Circular Migration in Zimbabwe and Contemporary Sub-Saharan Africa [Illustrated] 1847010237, 9781847010230

Citation preview

Potts_PPC_24mm:B+B

20.09.2010

13:03

Page 1

DEBORAH POTTS Deborah Potts The empirical core of this book illustrates these trends through a detailed examination of the case of Zimbabwe based on the author’s longstanding research on Harare. The political and economic changes in Zimbabwe since the 1980s transformed Harare from one of the best African cities to live in over this period to one of the worst. Harare’s citizens’ livelihoods exemplify, in microcosm, the central theme of the book: the reinvention throughout sub-Saharan Africa of circulation and rural-urban links in response to economic change. Deborah Potts is Senior Lecturer in Geography, King’s College London Contents: Introduction: re-inventing the wheel – Regional paradigms & approaches to circular migration: tropical Africa – Regional paradigms & approaches to circular migration: southern Africa – Harare & Zimbabwe: from formal city to outsourced urban livelihoods – Migrant livelihoods & migration trajectories in Harare: 1980s to 2000s – Harare migrants’ rural links & assets – Variations in migrants’ experience, perceptions & options – Social & cultural attachments to rural settings & homes – Attacking the urban poor & abusing rural links: Operation Murambatsvina 2005 – Conclusion Cover photograph: Residents of Mbare High Density Area in Harare made homeless by Operation Murambatsvina in 2005 (© Institute of War and Peace Reporting)

An imprint of Boydell & Brewer Ltd PO Box 9, Woodbridge, Suffolk IP12 3DF and 668 Mt Hope Ave, Rochester NY 14620, USA www.boydellandbrewer.com www.jamescurrey.com

Circular Migration in Zimbabwe & Contemporary Sub-Saharan Africa

Urban livelihoods in sub-Saharan Africa have become very fragile. Circular migration to cities, whereby rural migrants do not remain permanently in town, has particular significance in the academic literature on development and urbanization in Africa, often having negative connotations in southern Africanist studies due to its links with an iniquitous migrant labour system. Literature on other African regions often views circular migration more positively. This book reviews the current evidence on circular migration and urbanization in sub-Saharan Africa. The author challenges the dominant view that rural-urban migration continues unabated and shows that circular migration has continued and adapted, with faster out-migration in the face of declining urban economic opportunities.

Circular Migration in Zimbabwe & Contemporary Sub-Saharan Africa

Circular Migration in Zimbabwe & Contemporary Sub-Saharan Africa

Recent titles on Southern Africa Crossing the Zambezi JOANN McGREGOR Bulawayo Burning TERENCE RANGER Circular Migration in Zimbabwe & Contemporary Sub-Saharan Africa DEBORAH POTTS Zimbabwe’s Land Reform IAN SCOONES ET AL. War Veterans in Zimbabwe’s Revolution* ZVAKANYORWA WILBERT SADOMBA

* forthcoming

Circular Migration in Zimbabwe & Contemporary Sub-Saharan Africa DEBORAH POTTS Senior Lecturer in Geography King’s College London

James Currey is an imprint of Boydell & Brewer Ltd PO Box 9 Woodbridge, Suffolk IP12 3DF, UK and of Boydell & Brewer Inc. 668 Mt Hope Avenue Rochester, NY 14620, USA www.boydellandbrewer.com ©  Deborah Potts 2010 First published 2010 All Rights Reserved. Except as permitted under current legislation no part of this work may be photocopied, stored in a retrieval system, published, performed in public, adapted, broadcast, transmitted, recorded or reproduced in any form or by any means, without the prior permission of the copyright owner The right of Deborah Potts to be identified as the author of this work has been asserted in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988 ISBN  978–1–84701–023–0 (James Currey cloth) The publisher has no responsibility for the continued existence or accuracy of URLs for external or third-party internet websites referred to in this book, and does not guarantee that any content on such websites is, or will remain, accurate or appropriate. British Library Cataloguing in Publication Data Potts, Deborah (Deborah Helen)   Circular migration in Zimbabwe and contemporary sub-Saharan   Africa.   1. Migration, Internal--Economic aspects--Zimbabwe.   2. Migrant labor--Zimbabwe. 3. Harare (Zimbabwe)-  Economic conditions--20th century. 4. Migration,   Internal--Economic aspects--Africa, Sub-Saharan.   5. Migrant labor--Africa, Sub-Saharan.   I. Title   307.2'096891-dc22 Typeset in 10.5/11.5 Monotype Ehrhardt by Boydell & Brewer Papers used by Boydell & Brewer Ltd are natural, recyclable products made from wood grown in sustainable forests

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

Disclaimer: Some images in the printed version of this book are not available for inclusion in the eBook. To view these images please refer to the printed version of this book.

Contents







List of Maps, Figures, Tables, Boxes & Photographs List of Abbreviations Preface

vii x xi

1

Introduction

1

Re-inventing the Wheel

2

Regional Paradigms & Approaches to Circular Migration

30

Tropical Africa

3

Regional Paradigms & Approaches to Circular Migration

49

Southern Africa

4

Harare & Zimbabwe

74

From Formal City to Outsourced Urban Livelihoods

5

Migrant Livelihoods & Migration Trajectories in Harare

102

1980s to 2000s

6

Harare Migrants’ Rural Links & Assets v

133

Contents

7

Variations in Migrants’ Experience, Perceptions & Options

159

8

Social & CulturalAttachments to Rural Settings & Homes

188

9

Attacking the Urban Poor & Abusing Rural Links Operation Murambatsvina 2005

211

10

Conclusion

234



Appendix: Consumer Price Index (all items) for Urban Families 1973–2007 Bibliography Index

257 259 291

vi

Maps, Figures, Tables, Boxes & Photographs

Maps 1 2

Zimbabwe Harare

76 77

Figures 1.1 1.2 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8a 5.8b 5.8c 5.8d 5.8e 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 7.1 7.2 7.3 7.4 7.5 7.6 7.7 7.8 7.9

Push-pull model of circular migration Urban growth scenarios Migrants’ motives for leaving Migrants’ motives for coming to Harare Average real primary income for earning HHHs: 1985, 1994 and 2001 Activity status 1985, 1988, 1994 and 2001 (%) Unemployment rate and employment sector 1984, 1988, 1994 and 2001 Reasons given for expecting to leave Harare Future plans by relationship to household head Future plans by relationship to HHH Future plans by gender Future plans by tenure status Future plans by birthplace Future plans by current access to land Frequency of land ownership and average acreage Frequency of cattle ownership and average holding Rural assets by marital status and birthplace Contribution of rural food production to Harare consumption, 1988 and 2001 Reasons for not farming land Future destinations Future farming intentions Landless migrants’ intentions for obtaining rural land Income comparison by gender and land holding % Job comparison by gender and land holding % Income comparison by land holding by gender % Job comparison by land holding by gender % Living standards comparison by gender, landholding and birthplace 2001 What is bad about living in Harare What is good about living in Harare Good aspects of Harare: gender and position in household Negative aspects of Harare: gender and position in household vii

10 12 108 108 110 114 116 122 126 128 128 128 129 129 144 144 145 151 152 155 155 156 166 166 167 167 168 179 179 182 183

Maps, Figures, Tables, Boxes & Photographs

Tables 1.1 4.1 4.2 4.3 4.4 4.5 4.6 5.1 5.2 5.3 5.4 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9 6.10 6.11 6.12 8.1 8.2 8.3 9.1 9.2

Recent net migration evidence and urbanization in selected sub-Saharan 20–21 African countries Population and average annual growth rates for Zimbabwe, urban centres 78 and main towns 1960s to 2002 Primary source of household income: Harare HDAs mid-2005 93 Sources of urban household incomes in Harare Province, 2003 95 Sources of urban household incomes by income group: Zimbabwe 2003 95 Minimum wages and PDL 2007–08 (Z$ millions) 97 Demographic rates for Harare and Chitungwiza 1991–92 and 2001–02 97 Migrants’ birthplaces: 1985, 1994 and 2001 (%) 107 Last place of residence before moving to Harare: 1985, 1994 and 2001 (%) 107 Age and marital status of ‘dependent’ migrants eligible for work by 118 activity status, 1985, 1994, 2001 (%) Migrants’ future plans: 1985, 1988, 1994 and 2001 122 Access to land by relationship to household head (%) 135 Access to land by age (%) 135 Access to land by marital status (%) 137 Access to land by birthplace (%) 137 Cattle holding by relationship to household head (%) 139 Cattle holding by age (%) 139 Cattle holding by marital status (%) 140 Cattle holding by birthplace 140 HHHs with land or livestock: average amounts held 142 Amount of land held by age (%) 143 Composition of farming labour force on land farmed by migrant households 148 Migrants with land: household crop production and sales 149 Absentee family members reported: 1985, 1994 and 2001 (% of respondents) 202 Mobility and residence patterns of family members residing for at least half 202 of the year away from Harare: 1988 Plans for out-migration: ‘home’ or elsewhere (%) 202 Dwelling units demolished and designated urban population by province 216 (% of total) Households affected by Operation Murambatsvina in Harare’s High Density 221 Areas (%)

Boxes 3.1 5.1 5.2 5.3 5.4 5.5 8.1

Differing perspectives on post-apartheid circular migration in South Africa Prices and budgets for low-income households in Harare Secondary income households in 1994 Why I will leave Harare: views in 1985 I would stay in Harare only if … Migrants who will stay in Harare Cultural and social attachments to rural ‘homes’ and settings: typical views

viii

61 111 113 125 125 126 207

Maps, Figures, Tables, Boxes & Photographs

Photographs 9.1 9.2 9.3 9.4

Formal house left standing against rubble of demolished ‘informal’ backyard shacks and extensions to house, Mbare 2005 Kuwadzana plot a year after Operation Murambatsvina Abandoned People’s Market in Central Harare, 2006 Rebuilding at Hatcliffe Extension in 2006

218 218 219 219

Disclaimer: Some images in the printed version of this book are not available for inclusion in the eBook. To view these images please refer to the printed version of this book.

ix

Abbreviations

AIDS Acquired Immunity Deficiency Syndrome CA Communal Area Central Business District CBD CHRA Combined Harare Residents’ Association CSO Central Statistical Office DK (in tables) Don’t know Democratic Republic of Congo DRC ESAP Economic Structural Adjustment Policy Gross Domestic Product GDP GMB Grain Marketing Board Global Political Agreement GPA HDA High Density Area Household Head HHH HIV Human Immunodeficiency Virus HSRC Human Sciences Research Council IDP Internally Displaced People International Monetary Fund IMF LDC Less Developed Countries MDC Movement for Democratic Change NESMUWA Network of Surveys on Migration and Urbanization in West Africa NGO Non-Governmental Organization PASS Poverty Assessment Study Survey RLI Rhodes-Livingstone Institute SACN South African City Network SADC Southern African Development Community SAPs Structural Adjustment Policies SPT Solidarity Peace Trust STD Sexually Transmitted Disease UDI Unilateral Declaration of Independence UN United Nations USA United States of America ZANU(PF) Zimbabwe African National Union (Patriotic Front) ZNVAC Zimbabwe National Vulnerability Assessment Committee x

Preface

In 1984 I sent a research proposal on migration to Harare to Professor Chris Mutambirwa, Head of the Geography Department of the University of Zimbabwe. He responded with interest. Thus began a research collaboration which has been central to the development of my ideas and understandings of migration and urbanization in sub-Saharan Africa. In the following year, when our first project was undertaken, and during subsequent rounds of surveys of migrants and their livelihoods, and on many other occasions when I was in Zimbabwe, the Geography Department hosted and supported me. The results of that research form the core of this book and the context from which I have developed my broader analysis of cities and circular migration in sub-Saharan Africa. I should like to thank everyone involved in the department. I am also grateful to all the research assistants and interviewees involved in our surveys over the years; special mention must go to Ruth Masaraure. Above all, I am indebted to Chris and his wife Jane for their support and affection for both me and my family over the decades since our first correspondence all those years ago. In the 1990s and 2000s, often with my young children in tow, I could not have conducted my work in Harare without the hospitality and friendship of Richard and Lynette Owen. There are many other people who have contributed to my work on migrants and cities in Africa over the years, by providing intellectual stimulation and critique, and hospitality during research visits to other African cities. It is not possible to list them all here but in Johannesburg I would particularly like to thank Alan Mabin and Cynthia Kros who have welcomed me so many times. In Cape Town I must thank Sue Parnell and Owen Crankshaw. Liz Blunt kindly invited me to stay with her in Addis Ababa. Colin Murray, Kathy Baker and Liz Gunner deserve special thanks for their encouragement of this book as well as their wonderful support and friendship over the years. I am also grateful to Colin, and to Deborah James, for their comments on various drafts of the book. My research in Harare and in southern Africa more generally on the themes covered in this book has been funded by small grants from the School of Oriental and African Studies, King’s College London and the Nuffield Foundation. I am also grateful to King’s College London for a sabbatical, without which this book could not have been written. xi

1

Introduction

Re-inventing the Wheel?

Emerging new forms of resistance to the brutalities of global capitalism … must coexist with older forms, scrounged – like circular migration … – from the dustbin of history.  Ferguson (1999: 257)

This book is about trends in migration to, and from, African towns and cities, and the changing characteristics of migrants and migrancy. To some extent, therefore, it is also about changing urban population dynamics in sub-Saharan Africa as an outcome of these shifts and the nature of urban livelihoods which is a key driver of migration. The core of the book is based on empirical evidence from longitudinal research in Harare, Zimbabwe. In the 1980s, ordinary residents of Harare were probably the most economically and socially secure urban people in sub-Saharan Africa. By the mid2000s, they were among the least secure. Over the same period most ordinary urban people in sub-Saharan Africa had suffered significant falls in their living standards, leading to adaptations in their livelihoods and the nature of migration (Potts 1997; Beall et al. 1999). Harare’s experience provides in microcosm an extreme example of these falls and adaptations which can be traced through the evidence from directly comparable surveys conducted in the 1980s, 1990s and 2000s. A key finding of this research in Zimbabwe has been the strengthening and revitalization of circular migration and the rural–urban linkages upon which this is predicated. Similar processes have also occurred in many other African towns across the continent. This is one reason why this introductory chapter is entitled, ‘Re-inventing the wheel’, referring to this reinforcement of circularity as one element in ruralurban migration patterns. There are two more abstract aspects to the title, however, which relate to theorizing about urbanization and migration processes in African countries. The experiences of the last three decades have entirely undermined the utility of conceptualizing such processes in relation to unilinear models, as embodied in modernization theory generally, and urban transition models more specifically, whereby permanent migration inexorably replaces circulation as urban economies develop. An important contribution to this view was Ferguson’s book, Expectations of Modernity: Myths and Meanings of Urban Life on the Copperbelt (1999), which showed how the unilineal modernization model was meaningless in relation to what had really happened in urban Zambia since the 1970s. Further, it established how circular migration needed to be re-evaluated and incorporated into contemporary theoretical understandings of urbanization in southern Africa. 1

Introduction

The rejection of linear theorizing and the appreciation of cyclical and more variable trends as influences on African urbanization and migration points to another part of the reasoning behind this chapter’s title. Ferguson’s arguments related particularly to theoretical approaches typical of southern African academic traditions. This chapter’s title posits that these approaches have not taken sufficient cognizance of academic traditions and analyses of urban processes elsewhere in sub-Saharan Africa. These have much more comfortably accepted the many rationales for, and vitality of, circular migration over time. The eventual recognition in southern Africanist academic circles that circular migration is not simply fading away, despite the absence of any institutions enforcing it,1 is also ‘a reinvention of the wheel’ in that this has been understood, without surprise, in much of the rest of sub-Saharan Africa for decades. The questions of how circular migration has been evaluated in sub-Saharan Africa and, in particular, how different academic approaches and theoretical positions have influenced the tenor of these evaluations in different regions are considered in the next two chapters. This includes specific discussion of the experiences of, and academic debates within, South Africa where urbanization and migration exhibit some features in common with the rest of the continent but also some differences due to its particular political and economic history. The way in which circular migration has been defined and understood in broader migration theory is discussed in this first chapter. The general sub-Saharan African context within which migration trends and urban dynamics have developed is also examined, and related to how these compare with broader global (re)-evaluations of urbanization in developing countries. Particular emphasis is placed on the impact of African urban livelihood trends from the end of the 1970s. The key evidence for the two dynamics of more, and faster, circulation migration and of reductions in the rate of urban growth in many African urban areas is also presented and discussed.

African Internal Migration in Theory Migration within African countries has been analysed with reference to most of the general types of theories about migration which have been developed, including straight neo-classical economic treatments such as Todaro (1971), radical Marxist perspectives (e.g., Amin 1974; Meillassoux 1981; Arrighi 1973) and post-modernist approaches. The following discussion makes no attempt to review the general theoretical literature, which has been done elsewhere (Baker and Akin Aina 1995; van Binsbergen and Meilink 1978; Murray 1994).2 Instead, it outlines some key elements of theorizing which are of special relevance to this study. Theoretical approaches which have strongly influenced the different regional paradigms within which internal migration, including circular migration, have been analysed are highlighted. They are also cross-referenced in the subsequent two chapters in which these paradigms are 1

2

Discussed in Mabin (1990). Mabin’s awareness that circular migration processes had dynamics outside of institutionalized migrant labour systems and that unilinear transition models had limited applicability was a key intervention in the South African literature but much subsequent policy analysis ignored its message. Apart from some work in journals, the majority of literature on migration in Africa since the mid-1990s has focussed on international and/or forced migration flows rather than internal and ‘voluntary’ migration. Within this body of work, research output on cross-border migration to South Africa has been very significant.

2

Re-inventing the Wheel?

discussed in detail. Research conducted in Zimbabwe which relates to migration and urban livelihoods is highlighted, in both its theoretical and methodological aspects. At the end of the 1970s, as radical perspectives on migration were increasingly gaining sway, Gerold-Scheepers and van Binsbergen (1978) provided a typology of theoretical approaches for African migration which still has considerable utility. The methodological-individualist approach, which focuses on the individual motivations of the migrant, frequently emphasises the significance of economic motives but can also include various non-economic issues. This approach informed quantitative aspects of research surveys in Harare, which included simple inquiries into each migrant’s own choice of factors that had caused their migration. Neo-classical migration theory essentially falls into this grouping, with its emphasis on individual decision-making by rational homo economicus. The driving forces in neo-classical migration theory are geographical inequalities in incomes and employment and the desire to maximize income. As is well known, rural-urban migration in Africa was explained in these terms by Todaro (1969, 1971). At one level, the rates of, and rationale for, in-migration to African urban areas after independence are quite well explained by his model, as long as a crucial, but rarely noticed, implication of part of his theorizing is recognized. Todaro argued that continued migration to urban centres in the 1970s, when formal employment creation was evidently not able to provide enough jobs, could be rationalized as individual migrants maximizing their life-time income streams rather than merely comparing rural versus urban incomes in the short-term. Each migrant would thus ‘balance the probabilities and risks of being unemployed or underemployed for a considerable period of time against the positive urban-rural real income differential’ (Todaro, 1997: 280, emphasis added). This meant that, paradoxically, ‘more urban [formal] employment leads to higher levels of urban unemployment’ (Todaro, 1997: 282, original emphasis) because of in-migration. However, this latter point also implies that, as formal employment levels fell dramatically in the 1980s and 1990s, the model would logically predict that open unemployment and in-migration would also fall. In the case of Harare, the migration surveys of the 1980s and 1990s suggested that this corollary of the Todaro thesis was operating. In 1994, when formal employment had suffered a sudden downturn compared to the 1980s, migrant household heads were more likely to be formally employed than in the 1980s and there was evidence of increasing out-migration (i.e., falling net in-­migration) (ibid).3 Yet Todaro (1997: 276) persisted in blaming migrants as the primary cause of increasing urban unemployment in Africa despite the logical implications of his model. He paid insufficient attention to the potential circularity of rural-urban migration there and how such circulation influences net in-migration and employment demand. In relation to income maximization rationality, this misses the fact that when ‘the considerable period of time’ has elapsed, many migrants leave town if they have not succeeded in beginning their long-term earning ‘stream’. As will be shown, this is supported by evidence for falling net in-migration; furthermore, it is likely that open unemployment has not much increased in African towns (see Potts 2000). Structural-functionalist approaches to migration assume the existence of 3

The evidence for lower net in-migration by the time of the last survey round in 2001 was even greater (see Chapter 4). However, by this time the fundamentals of the national, as well as the urban economy, had become so unusual that the entire economy was informalizing. This was reflected in a somewhat higher rate of informal employment among migrants in Harare. Nonetheless, male household heads remained very largely formally employed (see Chapter 5).

3

Introduction

differentiated sectors within society (e.g. traditional versus modern; rural versus urban) created by various structural forces. These lead to what might be termed socioeconomic disequilibrium (Chapman and Prothero 1985a) and migration is one of the ways in which interactions between such sectors occur. The differences between the various sectors are largely taken as read, which often leads to studies of this type being ahistorical, as with methodological individualist approaches. Many of the works of the Manchester School of anthropology and sociology on African migrants in Central Africa conducted at the Rhodes-Livingstone Institute in the mid-twentieth century are ascribed as structural-functionalist (Gerold-Scheepers and van Binsbergen 1978; van Binsbergen and Meilink 1978). These include seminal works by scholars such as Gluckman (e.g., 1961), Mitchell (e.g., 1959, 1969) and Epstein (e.g., 1958, 1961), whose influence on southern African migration studies is considered in Chapter 3. The structural-functionalist approach was also central to Parkin’s (1975) edited volume on African migration, urbanization and rural-urban linkages. These types of studies, and neo-classical economic approaches, are often wedded to a meta-narrative of modernization, economic progression and ‘development’. In contrast, radical or Marxist approaches to internal migration in Africa take as their key task the explanation of how and why differentiated sectors have emerged, along with identifying whose material and political interests are best served by particular migration patterns. Explanations are usually sought in historical processes of how African societies have been incorporated into capitalism and the subsequent articulation of different modes of production. Many migration studies in southern Africa fall into this group, as will be discussed in Chapter 3. It is only such approaches, rooted in a historical perspective, which allow real insights into the causes of migration and can identify the conflicts frequently inherent in migration. These are often skated over by structural functionalists, and not even noted by methodological individualists. A central example for circular migration is how migrants are sometimes argued to ‘prefer’ this pattern because they are poorly paid and cannot afford to have their family with them, when the real explanation lies in the structural causes of low pay. Migrants’ preferences are self-evidently opposed to poor wages. Such radical approaches tend to be wedded to their own meta-narratives and can therefore suffer from the same problem of unilinearity as those rooted in ‘modernization’ theories. In migration studies, the rise of post-modernism as a reaction to the overly deterministic theorizing of the meta-narratives of modernization or Marxism, led to something of a reversion to the sort of focus found in methodological individualist studies. Certainly, the importance of what was now often termed the ‘agency’ of individuals has been promoted again. The associated literature has often been rich in case study material promoting the ‘voices’, or individual decision-making and life histories of migrants. Qualitative, ethnographic methodologies are often preferred. Although the material generated is often absorbing to read, discerning trends and patterns generalizable to other situations is frequently difficult and indeed may be specifically resisted as unsound or undesirable: thus agency is preferred over structure. My research in Zimbabwe has certainly been influenced in some of its methodologies and interest in migrants’ perceptions by some of these new directions in social science research, and these proved highly valuable. At the same time, it would have proved impossible to track the sorts of changes felt to be relevant without the base-line quantitative aspects of the surveys repeated in each round. Methodologically, therefore, like many other social scientists I would agree that a combination of approaches 4

Re-inventing the Wheel?

yields the most satisfactory results. Theoretically, the difficulties of situating a mass of individual narratives within a broader framework, allowing for more general insights, can be as much of a drawback for postmodernist approaches as the supposed inflexibility of single meta-narratives. The need to reconcile the contributions of structure and agency to the nature of migration was one of the main conclusions in Gerold-Scheepers and van Binsbergen’s (1978) review of African migration studies thirty years ago, before the advent of postmodernism, and doubtless these debates will continue. My own position is generally for the prioritizing of structure over agency, but in terms of the reconciling approach as in Murray (1994/5: 25)4 who argued that migration patterns in Africa have to be understood in terms of how different societies experience ‘particular historical processes of differentiation and integration’ as they are incorporated into capitalism. His approach falls generally within the more radical structuralist perspectives typical of southern African migration studies. However, the insistence on sensitivity to geographical and historical particularity helps to overcome the overly determinist analyses of migration associated with Marxist structural perspectives. My analysis of Zimbabwe’s migration experience therefore identifies both patterns and trends which exemplify experiences generally in sub-Saharan Africa or sometimes only more locally in southern Africa, as well as features which can only be understood in terms of its specific national trajectory. An alternative critique of African migration studies is that, far from the field being dominated by unilineal models, it has long been characterized by ‘an abundance of case studies’. This was a view that pre-dated postmodernism and was also levelled at the structural functionalist school for its failure to provide an ‘integrated theory of migration’ (Gerold Scheepers and van Binsbergen 1978: 24, citing Gould 1974, Magubane and O’Brien 1972). Twenty years later Gould (1995: 129) was still complaining that the ‘apparent anarchy in [African] migration studies remains, for there is no emerging or over-arching context into which these potentially insightful surveys can be set’. If anything, this critique is about too much, rather than too little, flexibility. The argument appeared in an edited volume on African internal nonmetropolitan migration (Baker and Aina 1995) which contained three very different theoretical takes on how such migration should be theorized: by Amin (1995) who remained wedded to the Marxist structuralist views he had propounded since the 1970s; by Krokfors (1995) who argued for the prioritization of cultural determinants of migration; and by Evans and Pirzada (1995) whose theorizing fell into the neo-classical economics school. Taken in the round, this volume provided evidence both of the continued existence of theoretical fault lines in African migration studies and of a wealth of empirical material,5 some of which is considered in the next two chapters.

4 5

Murray’s paper was originally presented at the Institute of Commonwealth Studies in 1979, but not published at the time. An overview chapter by Adepoju in the volume edited by Evans and Pirzada reviews a study by Oucho and Gould (1993) in which over 1500 individual studies of migration in Africa are categorized.

5

Introduction

Circular Migration: Theories, Models & Measurement Circular migration, as a special type of migration, has been subjected to much the same modelling and theorizing as discussed for African internal migration in general. The phenomenon is often incorporated into works on rural-urban linkages more generally (e.g., Tacoli 2006; Environment and Urbanization 1998, 2003; Beall et al., 1999) but, not being the main focus, is usually treated in a fairly general way. The most comprehensive and focussed contribution to this specific field of study remains M. Chapman and R. M.Prothero’s edited volume, Circulation in Third World Countries, which includes an excellent review, starting from the works of the nineteenth century French geographer, Vidal de la Blache, through Zelinsky’s migration transition model, to the 1980s (Chapman and Prothero 1985b). It revealed the continuing salience of circular migration in a wide range of countries in Asia, Latin America and Africa. The editors also squarely addressed one of the key debates about circular migration which is whether it is a type of movement which will inevitably disappear as time goes on – that is, is it a transitional element first created by modern social change which gradually and irreversibly dwindles as further change occurs? This was the essential position of an earlier review of the field by Nelson (1976: 732) who maintained that, ‘[c]urrent African and South Asian patterns of temporary migration will almost surely give way to permanent patterns’. For structural functionalists wedded to modernization theories, and the dualistic modelling this frequently implies, this pattern would ebb away as economies modernized and progressed under capitalism and residual, ‘traditional’ sectors were transformed. For Marxist structuralists it could play a key and sometimes institutionalized part in early stages of ­capitalist incorporation by shifting most of the costs of the reproduction of labour onto rural households. Thus it was part of the active process of underdevelopment, but it would be undermined by any processes which strengthened full proletarianization and should become unnecessary in a socialist future. Chapman and Prothero opposed these standpoints, arguing strongly against any position holding that the phenomenon could usefully be analysed in relation to western experience which, in turn, meant rejecting unilineal modelling. This was a position broadly recognized at the time in work on circular migration in Tropical Africa, but much less so in studies emanating from southern Africa. Such regional differentiation in interpretation was noted by Chapman and Prothero (ibid.: 18), and is further demonstrated in the following chapters which analyse this in detail and incorporate the contributions of more recent literature. Chapman and Prothero also related their review to the sort of typology of migration theories outlined in the previous section, and agreed with both Murray, and Gerold-Scheepers and van Binsbergen, to whose work they specifically referred, about the need to reconcile the contributions of structure and agency to the nature of circular migration. For Chapman and Prothero (ibid: 24), the problem with structural models was that they ‘emphasise discontinuities rather than continuities in the process of circulation over time, while the socioeconomic changes portrayed have the same remorseless and preordained character as do Zelinsky’s “phases of mobility’’’. They agreed with Bedford (1973, cited in Chapman and Prothero 1985a: 16) that: 6

Re-inventing the Wheel? [t]here is no transitional sequence [for circular migration] applicable to all societies, even though there is dependence between mobility and socio-economic changes associated with modernization. The manner in which these relationships are manifested in movement behaviour … is very much dependent on the society in which change occurs …

There are evident parallels between this position and Murray’s call for incorporating ‘particular historical processes of differentiation and integration’ (1994/5: 25) into any analysis of migration. Chapman and Prothero’s volume provided a fairly definitive debunking of the idea that circular migration could be theorized via any neat transitional model, ending with its demise. Nonetheless, the debate has continued, quite often without any recognition of the discussions and work that have gone before so that some theoretical and conceptual ‘wheels’ are reinvented. Various arguments are also rooted in different understandings of circular migration itself, with considerable differentiation in the definitions employed either implicitly or explicitly. For example, some studies focus on intra-rural circulation which may be strongly seasonal in character. Others are based on institutionalized circulation where labourers are contracted for a specific period to work on a mine (e.g., for one year in South Africa). Others look at rural–urban migrants who do not take their families with them, oscillating between their rural ‘homes’ where they are still regarded as part of the local household, and spending perhaps only a few years in town in total through their work careers. Yet others consider rural-urban migrants who have brought their immediate family to town and could not usefully be regarded as part of a rural household (even if they retain affective and economic links to their area of origin and are still linked by lineage), who have lived and worked in the town for many years, and intend nonetheless to return to a rural area in the future, possibly after a full working life in town. Geographical literature is more likely to recognize that there is a spatial and temporal spectrum of types of circular (and other) migration and therefore that interpretations of trends and shifts need to take this into account to avoid disagreements based on definitional misunderstandings. There are a number of descriptive typologies (e.g., Gould and Prothero 1975; Hance 1970: 162ff; Adepoju 1998) which provide guidance on such variations but usually acknowledge that they offer no explanatory power. All the migration types listed above (and others) can be understood as the outcome of structural forces acting on particular societies at particular times, and all of them are circular in terms of a fundamental characteristic of the migrant: that is, he or she does not remain permanently in their place of destination but returns to their place of origin. It is also quite possible that different trends can be simultaneously operating for these different types of circulation in any one society, and a failure to distinguish between types can then lead to confusion or contradiction. It is argued in Chapter 3 that this partly explains various contradictions in current South African literature on migration. A useful discussion of the confusion that can emerge from attempts to analyse trends in migration which involve simultaneous changes in the scale, the length of time people stay, and the proportion choosing permanent migration, can be found in O’Connor (1983: 67). He noted that in any African city: some migrants have come for only a few months. … some have come for a year or two … and may do so for a second and third time, some have come intending to stay for most or all of their working lives but then to return to their rural homeland, and some regard the

7

Introduction move as permanent. The uncertainties relate to the relative size of each group, and also to how far reality matches intentions.

Neither censuses nor small-scale surveys can capture all these variations adequately, if at all. O’Connor goes on to point out that conflicts between analysts who separately maintain that a city’s migrant population is mainly long-term or that most migrants only come for a short time can be more apparent than real, because these are not mutually exclusive. For: if each year 2,000 people came in for a long stay of ten years or more, while 5,000 came in for short stays of around one year, the migrant population would soon consist mainly of longstay individuals. Indeed, for many cities something like this is the case today, although of course the length of stay occupies a wide spectrum rather than two discrete categories. (ibid.)

Were these points more often kept in mind, some of the debates might be clarified and conflicts about migration to African cities avoided. Let us take another example, based on the figures above but assuming a significant downshift in the proximate forces facilitating or encouraging long-term migration (e.g. a serious fall in real urban incomes or rigorous enforcement of laws preventing informal housing solutions). Let us assume that this leads to a fall in long-term migrants to 500 per year, while short-term circulators remained at the same level. This would mean that at the end of twenty years, the net addition to the urban population would be 10,000 plus another 5,000 who would circulate out shortly (to be replaced the following year by another short-term 5,000). In the first example the respective figures would be 40,000 plus 5,000. Thus, not only do such differential mixes reflect shifting structural circumstances but they have very different impacts on the rate of urban population growth derived from migration. The definitional problem has another aspect because even the seemingly obvious issue of whether migration is permanent as the principle distinction between circular migration and other types may be treated differently (Gould and Prothero 1975: 41; Hance 1970). Sometimes the concept of stabilization, whereby a migrant has lived for many years in town probably with his or her immediate family, is used. This has conceptual value but may be problematic if it implies that an urban-rural move in the future is ruled out. The issue is illustrated by the discussion of regional literatures found in the next two chapters. The definition of circular migration may also vary in respect of whether it is assumed that the urban-rural move must be back to the original place of origin, therefore involving crucial kinship and cultural issues. In my view, while this is probably the most common type of circular migration in Africa, it is an overly narrow definition. Many migrants do not move straight to a particular town, stay there, and come back. They may move first to another rural area, then a town, and then yet another town. They may then plan to return to either of the rural areas, or to yet another. Women often move to their husband’s rural area, not their own place of origin. The research in Harare found that these sorts of trajectories were sufficiently common amongst migrants for it to be helpful to use a structural definition of circular migration. The essential defining characteristics were that the migrant did not anticipate remaining in Harare and expected to leave for a rural area at some future time. Where the main focus of the research is on the nature of urbanization in relation to urban livelihoods, economies and demography, rather than cultural affiliations, this broad definition is essential. Thus the crucial question was not whether migrants 8

Re-inventing the Wheel?

intended to return ‘home’, although this issue was fully incorporated, but whether they felt that an urban livelihood could sustain them until the end of their lives and, if not, whether they anticipated a rural alternative in the future. For this reason, the term ‘return’ migration is generally avoided in the analysis of the Harare research because of the numbers of migrants who planned eventually to farm somewhere other than in their birthplace. In sum, the definition used focussed on how the combination of social and economic structural forces influencing cohorts of in-migrants at particular times is translated into changes in their propensity to remain or circulate and the length of their expected stay. These factors not only reflect the nature of the city economy but also have important implications for the cities themselves because they strongly influence the nature of urban population growth which has major policy and resource implications. The combination of permanent migration and various types of circular migration across the temporal spectrum can thus vary over time for different cities or urban systems as they, and the rural areas from which they derive their in-migrants, are differentially incorporated into global capitalist processes. All this is then mediated by more local structural social, economic and political factors. There is nothing in such theorizing that assumes that migration will become more permanent, or that there will be particular trends in the mix of circulators. Nor, indeed, does neo-classical migration theory in its crudest income maximization format. If there is no presumption that the real rural-urban income gap (allowing for urban living costs) must always remain strongly in favour of towns, it is obvious that individual migrants may sometimes maximize their incomes in urban areas and at other times in rural areas. The structural factors that facilitate such moves back and forth (e.g., the nature of land tenure) will then influence the extent to which such ‘income maximization’ can be realized. Further, there is no need to assume that a ‘return’ move is permanent; ten years later the situation may reverse and urban livelihoods may yield more. It is only when migration theorizing is wedded to meta-narratives that increasing circulation is treated with suspicion and loaded with ideological baggage. The modelling (rather than broader theorizing) of circular migration between African rural and urban areas has often been conceptualized in terms of a push-pull model (e.g., see Wilson 1972). This is usually deemed to be at best old-fashioned, and at worst useless, because it is presumed these days to imply agreement with dualistic interpretations of societies along the lines embedded in much structural-functionalist theorizing of migration. The factors generally identified are typical of neo-classical economic and structural-functionalist approaches to migration. Although I am strongly opposed to dualistic theorizing which so frequently leads to misunderstandings and poor policy recommendations (see Potts 2008), in my view the push-pull model still has considerable value as an organizing framework for evaluating changing patterns of circulation. The dualism is not inherent and is overcome by recognizing that both rural and urban conditions are shaped by the same structural forces. The model is helpful both because of its extreme simplicity and its incorporation of circularity (see Figure 1.1). In the case of many sub-Saharan African countries in the 1960s and into the 1970s, for example, there was a real increase in urban incomes and employment opportunities compared to the colonial period, and a rise in education levels. These can be understood as urban ‘pulls’, while many of the factors which had pushed migrants out of urban areas, such as restrictive colonial influx controls and housing policies, and institutionalized segregation no longer operated. Rural 9

Introduction

Figure 1.1  Push-pull model of circular migration

‘push’ factors may not have changed as quickly or in such obvious directions, given the continued intervention of marketing boards set up in the colonial era. However, trends in taxation of peasant production via the pricing policy of such boards would become very important in determining rural–urban income gaps, with major differences between countries. On the other hand, the general trend in welfare indices in rural areas was positive at this time, reducing rural pushes from these factors; although since the improvements were even more marked in urban areas, the relative impact was to increase urban welfare pulls. Some urban push factors did not change much, since the general lack of a state-sponsored welfare net for the old, ill or unemployed remained. In the short term, there was little change in some of the pull factors back to rural areas stemming from social and cultural links, and the economic security presented by land. After a few years of structural adjustment policies in the 1980s, however, the balance between these forces went through major shifts. Urban opportunities and incomes dwindled, as did the benefits derived from better health, education and physical infrastructure, reducing urban pulls. A massive rise in the urban cost of living increased urban push factors and those derived from the lack of a welfare net persisted and became more pressing. The rural-urban income gap was sharply squeezed. Rural push factors became more geographically varied in some countries as areas far from roads and markets, once helped by pan-territorial pricing policies, were strongly disadvantaged by the privatization of trade. More advantageously located rural areas might benefit from liberalization depending on the degree to which they had previously been taxed under government pricing regimes, although in every case this had to be balanced against input price rises. In many societies, the broad social and cultural parameters of rural pull factors may not have changed much under structural adjustment, although urban migrants’ capacity to fulfil various obligations 10

Re-inventing the Wheel?

to their rural kin did reduce. On the other hand, the significance of the economic security aspect of rural land for farming and housing will have been much enhanced. The balance between these shifting forces in particular countries or regions helps to indicate probable changes in the direction, magnitude and speed of migration between rural and urban areas, as does the timing of the shifts. The shifts themselves, and the relative circumstances of the rural and urban areas, can only be understood in terms of the broader political economies within which they are situated: i.e., colonial or early postcolonial modernizing states or societies that are structurally adjusting and increasingly market-dominated. The nature of rural land tenure, which is a key element of any political economy, is one of the main determinants of circular migration by poor people. It is generally facilitated by indigenous tenures where purchasing power is not the main route to gaining access to land, and it is made more difficult where market forces predominate. Overarching these national structural conditions are global shifts in economic ideology, commodity prices, geo-political imperatives and economic growth. Trends in migration to and from cities are crucial elements of African urbanization. However, so is the impact of other demographic factors, especially since population growth in urban areas is not now generally driven primarily by net in-migration but by the natural increase of the city population: from the excess of births over deaths. Literature on African cities often assumes that city-born people are permanent urban residents (e.g., Macmillan 1996; Bryceson 2006c; Hance 1970; Mayer with Mayer 1961; Plotnicov 1967). This important issue will be discussed at various points in this book, as the situation is seemingly less definite, even if their propensity for permanence is much higher than for in-migrants. Adding natural increase to the urban equation makes the tracking of migration trends even more complex, especially in Africa where the data are often hard to come by. Any urban growth rate can be derived from a variable mix of birth and death rates, and in- and out-migration rates. This is depicted through a variety of scenarios in Figure 1.2. All three main scenarios assume the same rate of annual natural increase in urban areas of 3%, which was typical in many countries up until the 1990s, although there have been reductions since in many towns due primarily to falling fertility and rising adult mortality from HIV/AIDS (Potts 2006). In the first scenario, generally typical of sub-Saharan Africa urbanization in the 1960s and into the 1970s, the rate of in-migration is very high: at an annual rate of 8%. Some circular migration occurs, with out-migration at 3%. Taken together, migration and natural increase yield very high annual urban population growth of 8% which would lead to the population doubling in just under nine years. The second set of scenarios all yield a rate of population growth of around 3%, but assume very different patterns of in- and out-migration. Note that Scenario 2A, where in-migration occurs at an annual rate of 6% yields a slightly lower overall growth rate than 2D where there is very little in-migration (but no out-migration). In the third set of scenarios, urban growth falls below the rate of natural increase even though in-migration continues in 3B and 3C, and is negative in 3A and 3C. These scenarios are merely examples of the huge variety of mixtures of demographic factors that influence urban population growth, but serve to illustrate some key points about the analysis of African urbanization. First, they demonstrate both how towns growing at similar rates may be experiencing very different migration scenarios, implying that markedly different economic and social factors are at work. 11

Introduction

Figure 1.2  Urban growth scenarios

Second, they show how gross levels of in-migration can be misleading as predictors of urban growth. Third, they illustrate how the rate of out-migration, which in Africa is so often strongly influenced by the speed of circular migration, is a key determinant of actual growth outcomes. These points help to explain apparent contradictions which scholars of both rural and urban development, as well as city authorities, sometimes argue must be evidence of continued (often described as ‘unabated’), very rapid rates of urban growth which must be driven by in-migration. First, they point to evidence that the city is physically expanding, with the development of new (usually low-income) residential areas. Second, they may refer to studies and surveys that show that many people are migrating to town from rural areas and that many new in-migrants are to be found in town. However, while both these situations may be accounted for by high rates of net in-migration (e.g., Scenario 1A), they may derive from other scenarios. New household formation from within the existing urban population may be the primary cause of residential expansion; thus a town experiencing Scenario 2A where there is no in-migration would still double its population in 23 years. Evidence of significant rural-urban mobility could fit with Scenarios 2B, 2C and 3B and 3C, none of which produce an overall growth rate in excess of natural increase. It will be shown later in this chapter that the impact of economic decline in some African countries has led to periods when urban growth rates have dropped to levels not very different from natural increase, or even lower, due to some combination of raised levels of out-migration and falling (but possibly still high) in-migration, in line with these scenarios. It is, however, rarely possible to know the precise combination of factors for even one city, due to the perennial lack of adequate data in most African countries. Without comprehensive birth and death registration, or regular surveys of internal migration which distinguish rural and urban destinations, analysis must 12

Re-inventing the Wheel?

rest on periodic censuses triangulated with whatever other relevant survey material is available. Even censuses which are deemed reasonably reliable may not allow the disentangling of different components of urban growth. Their usefulness depends on whether it is possible to discern birth and death rates for rural versus urban areas, and what sort of migration data were collected and published. Some censuses include information on people’s whereabouts one year before the census and/or at the time of the last census. Depending on the scale at which these data are then aggregated, it may be possible to discern key trends and patterns in internal migration and to make some reasonable deductions about the nature of circular migration. Nonetheless, circular migration is worst served of all by almost any type of data gathering, be it a national census or a localized survey. A typical ten-yearly census round essentially misses all circulatory movements by individuals, although comparing in- and out-migration totals for the time intervals available may yield some crude surrogates. Even these are usually at the provincial level, making it even more difficult to assess rural-urban-rural flows except where there is a high degree of correlation between provincial and urban boundaries (as is sometimes the case for major cities, particularly capitals), or where some provinces are known to be largely rural. An alternative is merely to compare city growth rates derived from censuses with their natural increase rates which may be available from other sources, such as national Demographic Health Surveys. This provide a guide to the importance of net in-migration to overall growth and educated guesses can then be made about how this is derived from possible combinations of in- and out-migration. Small-scale surveys in one locality which include an interest in return or outmigration face a key methodological difficulty: out-migrants are, by definition, not available as research subjects. Questions about their motivations and characteristics, therefore, can never be answered directly. The best that can be managed is to ask migrants still present about their future plans and, if they are potential ‘leavers’, use their profiles as surrogates for recent out-migrants. This suffers from the obvious problem that intentions may not be translated into action so that, as with any surrogate methodology, it is an imperfect solution. This has been recognized for many decades in African migration research (e.g., Plotnicov 1967; Peil et al., 1988). Usually, it is suggested that there is a tendency to overestimate out-migration because it is felt that African urban dwellers are more likely to say they will leave than realize their plans. In itself this view, if correct, is an implicit acknowledgement of many people’s preferences for an eventual rural future and therefore indicative of the strength of rural-urban links. Yet, there is also the view that most urban dwellers view a rural livelihood very negatively and do not, and will not, contemplate this as a possibility. If this were the case, then it becomes logical to take stated future plans about outmigration as realistic or an under-estimation of what will occur, on the basis that some who eschew the possibility may be forced out, despite their plans. It is probably not possible to square this particular circle, especially since the balance between plans and eventual actions must surely shift back and forth as the structural factors influencing migration change. What can be stated, however, is that if there are very obvious shifts in the plans proffered by migrant cohorts in one urban area, sampled at different times using the same methodology, this definitely shows the direction in which the rate of out-migration is moving, and that there has been significant change in the structural conditions of migration. This has been the essential methodological basis of the research in Harare. 13

Introduction

Trends in Rural-urban and Urban-rural Migration in Sub-Saharan Africa: The Empirical Evidence Since the ending of colonialism, most countries in sub-Saharan Africa have experienced three broad economic phases which have shifted the balance of forces influencing urbanization and patterns of migration, as outlined in the discussion of push-pull modelling. In the 1960s and into the 1970s global economic conditions were generally positive and African governments, under the influence of prevailing modernizing ideology and advice, embarked on development paths in which they played a key role, directing and making investments into what were then seen as strategic productive sectors such as import-substituting industrialization, and investing heavily in government services like health and education. These policies generally encouraged rural-urban migration and facilitated family migration and longer stays in town, if not full permanence. The oil crises of the 1970s rapidly dismantled, then reversed, the upward trend in urban livelihoods. Real urban incomes and welfare dwindled as most non-oil-exporting nations became heavily indebted, forcing them to turn to the international financial institutions which were by then dominated by neo-liberal, as opposed to modernization, ideology. The market-dominated economic policies subsequently enforced deliberately unpicked the government’s central role in economic development, sharply reduced government spending and public sector employment and, by liberalizing trade and thereby concentrating production in areas of comparative economic advantage, led to the closure, reduction or stagnation of swathes of previously protected urban-based production and formal employment. Urban incomes, already on a downward trajectory before SAPs, plunged. These general global factors and African policy shifts and their intensely negative impacts on African urban livelihoods, incomes and welfare have been detailed and critically analysed in a wide array of literature on sub-Saharan Africa in general (e.g., Adepoju 1993; Baker 1997; Becker et al., 1994; Bryceson 2006a, 2006b; Hansen and Vaa 1994; Jamal and Weeks 1993; Meagher 1995; Nelson and Jones 1999; Potts 1995, 1997, 2006; Rakodi 1997; Rogerson 1997; Simon 1992, 1997, 1999; Simone 2004a; Simone and Abouhani 2005; Stren 1992; Zeleza 1999) and a host of individual country or settlement case studies.6 While these works are in general agreement on the severe increases in urban poverty which have occurred since the imposition of structural adjustment, not all of them consider the consequences for migration. Those which do are not always in agreement about how migration has been affected. To some extent, this is a function of sources: analyses which rest largely on institutional data compilations (i.e., World Bank or United Nations) on African urban populations and growth frequently concluded that, despite the negative transformations in urban economies and livelihoods, migration rates have been little affected. Consequently, it is argued that urban growth rates have not reduced (e.g., Jamal and Weeks 1993; Simon 1997; Jamal 1995) or, even if it is noted that there has been some reduction in growth in the largest cities, that the overall trend in the increase of national urbanization levels 6

These are too numerous to cite, but some studies incorporating the issue of rural-urban linkages and circular migration are discussed in Chapters 2 and 3.

14

Re-inventing the Wheel?

has not slowed (Bryceson 1996c). Others argue that net in-migration has reduced in response to urban economic declines (e.g., Becker et al. 1994; Zeleza 1999; Baker 1997b; van Dijk et al., 2001; Findley 1997; Tabutin and Schoumaker 2004). My own census-based research on trends in a range of mainly East and southern sub-Saharan African countries since the 1960s supports this view (Potts 1995, 1997, 2006). Other work based on large-scale migration surveys and some national censuses in francophone West Africa concurs (Beauchemin and Bocquier 2004; Beauchemin 2002a, 2002b, 2005, 2006; Beauchemin, Sabine and Schoumaker 2004). Even where it has been ascertained that urban growth has slowed in some African countries, the role of shifts in circular migration is not always clear due to the many methodological and statistical difficulties discussed earlier. The following section reviews a range of evidence on downward shifts in the growth of African urban settlements in different countries and what is known about the causes. It is helpful to place this in a broader international comparative context of debates and evidence about the nature of contemporary urbanization trends in developing countries, since these have undergone some important revisions. Many of these lend weight to the view that migration rates have been sensitive to economic change and that circular migration has played a role in reducing urban growth. First, there has been revision of the historical comparisons often made between urban growth in the developing and developed world. In essence, it has long been argued that the urban growth rates experienced in the last half of the twentieth century in many poor countries were unprecedented. It is claimed they greatly exceeded the rates typical of Europe or North America during the Industrial Revolution, when urbanization in those areas was at its fastest. However, direct comparisons between these very different societies can be misleading, as the underlying demography is different. For example, many European cities at that time had very much higher urban death rates and lower birth rates than those typical of contemporary African cities. They would scarcely have grown, and sometimes would have contracted, without net in-migration. Thus, most of their growth derived from migration from rural areas. By contrast, high urban natural increase rates in lower and middle income countries across the world is now recognized to have been the major component of urban growth for some decades. This has been confirmed in work by the UN and Chen, Valente and Zlotnik (1998). It is estimated that about 60% of population growth in the 1980s in the ‘median city’ in such regions was due to natural increase, with the rest attributable to net rural-urban migration and reclassification of rural settlements as urban (Montgomery et al., 2004). Furthermore, rates of net ruralurban migration in these regions have not been historically unprecedented; it has now been shown that the rates in Western countries in the late nineteenth and early twentieth century were not dissimilar to those in these poor regions more recently (Montgomery et al., 2004: 93). The issue can be illustrated by comparing the rate of growth of levels of ­urbanization (i.e., the urban share of the population, rather than the absolute numbers of urban dwellers). Such levels expanded from 29% to 41% in lower and middle income countries between 1975 and 2000, roughly equivalent to the experience of the developed world from 1900 to 1925 (Brockerhoff 2000, cited in Montgomery et al., 2004). Although in sub-Saharan Africa, per se, urban growth in the 1950s and 1960s in many cities was very often obviously mainly due to net in-migration, this began to shift, so that by the end of the 1970s, it was at least half due to natural increase. 15

Introduction

In many cities this became the main cause. For Africa as a whole, it has been estimated that the contribution of net in-migration to urban growth was about 40% in the 1960s and 1970s but fell to only 25% in the 1980s (Chen et al., 1998). An important differentiating issue for sub-Saharan African towns is that their natural increase rates often prove to be very similar to, or sometimes higher than, rural rates. This can be shown individually for many countries, including Zambia, Malawi, Tanzania and Kenya (Potts 2005, 2006a), Zimbabwe (see Chapter 4), Côte d’Ivoire, Central African Republic and the DRC (Potts 1997). This sometimes comes as a surprise, given the known tendency for urban fertility to be lower than rural. UN analyses of urbanization tend to emphasise this point and assume lower urban than rural natural increase. Yet, in many African countries, even into the 1990s, this was more than counterbalanced by the youthful and fertile urban age profile compared to rural areas which boosted birth rates, and significantly lower urban death rates (Potts 1995, 2006). A detailed proof of this point for Zambia has been demonstrated (Potts 2005) and is also shown for Zimbabwe in Chapter 4. The rate of urban natural increase for the whole of Africa is actually estimated to have increased from about 2.6% per year in the 1960s and 1970s to 2.8% in the 1980s (Chen et al., 2004). The regionally differentiated impacts of HIV/AIDS, which are far worse in southern African countries - where there have also been sharp falls in fertility since the 1980s (Potts and Marks 2001) – have altered these parameters to some extent. However, the gap between rural and urban HIV rates has also often narrowed in this region, reducing any differentiation derived from this tragic factor. Recent Demographic Health Surveys in the 2000s still indicate remarkably little difference between rural and urban crude birth rates in most (but not all) countries, but record significant differences in infant and child mortality rates which remain the chief factor in determining overall death rates in most African countries. This suggests that urban natural increase is often still as high or higher.7 There have also been major revisions of the forecasts made in the 1970s and 1980s of urbanization levels and individual city populations in developing countries. These have usually been significantly reduced in the following decades by the United Nations (Brockerhoff 1999; Satterthwaite 2002). Even sharper reductions are argued for by Bocquier (2005) on the basis of economic modelling, rather than population projections. He concludes that even the revised UN projections for the year 2030 may overestimate the world’s urban population by 19%, or roughly one billion people. Vast mega-cities have generally grown more slowly than projected: Mexico City, for example, had been projected to have around 31 million in 2001, but reached only eighteen million people (Montgomery et al 2004). In an analysis of large city growth in less developed countries in recent decades, almost a third of the cities were found to have grown more slowly than their respective national populations. It is argued that 7

Recent Demographic Health Survey data for two countries where there has been very steep fertility decline in recent decades are illustrative. In Kenya and Zimbabwe, the gap between rural and urban fertility rates has increased, which might be expected to have led to urban natural increase falling significantly below that in rural areas. Yet, the birth rate gap remains small: urban birth rates in Zimbabwe from the 2005–2006 DHS were 28.5 compared to 32.0 in rural areas, and the respective rates in Kenya from the 2003 DHS were 35.3 and 38.1 Disaggregated crude death rates were not published in these DHSs but, despite the impact of AIDS, these are (so far) invariably still lower in urban rather than rural areas due to the much better infant and child mortality rates in towns. For example, the Kenyan DHS reported that rural child mortality rates were 26% higher in rural compared to urban areas (117 to 93 per thousand). As yet, therefore, a major difference in rural and urban natural increase rates does not appear to have emerged.

16

Re-inventing the Wheel?

this suggests that ‘on the whole, there has not been net in-migration to LDC cities from smaller urban areas and rural areas; the net flow may be in the other direction’. This conclusion is explicitly related to the possibility of reductions in rural-urban migration rates and strengthened ‘return migration to [small] towns and villages’ (Brockerhoff 1999: 769). Such a scenario is equivalent to that illustrated by 3B in Figure 1.1. Authoritative reviews of African urbanization compared to other global regions in recent major studies of world urbanization acknowledge the changes in the role and scale of migration. Thus, net rural out-migration rates in Africa are reported to have declined from 1.07 per thousand in the 1960s to 0.5 in the 1980s (Chen et al., 2004), which certainly fits with the idea that circular migration has strengthened. The calculations are based on censuses from seventeen African countries, including five from North Africa, but their coverage of the time period is incomplete. Specific rates have to be treated with some caution, therefore, even if the direction of change can be regarded with greater confidence. A recent analysis of African urbanization by Bocquier (2004) argues not only that there has been a rapid slowing of urban growth but also that ‘de-urbanization’ is indicated by a range of demographic and economic indices. Furthermore, for sub-Saharan Africa, excluding southern Africa, it is argued that available data suggest that levels of urbanization may stabilize at under 50%, or even less if ‘urban’ is defined as settlements over 10,000 people. According to Bocquier, the explanation for this requires, inter alia, a better understanding of ‘neglected movements from urban to rural areas’ (p 149) which is largely a call for more research on circular migration. Such broad statistical continental reviews are always open to debates about the adequacy of the data and need to be triangulated with national or urban case studies for confirmation. However, the downward revision of individual mega-city populations which has been significant elsewhere is less relevant in the case of sub-Saharan Africa. First, there are few mega-cities. Outside the major urban agglomeration in Gauteng Province in South Africa, centred on Johannesburg, which has experienced strong in-migration since 1994, only Kinshasa and Lagos can be so considered. Second, there are major issues with data reliability for these latter two cities. There is no recent reliable estimate for Kinshasa, as there has been no census in the troubled DRC for decades, although the UN suggested 5.1 million for 2000 (Montgomery et al. 2004: 99). Censuses in Nigeria are all deeply contested for political reasons. The inability to evaluate the nature of urban growth in two of sub-Saharan Africa’s three largest cities, and for Nigeria as a whole (which has the largest number of large urban settlements in Africa), is a major problem for scholars in this field. They frequently lament the difficulties created by lack of reliable or recent data (e.g., Gould 1995; Oucho 1998). The Nigerian censuses have, however, indicated not only a sharp downward revision in the assumed national population in the early 1990s, but also led to the UN similarly reducing Lagos’s projected size from 13.5 million in 1994 to 8.7 million in 2001 (Montgomery et al., 2004). In 2006 the census reported a population of 9 million for Lagos State and 7.9 million for Metropolitan Lagos, well below the figure of around 15 million frequently still cited. This yields an annual metropolitan growth rate since 1991 of 2.9%, below the national rate of 3.2%. The rate for Lagos State as a whole was 3.1%. The census data for Lagos were immediately attacked as a ­political manipulation of the national regional balance between north and south, but the figures have been vigorously defended by the national statistical office as 17

Introduction

reasonable.8 If the data are broadly meaningful, and are set alongside the body of survey material indicating increased circular migration out of Lagos (see Potts 1995), they indicate that inflows have been countered by outflows and that circular migration has indeed been significant. Out-migration to other Nigerian towns almost certainly also played a part. Beyond the mega-cities, however, support for the general indications of the broad studies on African urbanization discussed above is plentiful. Reports from a wide range of individual countries across sub-Saharan Africa show that population growth in many large urban settlements has slowed and, in some cases, that growth in the levels of overall urbanization is also reducing. This is sometimes countered by an expansion in the numbers living in small, and sometimes very small, towns. For East and southern African countries, I have previously provided a detailed analysis of available census data from the 1960s until the early years of the 2000s. My studies have also indicated the frequent occurrence, from about the 1980s, of major towns growing at rates which imply that the role of net in-migration is only fairly small, and sometimes probably or certainly negative. This is explicitly related to shifts in global and national economic policies and their impacts on urban livelihoods and migration patterns, including circular migration (Potts 2006). When analysed on a country by country basis, rather than at the aggregated regional level, the vital significance for interpreting urban growth patterns of national variations becomes very clear (ibid.). These include factors such as the nature and timing of the de-colonization process, the imposition of structural adjustment, and the effects of wars and conflict (e.g., in Angola, Mozambique, Rhodesia, Uganda, Rwanda). Much of the literature on urbanization assumes a fairly stable political environment in which economic and social factors influence the nature and growth of towns. Frequently, this has not been true in sub-Saharan Africa. Virtually all migration or urban theory has to be abandoned in order to interpret the impacts of life-threatening conflicts. In addition to censusbased analysis, I have also previously reviewed a wide range of survey evidence from countries across sub-Saharan Africa, including Ghana and Nigeria, which identified increasing rates of return migration from towns or falls in the attraction of cities for rural migrants (see Potts 1995, 1997). In a more recent study from Nigeria on urban-rural migrants’ farming activities, for example, Arene and Mkpado (2002: 118) asserted that structural adjustment resulted in [urban] ‘poverty becoming the hallmark of the economy. Consequently, to avoid being crushed by the receding economy, the counter-urbanization process began.’ The strengthening of urban-rural migration as a response to structural adjustment has now also been specifically identified for Cameroon (Gubry et al., 1995) and suggested for Malawi (Chirwa 2001). Slow growth of individual towns, for long periods of time, sometimes below national rates or at rates close to those attributable to natural increase, has been identified in Ghana, Uganda, Kenya, Tanzania, Zimbabwe, Swaziland and South Africa (Potts 1995, 2006). The most extreme case is represented by Zambia where a clear trend of net urban–rural migration has been established for at least twenty years. The level of urbanization reduced from 40% to 36% between 1980 and 2000 and 8

Some of the dispute rests on misunderstandings about the differences between populations within state versus urban boundaries, and has easily been countered by the statisticians. The census is also generally in line with what is known about national demographic indices from other surveys (see National Population Commission 2007).

18

Re-inventing the Wheel?

circular migration has been very strong. There is also evidence of some urban-born people leaving towns and of absolute falls in the populations of some Copperbelt towns (Potts 2004, 2006). Studies of urbanization and migration in francophone West Africa have identified even clearer downward trends in urban growth in a number of countries, as well as increases in circular migration out of towns which have been explicitly related to the declines in urban living standards due to structural adjustment (e.g., Beauchemin and Bocquier 2004; Beauchemin 2002a, 2002b, 2005, 2006; Beauchemin, Sabine and Schoumaker 2004; Bocquier and Traore 1998). These studies are based on censuses and survey data from a large-scale migration project across the region covering 1988 to 1992 (NESMUWA – the Network of Surveys on Migration and Urbanization in West Africa). In addition, new census material from the end of the 1990s into the 2000s is available for a number of West African countries. These contain sometimes quite startling evidence of counter-urbanization at the top end of the urban hierarchy, although whether this extends to include the largest town varies between countries. A compilation of these census data from across the continent (excluding countries where conflicts have distorted migration patterns) supplemented by the NESMUWA results, is presented in Table 1.1. This indicates where strong or weak net out-migration, or negligible net migration, can be identified for individual cities or groups of towns for particular time periods. The data in this table form the backdrop to the following analysis of the accumulating evidence of major shifts in migration and urbanization trends in individual African countries. In Côte d’Ivoire, circular migration has been of central significance for decades. At the end of the 1970s, a national report calculated that only Abidjan, the largest town and former capital city, had net in-migration, while other towns had net out-migration. At this time, between a quarter to a third of all internal migrants including ruralrural, were return urban-rural migrants. Rural forest zones were more attractive to migrants than towns, except for Abidjan. By the end of the 1980s, as economic conditions became more difficult, even this city began to experience net out-migration to rural areas of national citizens (Beauchemin 2005) at a rate of 12,000 people per year (Bocqier and Beauchemin 2004). If foreign immigrants are included, its net migration rate from 1988–92 was zero.9 These trends continued through the 1990s, so that Côte d’Ivoire experienced actual counter-urbanization in the same way as Zambia: its level of urbanization, if defined as the population share of settlements with over 5,000 people, reduced from 46% to 43% between 1988 and 1998. The NESMUWA results indicate that rural out-migration rates to urban areas for both sexes and all ages declined from the 1970s to the 1980s (Beauchemin et al. 2004). Abidjan again experienced net out-migration in 1997–98 (ibid.), although calculating its average annual growth rate for 1988–98 from published census material shows that it was 3.8%,10 a little higher than the national rate of 3.3%. This must have been accounted for by high natural increase rates, or international immigration, since the city was not attracting and keeping local in-migrants. Three of the four next largest towns all grew at less than the national growth rate and thus certainly experienced net urban-rural   9 10

From 1988 to 1992, the country as a whole de-urbanized, as the population of urban settlements with over 5,000 people, taken together, diminished. Calculated from census data for 1988 and 1988 (published in Brinkhoff 2008: http://www.citypopulation. de/).

19

1989–1999

1987–1998

1988–2000

Kenya

Mali

Mauritania

2.7

1984–2000

2.4

2.6

3.0

3.3

2.9

National AAGR % 3.3

2.6

Time period 1992–2002

Burkina Faso 1990s 1996–2006 Cote d’Ivoire 1988–92 1988–98 1988–98 1997–98 1988–98 Ghana 1970–1984

Country Benin

Category of urban settlement(s)* Cotonou (capital) Porto Novo (2nd city) All towns >10,000 All towns All towns>10,000 excl capital city Abidjan (largest city) All towns >5,000 Abidjan Abidjan 3 of 4 next largest towns Accra (capital city) Kumasi (2nd largest) >100,000 (excl Kumasi) 5 of 10 largest towns Nairobi (capital) Mombasa (2nd city) Nakuru (4th city) Eldoret and Kisumu (3rd, 5th) Next three largest towns All towns Bamako plus 2nd largest town 3rd, 4th and 6th largest towns Nouakchott (capital) Nouadhibou: 2nd town 3 of 6 next largest towns 4.0 0–0.6 2.6 1.5 ≤2

< 3.3 3.2 2.6 3.3 2.1 4.9 3.7 2.9 4.9 2.0