commissioner for Refugees. ——— (1998) ‘Prospects for Voluntary Return of Afghan Refugees: A Feasibility Study.’ Islamabad: United Nations Commissioner for Refugees. ——— (2002) ‘UNHCR Statistical Yearbook.’ Geneva: United Nations High Commissioner for Refugees. ——— (2003a) ‘Towards a Comprehensive Solution for Displacement from Afghanistan.’ Geneva: United Nations High Commissioner for Refugees. ——— (2003b) ‘UNHCR Assisted Voluntary Repatriation to Afghanistan – Report from Daily Encashment Centres.’ Kabul: United Nations High Commissioner for Refugees. ——— (2004) ‘Afghanistan: Challenges to Return.’ Geneva: UNHCR. ——— (2005) ‘The Window Is Open: Achieving Durable Solutions for Afghanistan’s IDPs (Factsheet).’ Kabul: United Nations High Commissioner for Refugees. Urban, Mark (1990) War in Afghanistan. London: Macmillan. USCR (1995) World Refugee Survey 1995. Washington, DC: U.S. Committee for Refugees. ——— (1997) World Refugee Survey 1997. Washington, DC: U.S. Committee for Refugees. ——— (2001a) ‘Afghan Refugees Shunned and Scorned.’ September 2001. Washington, DC: U.S. Committee for Refugees. ——— (2001b) World Refugee Survey 2001. Washington, DC: U.S. Committee for Refugees.

220 Bibliography Van Brabant, Koenraad (2000) ‘Operational Security Management in Violent Conflicts.’ London: Overseas Development Institute. Van Hear, Nicholas (1994) ‘Migration, Displacement and Social Integration.’ Geneva: UNRISD. ——— (1998) New Diasporas: The Mass Exodus, Dispersal and Regrouping of Migrant Communities. London: University College London. ——— (2003a) ‘From “Durable Solutions” to “Transnational Relations”: Home and Exile among Refugee Diasporas.’ New Issues in Refugee Research, June. Geneva: UNHCR Evaluation and Policy Analysis Unit. ——— (2003b) ‘Refugee Diasporas, Remittance, Development and Conflict.’ The Migration Information Source, June. Washington, DC: Migration Policy Institute. Van Selm, Joanne (2002) ‘Foreign Policy Consideration in Dealing with Afghanistan’s Refugees: When Security and Protection Collide.’ Forced Migration Review 13: 16–18. Vertovec, Steven (2004) ‘Cheap Calls: The Social Glue of Migrant Transnationalism.’ Global Networks 4: 219–224. Vincent, Marc, and Birgitte Refslund Sørensen (Eds) (2001) Caught between Borders: Response Strategies of the Internally Displaced. London: Pluto Press. Vuorela, Ulla (2002) ‘Transnational Families: Imagined and Real Communities.’ In The Transnational Family: New European Frontiers and Global Networks, edited by Deborah Bryceson and Ulla Vuorela. Oxford and New York: Berg, pp. 63–82. Walker, Michael E., Stanley Wasserman, and Barry Wellmann (1994) ‘Statistical Models of Social Support Networks.’ In Advances in Social Network Analysis, edited by Stanley Wasserman and Joseph Galaskiewicz. Thousand Oaks: Sage. Wallerstein, Immanuel (1974) The Modern World System: Capitalist Agriculture and the Origins of European World Economy in the Sixteenth Century. London: Academic Press. Walsh, Declan (2004) ‘Day of Reckoning for the Emir of Herat: Next Month’s Elections Will Decide Fate of one of Afghanistan’s Most Powerful Warlords.’ Guardian, 4 September 2004. Wannell, Bruce (1991) ‘Report of a Trip to North-West Afghanistan, 1 September 1989–20 January 1990: Enjil and Zendajan Districts.’ Peshawar: unpublished. Wasserman, Stanley, and Katherine Faust (1989) ‘Canonical Analysis of the Composition and Structure of Social Networks.’ Sociological Methodology 19: 1–42. ——— (1994) Social Network Analysis: Methods and Applications. Cambridge: Cambridge University Press. Weiner, Myron (1992) ‘Security, Stability and International Migration.’ International Security 17: 91–126. Weiss, Thomas G. (2004) ‘The Sunset of Humanitarian Intervention? The Responsibility to Protect in a Unipolar Era.’ Security Dialogue 35: 135–153. Weitzmann, Eben A. (2000) ‘Software and Qualitative Research.’ In Handbook in Qualitative Research (2nd edition), edited by Norman K. Denzin and Yvonna S. Lincoln. Thousand Oaks: Sage, pp. 803–820. Wellman, Barry, and Scott Wortley (1990) ‘Different Strokes from Different Folks: Community Ties and Social Support.’ American Journal of Sociology 96: 558–588. Wellman, Barry, Renita Yuk-lin, David Tindall, and Nancy Nazer (1997) ‘A Decade of Network Change: Turnover, Persistence and Stability in Personal Communities.’ Social Networks 19: 27–50.

Bibliography 221 Williams, Holly Ann (1993) ‘Self-Settled Refugees in North-Western Zambia.’ In Selected Papers on Refugee Issues: II, edited by Mary Carol Hopkins and Nancy D. Donnelly: Committee on Refugee Issues, American Anthropological Association, pp. 135–155. Williams, Terry, Eloise Dunlap, Bruce D. Johnsen, and Ansley Hamid (1992) ‘Personal Safety in Public Places.’ Journal of Contemporary Ethnography 21: 343–374. Williamson, Oliver E. (1981) ‘The Economics of Organization: The Transaction Costs Approach.’ American Journal of Sociology 87: 548–577. Wimmer, Andreas, and Nina Glick Schiller (2002) ‘Methodological Nationalism and Beyond: Nation-State Building, Migration and the Social Sciences.’ Global Networks 2: 301–334. Wood, Elisabeth Jean (2008) ‘The Social Processes of Civil War: The Wartime Transformation of Social Networks’ Annual Review of Political Science 11: 539–561. Woodward, Bob (2002) Bush at War. New York: Simon & Schuster. Yamagishi, Toshio, and M. Yamagishi (1994) ‘Trust and Commitment in the United States and Japan.’ Motivation and Emotion 18: 129–166. Yin, Robert K. (2003) Case Study Research: Design and Methods (3rd edition). Newbury Park: Sage. Zackariya, Faizun, and Nadarajah Shanmugaratnam (2003) ‘Moving into the Extra Household Domain: Survival Strategies of Displaced Muslim Women in Sri Lanka.’ In In the Maze of Displacement: Conflict, Migration and Change, edited by Nadarajah Shanmugaratnam, Ragnhild Lund, and Kristi Anne Stølen. Kristiansand: Høyskoleforlaget, pp. 50–75. Zahedi, Ashraf (2007) ‘Transnational Marriages, Gendered Citizenship, and the Dilemma of Iranian Women Married to Afghan Men.’ Iranian Studies 40. Zetter, Roger (1988a) ‘Refugees and Refugee Studies: A Label and an Agenda.’ Journal of Refugee Studies 1: 1–6. ——— (1988b) ‘Refugees, Repatriation, and Root Causes.’ Journal of Refugee Studies 1: 99–106. ——— (2008) ‘More Labels, Fewer Refugees: Remaking the Refugee Label in an Era of Globalization.’ Journal of Refugee Studies 20: 172–192. Zinzer, A. A. (1991) ‘Refugee Repatriation in Central America: Lessons from Three Case Studies.’ In Repatriation under Conflict in Central America, edited by Mary Ann. Larkin, Frederick. C. Cuny, and Barry. N. Stein. Washington, DC/Dallas: Center for Immigration Policy and Refugee Assistance/Intertect Institute, pp. 187–212. Zolberg, Aristide R. (2008) How Many Exceptionalisms? Exploration in Comparative Macroanalysis. Philadelphia: Temple University Press. Zolberg, Aristide, Astri Suhrke, and Sergio Aguayo (1986) ‘International Factors in the Formation of Refugee Movements.’ International Migration Review 20: 151–169. ——— (1989) Escape from Violence: Conflict and the Refugee Crisis in the Developing World. New York: Oxford University Press.

Author Index Adelman, Howard, 142 Aguayo, Sergio, 37, 81 Allen, Rebecca, 19 Ask, Karin, 186, 198 Borchgrevink, Axel, 185–6 Burt, Ronald, 41 Centlivres, Pierre, 49, 78 Centlivres-Demont, Micheline, 49 Connor, Kerry M., 48–50, 182 Cuny, Fredrick C., 100, 137 Douglas, Mary, 38 Emirbayer, Mustafa, 43 English, Richard, 124–5 Faist, Thomas, 173 Giustozzi, Antonio, 146–7 Glatzer, Bernt, 103–7, 113, 130 Gould, Roger, 31–2 Granovetter, Mark, 24, 32, 193 Grasmuck and Pessar, 29, 35, 37, 192 Grønhaug, Reidar, 70, 182 Hammond, Laura, 135–6 Hansen, Art, 18 Hiller, Henry H., 17, 19, 47 Hugo, Graeme J., 24 Kasdan, Leonard, 50 Knoke, David, 36 Koser, Khalid, 32, 95, 101, 128–9

Lishcer, Sarah Kenyon, 37, 81, 109–110 Lomnitz, Larissa, 35 Malkki, Lisa, 4–6, 80, 191 Massey, Douglas, 28, 41 Merton, Robert, 39 Olson, Mancur, 37 Padgett, John F., 43 Ritchey, P. Neal, 24 Roberts, Bryan, 39 Stedman, Stephen John, 151 Schelling, Thomas, 40 Shahrani, M. Nazif, 81 Shami, Seteney, 29 Sliwinski, Marek, 49–50, 193 Spaan, Ernst, 35 Stein, Barry, N., 100, 137 Strand, Arne, 186, 198 Suhrke, Astri, 37, 81 Thomas, William I., 24 Taylor, Edward J., 28 Van Hear, Nicholas, 136 Wellmann, Barry, 26 Znaniecki, Florian, 24 Zolberg, Aristide, 37, 81

222

Subject Index absentee property management, see site-specific resources Afghan migration history, 6–9, 47–50, 78 escape, 7, 47–50 integration, 79–81 internal mobility, 48, 70 reintegration, 137–8 return, 101–3 to Iran, 47, 79, 196 to Pakistan, 47, 79, 105 Afzali, Safiullah, 83 agency, ix, 5, 18, 44, 183, 184 analysis, 189, 190 Angolan refugees in Zambia, 18 arbab, 57, 147, 161–2 Arbab Saidu, 10, 56, 143

types of displacement, 12, 58, 69, 133, 182 villages, vii, 11 competence (acquired in exile), 90, 119, 151 conscription, 48, 52, 57–8, 59–61, 158, 194 contagion, 40–2, 107, 114, 116, 133, 138, 173 contextual analysis, 42 coping, 2, 27–8, 77–8, 98, 157, 192 ‘corrosive community’ thesis, 3, 165–6, 177 credit, 22, 36–7, 67–8, 68–9, 122–3, 156–7, 170 cultural repertoires, 15, 25–6, 28, 35–6, 49–50, 80 ‘cumulative causation’, 173

Bosnians in Serbia, 66 boundaries (state), 29, 79, 122, 160 ‘bounded solidarity’, 30, 31 bridges (as network structure), 31, 32–6, 70, 129, 193 Cambodian refugees, 32, 155 camp-settlement versus Self-settlement, 18, 79–80, 170, 174 causality, 4–5, 17, 46–7, 48, 116 chain migration, 72, 193 change of preferences, 100, 118, 122, 139, 156, 162 collective action, 37 communication, 27, 36, 73, 93, 115, 127, 166, 170 ‘community alienation’, 137 comparison by host country (Iran, Pakistan), 8, 45, 77, 84, 122 leavers versus stayers, vii, 12, 46, 47, 66, 68, 134 other cases, 12 stages in flight cycle, 12, 99, 168–73, 173–6

data-collection, 186–9 anonymization, viii, 188, 189 on exile, 78 interviews and questions, 186–9 public access, 138 reliability, 184, 187, 198 self-evident facts and, 69–70 decision-making, 1, 13, 29, 72, 99–103, 114, 131–2, 168–73 diffusion, 33, 41 cohesion versus structural equivalence, 41–2, 173–4 see also contagion disarmament, demobilization and reintegration (DDR), 141, 151 disasters, 2–3, 25, 27–8, 177, 191, 193 Dominican migration to the US, 37, 40 Dostum, Abdul Rashid, 144 drought, 62, 121, 154 durable solutions, 5, 109–10, 194

223

224 Subject Index duration, socially expected, 38–40, 173, 176 escape and, 76, 173, 182 exile and, 22, 93, 98, 102, 115, 173 reintegration and, 176, 182 return and, 125 employment, 87, 88–9, 152 exile, 62, 88–9, 91–2, 97, 158, 171 job accidents, 89 niche economies, 91, 175 upon return, 147–8, 151, 152–5, 155, 171 work-teams, 90 ‘enforeceable trust’, 30 escape, 5, 51, 53–4, 168–73 costs and selectivity, 22, 66, 170 destination, 16–18, 24, 35, 55, 66, 72, 89, 170 internal pressure, 17, 18–19, 61 migration experience and, 25, 72, 94, 159, 170, 182 rationale for, 46–7, 48–50, 50–1 reactive versus proactive, 18–19, 32, 168 exclusion, 30 exile, see integration expulsion, 79, 86–7, 116–18, 127 deportation camps, 86, 116–17 fieldwork for book, 9, 11, 184 access, 11, 185, 190, 194, 196 ‘gradualist’ approach, 184, 186 security during, 11, 185, 186 ‘flight collectives’, 17, 23, 31–2, 37, 51, 101 decision-making and, 47, 168 in exile, 78–9, 83, 98 during return, 83, 101, 102, 108–9, 113 traditional leadership and, 50, 96 foci (of networks), 15–16, 26, 32, 76, 171 ‘forced migration’ studies, 4, 5–6, 181–3 fragmentation of networks, 3, 165–6, 177 gender differences, 37–8, 40, 90, 122, 130–1, 139

geographic space and networks, 16, 28, 76, 182 ‘group repatriation’, 102, 131–2, 197 hawala, x, 93, 119 Herat labor market in city, 140, 156 security belts around, 53, 56 uprising (1979), 52 Hezb-e Islami (Khales), 82 hijra, x, xi, 25, 36, 48–51, 80, 81, 194 host population, 18, 80, 87, 89, 102–3, 159–60 households, 15–16, 37, 65, 67 housing, 44, 82, 94, 138, 151–2, 152–5, 157 humanitarian assistance, 161, 192, 197 integration in exile, 62, 83, 88, 175 reintegration, 135–6, 137 see also refugee management human security, 21, 191 IDPs (internally displaced persons), 68, 170, 193, 194 inequality, 27, 31, 36–7, 62, 124–5 convergence with poor networks, 75, 151, 175 information, 19–20, 23–6, 129, 169–70, 174–5 assessments of future and, 71, 118, 125, 169 channels, 73, 90, 95–6, 128, 129–30, 131, 194, 198 escape and, 72, 128–9 in exile, 94–6, 97, 114, 169 reliability of, 125, 169 return and, 125–32 scarcity of, 94–5, 97, 126, 163, 169 see also communication integration, 5, 13, 85–8, 106–7, 174–6 Iran-Iraq war, 81, 85 Iran-US relations, 86 Islamist parties, 178–9 Izhaq Suleman, 10–11, 46, 52–9, 110, 115–16, 146–7, 150, 172 IDPs in, 58–9, 68, 142, 146, 149

Subject Index 225 Jamiat-e Islami, 82, 83 Javanese migration, 35 jobs, see employment Karzai, Hamid, 10, 101, 148 Khaliqyar, Fazel Haq, 53, 56 Khan, Amanullah, 148 Khan, Ismael, 52, 59–60, 84, 103, 110, 142, 143–4, 146–7, 148, 150, 155 kinship, 2, 15–16, 33–4, 75, 78, 113, 123, 137 landmines, 10, 60, 121, 141, 155 landowners, 30, 68, 121–2, 154–5, 157, 171 land (as private property), 120, 121, 154, 157 irrigation of, 64–5, 121, 154 recultivation, 105, 120, 121–2, 135, 151, 152 sale, 63, 65, 88, 120, 121–2, 155, 171 language interpretation, x, xi, 185–6 Malik, Abdul, 144 Mashad, 55, 82–3, 84, 144, 159–60 material resources, 122, 170–2, 175–6 capital accumulation, 65–6, 88, 91, 151, 157, 159 definition, 19, 22–3 long-distance property management, 23, 30 realization of, 67, 119, 170 see also credit, employment, land, site-specific resources mechanisms, 4–5, 17, 41, 44, 50, 61, 173 Mexican migration, 34, 35 migration studies, 28–9, 34, 41, 72, 162, 170, 176 diaspora studies, 6, 191 forced migration as integral to, 4, 33, 181–3 militia–mujahedin interaction, 57–8, 114–15, 128, 143, 172, 194 militia, x, 10, 52–61, 71, 74, 84–5, 145, 172, 194 recruitment, 56, 60–1, 147

mobilization, 18–19, 22, 31–2, 55, 81–2, 150, 172, 192 money, see material resources Mozambican refugees in Malawi, 33, 34, 95, 101, 109, 129–30, 196 Mozambicans in Malawi, 95, 101, 128–9 mujahedin, xi, 10–11, 52–9, 70–1, 83–4, 97, 110–11, 141–7, 147–8 government (1992–96), 79, 101, 121 landowners and, 68 see also militia multi–sited methods, 29, 184 networks, see social networks objectives of the book, 1, 2, 3–4, 9 Operation Enduring Freedom (OEF), 7, 111, 150, 198 Palestinian refugees, 100 Pashtuns in the north, 6, 7, 146 pioneer migrant, 34 policy-driven research, 4, 5, 8, 183 Polish Peasant, The, 24 power, 22, 26, 36–8 property, see material resources Qala-e Muhajerin, 10, 143–4, 147, 155, 195–6, 198 qawm, xi, 56, 61, 147, 194 reconciliation, 53, 83, 139–40 refugee management, 78–9, 95, 103, 125 registration, 79, 83, 91, 194 restrictions, 85–6, 87, 88, 93, 158, 172, 195 welfare provision, 83, 88–9 refugee studies, 4–6 refugee warriors, vii, 8, 55, 146, 148, 192, 194, 195 Iran and, 82–3, 84–5, 96, 176, 195 privileges upon return, 147–8, 155–6, 164, 171, 176 return of, 109–11, 121, 141–3 rotation at the front, 83

226 Subject Index reintegration, 6, 115, 123, 135–7, 173–6 absorptive capacity of networks, 137, 157 fighters and, 141 late versus. early returners, 123, 138–9, 140, 148, 164, 178 transnational networks and, 136–7, 158–60, 175–6 religious institutions, 35–6, 48–9, 80–1, 179 remittances, 30, 80, 92–4, 97, 158, 171 ‘repatration packages’, 79, 100, 101–2, 119, 123–4 retrospective data, 9, 127 return, 118, 121, 168–73, 197 misconceptions about, 99–102 pace of, 113, 124, 126–7, 140, 153 prevention of, 111, 197 reactive versus proactive, 34,100, 110 reversibility, 103, 159 risk, 24, 25, 27, 101–2, 107, 115 rumors, 24, 41 Salaffiya, xi, 104 Sara-e Nau, 9–11, 47, 52, 56–8, 71, 82, 142, 144–5 security, 5, 19, 20–2, 148–50, 172–3, 176 in exile, 85, 127 network scale and, 59, 109, 150, 172 shelter support, 138, 161 Shia politics, 80, 179, 196 shura, 71, 161–2, 194 site-specific resources, 22, 118–19, 121, 154, 171 definition, 23, 192 management of, 23, 30, 48, 63–4, 66, 87, 157, 194 sale of, 62, 63, 121–2, 171 smuggling (human), 67, 68–9, 75, 117, 119, 123, 170–1, 194 social capital, 28, 182, 192 social networks, 14–15, 16–17, 28 ‘coupling’ and ‘decoupling’, 31

interview questions about, 186–7, 188, 196 scale of, 21, 22, 51, 59, 108, 115, 165–6 social support, 20 strategic building of, 19, 178, 180 strong ties, 26–32, 134 transformation of networks, 34, 42–3, 162, 167, 177–81 split households, xi, 23, 30, 47, 58, 61, 63–5, 87–8, 90, 120, 122, 158–60, 175–6, 180 conscription and, 63–4 economy and, 90–1, 117, 133, 171 reintegration and, 160, 165, 171–2, 197–8 return and, 133, 153, 158, 169 spoilers, 151 state failure, 6, 59, 71, 107 strong ties, 26–7, 179–80, 193 uncertainty and, 27–8, 70, 76 Tajikistan war and refugees, 114–15, 130 Taliban, 7–9, 59, 83–5, 110–11, 145–50 in Herat, 110–11, 145–6 mujahedin and, 144, 145–6 ‘therapeutic community’ thesis, 3, 177 threats, 51, 59, 61, 108, 150 thresholds, see contagion transnational networks, 6, 23, 28, 41, 158–60, 165, 191 saturation, 72, 94 tribal organization, 105, 106 trust, 3, 44, 75, 93, 115, 125, 179 urbanization, 48, 139, 156 US integration of immigrants, 40 victimization, 18, 139, 142, 183 Vietnamese refugees, 17, 19, 51 ‘vintages’, 40–2, 45, 48, 49 see also contagion Zindajan, 53