Institutions and Emerging Welfare States

Social Assistance in South Africa and Brazil

A Dissertation

submitted in partial fulfilment of the

requirements for the degree

Doctor of Political Science (Dr. rer. pol.)

to the Department of Political and Social Sciences

of the Freie Universität Berlin

by

Leon Amos Schreiber

Berlin, June 2015

Supervisor

Prof. Dr. Susanne Lütz

Professor for International Political Economy

Department of Political and Social Sciences

Freie Universität Berlin

Second Examiner

Prof. Janis van der Westhuizen

Associate Professor

Department of Political Science

Stellenbosch University

Date of dissertation defence: 18 November 2015

1
Contents

Contents 2

Acknowledgements 7

List of Figures 10

List of Tables 12

Abbreviations 14

1 Introduction 19

1.1 Background and Rationale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

1.2 Research Question and Hypotheses . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

1.3 Case Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

1.4 Significance of the Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

1.5 Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

2 Conceptual and Analytical Framework 29

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

2.2 Emerging Welfare States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

2.3 Social Assistance: An Integral Part of Social Protection . . . . . . . . . . . . . . . . 31

2
 2.3.1 Labour Protection and Social Insurance . . . . . . . . . . . . . . . . . . . . . 33

2.3.2 Social Assistance: The Core of the Emerging Welfare State . . . . . . . . . . 34

2.4 Analytical Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

2.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

3 Literature Review 42

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

3.2 Classifying Emerging Welfare States . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

3.3 Social Assistance in the Developing World . . . . . . . . . . . . . . . . . . . . . . . . 48

3.4 Historical Institutionalism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

3.5 Explaining Variation in Social Assistance Provision . . . . . . . . . . . . . . . . . . . 55

3.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

4 Research Design 60

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

4.2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

4.3 Data Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

4.4 Limitations and Delimitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

4.5 Ethical Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

4.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

5 Comparing Social Assistance in South Africa and Brazil 69

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

5.2 Social Protection in South Africa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

5.2.1 Social Assistance in South Africa . . . . . . . . . . . . . . . . . . . . . . . . . 71

5.3 Social Protection in Brazil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

5.3.1 Social Assistance in Brazil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

3
 5.4 Comparing Social Assistance Systems . . . . . . . . . . . . . . . . . . . . . . . . . . 91

5.5 Dependent Variables and Social Assistance Typologies . . . . . . . . . . . . . . . . . 98

5.5.1 Dependent Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

5.5.2 Typologies of Social Assistance . . . . . . . . . . . . . . . . . . . . . . . . . . 99

5.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

6 The Development of Social Assistance in South Africa 103

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103

6.2 Deep Institutionalisation of Social Assistance Norms . . . . . . . . . . . . . . . . . . 105

6.2.1 Social Assistance before 1994 . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

6.2.2 The Legal Codification of Social Assistance Norms in South Africa . . . . . . 124

6.2.3 Social Assistance Norms in South Africa . . . . . . . . . . . . . . . . . . . . . 131

6.3 State Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132

6.3.1 Concentrated Institutional Structure . . . . . . . . . . . . . . . . . . . . . . . 133

6.3.2 Political Power Concentration . . . . . . . . . . . . . . . . . . . . . . . . . . . 135

6.4 The Institutional Setting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137

6.5 Institutions, Actors and Social Assistance in South Africa . . . . . . . . . . . . . . . 137

6.6 The Policy Development Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141

6.6.1 Initial Policy Design (1994-1999) . . . . . . . . . . . . . . . . . . . . . . . . . 142

6.6.2 Attempts at Comprehensive Reform (1999-2009) . . . . . . . . . . . . . . . . 154

6.6.3 Consolidation (2009-present) . . . . . . . . . . . . . . . . . . . . . . . . . . . 159

6.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162

7 The Development of Social Assistance in Brazil 164

7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164

7.2 Weak Institutionalisation of Social Assistance Norms . . . . . . . . . . . . . . . . . . 166

7.2.1 Social Assistance before 1990 . . . . . . . . . . . . . . . . . . . . . . . . . . . 166

4
 7.2.2 The Legal Codification of Social Assistance Norms in Brazil . . . . . . . . . . 174

7.2.3 Social Assistance Norms in Brazil . . . . . . . . . . . . . . . . . . . . . . . . . 179

7.3 State Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180

7.3.1 Diffuse Institutional Structure . . . . . . . . . . . . . . . . . . . . . . . . . . 180

7.3.2 Political Power Fragmentation . . . . . . . . . . . . . . . . . . . . . . . . . . 184

7.4 The Institutional Setting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187

7.5 Institutions, Actors and Social Assistance in Brazil . . . . . . . . . . . . . . . . . . . 187

7.6 The Policy Development Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191

7.6.1 Initial Policy Design (1990-1997) . . . . . . . . . . . . . . . . . . . . . . . . . 191

7.6.2 Federal Coordination (1997-2003) . . . . . . . . . . . . . . . . . . . . . . . . . 199

7.6.3 Unification and Universalisation (2003-2006) . . . . . . . . . . . . . . . . . . 203

7.6.4 Expansion and Consolidation (2006-present) . . . . . . . . . . . . . . . . . . . 209

7.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214

8 Explaining Variation 216

8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216

8.2 Alternative Explanations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217

8.2.1 The Functional Hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217

8.2.2 The Ideational Hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223

8.3 The Institutional Explanation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227

8.3.1 Coverage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228

8.3.2 Conditionality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230

8.3.3 Orientation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232

8.3.4 Productive versus Protective . . . . . . . . . . . . . . . . . . . . . . . . . . . 233

8.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235

5
9 Concluding Remarks 237

9.1 Summary of Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237

9.2 Generalizability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240

9.3 Opportunities for Future Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242

9.4 Recommendations for Policy Practitioners . . . . . . . . . . . . . . . . . . . . . . . . 243

Bibliography 245

Appendix A: Interviews – South Africa 285

Appendix B: Interviews – Brazil 286

Appendix C: Eligibility Requirements – South Africa 287

Appendix D: Eligibility Requirements – Brazil 291

Abstract 293

Zusammenfassung 295

Publications during the Course of Research 297

Biography 300

6
Acknowledgements

The completion of this project was only made possible through the generous support of a whole host

of individuals and institutions. Although I am sincerely thankful for having come to know so many

wonderful people all over the world during this journey of academic and personal development, I wish

to take this opportunity to thank a number of them in particular. My initial word of appreciation

goes to Professor Susanne Lütz at the Freie Universität Berlin for her unwavering commitment

in guiding my academic development throughout the last four years. I owe her a great debt of

gratitude not only for her role in aiding the completion of this dissertation, but also for the way in

which she continually supported me in all the endeavours I chose to pursue throughout our time

working together. It was a pleasure to be advised by such an incisive and committed academic.

A word of thanks also goes to Sibylle Schaefer for the tremendous administrative support which

facilitated our interactions, as well as to Professor Janis van der Westhuizen from Stellenbosch

University for his constant encouragement. I was further lucky enough to spend the final year of

the journey at Princeton University, working under the guidance of Professor Jennifer Widner. I

wish to thank her for making it possible for me to attend this exceptional institution, where I

was able to participate in coursework in addition to completing the writing process. The feedback

provided on the manuscript by her as well as by Professor Atul Kohli was invaluable. It was a true

privilege to be able to work with them.

None of this would however have been possible without the generous financial support provided

by a host of funding institutions. Perhaps the most decisive among these was the scholarship from

7
the European Commission, which initially enabled me to travel abroad and enrol at Berlin. The

journey would have been stillborn without this early faith they demonstrated in my abilities. Thank

you in this regard to Stefanie Böhler, Beatrice Lange and the rest of the coordinating team. Further

institutional support was subsequently provided by Princeton University, the Berlin Consortium for

Germanic Studies, as well as by the German Academic Exchange Service. I wish to also extend my

heartfelt thanks to these funders, and it is my hope that the project will go on to produce impacts

commensurate with the trust they have placed in me.

One of the greatest blessings of the journey came in the form of sharing the company of fellow

travellers. This includes longtime friends such as Aldu Cornelissen, Jeanie le Roux, Nelius Stephan,

Lappies Labuschagne and Egbert Loubser, who never wavered despite the years of distance between

us. I was further welcomed into a circle of truly exceptional people during my time in Berlin. To

Miguelangel Verde, Philani Mthembu, Anne Venhaus, Amrit Naresh, Benjamin Anderson, Daniel

Cardoso, Frank Odei-Addo, Julia Soeffner, Marc Venhaus, Raimonda Verde and Saud Al-Zaid:

thank you for your deeply treasured friendships.

At Princeton I wish to thank Emre Türköz, Alisya Anlaş, Aric Rousso, Bruce Perry, Dave

Sroczynski, Wes Reinhart, Korhan Koçak, Rachael McLellan, Alia Chaudri, Njeri Kimani as well as

the entire Kimani family. I also extend my thanks to Larissa Brezolin and Pedro Lara de Arruda for

their kind help during my time in Brazil. A special word of thanks furthermore goes to the Billerts.

You truly have become like a second family to me, and I will never be able to thank you for all that

you have done for me. From making my time at Princeton possible to always providing me with a

home in Germany, you have supported me throughout. Bernd, Heike, Vincent and Rouven: thank

you from the bottom of my heart.

Upholding all of this was the unwavering love of my family. Thank you to my parents, Jannie and

Suma, for providing me with the greatest role models a son could hope for. If any good ever comes

from my work, please know that your undying commitment to our family and your passionate belief

8
in the power of education is the reason. I am also proud to be a brother to such exceptional young

men as Harold and Hanno, and have no doubt that they will both go on to achieve great things. My

final word of thanks is reserved for my beautiful Julia, whose strength, love and dedication inspires

me every day. Ek is lief vir julle almal.

9
List of Figures

Figure 2.1 The Welfare State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

Figure 2.2 Social Protection Conceptualised . . . . . . . . . . . . . . . . . . . . . . . . . 32

Figure 2.3 Selected Social Assistance Programmes in the World . . . . . . . . . . . . . . 36

Figure 5.1 Social Protection in South Africa . . . . . . . . . . . . . . . . . . . . . . . . . 71

Figure 5.2 South African Lifecycle Risks . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

Figure 5.3 Social Protection in Brazil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

Figure 5.4 Brazilian Lifecycle Risks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

Figure 5.5 Composite Average Benefit Level . . . . . . . . . . . . . . . . . . . . . . . . . 93

Figure 5.6 Composite Means Test Threshold . . . . . . . . . . . . . . . . . . . . . . . . . 94

Figure 5.7 Composite Value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

Figure 6.1 Actor-Centred Institutionalism – South Africa . . . . . . . . . . . . . . . . . 104

Figure 6.2 Explanatory Framework – South Africa . . . . . . . . . . . . . . . . . . . . . 104

Figure 6.3 OPG Benefit Levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

Figure 6.4 South African Lifecycle Risks in 1994 . . . . . . . . . . . . . . . . . . . . . . 123

Figure 6.5 Electoral Support, General Elections (1994-2014) . . . . . . . . . . . . . . . . 136

Figure 6.6 Poverty Headcount Ratio by Population Group (1996) . . . . . . . . . . . . . 146

Figure 6.7 Percentage of SMGs received by Population Group (1993) . . . . . . . . . . . 147

Figure 6.8 Number of SMGs per 1 000 children aged 0-17 . . . . . . . . . . . . . . . . . 148

10
Figure 7.1 Actor-Centred Institutionalism – Brazil . . . . . . . . . . . . . . . . . . . . . 165

Figure 7.2 Explanatory Framework – Brazil . . . . . . . . . . . . . . . . . . . . . . . . . 165

Figure 7.3 Brazilian Lifecycle Risks in 1990 . . . . . . . . . . . . . . . . . . . . . . . . . 173

Figure 7.4 Electoral Support, Presidential Elections (1994-2014) . . . . . . . . . . . . . . 184

Figure 7.5 Number of Parties Represented in the Chamber of Deputies (1994-2014) . . . 186

Figure 8.1 Poverty and Unemployment Rates (1995) . . . . . . . . . . . . . . . . . . . . 219

Figure 8.2 Vaccination Coverage for Children 0-7 . . . . . . . . . . . . . . . . . . . . . . 220

Figure 8.3 Net Primary School Enrolment Rates (2000) . . . . . . . . . . . . . . . . . . 221

Figure 8.4 Family Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223

Figure 8.5 Public Attitudes towards Redistribution . . . . . . . . . . . . . . . . . . . . . 225

Figure 8.6 Public Attitudes towards Families . . . . . . . . . . . . . . . . . . . . . . . . 227

11
List of Tables

Table 1.1 Comparing South Africa and Brazil . . . . . . . . . . . . . . . . . . . . . . . . 23

Table 4.1 Dependent Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

Table 4.2 Independent Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

Table 4.3 Relevant Situations for different Research Methods . . . . . . . . . . . . . . . 62

Table 5.1 Social Grants in South Africa . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

Table 5.2 South African Means Test Thresholds . . . . . . . . . . . . . . . . . . . . . . . 77

Table 5.3 Social Assistance Programmes in Brazil . . . . . . . . . . . . . . . . . . . . . . 86

Table 5.4 Brazilian Means Test Thresholds . . . . . . . . . . . . . . . . . . . . . . . . . . 88

Table 5.5 Dependent Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

Table 5.6 Typologies of Social Assistance Policy Regimes . . . . . . . . . . . . . . . . . . 102

Table 6.1 Evolution of the pre-1994 South African Social Assistance System . . . . . . . 122

Table 6.2 South African Social Assistance in 1994 . . . . . . . . . . . . . . . . . . . . . . 123

Table 6.3 Independent Variables – South Africa . . . . . . . . . . . . . . . . . . . . . . . 137

Table 6.4 South African Social Assistance in 1999 . . . . . . . . . . . . . . . . . . . . . . 154

Table 6.5 South African Social Assistance in 2009 . . . . . . . . . . . . . . . . . . . . . . 159

Table 6.6 South African Social Assistance in 2014 . . . . . . . . . . . . . . . . . . . . . . 162

Table 7.1 Evolution of the pre-1990 Brazilian Social Assistance System . . . . . . . . . . 172

12
Table 7.2 Brazilian Social Assistance in 1990 . . . . . . . . . . . . . . . . . . . . . . . . 173

Table 7.3 Independent Variables – Brazil . . . . . . . . . . . . . . . . . . . . . . . . . . . 187

Table 7.4 Municipal Social Assistance Competency . . . . . . . . . . . . . . . . . . . . . 193

Table 7.5 Brazilian Social Assistance in 1997 . . . . . . . . . . . . . . . . . . . . . . . . 199

Table 7.6 Brazilian Social Assistance in 2003 . . . . . . . . . . . . . . . . . . . . . . . . 203

Table 7.7 Brazilian Social Assistance in 2006 . . . . . . . . . . . . . . . . . . . . . . . . 208

Table 7.8 The Evolution of the Bolsa Famı́lia from 2003 to 2012 . . . . . . . . . . . . . 211

Table 7.9 Brazilian Social Assistance in 2014 . . . . . . . . . . . . . . . . . . . . . . . . 214

13
Abbreviations

ACI Actor-Centred Institutionalism

ACVV Afrikaanse Christelike Vrouevereniging

AG Auxı́lio Gás

ANC African National Congress

BIG Basic Income Grant

BPC Benefı́cio de Prestação Continuada

CA Cartão Alimentação

CDG Care Dependency Grant

CEF Caixa Econômica Federal

CEME Central de Medicamentos

CESR United Nations Committee on Economic, Social and Cultural Rights

CLT Consolidação das Leis do Trabalho

CNAS Conselho Nacional de Assistência Social

COFINS Contribuição para o Financiamento da Seguridade Social

14
CSG Child Support Grant

CSGTT Child Support Grant Task Team

CSLL Contribuição Social sobre o Lucro Lı́quido

CT Cash Transfer

DA Democratic Alliance

DATAPREV Empresa de Processamento de Dados da Previdência Social

DG Disability Grant

DP Dependent Variables

DSD Department of Social Development

EBC Electronic Bank Card

EPWP Expanded Public Works Programme

FCG Foster Child Grant

FUNABEM Fundação Nacional do Bem-Estar do Menor

FUNRURAL Programa de Assistência ao Trabalhador Rural

FZ Fome Zero

GDP Gross Domestic Product

GEAR Growth, Employment and Redistribution

GIA Grant-in-Aid

GNU Government of National Unity

15
HDI Human Development Index

IAPAS Instituto de Administração Financeira da Previdência e Assistencia Social

ID Identity Document

ILO International Labour Organisation

INAMPS Instituto Nacional de Assistência Médica da Previdência Social

INPS Instituto Nacional de Previdência Social

INSS Instituto Nacional do Seguro Social

IPEA Instituto de Pesquisa Econômica Aplicada

IRM Internal Reconsideration Mechanism

ITSAA Independent Tribunal for Social Assistance Appeals

IV Independent Variables

LBA Legião Brasileira de Assistência

LOAS Lei Orgânica de Assistência Social

LOSS Lei Orgânica da Seguridade Social

MDS Ministério do Desenvolvimento Social e Combate à Fome

MESA Ministro de Estado Extraordinário de Segurança Alimentar e Combate à Fome

MP Member of Parliament

MPS Ministério da Previdência Social

MSSD Most Similar Systems Design

16
NA National Assembly

NCA National Constituent Assembly

NCOP National Council of Provinces

NDP National Development Plan

NGK Nederduits Gereformeerde Kerk

NP National Party

NPC National Planning Commission

OPG Older Persons Grant

PBE Programa Bolsa Escola

PBF Programa Bolsa Famı́lia

PBSM Plano Brasil Sem Miséria

PGRFM Programa de Garantia de Renda Familiar Mı́nima

PGRM Programa de Garantia de Renda Mı́nima

PMDB Partido do Movimento Democático Brasileiro

PNAS Polı́tica Nacional de Assistência Social

PSDB Partido da Social Democracia Brasileira

PT Partido dos Trabalhadores

RGPS Regime Geral de Previdência

RMV Renda Mensal Vitalı́cia

17
RPC Regime de Previdência Complementar

RPPS Regime Próprio de Previdência Social

SASSA South African Social Security Agency

SDG Sustainable Development Goals

SEAS Secretaria de Estado da Assistência Social

SMG State Maintenance Grant

SNAPS Sistema Nacional de Previdência e Assistência Social

SUAS Sistema Único de Assistência Social

UIF Unemployment Insurance Fund

UNDP United Nations Development Programme

WPSW White Paper for Social Welfare

WVG War Veterans Grant

18
Chapter 1

Introduction

1.1 Background and Rationale

The world has witnessed a revolution in the extension of social protection during recent decades.

The fundamental motivation for undertaking this research project is the fact that the provision of

non-contributory social assistance Cash Transfers (CTs) to impoverished people living in developing

countries has expanded from an almost non-existent base during the mid-1990s to around 1 billion

people today (Barrientos 2013: 4). The staggering growth of these programmes, where beneficiaries

receive monthly cash payments as a result of their poverty or vulnerability, has given rise to a

tremendous amount of popular and academic interest, with most of the attention having been

focused on assessing their impact on a wide range of social and economic outcomes.1

In the rush to proclaim the significance of these interventions, analysts have been quick to paint

an undifferentiated picture of developing countries implementing supposedly uniform policies in re-

sponse to shared problems of persistently high levels of poverty and inequality. While this approach

may have succeeded in drawing attention to the phenomenon, it has come at the significant price

of sacrificing nuance and masking the tremendous variation that exists in how these programmes
1
See section 3.3 for an overview of the literature on this subject.

19
are actually implemented in different countries. The overview of the academic literature on the

subject carried out in chapter three of this study additionally reveals that no systematic inquiries

have directly addressed the political economy of the domestic development of social assistance poli-

cies.2 In short, we know very little about how these programmes came to be established in the first

place, and why different countries have often taken different approaches to redistribution as entailed

through social assistance provision.

In seeking to address these issues, this study undertakes a comparative examination of the de-

velopment of such policies in two of the benchmark cases: South Africa and Brazil. Social assistance

programmes have come to occupy centrally important positions in the welfare and distributional

regimes of both countries since their respective transitions to democracy during the late-1980s and

early-1990s. In the case of South Africa, the provision of CTs to impoverished beneficiaries has

grown from a negligible level two decades ago to almost 16 million monthly beneficiaries at present

(29 percent of the total population). The same is true for Brazil, where about 54 million people

(27 percent of the country’s population) are supported through such programmes every month.

The transformation entailed by these developments is so monumental that the introduction of these

programmes are regarded as the ‘single most dramatic change’ to have affected these societies in

the twenty-first century (Luna & Klein 2014: 338).

1.2 Research Question and Hypotheses

The central research question posed by this study is explanatory in nature: what accounts for the

differences between the South African and Brazilian social assistance systems? Answering this ques-

tion however initially entails a descriptive procedure where the differences that are to be explained
2
This is in contrast to the political economy of social assistance programmes that are externally designed and

funded. Following the success of locally designed interventions in middle-income countries such as South Africa

and Brazil, international development organisations have played an active role in implementing similar schemes in

low-income regions.

20
are discussed in detail. Chapter five of this study concerns itself with this descriptive objective. In

addition to meticulously chronicling the different features of social assistance provision in the two

cases, the analysis shows that they exhibit significant variation in terms of which population groups

are incorporated (coverage), whether beneficiaries need to comply with any conditions in order to

receive benefit payments (conditionality), and the way in which they define and assess beneficiaries

(orientation).

The study’s dependent variables result from this descriptive exercise, with the technical differ-

ences being abstracted and incorporated into two distinct typologies. The result is that the South

African social assistance system is classified as productive, while the Brazilian regime is regarded as

being premised on a protective approach.3 The subsequent explanatory endeavour assesses the va-

lidity of the study’s central hypothesis that variation in institutional structure is best able to explain

this difference between the cases. The hypothesis incorporates two causal variables by positing that

the deep historical entrenchment of social assistance norms, combined with a highly concentrated

state structure, led to the development of a productive regime in South Africa. Conversely, the

comparatively weak historical entrenchment of norms, in combination with a highly diffuse state

structure, channelled the emergence of a protective system in Brazil.

This endeavour is concerned with examining the development of social assistance policies fol-

lowing democratisation. Even though the analysis is therefore exclusively focused on accounting

for the subsequent emergence of empirical variation in terms of policy outputs, it is important to

acknowledge that it does assume that the introduction of democracy itself played a significant role

in providing an initial impetus for reform in both cases. As shown in the case study chapters,

the pressures for greater inclusivity4 entailed through democratisation initiated the processes under

investigation. But while democratisation can thus perhaps be regarded as a necessary element, it
3
Chapter five elaborates on these typologies.
4
These pressures were perhaps most obvious in the South African case, where they were principally expressed in

the drive to deracialise the welfare state.

21
alone is simultaneously wholly insufficient in explaining why South Africa and Brazil ended up with

such different policy regimes. The near-synchronous adoption of democracy in fact only serves to

add to the puzzle, as it fails to explain why the two countries adopted greatly divergent approaches

even as they underwent broadly similar regime transitions. While democracy may have initiated

the reform processes in both countries, it did not shape it. Instead, the processes and the outcomes

they produced were ultimately moulded by much deeper variation on the institutional dimension,

as entailed through differences in norm institutionalisation and state structure.

The study further aims to balance the inherent endogeneity of this institutionalist approach

by assessing the accuracy of a set of popular alternative hypotheses premised on exogenous fac-

tors. Such explanations, usually based on functional and ideational claims, are widespread in the

academic literature. The investigation ultimately determines that, while exogenous functional and

ideational elements were certainly present during the policy development process, they alone are

unable to convincingly account for the cross-case variation. Instead, these elements were them-

selves moulded and channelled by the respective institutional structures, a conclusion which assigns

ultimate explanatory validity to the institutionalist hypothesis.

1.3 Case Selection

The cases of South Africa and Brazil were selected based on the principles of the Most Similar

Systems Design (MSSD) (Thomas 2011: 513). In the broadest sense, South Africa and Brazil are

two of the most sophisticated and high profile cases of developing countries implementing extensive

social assistance programmes. They certainly constitute the most similar instances out of the small

sample of potential cases,5 as measured at the beginning of the period under investigation. As
5
Aided by the abovementioned enthusiasm of international development organisations, a distinct feature of the

proliferation of social assistance schemes around the world has been the speed at which these programmes have

spread. An authoritative study that focused on social assistance schemes incorporating conditionality found that in

1997 there were only three such programmes in the world (in Brazil, Mexico and Bangladesh), along with South

22
indicated in table 1.1, the similarities however extend well beyond the contemporary prominence of

these policies. Levels of economic development were very similar by the mid-1990s, as both countries

were classified as upper middle-income with a Gross Domestic Product (GDP) per capita of around

US$3 000. The Gini coefficient of inequality was still identical across the cases in 2000, with large

sections of their populations living in poverty while a small group of elites controlled most of the

countries’ wealth.

Year South Africa Brazil

GDP per capita (US$) 1990 $3 152 $3 092

Gini Coefficient 2000 .58 .58

Poverty Rate (US$1.25 per day) 1995 21.4% 9.9%

Political Regime 1983 → 2000 4→9 3→8

(not democratic to democratic) (not democratic to democratic)

Table 1.1: Comparing South Africa and Brazil (Source: Adapted from Lieberman 2009: 111)

Both countries also experienced prolonged periods of authoritarian rule before transitioning to

democracy as part of the so-called ‘Third Wave’ during the early 1990s.6 This followed their coloni-

sation and settlement by European powers, a shared heritage which left both nations grappling with

the legacies of immigration and slavery. Their societies are also comprised of a wide mix of ethnic

and racial groups, with racial characteristics continuing to be correlated with wealth and income.

Both countries additionally had relatively well developed (if uneven) state apparatuses and largely

constructed their current social assistance systems under the governance of political parties with

broadly similar ideological persuasions. In recognition of the salience of these shared features, South

Africa and Brazil have since banded together on various public platforms by explicitly emphasizing
Africa’s unconditional system (Fiszbein et al. 2009: 4). By 2008, this figure had grown to at least 29. In addition to

thus currently being the largest, most sophisticated and most similar cases, this situation further justifies the focus

on South Africa and Brazil, given the fact that social assistance programmes have been operational in these two cases

longer than almost anywhere else.

6
The measurements indicated in table 1.1 are based on the Polity IV scores of democracy.

23
the commonality of their characteristics and challenges. In short, ‘the vast array of social, politi-

cal, and economic similarities between the two countries described above provides something of a

“natural experiment” which is rare in cross-national social science research’ (Lieberman 2003: 8).7

1.4 Significance of the Study

The theme of this research fundamentally deals with questions of redistribution. Perhaps more than

any other single policy arena, the particularities of social welfare provision plays a tremendously

important role in shaping the overall distributional regime8 of any given country. The study’s de-

pendent variables thus address the scope and mode of redistribution through social assistance in

South Africa and Brazil. Through the classification of their divergent features into distinct typolo-

gies (protective versus productive), the analysis makes a decisive contribution to the scholarship on

emerging welfare states by undertaking the first qualitative examination of the political economic

processes underlying their differentiated development.

But the relevance of the findings stretches well beyond these two (important) cases. The study

provides a striking account of the role played by institutional features in shaping social assistance

policy outputs in developing countries. It shows how the deep historical entrenchment of particular

norms can have profound path dependence effects even amidst the supposedly great transformation

entailed by democratic transitions. It further highlights the role of state structure in shaping political

incentives and subsequent policy development processes. At the broadest level, the outcome is a

powerful analysis of the role played by institutions in driving social welfare policy outputs and

ultimately influencing the scope and shape of redistribution.

Assessed purely from a scholarly perspective, these findings contribute to at least three distinct
7
Further justification for the comparability of South Africa and Brazil comes from the examinations undertaken

by authors such as Seidman (1994); Friedman and De Villiers (1996); Marx (1998) as well as Lieberman (2003; 2009).
8
These programmes are one of the most ‘direct and visible way in which states affect [distribution]’ (Seekings &

Nattrass 2005: 4) For more on the components of a distributional regime, see Seekings and Nattrass (2005).

24
streams of academic literature,9 including the recent rise of emerging welfare states, the concomitant

introduction of social assistance policies in developing countries, as well as the application of a

historical institutional theoretical framework to an exciting new empirical area. Regarded from

a somewhat more practical angle, the significance of the study is to emphasise the importance of

taking local institutional contexts into account when designing social assistance policy interventions.

This is of particular contemporary significance given the active role of international development

organisations in driving the proliferation of these programmes throughout the developing world.

The evidence suggests that these policies are likely to be both scalable and sustainable only in cases

where they are closely aligned with local institutional structures. In addition to its clear academic

relevance, this comparative study thus serves to affirm the decisive significance of ‘institutional fit’

in implementing workable social policy interventions.

1.5 Chapter Outline

This brief introductory chapter has served to illuminate the overall context wherein the project

is located, as well as to highlight its central themes. It achieved this by initially outlining the

background and rationale for undertaking the study through discussing the prolific recent rise of

social assistance programmes. This was followed by a specification of the central research question

and hypotheses, as well as a justification for the comparative focus on the cases of South Africa

and Brazil. The final section specified the substantial academic and practical significance of the

findings.

The following chapter introduces the project’s conceptual and analytical framework. Its primary

purpose is to provide consistent definitions in terms of the study’s central concepts. As such, the

initial part of the chapter situates social assistance policies within the broader framework of social

protection. It shows that the concept operates distinctly from other elements such as social insurance
9
See chapter three for a literature contextualisation.

25
and labour protection, thereby constituting an autonomous analytical subcategory. It then turns

its attention towards clearly articulating the way in which the concept of institutions is regarded in

constructing the study’s analytical framework.

The third chapter serves to further extend the contextualisation of the research by conducting

a detailed review of the existing literature. Its central aim is to illustrate the study’s contribution

to three primary streams of inquiry. The first of these flows from the outcome of the secondary

descriptive objective through the construction of qualitative typologies for classifying emerging wel-

fare states writ large. This is followed by a specification of the study’s contribution to the more

narrowly-defined scholarship on social assistance, an area which has been dominated by impact

assessments at the relative expense of explanatory inquiries. The chapter then discusses the signif-

icance of the project’s extension of a historical institutional framework to a fertile new empirical

area. The final section reviews the alternative functional and ideational claims which have emerged

as conventional wisdom in explaining the rise of social assistance.

Chapter four turns its attention to the question of research design. In addition to specifying the

research question and relevant variables, it meticulously discusses the methodology that was followed

in executing the project, with a particular focus on explaining the decision to conduct an inductive,

qualitative comparative case study. It additionally addresses issues such as information sources and

the data collection procedure that was followed, as well as clearly defining the delimitations of the

research focus. It concludes by examining the limitations and ethical implications of the research

process.

Chapter five commences the empirical analysis in earnest by conducting a highly detailed review

of the social assistance policies that are implemented in the two cases. The aim of the discussion is

to fulfil the study’s descriptive objective by clearly articulating the qualitative differences that exist

between the South African and Brazilian social assistance systems. It achieves this by carefully

reviewing all aspects of programme operation in both countries. This technical overview identifies

26
profound variation in terms of coverage, conditionality and orientation. Drawing from the latest

literature, these differences are then consolidated and abstracted in two distinct typologies which

serve as the study’s dependent variables.

The sixth chapter conducts the case study analysis for the South African case. It starts by

showing the significant extent to which particular norms regarding social assistance provision had

been entrenched as part of the construction of a segregationist and apartheid political economy, and

how these norms were subsequently transferred and incorporated into the institutional structure of

the post-apartheid state. This is followed by an account of the country’s highly concentrated state

structure, characterised by the centralisation of social assistance policymaking capacity and the

overwhelming power position of a single political party. The final section discusses the development

of the contemporary South African system by showing how the deep historical institutionalisation

of norms and a highly concentrated state structure ultimately shaped the emergence of a productive

social assistance regime in the country.

Chapter seven conducts an identical analysis of the Brazilian case. Its initial review of the his-

torical record indicates that social assistance programmes were almost entirely absent in the period

leading up to full democratisation in the early 1990s. This lack of embedded norms prohibited the

manifestation of path dependent effects and provided actors with a greater deal of policymaking

space. The subsequent discussion further shows that the country features a diffuse state structure

in terms of social assistance policy design capacity, as well as a highly competitive political sys-

tem. Taken together, the final section shows how these specific institutional features directed the

emergence of a protective social assistance system in Brazil.

Chapter eight consolidates the insights produced through the respective case studies by explic-

itly returning to the central research question and hypotheses. It initially conducts a systematic

assessment of the explanatory validity of the rival exogenous functional and ideational hypotheses,

determining that they are unable to satisfactorily account for the cross case variation. This is fol-

27
lowed by a similar procedure in terms of the central institutional hypotheses, which conclusively

demonstrates that the differences between the South African and Brazilian social assistance systems

fundamentally result from variation on the institutional dimension. The final chapter concludes by

summarising the findings and briefly addressing their generalizability, illuminating avenues opened

up by the study for potential future inquiries, as well as emphasising its significant contribution

to both the scholarly and practical knowledge on the role of institutions in redistribution through

social assistance provision in the developing world.

28
Chapter 2

Conceptual and Analytical Framework

2.1 Introduction

The preceding introduction has already alluded to concepts that are vitally important to the theme

of the research. This chapter undertakes a precise conceptualisation of these terms in order to

clearly articulate the focus of the inquiry. The initial section provides conceptual clarity on the

notion of emerging welfare states in the developing world. This is followed by a specification of

social assistance programmes through clearly differentiating this category from other constitutive

elements in the broader framework of social protection, emphasising the way in which social assis-

tance interventions have come to form the core of the welfare state in the developing world. The final

section provides an explication of the way in which institutions are conceptualised in constructing

the Actor-Centred Institutionalist (ACI) analytical framework for this study. The chapter thereby

equips the reader with a thoroughgoing understanding of what is implied when using the concept

of ‘institutional variation’ to explain differences in ‘social assistance provision’ across the ‘emerging

welfare states’ of South Africa and Brazil.

29
2.2 Emerging Welfare States

The literature on the emergence of nascent welfare states in the global South dates back no more

than a dozen years.10 In contrast, the notion of ‘welfare state’ itself can be traced back to 1834

Britain with the introduction of the New Poor Law Act and the Chadwick Report of 1842, while

Bismarck’s Prussian state created the world’s first health insurance programme and old-age pensions

in 1889 (Gough 2004: 19). The term Wohlfahrtsstaat first came into usage in Germany in the 1920s

before being incorporated into the English speaking world during the 1940s. Following the end of

World War II, welfare policies became ‘a massive feature of all Northern states’ (Gough 2004: 19).

At the most general level, a welfare state regime11 is an institutional matrix of market, state and

family forms that generate welfare outcomes (Gough 2004: 23). Figure 2.1 introduces the basic

conceptual composition of welfare state policies.

WELFARE STATE

Public Education Public Healthcare

Social Protection

Figure 2.1: The Welfare State

Social policy as an academic field consequently blossomed during this period, as scholars aimed

to construct typologies reflecting and accounting for the variation in the welfare state regimes of the

global North. But the recent spread of welfare policies throughout the developing world has posed

a fresh dilemma, leading to criticism that existing approaches were not applicable to the unique

policy challenges faced by the rest of the world (Midgley 2004). Although on average scholars have

paid very little attention to the question of welfare states in the developing world, some attempts
10
See section 3.2 for a comprehensive review of the literature on this subject.
11
The ‘regime’ aspect of the concept refers to a set of institutions, rules and structured interests that constrain

individuals through compliance procedures (Krasner 1983: 1-3; North 1990: 200-202; Gough 2004: 22).

30
have been made to conceptually recast the welfare regime approach as an analytical tool that is

appropriate to this growing empirical area (Rudra 2008: 78).

This regime approach is said to offer

a way out of a classic dilemma in understanding social policy and social development across the

world. By developing a variegated middle-range model it avoids both over-generalisation and

over-specificity. A regime approach can recognise, on the one hand, the commonalities across the

countries and regions of the South, while on the other hand identifying systematic qualitatively

distinct patterns within the South (Gough 2004: 239-240).

This study consequently uses the concept of ‘emerging welfare states’ to clearly delineate the field of

inquiry as being concerned with the welfare regimes contemporarily implemented in the developing

world.

2.3 Social Assistance: An Integral Part of Social Protection

The overarching arena of social protection refers to a range of potential policy interventions available

to governments aiming to manage, prevent and ultimately overcome situations that adversely affect

people’s wellbeing (UNRISD 2010: 135). The nominal goal of any country’s social protection system

is to enable and support citizens in maintaining an adequate standard of living when confronted

by events such as unemployment, illness, disability, old age, maternity and economic or natural

disasters (UNRISD 2010: 135). In conceptual terms, social protection systems aim to address

vulnerability, risk and chronic poverty by employing a set of public actions either directed privately

or by the state. These systems are generally divided into three constitutive categories: social

assistance, social insurance and labour protection, as illustrated in figure 2.2.

31
 WELFARE STATE

Public Education Public Healthcare

Social Protection

Social Assistance Social Insurance Labour Protection

 Non-contributory  Contributory  Minimum
Cash Transfer Pensions Workplace Safety
Programmes  Unemployment Standards
Insurance
 Health Insurance

Informal Economy Formal Economy

 Unemployed  Workers employed
 Workers not under formal
employed under arrangements
formal
arrangements

Figure 2.2: Social Protection Conceptualised (Source: Adapted from UKDFID 2005)

The figure also highlights the fact that these elements affect different societal groupings, with

access to social insurance and formal labour protection generally being limited to those working in

the formal economy,12 while the only form of social protection that is usually available to people

outside the formal economy (either due to unemployment or by virtue of working in the informal

economy) is social assistance programmes.

12
The terms ‘formal economy’ and ‘informal economy’ are used here instead of referring to the formal and informal

‘sectors’ in order to ‘get away from the idea that informality is confined to a specific sector of economic activity but

rather cuts across many sectors. “Informal economy” also emphasises the existence of a continuum from the informal

to the formal ends of the economy and thus the interdependence between the two sides’ (Becker 2004: 8). The concept

of ‘informal economy’ is employed in contradistinction to the ‘formal economy’ here to refer to ‘all economic activities

by workers and economic units that are — in law and in practice — not covered or insufficiently covered by formal

arrangements’ (Becker 2004: 11).

32
 Labour protection entails the enforcement of minimum standards to protect people within the

workplace, although this is largely applicable to the formal economy and presents significant enforce-

ment challenges in the case of the informal economy (ILO 2000). Meanwhile, both social insurance

and social assistance programmes are associated with some form of financial transfer to act as in-

come support (UNRISD 2010: 135). Contributory pensions, health insurance and unemployment

insurance are all classified as social insurance schemes, whereby individuals pool resources through

the regular payment of contributions to the state or a private provider in order to protect themselves

against a shock or permanent change to their livelihoods.

Social insurance can be considered more appropriate for the better-off sections of the popula-

tion, as its contributory nature presupposes the availability of a regular income, while its aims are

principally to prevent individuals from falling into poverty in the first place (Dekker 2008). Be-

cause of this limitation, social assistance programmes become vital in cases of already existing and

widespread poverty where many people may not be formally employed. Social assistance is defined

as non-contributory transfers to those deemed eligible by society on the basis of their vulnerability,

poverty or rights as citizens (UKDFID 2005: 6; UNRISD 2010: 135).13 The fundamental factor

therefore differentiating social insurance from social assistance in that of eligibility. The contribu-

tory nature of social insurance means that eligibility is determined by the ability to contribute in

the first place, whereas eligibility for social assistance is based upon needs or rights and previous

contributions (principally through taxation) are not a precondition to receiving support.

2.3.1 Labour Protection and Social Insurance

In the case of developed countries featuring large formal economies, the need for labour protection

and social insurance schemes are traditionally much greater than the marginal role often assigned
13
The term ‘social assistance’ in and of itself is a fairly vague expression, and it is worth emphasising that it is used

here exclusively to refer to non-contributory government programmes that transfer money to eligible beneficiaries on

the basis of need. It does not, for example, include the delivery of social services and other associated measures.

33
to social assistance (Dethier 2007). This is due to the prevalence of comparatively high employment

levels and low levels of poverty, deprivation and inequality. In financial terms, developed country

contexts are dominated by contributory social insurance programmes because an overwhelming

majority of the population has access to income-generating formal employment.

In addition to benefiting from labour protection measures, this means that they are tied to the

formal economy and can afford to contribute to social insurance schemes (both private and public).

In the event of unemployment or any other contingency, they become eligible to receive financial

support from the social insurance programmes they had previously contributed towards. At least

in basic theoretical terms, labour protection and social insurance is therefore largely considered

to be appropriate in these settings (Dekker 2008). The role of social assistance programmes is

much more limited here due to the simple fact that there is significantly less need for long term

non-contributory systems to financially support deprived people unable to find income-generating

formal employment.14

The outcome of this situation is that, under the rubric of social insurance, poverty and unem-

ployment are regarded as transient phenomena, meaning that developed country social protection

‘has traditionally been concerned with temporary or foreseeable income shortfalls and transitory

experiences of poverty in otherwise relatively stable life trajectories and acceptable living condi-

tions’ (UNRISD 2010: 135). Social insurance programmes generally include contributory medical,

unemployment, accident and other compensation funds.

2.3.2 Social Assistance: The Core of the Emerging Welfare State

The ability of social insurance to address the needs of society diminishes as access to employment

and regular income streams decline. Conversely, the more widespread poverty, unemployment and

non-formal employment are, the more significant social assistance becomes. This means that the
14
Although such programmes certainly do exist in developed countries, their scale is comparatively small in relation

to the overall significance of social insurance.

34
ability of social assistance to adequately address economic and social deprivation vis-à-vis that

of social insurance is highly context-specific. The role of social assistance becomes much more

significant in the context of high unemployment, high levels of informal employment and poverty

associated with many developing countries (UKDFID 2005). In the countries of the developing

world, social assistance has become the core of the welfare state.

Compared to the entrenched position of social insurance dating back at least to the above-

mentioned Bismarckian reforms of the 1880s, the elevated status of social assistance as a defining

component of social protection has emerged only recently (Hanlon et al. 2010). Contemporary de-

veloping country social assistance systems mostly date back to the 1990s or 2000s, and consist of a

range of CT programmes targeted at specific population segments. The fact that they are thus very

recent policy innovations that have become central features of the distributional regimes present

in many developing countries, including in the cases under examination, makes this an exceedingly

interesting and important new field to analyse.

These programmes demonstrate significant contestation regarding fundamental concepts such as

universality or targeting and the division of responsibility between markets, states and households

(Hanlon et al. 2010). Notwithstanding the wide variations that exist concerning the ways in which

these programmes are implemented, they do comprise a fundamentally separate category within

broader social protection — the category of social assistance.

In general terms, these are non-contributory programmes which transfer money from the state to

the poor directly for a specified period of time and are aimed at diverse population groups including

the elderly, children, unemployed adults and people with disabilities (Hanlon et al. 2010; Garcia

& Moore 2012: 18). The implementation of social assistance policies in developing countries has

been hailed as a ‘development revolution from the global South’ (Hanlon et al. 2010). Significantly,

these policies were initially neither imposed by developed donor nations nor inherited from colonial

administrations, but recently emerged as a policy response initiated by the countries of the South

35
themselves in an attempt to provide the poor with the boots required in order for them to ‘pull

themselves up by their bootstraps’ (Hanlon et al. 2010: 4).

Social assistance programmes have proliferated at a remarkable rate throughout the developing

world during the previous two decades, to the point where they currently reach more than a billion

people in low- and middle-income countries (Barrientos 2013: 4). The well documented examples

of CTs aimed at children and families illustrate just how rapid the growth in popularity of these

policies has been; in 1997 there were only four such programmes worldwide, but by 2008 this figure

had increased to at least 29 — and has continually grown further since then (Fiszbein et al. 2009:

31). This recent increase in the popularity of CTs, as illustrated in figure 2.3,15 has come to be

regarded as ‘one of the most significant developments in global social policy since the expansion of

social security in industrialised countries’ (Fajth & Vinay 2010: 1).

Figure 2.3: Selected Social Assistance Programmes in the World, (a) 1997 and (b) 2008 (Source: Adapted from

Fiszbein et al. 2009: 4)

Social assistance therefore represents a distinct component of states’ overall social protection
15
This illustrative figure highlighting the proliferation of social assistance programmes only includes information on

selected and currently active large-scale CTs directed at children and families. In some cases (as in the case of Brazil),

pre-existing programmes were recently merged or incorporated into ones with new names. This map of child-focused

programmes is obviously not nearly exhaustive, as social assistance also includes a range of other programmes aimed

at conditions such as disability, unemployment and old age.

36
frameworks, alongside labour protection and contributory social insurance. While non-contributory

programmes are found in many countries around the world, the relative significance of social assis-

tance is context specific, and it has recently proven immensely popular especially in middle-income

developing countries featuring high levels of poverty, inequality, unemployment or informal employ-

ment — to the extent that social assistance has become the central pillar of the emerging welfare

state.

2.4 Analytical Framework

The study furthermore applies a historical institutional framework in its analysis of the differen-

tiated development of social assistance policies in South Africa and Brazil. Given the fact that

neoinstitutionalism has spawned a variety of related approaches,16 this section endeavours to in-

troduce a concise specification of the way in which the framework is conceptualised in terms of

this particular project. Gough comments on the suitability of the approach for studying questions

related to welfare state regimes in the developing world by noting that the very adoption of a regime

approach automatically situates the research ‘within the historical institutional school of social re-

search’ (2013: 23). This allows for an analysis which ‘integrates structures and actors within a

framework that promises a comparative analysis of socioeconomic systems at different stages of

development and different positions in the world system’ (Gough 2013: 23).

It is important to emphasise that historical institutionalism per se does not amount to a specific

method or theory. Instead, it is to be regarded as a general approach that is interested in explaining

real-world empirical puzzles through a historical orientation that pays attention to the ways in which

institutions structure outcomes (Steinmo 2008: 118). The analysis conducted in this study therefore

uses the ACI approach to achieve a more precise operationalisation of historical institutionalism

(Scharpf 1991; Mayntz & Scharpf 1995; Scharpf 1997; Scharpf 2000; Boessen 2008; Pancaldi 2012).
16
Section 3.4 discusses the diversity of neoinstitutionalist approaches.

37
In a reflection of the fact that it is firmly rooted within the historical institutional school, this

analytical framework proceeds from the assumption that policy outputs are to be explained as ‘the

outcome of interactions among intentional actors, but that these interactions are structured, and

the outcomes shaped, by the characteristics of the institutional settings within which they occur’

(Scharpf 2007: 1).

The ACI focus on institutions enables researchers to utilise the same institutional information

with which actors themselves interact in order to reconstruct the policymaking process. The re-

sultant institutionalised expectations create conditions of common knowledge which reduces the

information costs associated with empirical research. In short, ‘once we know the institutional set-

ting of interaction, we know a good deal about the actors involved, about their options, and about

their perceptions and preferences’ (Scharpf 1997: 41). It assumes conditions of purposive actors17

operating according to bounded rationality, which implies that human action is not based on cog-

nition of real world data and causal laws, but on socially constructed beliefs about the real world

(Scharpf 1997: 21). The combination of purposive actors operating under conditions of bounded

rationality results in an analytic approach where actors strive to maximise their interests, but with

a more realistic view concerning matters surrounding incomplete information.

The framework thus emphasises the influence of institutions on defining the policy community,18

the modes of interaction, as well as the preferences and capabilities of actors. The institutional rules

of the game are firstly taken to define which actors constitute the relevant policy community.19 This
17
ACI conceptualises actors above the individual level into units of ‘composite actors’. The concept is consequently

employed in this study ‘in order to facilitate the task of explaining and predicting policy outcomes in actor-theoretic

terms’ (Scharpf 1997: 53).

18
While Scharpf (1997) prefers the term ‘actor constellation’, the concepts are used interchangeably throughout.
19
In terms of this project, the concept supposes that ‘policy communities comprise state and social actors, and

support a certain view as to desirable policy outcomes. State actors are individuals or groups that hold senior office in

government bodies, while the concept of societal actor refers both to social actors, from civil society, and market actors

associated with the market economy. The communities are made up of individuals and groups who hold positions in

the state and societal spheres, and participate in policy networks, endeavouring to affect decision-making processes.

38
concept is utilised in order to demarcate the process whereby policy decision making takes place, and

includes both state and societal actors active in the social assistance sphere (Heclo 1978; Jordan

& Richardson 1979; Rhodes 1986; Côrtes 2013). It also regards institutions as influencing the

relevant modes of interaction open to different groups of actors, with at least four such avenues

being theoretically possible: unilateral action, negotiated agreement, majority vote and hierarchical

direction (Scharpf 1997: 47). As discussed in chapters six and seven, the particularities of the two

countries’ institutional structures meant that the characteristic mode of interaction in the South

African case was hierarchical direction, while the Brazilian case was characterised by interactions

largely premised on negotiated agreement.

The framework additionally supposes that institutions shape the capabilities and preferences

held by actors during their policy interactions. Indeed, despite the clear predominance assigned to

institutions, the ACI framework does not hold that they create outcomes in a deterministic sense

(Scharpf 1997: 42). It thereby encourages analyses which are additionally focused upon examin-

ing strategic interactions between actors potentially holding different preferences as the proximate

process translating institutionally-derived preferences into policy outputs. The result is a histor-

ical institutional framework that is sensitive to the (bounded) agency of actors by involving two

sequential analytical steps. The first involves a detailed examination of institutional structure in

order to obtain information on the policy community, modes of interaction and the capabilities and

preferences of actors; followed by a second step of empirical process tracing examining strategic

interactions between purposive actors.

The present study incorporates the exact approach specified by Scharpf when he defines insti-

tutions as ‘systems of rules that structure the courses of actions that a set of actors may choose’

(1997: 38). This however includes ‘not only formal legal rules that are sanctioned by the court

system and the machinery of the state but also social norms that actors will generally respect’

(Scharpf 1997: 38). In the context of this project, the definition is composed of two independent
Their strategies are built up in processes closed to other communities and to the general public’ (Côrtes 2013: 137).

39
institutional variables: the level of historical entrenchment of social assistance norms20 (high versus

low) and the centralisation of state structure (high versus low). The case study chapters conduct

historical process tracing according to these variables. This initially involves an examination of the

degree to which social assistance norms were historically entrenched in both cases, and whether

these norms exhibited features of ‘institutional stickiness’ by constraining the options open to poli-

cymakers. In other words, it measures the degree to which the respective cases presented elements

of path dependence.

This concept is centrally important to many historical institutionalist analyses, and fundamen-

tally entails the assertion that ‘preceding steps in a particular direction induce further movement

in the same direction...This is because the relative benefits of the current activity compared with

other possible options increase over time’ (Pierson 2000: 252). This means that the deep historical

entrenchment of particular norms raise the cost of exit, leading to a self-reinforcing process (Pierson

2000: 252). The analysis shows that the South African case exhibited significant stickiness through

the deep historical institutionalisation of norms, while the Brazilian case featured relatively little

path dependence effects.

This is followed a specification of the second independent variable involving the respective state

structures. This variable again incorporates formal and informal institutional features in its exam-

ination of the distribution of power generated through both the constitutional-legal policymaking

framework, as well as the practical outcomes produced by the realities of electoral politics. It

concludes that the South African case has been characterised by a concentrated state structure

throughout the period under examination, in contrast to Brazil’s diffuse structure. The specifica-

tion of this variation in terms of the independent institutional variables is subsequently followed by
20
Even though they are in practise often subject to significant conceptual overlap, it is important to note the

difference between institutional norms and ‘ideas’ according to this approach. Fundamentally, ‘ideas’ become norms

once they are institutionally embedded and consequently approximate ‘something like basic templates upon which

other political decisions [are] made’ (Steinmo 2008: 130).

40
the introduction of the respective social assistance policy communities, along with an explication

of the available modes of interaction and actor preferences.

The final analytical phase involves the second step of the ACI procedure, which incorporates a

further round of process tracing which examines the contemporary development of the two countries’

social assistance systems. The narrative analysis, guided by the ACI analytical framework, thus

ultimately illustrates the centrally important role played by institutional variables in shaping the

differentiated development of social assistance provision in South Africa and Brazil.

2.5 Conclusion

This chapter has provided a clear articulation of the study’s central concepts. It initially illustrated

the project’s position within the greater scholarly paradigm of emerging welfare state research,

followed by a conceptualisation of social assistance policies as a distinct and centrally important

pillar of many nascent welfare states throughout the developing world. The chapter further sought

to articulate the way in which historical institutionalism is operationalised through the ACI ana-

lytical framework in order to conduct historical process tracing, premised upon an approach which

incorporates a broad definition of institutions and examines both the formal and informal features

of the South African and Brazilian institutional contexts.

41
Chapter 3

Literature Review

3.1 Introduction

The findings produced by this study respectively draw upon and ultimately contribute to at least

three distinct streams of academic inquiry. This chapter aims to elucidate the scholarly niche occu-

pied by the project through conducting a review of the current state of the relevant literature. The

first section surveys efforts to replicate in the developing world the considerable achievements of

scholars in comparatively chronicling and classifying the features of different welfare state arrange-

ments in developed countries. Overall, the literature on identifying varieties of welfare states in the

developing world remains a tremendously under-researched topic in comparison to the burgeoning

tomes that exist regarding their cousins in the global North. By explicitly undertaking an attempt

to differentiate and categorise the features of the South African and Brazilian approaches to social

assistance provision, this study thereby makes a significant qualitative contribution to the state of

research in this area.

The subsequent section examines the status of current knowledge claims related to the spe-

cific practise of social assistance provision in the developing world. Exploring this budding field of

research reveals that, while an impressive array of academic queries have focused on issues surround-

42
ing the implementation and impact of these programmes, very little is known about the underlying

factors that lead to their differentiated emergence in the first place. This study thus contributes to

the research on social assistance provision by taking up precisely that under-examined question.

The third part of the chapter shifts its attention towards reviewing the literature on the study’s

analytical framework. Relative to the other two fields this inquiry touches upon, research into the

effects of institutions on policy outputs is quite sophisticated. Nevertheless, the way in which this

study applies and extends a historical institutional ACI approach to the fledgling empirical field

of welfare provision in the developing world means that it also adds significant new value in this

regard.

The final section introduces a set of popular competing claims from the literature on the emer-

gence of social assistance provision. Despite the highly limited number of inquiries into the develop-

ment of these policies, most studies are content to simply point to exogenous functional or ideational

factors in explaining the design of programme features. These claims are consequently reviewed in

order to demonstrate their clear inability to account for the tremendous empirical variation that is

made manifest across countries. This literature review thereby introduces the reader to the current

state of scholarly understanding regarding emerging welfare states, social assistance provision and

the historical role of institutions, while simultaneously illuminating the study’s significant broader

relevance to scholarship in these fields and setting the scene for the subsequent empirical chapters.

3.2 Classifying Emerging Welfare States

Efforts to classify and interpret the nature of social welfare provision across different countries has

become one of the classic occupations of social science research. Scholars have applied an impressive

range of perspectives, including in fields such as sociology, history, political science, economics and

gender studies, in pursuit of understanding and explaining the profound variation between the

welfare regimes of different societies. This section provides an introduction to this area of inquiry

43
by initially examining the roots of welfare state research on the countries of the developed world.

It then looks at (limited) recent efforts to extend the classificatory approach to the newly emerging

welfare regimes of the developing world. It concludes by pointing to the fact that this study entails

one of the very first efforts to qualitatively assess, compare and explain the specific features of two

of the most prominent newly emerging welfare states.

The reference point for much of the work on welfare states in the developed world is Esping-

Andersen’s much celebrated book, The Three Worlds of Welfare Capitalism (1990). This ground-

breaking treatise, premised on the classification of developed welfare state regimes into ‘liberal’,

‘conservative’ and ‘social democratic’ ideal-types, has become nothing short of a modern classic.21

For Esping-Andersen, the principal issue distinguishing these three ideal-types from one another is

the extent of decommodification inherent to each approach. Decommodification is defined as

the degree to which individuals or families can uphold a socially acceptable standard of living

independently of market participation. The level of decommodification provided by a welfare

state depends on the stringency of eligibility rules, on the level of income replacement and on

the range of entitlements (Bonoli 1997: 353).

Esping-Andersen’s analysis built upon insights and decades of research produced by a stream of

scholarship that dates as far back as the 1940s, with Polanyi’s seminal The Great Transformation

setting the tone for the entire field of inquiry into social policy. This theoretical foundation was

further extended through work on social citizenship by Marshall (1950; 1963; 1965; 1981), and by

Titmuss on social policy and the philosophical underpinnings of the welfare state (1958; 1974).

Empirically, it drew upon research carried out during the 1970s and 1980s by a range of compar-

ativists, including Wilensky (1975); Flora and Heidenheimer (1981); Mommsen (1981); and Flora

(1983; 1986) (Arts & Gelissen 2002: 138).

The popularity of Esping-Andersen’s typology meant that it came under intense scrutiny, leading

to criticism that the specific categories were either flawed or lacked explanatory power (Lessenich
21
See Offe (1991); Cnaan (1992); Hicks (1991); Kohl (1993); as well as Arts and Gelissen (2002: 138).

44
& Ostner 1998; Baldwin 1996). These criticisms subsequently engendered a number of revisions.22

Its shortcomings notwithstanding, the book undoubtedly made a tremendous impact by going a

long way towards definitively establishing the practise of constructing comparative typologies to

make sense of the variations between different regimes of welfare provision. A host of contemporary

authors have subsequently followed in these footsteps by aiming to accurately represent and explain

the variations found between developed welfare states.23

The classification of welfare regimes into distinct categories has thus become a well established

academic practice. Indeed,

much of the existing literature on comparative social policy has been concerned with the clas-

sification of welfare states and the identification of ideal-types of welfare provision. This is

understandable: state welfare is a matter of high complexity, particularly when it comes to ex-

plaining differences between existing models of social protection. The classification of welfare

states and the resulting identification of ideal-types is thus a powerful tool for comparative social

policy, as it performs a significant reduction of complexity (Bonoli 1997: 351).

While much progress has therefore been made in terms of understanding the features of developed

welfare states, similar efforts to interpret welfare regimes in the developing world are still in their

infancy.

It is in fact only with recent work that similar analyses have been applied to emerging welfare

states (Gough 2004a; Gough 2004b; Gough & Wood 2004; Rudra 2007; 2008; Haggard & Kaufman

2008; Abu Sharkh & Gough 2009; Gough 2013). The fundamental premise of this growing field of

inquiry is that

existing scholarship overlooks the possibility of varieties of capitalism in the developing world.

In contrast, discussions abound regarding identifiable and systematic differences in domestic

institutional arrangements within the advanced capitalist countries, particularly with respect to
22
These are contained in Esping-Andersen (1993; 1994; 1996; 1997; 1999).
23
These include Pierson (2000; 2001); Huber and Stephens (2001); Arts and Gelissen (2002); Blekesaune and

Quadagno (2003); Scruggs and Allan (2006); Pestieau (2006); Pfeifer (2010); as well as Bonoli and Natali (2012).

45
 their distributional regimes. Is it feasible that distribution regimes of relatively poor countries

also fall into distinct patterns (Rudra 2007: 378)?

The application of sophisticated multivariate statistical cluster analyses have led scholars to

answer that question in the affirmative, concluding that ‘there is thus prima facie evidence to sug-

gest that...[there are] distinct welfare regimes based on different institutions and cultural-historical

antecedents and following different paths of development’ (Abu Sharkh & Gough 2009: 22). Two

basic typological models have been produced from this largely quantitative research agenda. The

first is based on work by Abu Sharkh, Gough and Wood (Gough 2004a; Gough 2004b; Gough &

Wood 2004; Abu Sharkh & Gough 2009; Gough 2013), while the second stems from Rudra’s (2007;

2008) research.

The Abu Sharkh and Gough typology identifies four regime clusters in the developing world:

‘externally dependent insecurity regimes’, ‘less effective informal security regimes’, ‘more effective

informal security regimes’, and ‘actual or potential welfare states’ (Gough 2004a; Gough 2004b;

Gough & Wood 2004; Abu Sharkh & Gough 2009; Gough 2013). The primary measurement indi-

cators used in these analyses relate to public spending, international aid and remittance flows, as

well as welfare outcomes expressed through the Human Development Index (HDI). The accuracy

of this typology has been questioned for its focus on outcomes data, as it erroneously assigns the

effects of events like conflict and the HIV/AIDS epidemic to ineffective welfare provision (Seekings

2014).

In turn, Rudra’s analysis identifies ‘the existence of two ideal-types of welfare states in the

developing world...[while] cluster analysis reveals a third group with elements of both’ (2007: 379).

The first approach is classified as productive, in which policies are geared towards promoting the

market dependence of citizens. This model shares certain elements with Esping-Andersen’s liberal

model by embracing some of the ‘enthusiasm for the market and self-reliance’ (Rudra 2007: 384).

Productive emerging welfare states thus emphasise the commodity status of labour. In contrast,

46
the protective model is aimed at protecting individuals from the market through relatively greater

decommodification.

Rudra finds that protective emerging welfare states constitute ‘a curious fusion of elements of

socialism and conservatism’ by incorporating a strong distrust of markets alongside a conservative

emphasis on the preservation of authority (2007: 379; 384). The protective typology accordingly

involves a combination of features from Esping-Andersen’s original social democratic and conser-

vative categories, while the productive moniker strongly equates, within a different context, to

Esping-Andersen’s liberal category.

The descriptive classificatory exercise undertaken in chapter five of this study builds upon this

previous work through the incorporation of the conceptual categories recently developed by Rudra

(2007; 2008). The key difference between this analysis and the literature cited above is that instead

of taking a global perspective which examines the entire welfare state apparatuses of dozens of

developing countries, the present study exclusively focuses on the single most important aspect of

welfare policy in these two cases: social assistance. The resultant classification of the South African

social assistance system as productive and the Brazilian approach as protective thereby amounts

to the study’s initial noteworthy contribution to the existing scholarship by undertaking the first

documented attempt at constructing typologies for social assistance provision in the developing

world.

This descriptive procedure is followed by an in-depth qualitative evaluation of the ‘different

institutions, cultural-historical antecedents and different paths of development’ that underlay their

differentiated development (Abu Sharkh & Gough 2009: 22). In addition to therefore making a

valuable descriptive contribution, the study also constructs a theoretically-grounded explanation

for why these differences emerged in the first place. It thereby extends the proud scholarly tradition

of comparative inquiry into welfare provision to these exciting and fertile new cases through a focus

on social assistance provision.

47
3.3 Social Assistance in the Developing World

This exclusive focus on social assistance (as a prominent subcategory of social welfare provision

in these cases) means that the analysis also ties into the related literature on the rise of social

assistance CTs to the poor in developing countries. The tremendous recent proliferation of these

schemes throughout the global South has been accompanied by a rush of academic interest. This

interest has however been almost exclusively preoccupied with measuring the impact that these

programmes have had on a dizzying range of outcomes, while research into why these schemes took

on different shapes in different countries has been neglected. This section introduces the state of

knowledge in these regards, while also demonstrating the present study’s value by constituting one

of the first attempts to actually explain why social assistance CTs took on certain specific features

in the cases under examination.

The practise of scientifically measuring the effects of CTs in the developing world dates back to

the 1990s, when the first of these programmes were introduced at scale. The first such evaluations

were in fact carried out in Brazil by the World Bank, Inter-American Development Bank, the

United Nations Development Programme (UNDP), the International Labour Organisation (ILO),

UNESCO, as well as UNICEF shortly after the introduction of the country’s first subnational

CT schemes.24 These early assessments were particularly focused on measuring the impact that

conditional CTs had on school attendance and achievement.

This was followed by an explosion of similar inquiries by economists, sociologists, social policy

experts and international development organisations that attempted to measure the effects of CTs

on outcomes including poverty, child labour, human capital investment, schooling, nutrition, health,

gender relations, crime, inequality, domestic violence and entrepreneurship.25 Much of this work
24
These assessments are contained in Moraes and Santana (1997); Abramovay, Andrade and Waiselfisz (1998);

Amaral and Ramos (1999); also see Buarque (2013b).

25
These themes are covered in Cardoso and Souza (2004); Gertler (2004); De Janvry and Sadoulet (2005); Rawlings

and Rubio (2005); De Janvry, Finan, Sadoulet and Vakis (2006); Schady and Araujo (2006); Soares, Soares, Medeiros

48
was carried out during the early-2000s with a regional focus on Latin America, where many of the

first programmes were implemented. More recent investigations have extended the endeavour of

measuring the impact of social assistance programmes to regions including Africa and the Indian

subcontinent.26 The vast majority of these assessments have found that CTs have a range of positive

impacts. Useful overviews integrating some of the findings from this expansive body of literature are

provided by Fiszbein et al. (2009); Hanlon, Barrientos and Hulme (2010); as well as by Barrientos

(2013).

It is thus clear that our understanding of the impacts produced by social assistance CTs is

relatively advanced. But there are as yet no systematic studies aimed at interrogating and account-

ing for the tremendous amount of variation that exists in how these programmes are implemented

across cases. The only literature that even alludes to this variation (which includes features such as

coverage, conditionality and systemic orientation) spans a total of three published journal articles

and one review on the relative effects of conditional versus unconditional CTs.27 But by again

focusing purely on outcomes, these also do not take up the question of why such variation exists in

the first place.

The scholarship on social assistance provision in the cases of South Africa and Brazil reflect

this same reality. In the case of South Africa, authors have meticulously analysed the outcomes

produced by the country’s social assistance policies,28 while a handful of studies have documented
and Osório (2006); Lagarde, Haines and Palmer (2007); Fernald, Gertler and Neufeld (2008); Adato and Hoddinott

(2009); Fiszbein, Schady, Ferreira, Grosh, Keleher, Olinto and Skoufi (2009); Leroy, Ruel and Verhofstadt (2009);

Soares, Ribas and Osório (2010); as well as in Soares and Silva (2010).
26
This includes work by Ahmed (2005); Kakwani, Soares and Son (2005); Hossain (2010); Lim, Dandona, Hoisington,

Jamers, Hogan and Gakidou (2010); UKDFID (2011); Ellis (2012); as well as by Garcia and Moore (2012).
27
Individual studies were carried out by Baird, McIntosh and Özler (2011); Akresh, De Walque and Kazianga

(2013); as well as Robertson, Mushati, Eaton, Dumba, Mavise, Makoni, Schumacher, Crea, Monasch, Sherr, Garnett,

Nyamukapa and Gregson (2013). These were subsequently reviewed in Baird, Ferreira, Özler and Woolcock (2013).
28
Including Case and Deaton (1998); Duflo (2003); Woolard (2003); Samson, Lee, Ndebele, MacQuene, Van Niekerk,

Gandhi, Harigaya and Abrahams (2004); Aguëro, Carter and Woolard (2006); Hall (2007); Williams (2007); UNICEF

49
the evolution of the system.29 The same is true in the case of Brazil, where the effects produced

by especially the Programa Bolsa Famı́lia (PBF – Family Grant Programme) have been loudly

proclaimed in both the academic and popular press.30 But the lack of rigorous comparative in-

vestigations into the political economy of social assistance provision means that there is very little

understanding of the processes whereby the specific features of these programmes actually emerged.

The present study thus directly addresses this shortcoming by conducting a detailed comparative

investigation into these two benchmark cases. While the current state of research on CTs reflects

an understandable preoccupation with the social and economic consequences produced by these

programmes, the time is ripe for more nuanced analyses that take up questions regarding the specific

features of social assistance programmes. By doing precisely that, this study compliments the

existing knowledge on programme effects by breaking new ground with an approach that explicitly

focuses on the emergence of such programmes in the first place.

3.4 Historical Institutionalism

This inquiry adopts a historical institutional theoretical approach to explaining the variation across

cases. Examining the role played by institutional configurations in shaping policy outputs is a well

established approach in the political science literature. In fact, the institutional school represents

perhaps the oldest theoretical tradition in the field: Plato’s Republic investigates the role played

by institutions in shaping political behaviour, while Aristotle examined the effect of institutional

structures on shaping political incentives in Politics. These classical works went on to greatly

influence social science as it began to emerge as a modern academic discipline in the late nineteenth
(2008); Woolard, Harttgen and Klasen (2010); as well as Woolard and Leibbrandt (2010).
29
See Kruger (1992); Duncan (1993); and Seekings (2000; 2007).
30
Prominent examples include Medeiros, Diniz and Squinca (2006); Legido-Quigley (2009); Sedlacek, Gustafsson-

Wright, Ilahi and Lannon (2001); Lindert (2006); Soares et al. (2006) Lindert, Linder, Hobbs and De la Brière (2007);

Haddad (2008); Hall (2008); The Economist (2008); and Watts (2013).

50
and early twentieth century (Almond 1990). The earliest ‘institutionalists’ thus included celebrated

theorists such as Montesquieu, Weber, Polanyi and Veblen. Much of the research carried out during

this fledgling period of political science focused on the design of constitutions (Steinmo 2008: 119).

This ‘old institutionalism’ soon came to dominate the field of comparative politics. Studies

largely consisted of highly detailed configurative inquiries into varying formal legal, political and

administrative structures (Thelen & Steinmo 1992: 3). By strictly contrasting and comparing

institutional configurations across different countries, the overwhelming majority of these studies

were descriptive in nature; and often deeply normative in orientation. As a result, ‘this approach

did not encourage the development of...concepts that would facilitate truly comparative research

and advance explanatory theory’ (Thelen & Steinmo 1992: 3).

The social and historical context of the post-war years gave rise to deep scepticism and even

hostility towards the focus on institutions, as carefully designed democratic institutions fell to

dictatorship and chaos throughout the developing world. As a result, political scientists increasingly

came to view institutions as mere ‘vessels in which politics took place; what mattered was what filled

the vessels’ (Steinmo 2008: 119). This was further compounded by the ‘physics envy’ that emerged

in the wake of the progress made by the physical sciences, as well as a related desire to move away

from ‘thick description’ towards more theoretical approaches (Steinmo 2008: 120). The result was

that institutional approaches largely fell out of favour, to be replaced by the rise of behaviouralism.

Political science consequently witnessed the behavioural revolution in the 1950s and early 1960s as

a reaction against and a rejection of the old institutionalism (Thelen & Steinmo 1992: 3).

Instead of examining the formal attributes of governmental institutions, behaviouralists postu-

lated that politics could only be understood through analysing political attitudes, behaviours and

distributions of power. The perceived atheoretical bent of the old institutionalism thus came to

be replaced by the explicitly theoretical behaviouralist project (Thelen & Steinmo 1992: 4).31 In
31
It is important to note that despite the general trend away from institutional analyses and towards behaviouralism,

‘dissident’ scholars such as Barrington Moore (1966) and Samuel Huntington (1968) continued to strongly incorporate

51
turn, by the 1970s, a clear institutionalist critique of behaviouralism had congealed around a central

theme: precisely due to the fact that behaviouralist theories focused on the attitudes, characteris-

tics and behaviours of groups and individuals in explaining outcomes, they consistently overlooked

questions about why these same attitudes, behaviours and the distribution of resources between

groups themselves differed between countries (Thelen & Steinmo 1992: 5).

Kenneth Shepsle provided what amounts to perhaps the most succinct critique of this approach

when he stated that

the price we have paid for the methodological and theoretical innovations of the post-World War

II era, however, is the inordinate emphasis now placed on behaviour. Our ability to describe (and

less frequently, to explain) behaviour...has diminished the attention once given to institutional

context and actual outcomes. On net, the behavioural revolution has probably been of positive

value. But along with the many scientific benefits, we have been burdened by the cost of the

restricted scope in our analyses (1986: 52).

The so-called neoinstitutionalism subsequently arose in response to these perceived shortcomings

with the aim of producing institutionalist analyses that simultaneously contributed to theory build-

ing.

In contrast to the abstract grand theories pursued during the 1960s and early 1970s, neoin-

stitutionalists became focused on intermediate institutions, meso-level analysis and middle-range

theory (Steinmo 2008: 123). But it soon became clear that at least two distinct strands32 had

emerged within neoinstitutionalism: rational choice institutionalism33 and historical institutional-

ism. Whereas rational choice institutionalism was focused on uncovering general laws of political

behaviour and action by deductively testing theoretical models, historical institutionalist scholars

were more interested in explaining real world outcomes inductively (Steinmo 2001). In rational
institutional variables into their work during this period.
32
See Peters (2012) for a full elaboration on the different neoinstitutionalist approaches.
33
Some of the foundational rational choice works include Shepsle (1986); Levi (1988); North (1990); and Bates

(1981; 1989).

52
choice, the primary priority is creating, elaborating and refining a theory of politics; in historical

institutionalism it is understanding and explaining real outcomes (Weingast 1996).

This study is situated within the historical institutionalist strand of neoinstitutionalism, with

a focus on inductively explaining real world variation. It thereby draws on a stream of scholarship

which argues that the role played by institutions in shaping politics is far greater than that suggested

by a narrow rational choice model, with the core feature being historical institutionalism’s treatment

of preference formation as endogenous (Thelen & Steinmo 1992: 5). Historical institutionalism

assumes that political struggles ‘are mediated by the institutional setting in which [they] take

place’, and incorporates a definition of institutions that includes both formal organisations and

informal norms that structure conduct (Ikenberry 1988; Hall 1986: 19; Thelen & Steinmo 1992: 2).

In sum, for historical institutionalist scholars,

institutions are not just another variable, and the institutionalist claim is more than just that

‘institutions matter too’. By shaping not just actors’ strategies (as in rational choice), but their

goals as well, and by mediating their relations of cooperation and conflict, institutions structure

political situations and leave their own imprint on political outcomes (Thelen & Steinmo 1992:

9).

The late 1970s witnessed the first great flourishing of historical institutionalist scholarship,

including groundbreaking work by Peter Katzenstein (1978); Theda Skocpol (1979); Peter Hall

(1986); as well as James March and Johan Olsen (1984; 1989). This was followed by an avalanche of

historical institutionalist investigations into topics as diverse as the intersections between domestic

and international politics,34 American politics,35 the causes and consequences of revolutions and
34
See Friedberg (2000); Gourevitch (1986); Ikenberry (2001); Katzenstein (1978); Krasner (1978); and Simmons

(1994).
35
Examples include Skowronek (1997); and Hansen (1991).

53
social movements,36 the origins and development of economic regimes,37 transitions to democracy,38

the rise and fall of authoritarian regimes,39 and the role of social identity in politics40 (Pierson &

Skocpol 2002: 695). Of particular relevance to this study are the rigorous historical institutionalist

analyses that have been carried out in accounting for the development of welfare states.

Indeed, the work by Esping-Andersen (1990) on the classification of developed country welfare

states cited above represents the most famous example of an historical institutionalist approach

to explaining variations in welfare provision. Other authors who have contributed to the ‘striking

accumulation of knowledge’ in this field include Flora and Heidenheimer (1981); Hacker (1998);

Huber and Stephens (2001); Immergut (1992); Maioni (1999); Pierson (1994); Skocpol (1992); and

Steinmo (1996) (Pierson & Skocpol 2002: 695). The result of these previous inquiries has been to

firmly establish historical institutionalist approaches to studying welfare policies as a legitimate and

accepted scholarly practise.

The study’s specific analytical approach of ACI also belongs to this contemporary school of lit-

erature (Scharpf 1991; Mayntz & Scharpf 1995; Scharpf 1997; Scharpf 2000; Boessen 2008; Pancaldi

2012). ACI emerged during the 1990s and 2000s as a precisely oprationalised framework which en-

ables researchers to consistently define and study the processes and mechanisms whereby historical

institutional factors shape policy outputs. While it consequently regards institutions as the primary

force shaping such outputs, it additionally allows for the construction of a rigorous empirical pic-

ture through its incorporation of agency by regarding policy not as deterministic but as ‘intentional
36
See Goldstone (1991); Goodwin (2001); McAdam (1982); McAdam, Tarrow, and Tilly (2001); Skocpol (1979);

Tarrow (1998); and Wickham-Crowley (1992).

37
Important examples include Karl (1997); Richards and Waterbury (1990); Streeck (1992); Thelen (1993; 1994);

Zysman (1994); and Acemoglu and Robinson (2006).

38
Refer to work by Baloyra (1987); Bratton and Van de Walle (1997); Diamond (1999); Downing (1992); Gould

(1999); Haggard and Kaufman (1995); Rueschemeyer, Stephens, and Stephens (1992); Yashar (1997) and Mahoney

(2000).
39
Examples include Doyle (1986); Ekiert (1996); Im (1987); Mahoney (2001); Snyder (1998) and Mahoney (2000).
40
See Hattam (1993); Katznelson and Zolberg (1986); Kryder (2000); Lustick (1993); and Marx (1998).

54
action by actors who are most interested in achieving specific outcomes’ (Scharpf 1997: 36).

The contribution of the present study is thus to extend what has become an accepted analytical

convention into an exciting new empirical area. In keeping with established academic norms, it

applies a historical institutionalist explanation (operationalised through the ACI framework) to

account for the variation in social assistance provision to the as-yet unexamined cases of South

Africa and Brazil. This pioneering project thereby hopes to encourage the further investigation of

emerging welfare states through the application of a historical institutionalist approach. It is a field

that is ripe for the picking, and institutionalist scholars are perfectly positioned to add great value

to our understanding of the recent domestic emergence of welfare policies in the global South.

3.5 Explaining Variation in Social Assistance Provision

Owing to the abovementioned lack of comparative inquiries into the emergence of social assistance

provision in the developing world, there remains a relative dearth of theoretical claims about what

accounts for the variation across cases. Despite the lack of rigorous explanatory analyses, a handful

of authors (usually of descriptive studies) have nevertheless proffered two primary hypotheses for

why social assistance programmes in any given country has taken on its particular shape (Graham

2002; Fiszbein et al. 2009; Hanlon et al. 2010; UNRISD 2010; Liesering & Barrientos 2013;

Mothiane 2014). These exogenous claims generally either lurk below the analytical surface, or are

often introduced in a rather matter-of-fact fashion without citing satisfactory supporting evidence

— usually in studies otherwise devoted to exalting the benefits of social assistance. This section

briefly reviews this conventional wisdom’s capacity to amount to potential competing claims for the

study’s central institutionalist hypothesis.

The first explanation usually introduced in accounting for the emergence of social assistance

systems is functional in nature, and it is nearly ubiquitous in the literature. A striking majority

of evaluations appear happy to loosely argue that programme features mechanistically emerged as

55
a result of high levels of poverty and inequality (see, for example, Fiszbein et al. 2009: 9; Garcia

& Moore 2012: 1). It is the assertion of proponents of this view that domestic social assistance

programmes arose in response to ‘economic crisis, structural adjustment and global integration,

where the limitations of residual, ad hoc safety nets to address the social consequences of neoliberal

policies became painfully apparent’ (UNRISD 2010: 135).

A representative example is provided by one analysis on the rise of social assistance in Sub-

Saharan Africa, where it is proposed that

cash transfers arose...as recognition grew that some other types of aid were not effectively achiev-

ing their goals...Increasing migration, urbanization, and the evolution of traditional family struc-

tures have also weakened traditional safety nets...The ability of informal safety nets to protect

individuals has weakened considerably in the face of increased demands brought on by the

HIV/AIDS crisis...These problems are compounded when considered jointly with other sources

of vulnerability and poverty in the region, such as exclusion, patronage politics, insecure property

rights and landlessness, environmental degradation, and conflict stemming from ethnic differences

(Garcia & Moore 2012: 12).

In addition to apparently assigning explanatory power to a stunning array of factors, the crucial

shortcoming of this interpretation is not that its diagnosis of social and economic challenges is

factually incorrect. It is rather that this functional view is not able to analytically accommodate the

empirical fact of significant variation across different countries. It paints a picture of a mechanistic

and uniform policy response to a panoply of shared crises (which have often persisted for many

decades prior to the adoption of social assistance) that leaves no room for nuance and is simply not

borne out in reality. Based on this logic, applying the functional hypothesis to this study’s research

question thus suggests that differences between the social assistance systems of South Africa and

Brazil must be the result of considerable ‘objective’ differences in the underlying policy problems

faced by the two countries.

The second popular claim is related to the first in that it argues that the failure of traditional

56
approaches to development and the consequent persistence of high rates of poverty in developing

countries has led to ‘changed thinking’, with the introduction of social assistance therefore represent-

ing ‘a paradigmatic shift in poverty reduction’ (Barrientos & Hulme 2010: 4). This interpretation

therefore suggests that ideational factors constitute the primary explanatory variables in accounting

for the design of social assistance policies in developing countries.

Reviewing the literature on this claim suggests that it is both the ‘ideology of the state’ and

‘public attitudes’ towards social assistance provision that supposedly determines the shape of policy

interventions (Mothiane 2014: 4; Graham 2002). In the simplest terms, the logic of this interpre-

tation means that ‘differences in political attitudes about redistribution’ are decisive, with a state

featuring a more ‘neoliberal’ orientation expected to yield a reduced scale of provision compared to

that of a supposedly more ‘social democratic’ state and polity (Graham 2002: 2). Operationalising

this approach in terms of the current study’s research question thus results in a hypothesis sug-

gesting that any differences between the two cases must have resulted from preceding differences in

their underlying ideational orientations.

While acknowledging that functional and ideational factors were certainly relevant during the

policy development processes in both cases, the outcome of this study’s careful evaluation shows

that these exogenous factors are not able to satisfactorily account for the variation in the approaches

adopted by South Africa and Brazil. Functional and ideational factors were plainly too similar across

the cases to be of any real explanatory relevance. The failure of these approaches to adequately

account for the policies implemented in these two benchmark cases serves to confirm the existence

of an alarming gap in our understanding of the political economy of social assistance provision in

developing countries.

In addition to the extension it entails to the scholarship on the classification of emerging welfare

states, social assistance provision, and historical institutionalism, addressing this explanatory void

in the literature is the primary objective of this study. Proceeding inductively from the empirical

57
observation that ‘in developing countries, social transfer programmes show considerable innovation

and diversity’ (Barrientos 2013: 6; Steinmo 2001: 571), it applies an analytical framework that

emphasises endogeneity and the way in which varying historical institutional configurations influ-

enced policy outputs by channelling functional and ideational factors through the creation of specific

institutional incentives and constraints.

3.6 Conclusion

This chapter has demonstrated the way in which the current study directly contributes to expanding

the state of knowledge in at least three important streams of scholarly literature. It initially reviewed

each of these in turn with the aim of illustrating the particular niche occupied by the project, followed

by the introduction of a set of competing explanatory claims regarding the domestic emergence of

social assistance provision in developing countries. The first section thus illuminated the study’s

contribution to the established academic practise of classifying the features of different welfare state

arrangements by qualitatively extending this endeavour to the important emerging cases of South

Africa and Brazil. The significance of this contribution was emphasised by pointing to the way

in which the identification and chronicling of systematic divergence in the welfare regimes of the

developing world remains a hugely under-researched area.

This was followed by a section introducing the literature on the emergence of social assistance

CTs throughout countries of the developing world. The review concluded that, while a great deal

of scholarly interest has been directed towards examining the (important) impacts produced by the

introduction of these programmes, scant attention has been paid to explaining their differentiated

development in the first place. It was subsequently argued that it is crucial for analysts to com-

pliment this descriptive knowledge by taking up explanatory questions which directly address the

tremendous amount of nuanced cross-case variation in how social assistance policies are actually

implemented. The contribution of this project is to do precisely that with regard to the cases of

58
South Africa and Brazil.

The review then shifted towards an explication of the literature on the study’s analytical frame-

work of historical institutionalism. It showed that there is a deep scholarly tradition of applying

this approach to analyse questions surrounding the development of social policy. The novelty of

the current study lies in its extension of this established interpretive framework — through an ACI

analytical approach — to a new empirical field. This was followed by a final section introducing a

set of popular competing claims aimed at accounting for the recent domestic development of social

assistance systems in many developing countries. It showed that these functional and ideational

explanations are often somewhat carelessly proffered, and concluded with a juxtaposition of these

claims relative to the study’s central institutional hypothesis. The review thereby sets the stage for

the empirical analysis conducted throughout later chapters.

59
Chapter 4

Research Design

4.1 Introduction

This chapter introduces the study’s research design and offers guidance on the analytical procedure

that follows. The initial section specifies the research question and dependent and independent

variables. This is followed by a discussion of the study’s methodology that illustrates the suitability

of the comparative case study method for addressing the research question. A subsequent section

reviews the procedure that was followed throughout the process of data collection, discussing sources

of both primary and secondary information. The chapter also addresses the limitations and delim-

itations of the present study. It articulates the procedures that were followed in overcoming the

practical challenges which arose, while simultaneously clearly framing the precise themes addressed

by the research. The final section reviews the ethical considerations which arose during the research

process, particularly with regards to the participant interviews. The chapter thereby provides a

concise overview of the primary methodological and practical aspects of the research project.

60
4.2 Methodology

As mentioned briefly in the introduction, the study’s research question is: what accounts for the

differences between the South African and Brazilian social assistance systems? In effect, the anal-

ysis aims to explain why South Africa adopted a productive social assistance regime, while Brazil

implemented a protective approach. These typologies, summarised in table 4.1 below as the study’s

dependent variables, are operationalised along three dimensions: coverage, conditionality and orien-

tation. The subsequent explanatory endeavour is conducted according to the independent variables

identified in table 4.2.

Variable South Africa Brazil

Policy Regime Productive Protective

Coverage: Limited Coverage: Universal

Conditionality: Unconditional Conditionality: Conditional

Orientation: Individualist Orientation: Familial

Table 4.1: Dependent Variables

Variable South Africa Brazil

Institutionalisation of Norms High Low

Concentration of State Structure High Low

Table 4.2: Independent Variables

As hinted at by the nature of the theme under investigation, the study adopts a qualitative

orientation to answering the research question. The rationale for undertaking a qualitative inquiry

is fundamentally related to the way in which this allows for the interpretation of phenomena within

their overall contexts (Denzin & Lincoln 2005: 3). As such, the research was premised upon

‘the collection of data in a natural setting sensitive to the people and places under study, and

data analysis that is inductive and establishes patterns and themes’ (Creswell 2007: 37). Further

61
justification for the suitability of a qualitative orientation aimed at developing hypotheses emerged

from the preceding review of the literature, where it was clearly established that ‘existing theories

do not adequately capture the complexity of the problem we are examining’ (Creswell 2007: 40).

Having settled upon a qualitative orientation, the next methodological consideration was to

decide upon a specific approach out of what initially appears to be a ‘baffling number of choices’

(Creswell 2007: 6). Creswell’s exhaustive survey on the nature of qualitative inquiry assisted greatly

in this pursuit, as it fundamentally narrowed the amount of options down to a limited number of

qualitative approaches that feature ‘systematic procedures for inquiry’ (2007: 9-10). Yin (2003)

further introduces the matrix depicted in table 4.3 to assist in identifying an appropriate research

method.

Method Form of Requires Control of Focus on

Research Question Behavioural Events? Contemporary Events?

Experiment how, why? yes yes

Survey who, what, where, no yes

how many, how much?

Archival who, what, where, no yes/no

Analysis how many, how much?

History how, why? no no

Case Study how, why? no yes

Table 4.3: Relevant Situations for different Research Methods (Source: Yin 2003: 111)

He goes on to note that choosing the case study method

depends in large part on the research question. The more that the question seeks to explain some

present circumstance (e.g. ‘how’ or ‘why’ some social phenomena works), the more that the case

study method will be relevant. The method also is relevant the more that your questions require

an extensive and ‘in-depth’ description (2003: 85).

62
This study’s research question deals with precisely these themes. It firstly contains a distinct and in-

depth descriptive (‘how?’) element in terms of articulating the cross-case variation. When combined

with the explanatory (‘why?’) essence of the research question, the comparative nature of the study,

the fact that behavioural control was not required, as well as the contemporality of the inquiry, it

became abundantly clear that the case study method was ideally suited for addressing the subject

matter of the research.

The use of this strategy is a well-accepted convention in the quest to ‘understand the complex

social phenomena’ constituted by social welfare provision (Yin 2003: 2). A detailed recent review

concluded that the practise of employing case studies in social science research is in fact ‘solidly

ensconced and, perhaps, even thriving’ (Gerring 2004: 341). It is a particularly useful method when

a holistic and in-depth examination is required, as with the present topic (Feagin, Orum & Sjoberg

1991). Case studies can however be further subdivided into three distinct types: exploratory,

descriptive and explanatory (Yin 2003: 102).

The present study is y-centred and explanatory in nature, with a secondary descriptive element,

as it firstly attempts to identify ‘patterns of association within the data, [combined with an] attempt

to account for why those patterns occur’ (Ritchie & Lewis 2003: 215). It incorporates multiple cases

into the subject, thereby constituting cross-case analysis. The study is carried out on the macro

level through the investigation of the policy regimes of two cases (the units of analysis): South Africa

and Brazil. The analysis is conducted on the basis of a parallel study, as the timelines utilised for

examining the cases run concurrently from approximately 1990 to the present (Thomas 2011: 517).

The study incorporates an inductive approach in carrying out the case analysis. The process

initially involved an observation of the empirical world through measuring the differences between

the cases. This was followed by ‘thinking in increasingly more abstract ways, moving toward the-

oretical concepts and propositions’ (Neuman 2005: 60). The analysis of each case takes the form

of a historical process tracing narrative, guided by the effects produced by the independent vari-

63
ables. Systematisation is achieved through the application of the historical institutionalist analytical

framework of ACI. This approach is ideally suited to account for the differentiated developmental

processes due to its focus on the role played by specific institutional incentives and constraints in

the course of policy formulation, enabling a parsimonious exposition of the underlying processes.

4.3 Data Collection

The application of the case study method often calls for the assembly of a large amount of data.

In order to increase the validity of the findings, every attempt was made to obtain data from a

multiplicity of sources. This list ultimately included scholarly publications (books and journal ar-

ticles), information sourced through the popular press, statistical records, government documents

and regulations, annual reports, records of parliamentary proceedings as well as participant inter-

views. This diverse set was comprised of both primary and secondary sources. The initial part

of the research process was focused on reviewing the existing literature on the emergence of social

assistance provision in the developing world. This phase, which was principally carried out based on

secondary data from books, journal articles and working papers, allowed for the initial identification

and refinement of the study’s research question.

It was also during this period that the shortcomings of existing accounts on the development

of specific programme features were first identified. It became clear that most analyses were only

interested in examining the effects produced by these policies, while largely ignoring the significant

cross-country variation with regards to programme implementation. Accounts of the emergence

of the policies were usually limited to sweeping statements premised on functional or ideational

interpretations. The identification of South Africa and Brazil as the most suitable candidates for an

analysis premised on the MSSD approach was followed by intensive research into the specific details

of social assistance provision in these two countries. Working papers from research institutions and

think tanks, as well as governmental information portals, statistical reports and legal regulations

64
proved critical in this regard.

The data gathered through these secondary sources were subsequently complimented by inter-

views with policymakers, civil society actors and social policy experts. In-depth interviews with

South African actors were carried out during a fieldwork trip to that country in November and

December 2012, while interviews with Brazilian policy actors were conducted during a research stay

in Brası́lia from October to December 2013. Given the intensive nature of the interviews carried out

for this study, as well as the generally small network of actors involved in social assistance policy

development, a relatively small group of 17 actors were ultimately interviewed.41

In the case of South Africa, the first interviewee was the head of a prominent government

commission on social assistance, while the first participant identified for the Brazilian interviews

was the instigator of the initial programmes in that country. This was followed by a procedure of

purposive and snowball sampling, where the predetermined list of participants was complimented

through interviewees providing links to other potential informants. At the same time, these formal

data gathering exercises were supplemented by the invaluable informal dialogue fostered through

participation in internships at the South African Parliament and the UNDP’s International Policy

Centre for Inclusive Growth in Brazil.

The final phase of the data collection process involved the triangulation of data from these

different sources. Information collected through document analysis, press clippings and interviews

were combined in order to ‘clarify meaning by identifying different ways in which the phenomenon

is being seen’ (Stake 1995: 97). Discussions with numerous experts in the policy community as

well as the feedback generated through presentations at a range of academic conferences further

strengthened these efforts. The result is that sources have been drawn from as wide a spectrum

as possible, while every attempt was made to confirm and validate the narratives and findings

produced by this study.

41
See Appendices A and B for more information on the interviewees.

65
4.4 Limitations and Delimitations

The nature of qualitative research implies a number of inherent limitations that must be acknowl-

edged. Despite the fact that a rigorous procedure was followed to assure the validity of the findings,

a methodology premised on written sources and participant interviews automatically raises ques-

tions regarding sample selection. With regards to the interviews and given the specificity of the

topic, it was determined that representative or randomised sampling approaches would not be ap-

propriate. The theme of the research necessitated discussions with very specific groups of policy

actors.

Drawing from general precedent, the decision was made to follow a method that combined

purposive and snowball sampling. In general, the main potential limitation of this approach is

that participants are largely drawn from the same network. However, the limited size of the social

assistance policy communities meant that in-depth interviews with a small group of interrelated

actors were deemed appropriate. Nevertheless, a keen awareness of this potential limitation meant

that additional purposive steps were taken to gather the views of actors from different networks,

including government, civil society, academia and opposition parties. Despite these efforts, it is

nevertheless still possible that important potential participants were simply unavailable during the

roughly two months spent conducting fieldwork in the respective countries.

A further practical limitation worth mentioning concerns the data gathered from various written

sources on the Brazilian case. In spite of the generous assistance received from colleagues and the

utilisation of the latest translation software, it is possible that some important Portuguese language

sources were omitted. The widespread global coverage of Brazilian social assistance programmes

however offered a partial remedy to this, as it was generally not significantly more problematic to

gather relevant information on this case than it was for South Africa.

It is also important to note a few facts regarding the delimitations of the project. In contrast to

the vast majority of work on this topic, this study does not undertake an evaluation of the effects

66
produced by social assistance interventions in these two cases. Neither does it explicitly argue that

one approach is necessarily normatively superior to the other. This is in spite of the fact that the

analysis shows that the Brazilian system qualitatively fosters somewhat greater decommodification

and provides greater redistributive coverage than its South African counterpart. It is however up

to the reader to judge the normative implications of the described variation.

Even though the study examines the effects produced through path dependence (or the lack

thereof), it is also not primarily concerned with examining the reasons for certain practices and

norms being transferred (or not) into the institutional-legal configurations of the countries following

their transitions to democracy. This would entail an examination of constitution-making processes,

an endeavour which falls outside the scope of the present inquiry. The focus is instead on illustrating

the effects that these configurations had on subsequently shaping social assistance policy outputs.

The general point to keep in mind is that this study is aimed at explaining variation in terms of a

set of clearly defined outcome variables.

While it is forthrightly acknowledged that social assistance provision touches upon and is inti-

mately related to a number of other policy arenas that ultimately shape a country’s distributional

regime, this study does not address policies related to social insurance, labour protection, healthcare

or any other fields traditionally associated with welfare provision. It is instead resolutely focused

on one crucially important but specific aspect of welfare policy: programmes that transfer cash to

beneficiaries on a non-contributory basis. While it is certainly hoped that this strict focus and the

results produced thereby will spur similar comparative studies into additional aspects of redistribu-

tive policies in newly emerging welfare states, this study presently occupies itself only with this one

specific theme.

67
4.5 Ethical Considerations

The process of carrying out the research did not present any extraordinary ethical quandaries.

The nature of the project meant that the primary consideration was ‘ensuring that [interview]

participants are not harmed, privacy is maintained, and the participants have provided informed

consent’ (Lichtman 2013: 51). These conditions were all met, including the crucially important

issue of informed consent. All participants agreed to conduct the interviews on the record, including

recording of the audio for subsequent transcribing purposes. Publications that emerged during the

research process were sent to all cited participants for verification.

In addition to meeting these specific requirements, the research procedure adhered to all the

requirements of good scholarly practise. Throughout the research process, the author was certainly

‘as concerned with producing an ethical research design as [with] an intellectually coherent and

compelling one’ (Bloomberg & Volpe 2008: 76). No aspect of the inquiry necessitated formal

intervention by an institutional review board. The research thus conforms with standards of ethical

scholarly conduct.

4.6 Conclusion

This chapter has introduced the reader to the primary methodological and practical considera-

tions relevant to the project. In addition to specifying the research question and variable sets, it

expounded upon the study’s methodology with a particular focus on illuminating the rationale be-

hind conducting comparative case studies. This was followed by a discussion on the data collection

procedure, information on the steps that were taken to address potential limitations, as well as a

clear delimitation of the themes under examination. The final part of the discussion addressed the

ethical implications of the research procedure. The outcome is a chapter which effectively deals

with the relevant methodological and practical aspects of the research project.

68
Chapter 5

Comparing Social Assistance in South

Africa and Brazil

5.1 Introduction

This chapter is oriented towards addressing the study’s descriptive objective concerning the qual-

itative features of, and differences between, the South African and Brazilian systems of social

assistance. This initially involves a conceptual specification of each country’s social protection

framework, as well as detailed technical analyses of the exact features that constitute the social

assistance systems in South Africa and Brazil. This section includes a focus upon the coverage of

the systems through the prism of lifecycle risks, the legal and institutional frameworks governing

them, the individual programmes that constitute the two approaches, benefit levels, targeting and

eligibility requirements, the delivery of benefit payments and possible conditionalities.

Moving from a technical to an abstract discussion, the subsequent section of the chapter un-

dertakes a direct comparison that identifies significant variation across the cases in terms of sys-

temic coverage, the enforcement of conditionalities and orientation. These differential features are

subsequently synthesised into two distinct typologies, where it is argued that South Africa has

69
implemented a productive social assistance system, while Brazil’s approach is classified as protec-

tive. This synthesis of the key variations into distinct typologies — which constitute the study’s

dependent variables — sets the stage for the following chapters analysing the development of both

systems. The chapter therefore plays a crucial role in defining the exact parameters of the ensuing

investigation through inductively specifying the dependent variables and addressing the descriptive

research objective.

5.2 Social Protection in South Africa

In terms of the conceptual approach outlined in chapter two, the South African social protection

framework is composed of a labour regime governed through legislation such as the Labour Relations

Act (RSA 1995a), the Basic Conditions of Employment Act (RSA 1997a), and the Employment

Equity Act (RSA 1998), as well as a large social assistance system and a relatively small social

insurance system.42 These three components are in turn composed of various different individual

programmes, statutory bodies and enforcement mechanisms. While the focus here is evidently on

the social assistance aspect, figure 5.1 below introduces a brief conceptualisation of the overall social

protection framework, including the relevant funding mechanisms.

A crucial feature of this framework is the way in which it approaches eligibility. By definition,

labour protection is only applied to workers in the formal economy, leaving those in the informal

economy exposed. Social insurance funds, both statutory and voluntary, overlap to a large degree

with those people who are formally employed and covered by labour regulations, due to the fact

that eligibility is determined on the basis of previous contributions. This leaves social assistance as

the only part of the South African framework that supports those people who do not contribute to,

and are therefore ineligible for, social insurance.

42
When measured in terms of the relative number of beneficiaries supported.

70
 SOCIAL PROTECTION IN
SOUTH AFRICA

Social Assistance Social Insurance Labour Protection

 Child Support Grant  Statutory Funds  Labour Relations
 Older Persons Grant  Voluntary Funds Act, 1995
 Disability Grant  Basic Conditions of
 Foster Child Grant Employment Act,
1997
 Care Dependency
Grant  Employment Equity
Act, 1998

Statutory Funds Voluntary Funds

 Unemployment  Medical Schemes
Insurance Fund (Private Health
 Compensation Insurance)
Funds  Retirement Funds

Road Users

Government Tax Funds from Workers and

Employers

Figure 5.1: Social Protection in South Africa (Source: Adapted from NPC 2011: 329)

5.2.1 Social Assistance in South Africa

Having contextualised the position of social assistance within the general social protection frame-

work, this section will focus exclusively on social assistance programmes in an attempt to specify

the exact contours of these policies in South Africa. It carries out an in-depth review of the sys-

tem’s principal features, including its coverage, legal and institutional framework, the individual

programmes that constitute the system, benefit levels, targeting and eligibility requirements, the

delivery of benefit payments and possible conditionalities.

71
Coverage

An important initial approach to conceptualising the system involves examining it through the

prism of lifecycle risks. This entails mapping the social assistance framework in accordance with

the various life stages and their associated basic risks, in order to assess the extent of overall

coverage; figure 5.2 below provides such an overview.43

Birth Time Death

Education Phase (age Accumulation Phase Dependent Phase (age

0-18) Risks: (age 18-60) Risks: 60+) Risks:
Poverty/Education Poverty/Unemployment Poverty/Old Age
Funding Risk: Risk: Risk:

 ? 
Health/Disability Health/Disability Risk: Health/Disability


Risk: Risk:

 

Figure 5.2: South African Lifecycle Risks (Source: Adapted from Smith 2011)

This figure reveals that

there [is a] critical gap [in the South African social assistance system, through] the lack of

protection for many working-age people. . . For those who are willing and able to work, but who

are locked out of the economy, there is no meaningful level of social protection (NPC, 2011: 331).

The Unemployment Insurance Fund (UIF – a contributory social insurance programme) provides

some support to workers who have lost their sources of income; in 2010/11 average monthly benefits
43
A checkmark indicates that there is a programme in existence aimed at addressing a particular risk, while a

question mark indicates that there is no programme protecting against a specific risk. In the education phase, the

programmes providing support are the Child Support Grant and the Foster Child Grant (poverty; education funding

risk), as well as the Care Dependency Grant (health/disability risk). During the accumulation phase there is no social

assistance programme aimed at the poverty/unemployment risk, while the Disability Grant covers the health/disability

risk. Finally, for the dependent phase, there is the Older Persons Grant (poverty/old age risk and health/disability

risk).

72
amounted to R466.8 million paid to 207 646 beneficiaries (NPC 2011: 331). However, this figure

represents only a fraction of the total unemployed, which stood at between 4.6 million and 17.7

million by the end of 2013 (Stats SA 2013: XI). As a result, only 3 percent of the unemployed have

access to these unemployment benefits at any given time (Klasen & Woolard 2008). This is due

to the fact that fully 55 percent of the unemployed have never worked and, by implication, have

never contributed towards the UIF, while 44 percent of unemployed people who have previous work

experience have been unemployed for more than a year and ‘would have exhausted their benefits if

they were ever eligible for them’ (NPC 2011: 331).

This means that there is a yawning gap right in the middle of the South African social assistance

system, as there are no non-contributory programmes aimed at supporting able-bodied unemployed

adults. In effect, once you turn 18 years old in South Africa, you are ‘on your own’ (Ulriksen 2014:

30). The implications are significant, as this shifts the considerable burden of unemployment onto

workers and those lucky enough to have access to social grants44 meant for other purposes, as the

able-bodied impoverished have no choice but to depend on ‘goodwill transfers’ from them. This

in turn dilutes the anti-poverty effects of social assistance and exacerbates existing wage pressures,

pushing households that are already financially constrained even further into poverty (NPC 2011:

332-333).

In lieu of implementing social assistance schemes to support this vulnerable societal group,

South Africa has instead created the Expanded Public Works Programme (EPWP). The creation

of a large, national public employment programme45 is relevant here because many of the people
44
Social assistance programmes are commonly known in South Africa as ‘social grants’, and ‘social assistance

programmes’ is used interchangeably with this term throughout the study.

45
The EPWP aims to create 4.5 million short-term ‘work opportunities’ between 2009 and 2014 (DPW 2012: 4).

The programme has been severely criticized because it does not create any ‘real’, permanent jobs (and income) for

beneficiaries and offers only meagre, short-term employment to those lucky enough to be enrolled (Meth 2011; McCord

2012). Its potential in providing any kind of safety net to the unemployed is further diminished due to the fact that it is

not an employment guarantee scheme, and therefore does not provide income security through a guaranteed minimum

73
who end up working on EPWP projects are precisely those who would otherwise have been eligible

for support through social assistance. The state has thereby effectively chosen to focus its anti-

poverty efforts in relation to poor adults on public works programmes which supposedly provide

the poor with the ‘dignity of work’ (Seekings 2008: 33). The design of the South African social

assistance system thus reflects a ‘classic northern conception of desert...[that] those poor who are

unable to work should be assisted, but those of working age must go out and earn a living (or be

dependent on a breadwinner, for example, through marriage’ (Seekings 2008: 33).

The lack of universality in the South African system has two very important implications. The

first is that in spite of one of the world’s highest rates of unemployment, all adults who are physically

able are compelled to rely on selling their labour to earn an income through working (Seekings 2008:

33). The second, related, implication is that some people — adults who are unable to find work —

should be allowed to live in poverty. Thus emerges the first important feature of the South African

system: the coverage gap resulting from the lack of any social assistance programmes supporting

able-bodied unemployed adults.

Legal and Institutional Framework

Moving beyond these key framing contours, it is important to identify the legal and institutional

framework governing the administration of social grants. This is derived from provisions contained

in the following set of laws and regulations: sections 27 (1) (c) and 27 (2) of the Constitution of

the Republic of South Africa (RSA 1996a); the Social Assistance Act, as amended (RSA 2004); the

South African Social Security Agency Act (RSA 2004); the Department of Social Development’s

(DSD) Regulations (RSA 2009); the Promotion of Access to Information Act (RSA 2000a); the

Promotion of Administrative Justice Act (RSA 2000b); the Promotion of Equality and Prevention
number of days of work for the unemployed (Mpedi 2014: 32). These features further underscore the fact that the

implementation of the EPWP instead of a social assistance programme aimed at working age adults constitutes a

significant systemic feature.

74
of Unfair Discrimination Act (RSA 2000c); and the Public Finance Management Act (RSA 1999a).

The social assistance system is centrally administered by the national DSD, which is responsible

for formulating policies, regulations and guidelines. However, a subsidiary of the department, the

South African Social Security Agency (SASSA), is the implementing agent. SASSA is a section 3

(a) entity under the Public Finance Management Act, and its stated aim is to ‘[pay] the right social

grant, to the right person, at the right time and place’ (Interview 2; SASSA 2014). Prospective

beneficiaries therefore have to apply for grants at local SASSA (not departmental) offices. In

addition to the head office in Pretoria, SASSA currently has 902 service points, 335 local offices

and 9 937 pay points spread throughout a total of 44 districts across the country, with more than

8 000 staff members (Interview 2).

Individual Programmes

This institutional overview sets the scene for an introduction of the individual social grants which

constitute the system, as administered by the DSD through SASSA. Table 5.1 provides information

on the respective grants, the beneficiaries and age brackets they target, their monthly monetary

value, and the total number of beneficiaries currently benefiting from each programme.

Grant Name Target Group Age Bracket Monthly Value Beneficiaries

Child Support Impoverished Children 0-18 R300 (US$30) 10 898 923

(max 5)

Foster Child Foster Children 0-18 R800 (US$80) 478 781

Care Dependency Disabled Children 0-18 R1 270 (US$127) 119 575

Disability Impoverished 18-59 R1 270 (US$127) 1 208 301

Disabled Adults

Older Persons Impoverished 60+ R1 270/R1 290 2 938 693

Elderly Persons (US$127/US$129)

Total - - - 15 644 273

Table 5.1: Social Grants in South Africa (Source: RSA 2014; SASSA 2013)

75
 Social assistance in South Africa is composed of five46 primary social grants currently paid out

to 15 644 273 total beneficiaries on a monthly basis (SASSA 2013). The system can broadly be

divided into providing support for children on the one hand, and the elderly and disabled on the

other. The first element of the social assistance system is the Child Support Grant (CSG), which

forms the single biggest social assistance intervention with almost 11 million beneficiaries (SASSA

2013). It currently pays R300 (US$30)47 per child on a monthly basis to the caregivers of children

younger than 18 who annually earn less than R34 800 (US$3 480) if single, or R 69 600 (US$6 960)

if married for a maximum of six children.

The state also provides a Foster Child Grant (FCG) of R800 (US$80) per month and a Care

Dependency Grant (CDG) of R1 270 (US$127) to the caregivers of foster and disabled children

(younger than 18). The other end of the spectrum provides for the elderly and disabled adults. The

Older Persons Grant (OPG) targets the first group. It is available to all citizens above the age of 60

earning less than R49 200 (US$4 920) per year and who own assets worth no more than R831 600

(US$83 160) if single, or, if married, less than R99 840 (US$9 984) and R1 663 200 (US$16 6320),

respectively. The grant value is R1 270 (US$127) per month, with 2.9 million recipients (SASSA,

2013). Finally, a Disability Grant (DG) for adults is available in the amount of R1 270 (US$127)

per month.
46
Two additional niche grants are not explicitly listed here: the so-called Grant-in-Aid (GIA) and the War Veterans

Grant (WVG). The GIA is only available as a ‘top up’ in the amount of R300 (US$30) per month to severely disabled

people who are already receiving another social grant (SASSA 2013). For the purposes of this analysis it has thus

been incorporated into the DG. The WVG (which has a mere 479 recipients) is subsumed under the Older Persons

Grant, as it is essentially an identical social pension for war veterans above the age of 60 (SASSA 2013).
47
Current exchange rates are calculated according to the average for the third quarter of 2013: USD1.00 = ZAR10.00;

USD1.00 = BRL2.30. All exchange rates are based on data from Feenstra, Inklaar & Timmer 2013.

76
Eligibility

Each of these grants is by definition non-contributory in nature, with eligibility being determined on

the basis of need. However, a specific set of targeting requirements ultimately determines whether

any given individual is eligible.48 The most significant aspect determining eligibility involves means

testing, where the income of an individual (or household) needs to be lower than a specified threshold

in order for that potential beneficiary to qualify for enrolment into the programme. The means test

essentially ‘defines [the] criteria of a target group. Ideal-typically the means test selects a group

of people who are identified as being in need and, in order to be cost efficient, the means test is

supposed to exclude those who do not need support’ (Haarmann 1998: 178). Table 5.2 introduces

the means test thresholds for each type of grant.

Grant Name Individual Means Test Threshold

Child Support Single: R34 800 (US$3 480)

Married: R69 600 (US$6 960)

Foster Child N/A

Care Dependency Single: R151 200 (US$15 120)

Married: R302 400 (US$ 30 240)

Not applicable to foster parents

Disability Single: R49 200 (US$ 4 920);

assets worth less than R831 600 (US$83 160)

Married: R98 840 (US$9 884);

assets worth less than R1 663 200 (US$166 320)

Older Persons Single: R49 200 (US$4 920);

assets worth less than R831 600 (US$83 160)

Married: R99 840 (US$9 984);

assets worth less than R1 663 200 (US$166 320)

Table 5.2: South African Means Test Thresholds (Source: RSA 2014)

48
An exhaustive list of the full set of eligibility requirements for each grant is attached in Appendix C.

77
Orientation

These registration requirements make it clear that the orientation of the South African system is

individualist in nature. This is in contrast to a familial orientation, where the means and eligibility

of beneficiaries are assessed based upon their position within defined family or household structures.

Instead, the South African system identifies and assesses each beneficiary as an individual.49 This

becomes especially clear with regards to the CSG, where the focus is explicitly on ‘following the

child’ by simply identifying the child’s ‘primary caregiver’, independent of familial or household

composition (Lund 2008). This ‘accommodates the mobility of [individual] children moving between

different households’ (Patel 2011: 371; Lund 2008: 53). An individualist orientation is another

important feature of the overall South African approach.

Appeals Mechanism

There are two sequential remedial options available to applicants who have had their social grant

applications rejected. The first is through an internal reconsideration in terms of section 18 (1) of

the Social Assistance Act, which stipulates that

if an applicant or a beneficiary disagrees with a decision made by the Agency in respect of a

matter regulated by this Act, that person or a person acting on his or her behalf may, within 90

days of his or her gaining knowledge of that decision, lodge a written application to the Agency

requesting the Agency to reconsider its decision in the prescribed manner (RSA 2010: 3).

This is an internal application handled by SASSA through its Internal Reconsideration Mechanism

(IRM), and the official reconsidering the application must be of a higher rank than the official

who initially rejected it. SASSA must inform the applicant within 90 days about the outcome of
49
This is true even though the means test does make provisions for assessing the income and/or assets of individuals

who are married. However, this is achieved by simply doubling the means test’s income threshold, and crucially does

not involve a means assessment based on per capita ‘family’ or ‘household’ income. The means test regards beneficiaries

as individuals, independent of their family or household situation.

78
reconsideration.

Should an applicant disagree with the outcome of internal reconsideration, their next option is

to lodge an external appeal in terms of section 18 (1) (a), which specifies that

a person or a person acting on his or her behalf may, within 90 days of his or her gaining

knowledge of that decision, lodge a written appeal with the Minister against that decision, setting

out the reasons why the Minister should vary or set aside that decision. (2) The Minister may —

(a) upon receipt of the applicant’s or beneficiary’s written appeal and the Agency’s reasons for

the decision confirm, vary or set aside that decision; or (b) appoint an independent tribunal to

consider an appeal contemplated in subsection (1) (a) in the prescribed manner and that tribunal

may, after consideration of the matter, confirm, vary or set aside that decision. (3) If the Minister

has appointed an independent tribunal. . . all appeals contemplated in subsection (1) (a) must be

considered by that tribunal. (4) Notwithstanding subsection (1) (a), the independent tribunal

may in the prescribed manner condone any late application by an applicant or a beneficiary

(RSA 2010: 3-4).

The tribunal alluded to in the legislation is the Independent Tribunal for Social Assistance Appeals

(ITSAA), which is composed of legal practitioners. SASSA must abide by any ruling made by the

ITSAA within 14 days.

Delivery

Once beneficiaries have been enrolled onto the system, the next step in the administrative process

involves delivering the actual cash payments. This is another core competency of SASSA. However,

this phase of the procedure is currently executed by a private contractor through a R10 billion (US$1

billion) government tender that will expire in March 2017, with SASSA aiming to take over full

control and administer payments in-house from that date onwards (Monama 2013). Beneficiaries

have three options in collecting their monthly grant payments: through having it uploaded onto

a SASSA biometric Electronic Bank Card (EBC) for withdrawal at selected financial institutions;

79
through collection at a one of 9 937 pay points across the country; or through delivery to the homes

or institutions for the aged and infirm (SASSA 2012).

Conditionality

SASSA also has the authority to conduct periodic reviews regarding the medical and financial status

of grant beneficiaries. The decision to review is based upon ‘evidence [existing] that changes in the

medical or financial circumstances [of a beneficiary] may have occurred’ (RSA 2009). The agency

must inform any beneficiary who is up for review three months in advance of the date on which the

review is scheduled to take place. SASSA conducts two basic types of reviews; a medical review

where medical proof of continued disability must be submitted, and an administrative review where

financial documentation and a life certificate (proof that the beneficiary is still alive) is reviewed.

Additionally (and if applicable), beneficiaries who receive their grant payments through a bank

or similar institution are required to fill out a life certificate for their children at SASSA offices every

year (RSA 2014). Although the review process (and anti-fraud measures) specify the circumstances

under which a grant may be terminated, it is important to note that the South African system

does not enforce any conditionalities. This is especially noteworthy with regards to the CSG, as the

norm throughout most of the developing world in relation to child-focused CTs is to place certain

concomitant responsibilities upon beneficiaries in order for them to continue qualifying.

5.3 Social Protection in Brazil

This section shifts the focus towards the Brazilian case. The first step in this endeavour involves

the introduction of the country’s basic social protection framework. The labour regime component

of the Brazilian social protection framework is largely governed through the Consolidação das Leis

do Trabalho (CLT – Consolidation of Labour Laws), as amended (FRB 1943). In addition to the

respective statutory bodies and enforcement mechanisms, social insurance is divided into three

80
pillars, encompassing private sector, public sector as well as complimentary funds (MPS 2009:

13). In turn, social assistance is composed of two overarching programmes. Figure 5.3 provides a

conceptual overview of social protection in Brazil.50

Similar to the case of South Africa, the unemployed and workers in the informal economy are

left vulnerable by a labour protection regime that largely applies to the formal economy, while

eligibility for social insurance benefits depends on previous contributions. This means that, just as

in South Africa, social assistance is the only component of the Brazilian framework which provides

significant protection to people who are not eligible to be supported through formal, contributory

programmes and policies.

50
The official name for the General Social Security fund is Regime Geral de Previdência (RGPS); for the Public

Servants’ Social Security fund it is Regime Próprio de Previdência Social (RPPS); and for the Complimentary Social

Security fund it is Regime de Previdência Complementar (RPC).

81
 SOCIAL PROTECTION IN
BRAZIL

Social Assistance Social Insurance Labour Protection

 Bolsa Família  Statutory Funds  Consolidation of
 Benefício de Prestação  Voluntary Funds Labour Laws (as
Continuada amended), 1943

Statutory Funds Voluntary Funds

 General Social  Complimentary Social
Security Security
 Public Servants’
Social Security

Government Tax Funds from Workers and

Employers

Figure 5.3: Social Protection in Brazil (Source: Adapted from NPC 2011: 329; MPS 2009: 13)

5.3.1 Social Assistance in Brazil

This initial contextualisation now allows for the introduction of the precise technical features of

the Brazilian social assistance system. This involves an examination of systemic coverage, the legal

and institutional framework, the individual programmes comprising the system, the benefit levels

of these programmes, their targeting and eligibility requirements, delivery mechanisms as well as

relevant conditionalities.

82
Coverage

It is highly instructive to map the Brazilian system in a similar way to the South African one in

terms of lifecycle risks, in order to gain a better understanding of the overall approach to coverage.

This is done in figure 5.4 by assessing whether there is a social assistance programme in existence

aimed at protecting against each of these basic risks.51

Birth Time Death

Education Phase (age Accumulation Phase Dependent Phase (age

0-18) Risks: (age 18-65) Risks: 65+) Risks:
Poverty/Education Poverty/Unemployment Poverty/Old Age
Funding Risk: Risk: Risk:

  
Health/Disability Health/Disability Risk: Health/Disability


Risk: Risk:

 

Figure 5.4: Brazilian Lifecycle Risks (Source: Adapted from Smith 2011)

This figure reveals that, in contrast to South Africa, the Brazilian social assistance system does

provide coverage throughout the entire lifecycle. Importantly, this includes a basic social assistance

benefit protecting able-bodied unemployed adults. Indeed, a full 15 percent of PBF beneficiaries

are impoverished adults who do not have any children (Interview 12).

The Plano Brasil Sem Miséria (PBSM – Brazil Without Extreme Poverty Plan) further illumi-
51
A checkmark indicates that there is a programme aimed at addressing a particular risk, while a question mark indi-

cates that there is no programme in existence protecting against a specific risk. In the education phase, the programmes

providing support are the basic, variable 1 and variable 2 Programa Bolsa Famı́lia benefits (poverty/education funding

risk) and the Benefı́cio de Prestação Continuada (health/disability risk). During the accumulation phase the basic

Programa Bolsa Famı́lia benefit is aimed at poverty/unemployment risk, while the Benefı́cio de Prestação Continuada

covers the health/disability risk. Finally, for the dependent phase, the Benefı́cio de Prestação Continuada covers both

the poverty/old age risk and the health/disability risk.

83
nates the the system’s approach through instituting a ‘residual’ benefit52 which is explicitly aimed

at making sure that all social assistance beneficiaries live above the extreme poverty line, while

prioritising the inclusion of all eligible families by mandating that the state actively search for un-

registered potential beneficiaries (FRB 2011b). The universality of the system, as well as the stated

focus on using it to eliminate extreme poverty,53 means that the Brazilian regime has some very

different implications from the South African approach.

The first is that able-bodied adults should not be left solely to rely on selling their labour to earn

an income. The second implication is that no person should be allowed to live in ‘extreme’ poverty.

In fact, Brazil was the first country in the world to adopt legislation towards the implementation

of a Basic Income Grant (BIG), which would guarantee a minimum basic income to every citizen

regardless of need. President Luiz Inácio ‘Lula’ da Silva signed the bill establishing the BIG into

law on 8 January 2004 (FRB 2004a). The law envisions a progressive rollout of the policy, with

the focus initially being on those most in need of income support. Some within the government

in fact regard the PBF to be the first step towards the full-scale implementation of a universal

BIG in Brazil (Suplicy 2008: 4). The fact that coverage is universal across the lifecycle, as well as

the implications flowing from this, therefore constitutes the first important feature of the Brazilian

system.

Legal and Institutional Framework

The analysis now turns towards the legal and institutional framework, with the following set of

laws and regulations governing the overall system: articles 203 and 204 of the Constitution of the

Federative Republic of Brazil (FRB 1988); the Lei Orgânica de Assistência Social (LOAS – Organic

Social Assistance Law), as amended (FRB 1993; 2011); the Regulations as set forth in decree number
52
See table 5.3.
53
There is some controversy over the (inherently political) ways in which the extreme poverty line is determined in

Brazil, with the current level of R$70 having been set in July 2009 (Interview 12).

84
6.214 (FRB 2007); the Polı́tica Nacional de Assistência Social (PNAS – National Social Assistance

Policy) (FRB 2004b); and the ‘Bolsa Famı́lia Law’ number 10.836, as amended (FRB 2004c).

Overall coordination responsibility for the decentralised social assistance system is situated

within the national Ministério do Desenvolvimento Social e Combate à Fome (MDS – Ministry

of Social Development and Hunger Alleviation) through the Sistema Único de Assistência Social

(SUAS – Unified Social Assistance System). However, the ministry has no field offices and only

coordinates and transfers funds to states and municipalities, who are then responsible for disbursing

the funds.

Individual Programmes

This institutional structure governs the individual programmes that constitute the Brazilian system.

Table 5.3 introduces these programmes, and provides information on the target groups and age

brackets, monthly monetary value as well as the means test thresholds for each programme. The

Brazilian system, with more than 54 million total beneficiaries, can be meaningfully divided into a

system of support for older and disabled people on the one hand, and all other households on the

other.

The first social assistance programme to be implemented in post-authoritarian Brazil was the

Benefı́cio de Prestação Continuada (BPC — Continuous Cash Benefit), which was effected in 1995.

It is the direct result of constitutional article 203 (5), as it transfers a monthly cash amount equal

to the minimum wage (currently R$677 or US$294) to disabled individuals of any age and to those

above the age of 65 whose household per capita income is less than one quarter of the minimum

wage (R$169 or US$73) per month. By September 2013, the BPC went to 3.9 million elderly and

disabled beneficiaries, making it a core element of the overall social assistance system (MPS 2013:

11).

The other major component of Brazilian non-contributory income support is the PBF. The

PBF was created in January 2004 through the merger of four smaller pre-existing programmes

85
Programme Target Group Ages Monthly Value

Bolsa Famı́lia (Basic) Households in extreme poverty All R$70 (US$30)

Bolsa Famı́lia (Variable 1) Households in poverty (R$70- 0-16 R$32 (US$14) per child (max

R$140/capita) with children 5)

Bolsa Famı́lia (Variable 2) Households in poverty (R$70- 16-18 R$38 (US$17) per adolescent

R$140/capita) with adolescents (max 2)

Bolsa Famı́lia (Residual) Households still in extreme poverty All Per capita value of residual

(<R$70/capita) after receiving other extreme poverty gap

PBF benefits

Benefı́cio de Prestação Contin- Impoverished and disabled All R$677 (US$294)

uada (Disability)

Benefı́cio de Prestação Contin- Impoverished elderly 65+ R$677 (US$294)

uada (Old Age)

Table 5.3: Social Assistance Programmes in Brazil (Sources: CEF 2014; MPS 2013: 11)

which were all aimed at supporting families earning less than half the minimum wage. The PBF

therefore initially represented a consolidation of the social assistance administration targeted at

those citizens not covered by the BPC (Bither-Terry 2013: 114). It has since undergone continued

expansion and maturation, with a total of 13.8 million families (representing more than 50 million

individuals) currently enrolled (FRB 2013). This makes it the largest CT programme in the world.

The programme consists of a fixed and a variable component. The first is a basic monthly

benefit of R$70 (US$30) paid to households in extreme poverty, defined as a monthly income of

less than R$70 (US$30) per capita (this applies to all households, even those without children or

youths). The second component is a variable benefit, which is paid to poor families with children

who have monthly earnings between R$70.01 (US$30) and R$140 (US$61) per person. For families

with children younger than 16, the benefit amount is R$32 (US$14) per child, up to a maximum

of five children. For households with children aged 16 or 17, the benefit is R$38 (US$17) for a

maximum of two children (CEF 2014).

86
 The residual benefit is aimed at all households that are still in extreme poverty even after

receiving the PBF’s other benefits. This means that if a family’s per capita income is still below R$70

(US$30) per month after receiving the variable PBF benefit, the household will additionally receive

the residual benefit equal to the remaining ‘extreme poverty gap’. For example, if a family’s per

capita income is equal to R$42 (US$18) per person after the PBF’s variable benefit, the household

will receive an additional R$28 (US$12) per capita to bring income levels above the extreme poverty

line (Osório & Ferreira de Souza 2013).

The existence of this benefit highlights the dual nature of the PBF. On the one hand there

is an explicit focus on the PBF’s role as a vehicle for universally eliminating extreme poverty in

Brazil through the unconditional transfer of cash (via the basic and residual benefits) to all families

living in extreme poverty. On the other hand there is an additional emphasis on breaking the

intergenerational cycle of poverty through the conditional variable benefits aimed at children and

adolescents (Fiszbein et al. 2009). The combination of these elements has led some analysts to

regard the PBF as an ‘unconditional CT nestled within a conditional CT’ (Interview 12).

Eligibility

The Brazilian system also incorporates means testing, with a specific set of eligibility requirements

for each programme. The means test threshold for each individual programme is introduced in table

5.4.54
54
Refer to Appendix D for a list of additional requirements.

87
 Grant Name Individual Means Test Threshold

Bolsa Famı́lia (Basic) Up to R$70 (US$30)

Bolsa Famı́lia (Variable 1) Between R$70.01 (US$30) and R$140 (US$61)

Bolsa Famı́lia (Variable 2) Between R$70.01 (US$30) and R$140 (US$61)

Bolsa Famı́lia (Residual) Below R$70 (US$30) even after receiving other Bolsa Famı́lia

benefits

Benefı́cio de Prestação Continuada R$169 (US$73)

(Disability)

Benefı́cio de Prestação Continuada (Old R$169 (US$73)

Age)

Table 5.4: Brazilian Means Test Thresholds (Sources: CEF 2014; FIAP 2011)

Orientation

In contrast to South Africa, the requirements related to eligibility highlight the fact that the orien-

tation of the Brazilian system is familial in nature, because ‘the [fundamental] assistance unit is the

family’ (Lindert et al. 2007: 54). The LOAS law clearly specifies that social assistance is aimed at

supporting those people who do not have the ‘means to provide maintenance of their own or have

it provided by their family’ (FRB 2011a). Indeed, ‘the concept of family was first defined in [article

20] of the LOAS as a “mononuclear unit, living under the same roof, which is maintained by its

members’ contributions”’ (Legido-Quigley 2009: 5). Subsequent amendments have slightly altered

the definition, with it currently reading that the ‘family consists of the applicant, [their] spouse or

partner, parents and, in their absence, a stepmother or stepfather, unmarried siblings, unmarried

children and stepchildren, provided they live under the same roof’ (FRB 2011a).

The result is that eligibility is determined through measuring the per capita income of the

entire household (total household income divided by the total number of family members). This

is distinct from an individualist orientation, where only the individual income of the applicant or

primary caregiver (and his or her spouse, if applicable) is taken into consideration when applying

88
the means test, with total household income and family composition not featuring at all. This

familial orientation is a further important characteristic of the Brazilian system.

Appeals Mechanism

Turning to the issue of controls and appeals in Brazilian social assistance, it is important to note that

qualification for PBF benefits does not automatically guarantee that an applicant will receive them,

as the PBF is technically not a social right (Interview 12). However, an applicant is still allowed to

lodge a complaint or appeal against rejection. The inherent decentralisation of the system means

that municipalities are the primary entities responsible for handling complaints and appeals, and

they must communicate decisions on outcomes within 15 days to appellants (Lindert et al. 2007:

54; 79). It is only in cases where the municipalities are unable to resolve a dispute that the matter is

escalated to the MDS. BPC applicants have 15 days to appeal against a rejection and — unlike the

PBF — an applicant does have the option of going to court in order to have an appeal overturned,

because the BPC is considered to be a constitutional right (Legido-Quigley 2009: 5; Interview 12).

Delivery

The process of delivering the payments for the PBF and BPC takes place through the regular

banking system, with some branches also operating in post offices, lottery agencies and commercial

establishments (Medeiros, Diniz & Squinca 2006: 4). While the procedure for the BPC is managed

by the Instituto Nacional do Seguro Social (INSS – National Institute for Social Security), it follows

the same pattern as the PBF. In the case of the PBF, the Caixa Econômica Federal (CEF – Federal

Savings Bank) credits payments to beneficiaries’ EBCs on a monthly basis. Although the payments

are implemented by the CEF, the process if overseen by the MDS. The ministry firstly validates

the monthly beneficiary list and then sends a financing proposal to the Treasury for approval. The

CEF then generates the monthly payroll and, following approval by the Treasury, issues a bank

order and transfers funds from a Treasury account located at the Central Bank to the CEF’s PBF

89
account. The final step in the process then involves the crediting of beneficiary accounts (Lindert

et al. 2007: 51-52).

Beneficiaries can withdraw this money from any of the 2 000 CEF branches and 30 000 other

participating locations throughout the country. Fully 97.8 percent of beneficiaries use the EBC

to withdraw their monthly benefits, with 65 percent preferring lottery points, 30 percent using

automated teller machines55 and 5 percent making withdrawals from stores or supermarkets. The

remaining 2.1 percent who do not use EBCs withdraw their benefits from CEF agencies. Beneficia-

ries have 90 days to withdraw the funds, after which time it is returned by the CEF to the original

ministerial account (Lindert et al. 2007: 52-53).

Conditionality

While beneficiaries are re-evaluated every two years, the BPC is an unconditional programme. In

fact, it can be regarded as more than that, as its constitutional position enshrines it as a universal

citizenship right that cannot be denied and ‘any individual who meets the eligibility criteria can

receive the benefit and might demand it judicially’ (Medeiros, Britto & Soares 2008: 2). This is in

contrast to the child-focused components of the PBF, where conditionality is deeply enshrined as

part of the programme’s operation. Here the Brazilian system stands in sharp contrast to its South

African counterpart (the CSG), as the PBF endows beneficiaries with a set of responsibilities which

they must fulfil in order to continue receiving the benefit payments.

These include an 85 percent school attendance requirement for children between six and 15

years old and 75 percent for 16 and 17 year olds; prenatal monitoring of pregnant and lactating

mothers between 14 and 44; immunisation and growth monitoring for children between zero and

seven; and an 85 percent attendance rate of socio-educational services by children and adolescents

up to 15 years of age who are at risk or have been liberated from child labour (Jaccoud, Hadjab

& Chaibub 2010: 12). Failure to adhere to these conditions initially results in an intervention by
55
Beneficiaries are however not required to have a bank account.

90
a social worker, temporary suspension of the benefit and ultimately, following a third instance of

non-compliance, permanent cancellation. This therefore highlights a further important feature of

the Brazilian social assistance system: the incorporation of conditionalities in programmes aimed

at supporting children and adolescents.

5.4 Comparing Social Assistance Systems

It is clear from this technical discussion that a number of similar fundamental features emerge across

the cases. However, these similarities are all related to the basic composition of social assistance as

a distinct category within social protection. The fact that both systems are non-contributory CTs

where eligibility is determined on the basis of need only tells us that they are in fact both social

assistance regimes, and does not tell us anything about different policy choices or dissimilar overall

approaches to social welfare.

The previous sections therefore set the scene for a more abstract extrapolation of the different

approaches taken by South Africa and Brazil. This section will undertake such an extrapolation

by directly contrasting the relevant features. This involves an assessment of possible differences

regarding the overall benefit values of the systems, their overall coverage, conditionality and systemic

orientation. The analysis indicates that there are no significant differences in terms of overall value,

while vitally important differences emerge with regards to coverage, conditionality and orientation.

Comparing Value

Perhaps the most intuitive area to commence with a direct comparison of the two systems is assessing

their overall value. Significant differences in the amount of money that actually ends up in the

hands of beneficiaries would clearly constitute an important feature in need of explication. The

same applies to significant differences in targeting through means test thresholds. However, an

analysis based upon the available data suggests that the overall benefit and targeting levels of the

91
South African and Brazilian systems are quite similar.56

A headline figure which is often used to assess and compare spending levels across countries

is total social assistance expenditures as a percentage of GDP. In 2008, this figure (calculated on

the basis of spending reported by individual countries to the World Bank) stood at 1.4 percent for

Brazil, while the corresponding number for South Africa was 3.2 percent (Weigand & Grosh 2008:

7). While this seemingly indicates a notable difference between the two countries, there are myriad

problems with simply using this all-encompassing figure as the basis for analysis.

This includes the fact that there are no comparable data available on what fraction of these

figures are purely administrative and other spending. Some countries may include social service

provision under social assistance spending, while others do not. Another obvious problem is the

fact that different countries (especially ones with different institutional structures) can have vastly

different reporting frameworks and definitions. In sum, concentrating only on this expenditure

would be misleading, as ‘not all spending counts equally’ and ‘expenditures are epiphenomenal to

the theoretical substance of welfare states’ (Esping-Andersen 1990: 19). The result is that this

simple figure does not tell us much about the strictness of targeting or the amount of money that

beneficiaries actually receive.

The only way to determine whether there are significant differences between benefit and targeting

levels is therefore to directly compare the actual value of social assistance benefits, as well as the

relative severity of the means tests. This is achieved through the construction of a weighted value

measure which takes into account both the benefit amounts paid to beneficiaries as well as the

relevant means test thresholds, to enable a viable side-by-side comparison of the cases. The first

step in compiling this measurement involves calculating the composite benefit values paid out to

beneficiaries. Figure 5.5 indicates that the weighted average monthly benefit value paid out per

beneficiary is US$39.50 in Brazil and US$57.98 in South Africa.57

56
See figure 5.7.
57
Calculating these figures involved weighting each individual social assistance programme through multiplying
the number of beneficiaries that receive it by its benefit value. This procedure was repeated for each programme,

92
 80

Weighted Average Benefit Level (US$)

70

60

50

40

30

20

10

0
Brazil South Africa

Figure 5.5: Composite Average Benefit Level

A higher level in figure 5.5 indicates higher payout amounts, with composite benefit levels there-

fore being slightly higher in South Africa. But this is only one element in the overall measurement

of programme value. The second feature to take into account is the strictness of the respective

means tests. A direct comparison of the rigidity of means test thresholds is introduced in figure 5.6;

it indicates that the composite average means test threshold in Brazil is US$57.50 and US$55.71 in

South Africa.58
producing a final figure encompassing the total expenditure on social assistance benefit payouts every month by each

country. The corresponding amount for Brazil is US$21 292 560 54 and US$907 078 561 for South Africa. These

totals were then divided by the total number of beneficiaries paid each month (53 907 804 in Brazil and 15 644 273

in South Africa) to produce a weighted average of U$39.50 received per beneficiary in Brazil and US$57.98 in South

Africa.
58
Calculating these composite averages involved a slightly more complicated procedure than the one that was

required to produce figure 5.5. The complication stems from the different orientations inherent to the systems.

Whereas Brazil’s familial orientation measures eligibility in terms of total monthly per capita household income,

South Africa’s individualist approach only assesses the income of individual beneficiaries without regards to household

structure. The only workable solution in enabling a meaningful comparison thus involves firstly converting the South
African means test into one similar in shape to its Brazilian counterpart. This is achieved by firstly dividing the means

93
 80

Weighted Average Means Test Threshold

70

60

(US$) 50

40

30

20

10

0
Brazil South Africa

Figure 5.6: Composite Means Test Threshold

As with figure 5.5, a higher level in figure 5.6 indicates a more generous means test threshold.

The result is that in this case Brazil has a slightly more generous approach. However, the only way

to draw any concrete conclusions from this comparison is by constructing a composite measure of

overall value by combining both benefit levels and means test thresholds into one weighted average

test thresholds indicated in table 5.2 by 12 to produce a ‘monthly means test’ figure for South Africa. The next step

entails a further division of this figure based upon the average household sizes of social grant beneficiaries. Research

into the household composition of social grant recipients in South Africa indicates that the average household size

for beneficiaries of the CSG is 6.4, while the corresponding figure for OPG beneficiaries is 5 (DSD et al. 2012: 15;

Ambler 2011: 54). These are the household sizes that were applied to the respective grants in the conversion of the

South African means tests. Once this conversion was achieved to enable direct comparison, the same basic procedure

as the one employed in figure 5.5 was followed. This involved weighting each individual social assistance programme

by multiplying the number of beneficiaries that receive it by its means test threshold. This procedure was repeated

for each programme, producing a final figure encompassing the combined value of monthly means test levels in each

country. These totals were subsequently divided by the total number of beneficiaries paid each month (53 907 804 in

Brazil and 15 644 273 in South Africa) to produce a composite average means test threshold of U$57.50 per beneficiary

in Brazil and US$55.71 in South Africa.

94
value measurement. This measure is introduced in figure 5.7, which indicates that the composite

average value of the Brazilian system is US$48.50 per beneficiary, while the corresponding figure

is US$56.85 for South Africa.59 This analysis of the benefit levels and means test thresholds of all

80

70
Weighted Average Value (US$)

60

50

40

30

20

10

0
Brazil South Africa

Figure 5.7: Composite Value

social assistance programmes in South Africa and Brazil thus indicates that the overall values are

similar. The only conclusion to be drawn from this is that there is fundamentally no real variation

in the overall benefit values of the two systems.

Comparing Coverage

The first important feature where significant contrasts do emerge is in relation to the issue of which

people are actually covered by the social assistance systems in both countries. The mapping of basic

lifecycle risks indicates that both countries have provisions to support children, disabled people of

all ages, as well as the elderly. An important contrast however emerges regarding the issue of

impoverished able-bodied adults. Whereas Brazil provides some support to such families who live
59
This consolidated measurement of overall generosity is created by adding the weighted average benefit value (figure

5.5) and the weighted average means test threshold (figure 5.6) together, and dividing the resulting number by two.

95
in extreme poverty through the PBF’s basic and residual benefits, South Africa’s system offers no

such protection.

This leads to some very different implications: while the South African approach implies that

able-bodied adults should solely rely on selling their labour to earn an income, its Brazilian coun-

terpart implies that those people should not solely rely on earning an income through the labour

market; and while the Brazilian approach conceptually emphasises that no person who qualifies for

social assistance should be allowed to live in extreme poverty, the South African system accepts

that at least some people should be allowed to live in extreme poverty. This is very significant, as it

means that between 4.6 million and 17.7 million impoverished people are left unprotected in South

Africa (Stats SA 2013: XI). This distinction alludes to fundamentally dissimilar policy choices in

each country, constituting the first major feature in need of explanation: why is it that the social

assistance system provides universal60 coverage in Brazil, but only limited coverage in South Africa?

Comparing Conditionality

Another important element of social assistance policy design relates to conditionality. South Africa

— dissimilarly from the approach taken by a number of its peers — does not implement any social

assistance conditionalities. And while conditionalities do not feature in relation to programmes

aimed at the elderly and disabled in Brazil, they form an integral part of the PBF’s variable benefits

aimed at supporting children and adolescents. As mentioned above, these conditions include that

children between six and 15 years of age must achieve an 85 percent school attendance rate; 16
60
The characterisation of a welfare system as ‘universal’ is traditionally employed in debates to indicate a system

where everyone, regardless of income level, receives regular income support. It is usually contrasted to a ‘targeted’

approach where means testing is used to identify the ‘deserving poor’. This is not the context of its usage in this

study. Instead, a social assistance system is characterised here as ‘universal’ if some support is available across the

entire lifecycle, regardless of the fact that such support may be means tested (as is the case in Brazil). This stands

in contrast to a ‘limited’ approach, where certain segments of the poor are entirely excluded from even means tested

income support (as is the case with unemployed able-bodied adults in South Africa).

96
and 17 year olds must achieve a 75 percent attendance rate; prenatal monitoring of pregnant and

lactating mothers between 14 and 44; immunisation and growth monitoring for children between

zero and seven; and an 85 percent attendance rate of socio-educational services by children and

adolescents up to 15 years of age who are at risk or have been liberated from child labour (Jaccoud

et al. 2010: 12).

This marks a vital point of distinction between the two approaches, as the literature overwhelm-

ingly treats this incorporation of conditionalities as a transformative difference distinguishing a

straightforward ‘income transfer’ from a ‘transfer plus interventions aimed at human. . . capital ac-

cumulation’ (Barrientos, Niño-Zarazúa & Maitrot 2010: 7). Others argue that the incorporation

of conditionalities means that, in contrast to unconditional programmes, conditional CTs address

both ‘present and future poverty’ through attempting to additionally break the ‘intergenerational

cycle of poverty’ by ‘encouraging investment in human capital’ (Das, Do & Özler 2005; Lindert et

al. 2007; Fiszbein et al. 2009; Santos, Paes-Sousa, Miazagi, Silva, & Da Fonceca 2011). The scope

of these claims means that this constitutes another significant point of differentiation between the

two policy regimes.

Comparing Orientation

A final interesting area of contrast emerges with regards to systemic orientation. This involves the

ways in which the societal position of beneficiaries are defined. Are beneficiaries defined based on

their positions within a specific household or family unit, or are they strictly defined as individuals

regardless of household context? In practical policy terms, the implication of this fundamental

distinction is between providing family-based income support and individual-based support (Lund

2008: 53).

While ‘the five major social assistance policies in [South Africa] focus only on specific individ-

uals’, Brazil’s emphasis on families is made explicit even by its incorporation of the word ‘family’

into the title of the PBF (RSA 2012: 7). Through the preceding examination, it becomes clear that

97
South Africa has adopted an individualist approach, while Brazil follows a familial orientation. The

issue of systemic orientation therefore constitutes another significant differentiating feature between

the cases.

5.5 Dependent Variables and Social Assistance Typologies

This section synthesises the insights generated throughout the preceding parts of the chapter by

identifying and consolidating cross-case differences as the study’s dependent variables. It essentially

categorises the variation across the cases in terms of coverage, conditionality and systemic orien-

tation into distinct typologies which serve as the study’s dependent variables. The construction of

these typologies is informed by the existing literature, resulting in the South African case (as out-

lined in section 3.2.1) being regarded as a protective social assistance system, while Brazil (section

3.3.1) is classified as a productive system.

5.5.1 Dependent Variables

Addressing the descriptive research objective thus leads to the multi-dimensional identification of a

number of contrasting features regarding the social assistance systems of South Africa and Brazil.

Variation in terms of these features — coverage, orientation and conditionality — are consolidated

into the policy regime typologies which constitute the study’s dependent variables, as indicated in

table 5.5. The aim of subsequent chapters is thus to account for the emergence of the variations in

these features through examining the development of the social assistance systems in South Africa

and Brazil.

Variable South Africa Brazil

Policy Regime Productive Protective

Table 5.5: Dependent Variables

98
5.5.2 Typologies of Social Assistance

Crucially, these differences are not merely isolated contrasts nestled within contexts of otherwise

overwhelming similarity. Instead, variation on such profoundly important aspects as systemic cov-

erage, conditionality and orientation are unambiguous indicators that these cases constitute two

different ideal-types of developing country social assistance provision. This section therefore con-

solidates the variations in terms of the dependent variables into coherent classification typologies in

an attempt to make analytic sense of this significant cross-case variation and enable a subsequent

analysis of the development processes that led to their emergence.

The ideal-types introduced in chapter three provide the basis for classifying the two cases.

Rudra’s (2007; 2008) work forms the foundation of this endeavour, and it is worth recalling that

her analysis identifies the existence of two ideal-types of welfare states in the developing world.

The first, the productive variant, features efforts ‘primarily directed towards promoting the mar-

ket dependence of citizens’ and shares ‘certain elements with [Esping-Andersen’s] liberal model...by

embracing...enthusiasm for the market and self-reliance’ (Rudra 2007: 379). The particular prop-

erty of the liberal paradigm which ultimately distinguishes this productive typology is the way in

which it serves to strengthen ‘the commodity status of labour’ (Rudra 2007: 379). The second

variant is the productive approach, which amounts to a ‘curious fusion of elements of socialism and

conservatism...[featuring] a strong distrust of markets’ (Rudra 2007: 384).

By applying this typological framework to the cross-case variation in terms of coverage, condi-

tionality and orientation, it becomes clear that the South African social assistance system best fits

the productive classification, while Brazil reflects the logic of a protective system.61 An examination
61
It is important to note that this classification not only involves borrowing terminology from Rudra (2007; 2008),

but also entails a modification of the features used to classify a system as productive or protective. This is because

her focus is upon the overall distributional regimes, while the focus of this study is only upon one aspect of the

distributional regimes, namely the social assistance element. The different scope involved thus means that, while the

logic and titles of the typologies are preserved, they are strictly applied only to the social assistance aspects of each

99
of the system’s features reveals that the South African productive approach certainly contains (in

a different overall context) features of the original ‘liberal’ category, wherein

the progress of social reform has been severely circumscribed by traditional, liberal work-ethic

norms: it is one where the limits of welfare equal the marginal propensity to opt for welfare

instead of work, [and where] the state encourages the market (Esping-Andersen 1990: 26).

The classification of the South African system as productive is based upon the fact that it as-

sumes that people should rely solely on selling their labour to earn a minimum income and that at

least some people should be allowed to live in extreme poverty through providing only limited cov-

erage. In a reflection of eminently liberal ‘work-ethic norms’, South Africa in fact actively ‘pushes’

impoverished people out of the social assistance system and into the realm of the labour market

and its associated policies through highly limited public works schemes such as the EPWP, which

supposedly provides them with the ‘dignity of work’ (Seekings 2008: 33). South Africa has further-

more adopted a more traditionally liberal approach in not prescribing behavioural conditionalities

and strictly defining beneficiaries as individuals without regards to family structure, as well as by

‘switching from biological parent to practical carer’62 in terms of its ‘follow the child’ policy (Lund

2008: 64).

In contrast to this, Brazil’s system is classified as protective. Consistent with the above defi-

nition, it certainly involves a ‘curious fusion of elements of socialism and conservatism’. It can be

convincingly argued that by not assuming that people should solely rely on selling their labour to

earn a minimum income and emphasising that no person should be allowed to live in extreme poverty

through providing universal coverage (as well as at least legally adopting a BIG), the Brazilian sys-

tem reflects a certain ‘distrust of markets’. An interesting conservative element also emerges in the

Brazilian case, as the state both actively intervenes in the choices made by beneficiaries through
country’s overall distributional regime.
62
Lund explicitly expresses concern about the reaction from conservatives to these liberal policies, who may regard

them as undermining family life (2009: 64).

100
its conditional approach and defines beneficiaries in terms of their family structure. Both of these

features reflect Esping-Andersen’s original ‘conservative’ model,63 which is ‘strongly committed to

the preservation of traditional family-hood’ (Esping-Andersen 1990: 27). Brazil’s system is there-

fore indeed an interesting mixture of social democratic and conservative essentials, in accordance

with the protective model.

Esping-Andersen argues that

the outstanding criterion for social rights must be the degree to which they permit people to

make their living standards independent of pure market forces. It is in this sense that social

rights diminish citizen’s status as ‘commodities’ (1990: 3).

The analysis suggests that the South African approach certainly permits people to ‘make their living

standards independent of pure market forces’ to a lesser extent 64 than in Brazil. This compara-

tively lesser decommodification in South Africa is complemented by liberal policies with regards to

autonomous decision making and defining beneficiaries as individuals. This enables classification of

the South African social assistance regime as productive. In contrast, the social democratic element

in Brazilian policy is complemented by a conservative streak concerned with modifying behaviour

through conditionalities and defining family structures, thus allowing it to be classified as a pro-

tective developing country social assistance regime. This synthesis of the different policy outputs

regarding coverage, conditionality and orientation is summarised in table 5.6.

This analysis thus serves to integrate the variations which were inductively identified throughout

previous sections of this chapter into distinct social assistance typologies, which in turn constitute

the study’s dependent variables. This inductive approach to classifying the dependent variables

enables a focus in subsequent chapters on addressing the explanatory research question: what
63
Although clearly in a very different context from the developed welfare states he examined.
64
There are certainly no developing countries that achieve levels of decommodification comparable to those achieved

by some OECD nations, meaning that there are no developing countries where people can truly ‘make their living

standards independent of pure market forces’. However, some are comparatively more decommodifying than others.

101
 Feature Productive (South Africa) Protective (Brazil)

Coverage Limited Universal

Conditionality Unconditional Conditional

Orientation Individualist Familial

Table 5.6: Typologies of Social Assistance Policy Regimes

accounts for these differences between the South African and Brazilian social assistance systems?

5.6 Conclusion

The focus in this chapter has been on addressing the study’s descriptive objective concerning the

qualitative features of, and differences between, the South African and Brazilian systems of social

assistance. The process involved an initial contextual overview of social protection in South Africa

and Brazil, followed by a subsequent in-depth discussion on the precise features of social assistance

in each country. This discussion included a focus upon a range of fundamental elements, including

an examination of coverage, benefit levels, the legal and institutional frameworks governing each

system, the individual programmes that constitute the two approaches, targeting and eligibility

requirements, the delivery of benefit payments and possible conditionalities.

The final sections of the chapter moved away from technicalities towards an abstraction of the

cases by specifying divergence in terms of coverage, conditionality and systemic orientation. These

differences were consequently incorporated into the construction of distinct typologies, with the

South African social assistance system being classified as productive, and the Brazilian system as

protective. The empirical fact of these qualitatively distinct policy typologies constitute the study’s

dependent variables. By undertaking this descriptive overview, the chapter has thus contributed

towards the identification of important variations across the cases. The focus in subsequent chapters

shifts towards explaining the emergence of these different typologies and their associated features

through addressing the explanatory research question.

102
Chapter 6

The Development of Social Assistance

in South Africa

6.1 Introduction

This chapter addresses the explanatory research question for the South African case. As such, it

aims to account for the emergence of a productive social assistance system in South Africa through

undertaking an in-depth process tracing examination of policy development. This examination is

carried out inductively, guided by the study’s analytical framework of ACI. Figure 6.1 describes the

ACI analytical procedure for the South African case. The application of this approach initially en-

tails a detailed examination of the institutional context according to the two independent variables.

This is followed by an examination of the ways in which the resulting institutional configuration

channelled interactions among policy actors into concrete policy outputs. Figure 6.2 introduces the

overall explanatory framework.

The chapter commences by addressing the first independent variable through analysing the his-

torical emergence and institutionalisation of social assistance norms in South Africa. It examines

both the historical process that initially led to the entrenchment of a productive approach to social

103
 Institutional Setting
(1) Norms deeply Institutionalised
(2) Concentrated State Structure

Problems Actors Constellation Modes of Interaction Productive

- Preferences - Hierarchical System
- Capabilities - National State Actors
Direction
- Civil Society Actors

Policy Environment

Figure 6.1: Actor-Centred Institutionalism – South Africa (Source: Adapted from Scharpf 1997: 44)

IV1:
Institutionalisation of
Social Assistance
(High)
Constitutional- Preferences of Interactions Policy
Legal Policy Actors between Actor Sets Outputs
Framework (Productive
IV2: Concentration of System)
State Structure (High)

Figure 6.2: Explanatory Framework – South Africa

assistance, as well as the way in which this approach and its associated norms were subsequently

constitutionally transferred into the political economy of the post-apartheid state. It is essentially

an eight decade story of path dependence which saw the survival and entrenchment of productive

social assistance norms codified during the transition to democracy in the early 1990s. This is

followed by an examination of state structure as the second independent institutional variable. The

discussion focuses on both the formal institutional construction of the state and the distribution

of political power during the period under examination. It concludes that South Africa featured a

104
highly concentrated social assistance policymaking structure, with the capacity to formulate and

implement policy having been entirely centralised. Analysing these two independent variables there-

fore effectively maps the South African institutional context.

The subsequent section proceeds to examine the way in which interactions between actors from

the policy community, embedded within the previously specified institutional context, effected the

distinctive policy outputs constituting the contemporary South African social assistance system. In

accordance with the ACI analytical framework, this is achieved through specifying actor preferences

and analysing the way in which interactions between different actor sets determined the shape of

the country’s social assistance policies. The chapter thus demonstrates that an examination of

institutional structure is able to account for the fact that post-apartheid South Africa features a

productive social assistance system.

6.2 Deep Institutionalisation of Social Assistance Norms

This section initially conducts an overview of the emergence of distinct social assistance norms in

South Africa following unification in 1910. The analysis shows that by the time of the transition to

democracy in the 1990s, institutional norms regarding limited coverage, an unconditional approach

and an individualist orientation had become deeply entrenched in the South African political econ-

omy. This is followed by an examination of the extent to which these norms were subsequently

legally codified and institutionalised through the current Constitution and other pieces of flagship

social welfare legislation. The section thus addresses the first independent variable by analysing the

historical emergence and institutionalisation of distinct social assistance norms in South Africa.

6.2.1 Social Assistance before 1994

By the time of South Africa’s transition to democracy in the 1990s, social assistance policies had

become deeply institutionalised in the country’s overall political economy (Kruger 1992; Duncan

105
1993; Van der Berg 1997; Seekings 2000; 2006; 2007). In fact, already as early as the 1930s, ‘South

Africa had developed a welfare state that was remarkable in terms of both the range of risks against

which it provided and its coverage of the poor’ (Seekings 2006: 1). Even more remarkable were

the specific features of this welfare state, as ‘unusually in the world, and exceptionally in the global

South, South Africa developed in the course of the twentieth century a public system based primarily

on social assistance rather than social insurance’ (Seekings 2007: 376).

An overview of social assistance since the unification of South Africa can be meaningfully di-

vided into three main epochs: an initial period seeing the introduction and elaboration of the first

programmes: the OPG,65 DG and State Maintenance Grant (SMG); followed by the introduction

of apartheid and the associated reracialisation and retrenchment of social assistance programmes;

as well as a period of later apartheid featuring systemic expansion and reduced inequality. This

subsection carries out a review of the pre-1994 period in order to demonstrate the extent to which

certain features became deeply entrenched in the country’s approach to social assistance. These

features subsequently played a crucial role in shaping the institutional context and preferences held

by policymakers during the contemporary period.

First Steps and Consolidation

The scarcity of historical examinations into the provision of social assistance (and social security

more broadly) in pre-colonial Africa is a serious shortcoming which hampers attempts at analysing

their potential path dependence effects. While the absence of formal institutions in pre-colonial

Africa certainly does not preclude the presence of strong, informal institutions catering for impov-

erished people, it does make it very difficult ‘to trace the development of [social assistance] before the

arrival of formal institutions from Europe and has made it necessary to start in the. . . Eurocentric

way with the first European settlement at the Cape in 1652’ (Kruger 1992: 108).
65
For much of its life, this grant was known as the Old Age Pension. The name was changed to Older Persons

Grant after 1994.

106
 The early period of European settlement66 during the 17th and 18th centuries featured very little

in the way of even rudimentary ‘poor relief’67 and social services, with provision being dominated by

the Nederduits Gereformeerde Kerk (NGK – Dutch Reformed Church), and the state only featuring

in a minor piecemeal capacity (Van der Berg 1997: 485). What little role there was for the state

was dominated by extremely modest and stigmatising poor relief along the British model, including

both outdoor relief (grants in cash or in kind) which was often also administered by churches even

when funded by the state, and indoor relief in the shape of almshouses (Seekings 2007: 380).

Additionally, the common law made children directly responsible for maintaining aged and indigent

parents (Seekings 2007: 379). This situation largely persisted for most of the first two decades

following the formation of the Union of South Africa in 1910, as ‘poor relief [remained] the most

important state programme in the social welfare field until about 1928’ (Kruger 1992: 157).

In 1913 the Union government introduced the Children’s Protection Act (USA 1913). While

the Act ‘gave the powers to refer children to institutions and appropriate care’, it did so without

making provision for the financial implications of such referral (Kruger 1992: 161). Such provisions

only emerged in 1921 through the Children’s Protection Amendment Act, which authorised and

introduced the payment of the SMG in South Africa (USA 1921; Kruger 1992: 162; Haarmann

1998: 82). The SMG was payable to mothers, stepmothers and grandmothers who were taking care

of a child who had been committed into their care under the Act, if she was a widow, or if the child’s

father was unable to care for the child due to reasons beyond his control (Kruger 1992: 162). The

SMG can therefore be regarded as the first modern social assistance programme in South Africa,

and its aim was largely to ‘protect white family life’ (Lund 2008: 15).
66
By definition, this entire review of the pre-democratic era focuses on the role of a state where only whites (initially

only men) were enfranchised with political and voting rights, while all other groups were practically excluded from

the political process.

67
This term has become almost synonymous with paternalistic, stigmatising and generally inadequate patterns of

social welfare provision, taking its name from the 1662 enactment of the ‘Poor Relief Act’ by the British Cavalier

Parliament (Spicker 1984).

107
 In a sign of things to come, the grant was only paid in support of black children under exceptional

circumstances. Black people living in rural areas were entirely excluded, while the law stipulated

that all other avenues, such as the ‘repatriation’ of children to relatives in the rural areas or the

provision of day-care services, had to be exhausted before an application for an SMG could be

considered in respect of an urban black child (Kruger 1992: 162). The introduction of the SMG was

followed in 1924 by the proposals contained in the Collie Memorandum. This document featured the

results of an investigation into old age pension schemes and signalled a shift towards the acceptance

of state responsibility in the provision of welfare. The acceptance of this situation was deemed

necessary by the government primarily due to the extent of the problem of ‘poor whites’,68 as well

as the fear of the so-called swart gevaar (black danger) (Seekings 2007).

The early 1920s was a period of intense social, class and racial conflict in South Africa that saw

the initial emergence of poor whites: high levels of unemployment and poverty among white people,

a situation which affected about 10 percent of the white population (USA 1922; 1925; 1929; Berger

1983; Davie 2005). The period also featured a relaxation of the strict racial hierarchy69 through a

weakening of the ‘colour bar’70 and the associated practice of job reservation, due in large part to

the demand for black labourers on the gold mines of the Witwatersrand (Seekings 2007). The result

was that some rural white people had become impoverished, while some urban blacks had become
68
The pre-apartheid, apartheid as well as post-apartheid period sees the classification of the South African popu-

lation into different racial groupings, and the use of these terms is relevant to the subject of this study. The terms

‘black’, ‘coloured’ (people of mixed race and/or descendents of former Malay slaves), ‘Indian/Asian’ and ‘white’ are

therefore used throughout in this context. The terminology has been standardised where other authors have used

different terminology to refer to these population groups. Poor whites were ‘far from being the poorest people in

South Africa, but they were poor in the sense that they were poorer than growing numbers of [non-whites]’ (Seekings

2007: 381).
69
This racial hierarchy was effectively defined as 1) whites, 2) coloureds, 3) Indians and 4) blacks.
70
The colour bar was a systematic definition of racial hierarchy mainly expressed through the practise of job

reservation. It principally emerged from the 1911 Mines and Works Act, where skilled jobs were reserved for whites

and, in some cases, coloureds (USA 1911).

108
richer (in spite of pervasive economic and social restrictions).

In turn, this led to protests from white workers in favour of stricter job reservations, with

tensions being further exacerbated through a sharp drop in the gold price in late 1921 (Krikler

2005). The situation finally exploded during the insurrectionary Rand Revolt of 1922, where white

workers attacked blacks but were in turn the victims of state repression (Krikler 2005). The Collie

investigation into old age pensions (and subsequent legislation) can therefore be regarded as having

essentially formed part of the government’s response to the crisis and the problem of poor whites, as

the elderly and invalid comprised ‘the most pressing category of “deserving poor”’ (Seekings 2007:

384).

The Collie Memorandum explicitly made the conceptual distinction between contributory and

non-contributory models of old age pension provision, setting out eight possible non-contributory

schemes for South Africa. While men and women would receive equal benefits and all of the

proposals covered non-white groups, they also discriminated in terms of benefit levels (Seekings

2007: 384). The government did not immediately act upon these recommendations. But further

investigations by a departmental committee examining unemployment and poor relief, as well as by

the Economic and Wage Commission, continued to argue that prevailing welfare provision for the

(white) elderly and disabled was inadequate (USA 1924: 201; 1925: 112; 339).

This emerging consensus around the need for greater welfare measures led to the formation of

the Commission on Old Age Pensions and Social Insurance in February 1926. It was chaired by

senior National Party (NP) Member of Parliament (MP) Ben Pienaar and came to be known simply

as the Pienaar Commission. Its mandate featured a call to investigate ‘the payment of pensions by

the state to necessitous aged and permanently incapacitated persons who are unable to maintain

themselves and for whom no provision at present exists’ (USA 1927). The final recommendations

of the Commission featured two significant differences from all eight of Collie’s suggested schemes:

it made no provision for blacks, while the pension benefit level was set at a much lower level that

109
amounted to only £26 per year — lower than the wages paid to many black labourers (Seekings

2007: 386). The programme would also be means tested, and the Commission estimated that old

age pensions would be paid to 15 500 whites and 14 000 coloureds and Indians (Seekings 2007: 386).

In July 1928 the government officially enacted the Old Age Pension Act. It generally accepted

the recommendations of the Pienaar Commission, setting benefit levels for whites at £30 per year,

which was slightly higher than the £26 recommended by the Commission. Coloured pensioners

would receive £18 per year, while blacks and Indians were excluded. The programme was to be

means tested, with the age of eligibility set at 65 for both men and women, who were to receive

equal benefit amounts (Seekings 2007: 387). Very significantly, the law did not take into account

familial circumstances. If an applicant’s child ‘was in a position to maintain the applicant, then

the Commissioner of Pensions might recover from the child part or all of the pensions paid to the

parent, but this would not preclude the initial payment of a pension (emphasis added)’ (Seekings

2006: 6). The South African OPG, which would go on to become a deeply institutionalised feature

of social assistance in the country, was subsequently introduced on 1 January 1929.

The purposive enactment of a non-contributory programme in the face of the government’s

knowledge about contributory alternatives meant that ‘the [social] insurance option was, quite ex-

plicitly, rejected in South Africa with respect to old age pensions’ (Seekings 2007: 387). There were

essentially two reasons for this fateful choice of social assistance over social insurance: the perceived

urgency of assisting poor elderly white people and the difficulty entailed in establishing insurance

schemes that were racially discriminatory (Seekings 2007: 388). A social insurance approach would

have meant that the country would need to wait as beneficiaries first accumulated (contributory)

benefits, which would have caused a significant delay in implementation while impoverished elderly

(white) people continued to suffer. It would also have been significantly more expensive to exclude

black workers from a contributory scheme, while simultaneously raising the cost of employing white

workers.

110
 These problems did not arise under a social assistance rubric, as ‘non-contributory pensions made

it easier to establish a clear and unambiguous racial hierarchy’ (Seekings 2007: 388). As a result, the

implementation of a social assistance-based OPG partly reflected the urgency of addressing white

poverty in order to ‘raise all white people to “civilised” standards of living’ (Seekings 2007: 378).

This entailed a clear response to the voting power of whites, who were the only enfranchised group

in the country. But it also revealed ‘the urgency of re-establishing a racial hierarchy [in response to

fears of the swart gevaar ] by raising the incomes of white (and coloured) people whilst excluding

black (and Indian) people’ (Seekings 2007: 378).

The next significant step in the institutionalisation of social assistance in South Africa was the

set of five reports produced in 1932 by the non-governmental Carnegie Commission of Inquiry into

The Poor White Problem in South Africa. The Commission was funded by the Carnegie Corporation

of New York, and was composed of academics, civil society groups and representatives of the NGK.

Their work built upon prior inquiries into the poor white problem and focused renewed attention

on it,71 inaugurating a period of accelerated state social spending which featured the expansion of

existing social assistance programmes and the introduction of new ones (Seekings 2006: 19).

The expansion of the social assistance system was further propelled by the realities of the Great
71
Seekings challenges this popular interpretation by pointing out that many of the Carnegie Commission’s proposals

amounted to ‘a deeply conservative reaction against the modernising impulses represented by the Pienaar Commission

and the Old Age Pension Act’ (2006: 12). As such, he regards it as an ‘emphatic backlash against the existing

programmatic, state-based responses to poverty among white South Africans’ (Seekings 2006: 6). Whereas the

Pienaar and Economic and Wage Commission investigations of the 1920s interpreted the problem in structural terms,

the Carnegie Commission’s analysis was essentially psychological, as it stated that the redress of poverty would require

‘eliminating vice and ignorance’ among poor whites (Seekings 2006: 7). This approach was opposed by a civil society

coalition led by Hendrik Verwoerd and the Afrikaanse Christelike Vrouevereniging (ACVV – Afrikaans Christian

Women’s Organisation). Seekings further notes that, by the late-1930s, the head of the Department of Social Welfare

‘became convinced that the ACVV and Verwoerdian proposals were better than those of the Carnegie Commission

and NGK’ (Kruger 1992: 167; Seekings 2006: 25). The result was the consolidation of the emerging programmatic

system.

111
Depression, a significant improvement in public finances and a change of government in 1933. The

onset of the Great Depression fuelled the growth of poverty among ‘morally upstanding’ and ‘hard-

working’ families, which had the effect of undermining arguments about their ‘moral inferiority’

as causes of poverty. This effect was further enhanced through the concerns it generated in South

Africa about the security of white supremacy and national decline (Klausen 2004: 23). Somewhat

paradoxically, the improvement in public finances during this period was greatly facilitated by the

rise in the gold price following the collapse of the gold standard (Seekings 2006: 27).

These developments were finally consolidated by the incoming ‘fusion government’72 in 1933,

which saw the responsibility for social welfare designated to Jan Hofmeyr, who would go on to play a

decisive role in further systemic expansion (Seekings 2006: 30). The effects of the Great Depression

therefore generated a consensus on the need for a programmatic approach to social assistance, while

the government had the surplus money to spend as a result of the rise in tax revenues generated by

spiking gold prices. When combined with the incoming government Ministers’ deeper commitment

to programmatic reform, it becomes clear that this was the perfect political recipe for the expansion

of the fledgling social assistance system (Seekings 2006: 27).

This resulted in the introduction of a means-tested grant for blind coloured and white people

older than 19 in 1936; the extension of this DG to physically disabled whites in 1937; the introduction

of war veterans’ pensions and ‘cost-of-living allowances’ to protect the beneficiaries of the OPG

against rising war-time inflation in 1941; as well as ‘a number of measures. . . which to some extent

implied an awareness by the authorities of family poverty and poverty among working people’

(Kruger 1992: 169). In 1937 the state also lowered the OPGs eligibility age to 60 years for women

(while it remained 65 for men) (USA 1937a). In addition to thus generally beginning to provide for

more contingencies, the first step towards broader population coverage was also taken in 1937 with

the replacement of the 1921 Children’s Act.

The Act made the SMG payable under wider conditions that left more room for administrative
72
This period saw the ‘fusion’ of the South African Party and the National Party into a new United Party.

112
discretion, as the SMG could now be paid for ‘the maintenance of any child by any person in

whose custody the child has been placed under this Act or by its parent or guardian’ (USA 1937b).

Administration of the SMG was furthermore transferred from the Department of Education to the

newly established and autonomous Department of Social Welfare in 1937 (Kruger 1992: 170). While

the SMG system was still highly racially discriminatory — with 13 276 white, 5 816 coloured, 3 034

Indian and only 190 black children receiving benefits in 1942 — the amendments and administrative

consolidation broke the traditional approach where services for blacks were provided separately

through the Native Affairs Department (Iliffe 1987: 141). Kruger regards this move as inaugurating

‘a period of more coordinated, ambitious and equalizing action in the provision of social [assistance]’

(1992: 159).

Perhaps the most important step in this regard was the de jure deracialisation of the OPG

and the DG by the government of Jan Smuts. This entailed the legislative extension of pensions

for the blind and aged to blacks and Indians73 in 1944, while the DG was subsequently extended

to all disabled groups through the Disability Grants Act in 1946 (USA 1946; Kruger 1992: 171).

By 1947/48, the OPG was being paid to 196 846 black people, while 49 128 were receiving a DG

(Kruger 1992: 171). Despite the fact that this signalled the legal inclusion of all population groups

in the social assistance system for the first time, benefit levels and means test thresholds were still

highly discriminatory.

In general, the ‘means test was stricter and benefits lower for the other races than for whites’

(Kruger 1992: 171). Indeed, in 1947 the maximum monthly OPG benefit for coloureds and Indians

was 54 percent of the level for whites, while the maximum benefit level for blacks was a mere 33

percent; the corresponding ratios for the SMG stood at 34 percent and 25 percent respectively

(Pollak 1981: 157). Additionally, benefit levels were lower for rural inhabitants than for urban

beneficiaries in terms of both the OPG and SMG. While racial discrimination thus persisted, the
73
Although ‘non-statutory payments’ were already being paid since 1935 to the Indian ‘aged, infirm and blind’ and

to blacks who were blind or war veterans (USA 1944: 56-57).

113
1940s was nevertheless a period of expansionary reform.

This leads to the conclusion that

social citizenship in South Africa thus has its origins in the period immediately prior to the NP’s

electoral victory in 1948. . . because the state had. . . accepted some responsibility for the welfare,

broadly understood, of the black population. It was in the early and mid-1940s that the state

first assumed this responsibility, with a series of reforms that established a multi-racial welfare

system. . . Some of these reforms were undone after 1948, but others (including especially the

OPG) survived (Seekings 2000: 5).

As a result, South Africa had created the ‘basis of a remarkable welfare state’ by the late 1940s

(Seekings 2006: 2). The fledgling social assistance system covered all population groups (albeit in

a highly discriminatory manner) and was composed of the SMG to support children, the DG to

support the disabled and the OPG to support the elderly. All programmes were unconditional,

their orientations were individualist through defining beneficiaries without regard to familial cir-

cumstances, while overall coverage was strictly means tested and limited to the ‘deserving poor’74

and excluded impoverished people of working age.

Early Apartheid, Reracialisation and Retrenchment

The trend towards increasing deracialisation and consolidation that had clearly emerged towards

the end of the 1940s came to an abrupt halt with the electoral victory of the NP in 1948, and the

resulting implementation of apartheid. The period was ‘characterised by a direct and purposeful

assault on aspects of the new fragile welfare construction that had begun to emerge during the

previous decade’ (Bromberger 1982: 175). In terms of social assistance, the assault was expressed

principally through increasing racial inequalities and a general contraction in coverage.

Means testing and benefit levels became more potent tools in the government’s quest to increase

racial discrimination (Kruger 1992: 175-176). Indeed, in 1944 the means test threshold for social
74
In effect this was impoverished children, as well as the disabled and elderly poor.

114
pensions in terms of coloured and Indian beneficiaries was 60 percent that of the white level, while

the figure stood at 40 percent for blacks. By 1966 this ratio was further reduced to 50 percent

for coloureds and Indians, and to 12.5 percent for blacks (Kruger 1992: 175). As noted above, in

1947 the maximum monthly OPG benefit for coloureds and Indians was 54 percent of the level for

whites, while the benefit level for blacks stood at 33 percent. By 1960 these maximum levels had

been reduced to only 40 percent for coloureds, 35.9 percent for Indians and 14.9 percent for blacks

(Pollak 1981: 156-158). In terms of average monthly OPG payouts, by 1962 whites received R267.19,

while coloureds were paid R118.01, Indians R102.24 and black people a mere R25.57 (Kruger 1992:

175). This trend held true for the SMG and DG as well (Kruger 1992: 175).

This systemic contraction and increase in discrimination was further enhanced through the

return to the ‘self-balancing principle’75 in welfare administration from 1948. It resulted in the

racial fragmentation of service provision and attempts to eliminate or at least minimise the financial

cross-subsidisation of segregated social assistance services (Kruger 1992: 175). In this regard, the

logic of the apartheid political economy resulted in an attempt to move dependent blacks, including

the old and infirm, out of areas under ‘white administration’ and to the homelands.76

The aim was to ‘transfer welfare institutions and provisions to the homelands’ and to organise

welfare provision

along ethnic and tribal lines. The lack of effective administration and financial constraints on

homeland governments led to de facto pension coverage far below what legal entitlement would

indicate. The process of fragmentation of welfare provision gained further momentum after

control of coloured welfare was transferred to the Department of Coloured Affairs in 1958 and

responsibility for [blacks] to the Native Affairs Department in 1960. The involvement of other
75
A central premise of apartheid, this principle held that ‘every group should be regarded as a separate political

and economic unit’ (Van der Berg 1989a: 197). In practice, this meant that ‘white taxes’ were not to be spent on

‘black services’ and that administration had to be kept separated.

76
Homelands, also known as Bantustans, were areas administratively designated by the government as pseudo-

national countries for the black population.

115
 population groups in the provision of black welfare was also declared as ‘contrary to policy’ in

1957, necessitating expensive and inefficient duplication of voluntary and state welfare structures

(Kruger 1992: 176-177).

The resulting inequality was clear: by 1958 blacks composed 60 percent of the 347 000 people receiv-

ing the OPG, although they received only 19 percent of the total amount spent on the programme

(Van der Berg 1997: 487).

By this time, the underlying nature of the South African social assistance system was becoming

clear. As shown above, it had principally been created prior to the onset of apartheid in order

to protect the white population. Given the fact that white employment was now largely secure

(in contrast to the situation during the Great Depression), the need for social assistance continued

to be limited to the most vulnerable groups: the elderly (the OPG), children (the SMG) and

the disabled (the DG) (Van der Berg 1997: 488). The retrenchments of the early apartheid period

again confirmed that despite significant poverty among particularly rural blacks in this group, social

assistance would not be forthcoming to support impoverished people of working age, as there was

simply no need for it amongst whites.

In addition to therefore ignoring the needs of the majority of impoverished unemployed peo-

ple, early apartheid policies actively retrenched the social assistance safety net open to non-whites

through increasing discrimination and a concomitant contraction in coverage. All of these develop-

ments therefore served to further reaffirm the limited nature of social assistance provision through

excluding almost all but the deserving poor (here perhaps best understood simply as young, el-

derly and disabled whites), while the principles of unconditionality and an individualist orientation

remained intact.

Late Apartheid, Expansion and Reduced Inequality

Although it was not yet being reversed, the trend towards retrenchment and increasing inequality

in terms of social assistance started to slow during the early 1960s. This was initially made clear

116
through the fact that the positions of coloured and Indian OPG beneficiaries started to improve,

as benefit levels began to increase relative to whites (Bromberger 1982: 183). This was eventually

followed by a similar halt in the deterioration of relative benefit levels for black OPG beneficiaries

(Bromberger 1982: 183). These developments further went along with an overall increase in absolute

and per capita expenditure on black people from 1962/1963 (Kruger 1992: 178). This revealed ‘the

waning of determined efforts to reduce expenditures on black social services, possibly illustrating the

realisation of the need for cross-subsidisation between income groups and across the racial divide’

(Kruger 1992: 177).

In spite of this implicit acknowledgement of the need for social assistance spending on all groups,

the idea that ‘separateness and equality can co-exist’ remained deeply entrenched (Kruger 1992:

178). The continued salience of this approach was demonstrated through the fact that administra-

tion of ‘black areas’ was taken over by the Administration Boards77 in 1972, which implied ‘an end

to all direct and indirect subsidies within municipal areas and thus reflected a continued belief in

the ideal of financial self-sufficiency of the different groups’ (Kruger 1992: 178).

Despite this shift, a more distinctive trend towards reincorporation and reduced inequality did

start to emerge from the mid-1970s, although it developed in a highly uneven and ambivalent

fashion that continued to reflect the logic of preceding periods (Bromberger 1982: 183). The

trend was expressed most clearly in the area of social assistance through a reduction in OPG and

DG spending differentials due to an increased awareness of poverty and the need for corrective

policies (Bromberger 1982: 185-187; Kruger 1992: 179). Indeed, in 1970 the maximum benefit

level for coloureds and Indians amounted to 43 percent of the level for whites, while blacks received

13 percent of the white level. By 1991, these ratios had improved to 86 percent for coloureds
77
In 1972, ‘the country was divided into twenty-two regions, in each of which an Administration Board was set

up. The Boards were to take on most of the functions previously exercised by municipalities in the administration

of urban [blacks]’ (Maylam 1990: 68). The Boards were expected to be financially self-sufficient (principally through

controlling municipal liquor monopolies), which implied further administrative fragmentation.

117
and Indians, and 74 percent for blacks (Kruger 1992: 180). Furthermore, in 1978 black people

accounted for 70 percent of the 770 000 pensioners and received 43 percent of the amount spent

on OPG benefits. By 1990, 67 percent of the total amount spent on the OPG was going to black

pensioners (Van der Berg 1997: 487).

The slow movement towards deracialisation reflected a reluctant acceptance of the fact that

the apartheid fiscal paradigm, premised on features such as the self-balancing principle and limited

cross subsidisation, was failing (Van der Berg 1989a: 198). This meant that ‘once it was recognised

that fiscal apartheid cannot be legitimately retained in an integrated economy and society, the

implication was that horizontal fiscal justice across race boundaries was required’ (Van der Berg

1989b: 9). In the area of social assistance, practical ‘horizontal fiscal justice’ entailed taking steps

towards deracialisation and equalisation. Despite the clear emergence of a such a trend78 in the

late apartheid period, change remained highly uneven. This is reflected by the fact that despite the

progress in terms of the OPG and DG, average SMG benefit levels in 1987 for blacks was still only

17 percent of the white level (Kruger 1992: 180).

Nevertheless, the trend towards equalisation demonstrated an increased awareness of poverty

among all groups, with the 1976 report of the Theron Commission on Coloured Affairs being

regarded as the first sign of this increased recognition (Bromberger 1982: 186). The growing

protest voice of the disenfranchised poor put further pressure on the state to carefully consider the

impact of social assistance policy choices (Kruger 1992: 182). However, the government’s evermore

heightened concern with equality in terms of social spending did not immediately translate into

abandoning the ideal of apartheid.79 Indeed, it can be argued that the period featured renewed

efforts to consolidate ‘separateness’ through the introduction of the Bantu Homelands Citizenship
78
Kruger rightfully cautions against over-emphasising this trend, as means tests were probably not adjusted as

rapidly as benefit levels (1992: 180). The implication is thus that weighted average composite benefit and means test

levels most likely remained highly unequal.

79
The struggle against apartheid indeed rose to a crescendo during the 1970s and 1980s because racial discrimination

remained firmly entrenched.

118
Act in 1970, which compelled all black people to become citizens of the homeland that purportedly

corresponded with their own ethnic group, regardless of whether they had ever lived there or not.

It also removed their South African citizenship (RSA 1970).

The concept of separate governance was additionally strengthened through the introduction of

the Tricameral Parliament80 in 1984. However, the effect of these attempts to legitimate separate

governance was ironically to further equalise social assistance spending. Indeed, a major impetus

for expanding social assistance resulted from attempts to politically legitimate the homeland system

and the Tricameral Parliament. The social assistance funds flowing to the homelands consequently

rapidly increased, especially in terms of the OPG. Both coverage and benefit values for the elderly

black population increased markedly, and ‘in 1993 there were almost twice as many black pensioners

inside the homelands as outside’ (Van der Berg 1997: 487). Additionally, social assistance funding

directed at the coloured and Indian communities also increased rapidly, with both coverage and

benefit levels improving (Van der Berg 1997: 488).

Incorporating these small and relatively marginal groups into the social assistance system

throughout the 1970s and 1980s was financially manageable (Van der Berg 1997: 488). The much

bigger fiscal challenge arose once the principal of parity in terms of both benefit levels and means

test thresholds across all groups was accepted in the late 1980s. The situation was further compli-

cated by the rise of neoliberal economics and the discourse of fiscal conservatism. In South Africa,

this resulted in an ‘increased insistence from government sources on the need to limit and even

reduce the role of the state in the economy’ –– also in terms of social assistance provision (Kruger

1992: 183). This resulted in a shift towards ‘residualism’ as part of the wider strategy to reduce

state expenditure and shift the burden of welfare to individual consumers (Van Niekerk 2003: 366).

Because fiscal constraints thus precluded simply increasing expenditure on black social policies to

match white levels, the focus instead shifted towards reducing entitlements traditionally earmarked
80
While the Tricameral Parliament theoretically granted limited representation to coloured and Indian South

Africans, it also served to further embed the approach of separate governance for different groups.

119
for whites. This was achieved most readily in the area of social assistance, where the number of

elderly and disabled poor whites who qualified under a regime of limited coverage was politically

marginal (Van der Berg 1997: 488). Equalisation in terms of the OPG was thus achieved through a

combination of enhancing black benefit levels (by an average of 7.3 percent each year from 1970 to

1993) and systematically eroding white levels. While in 1980 white pensions displaced more than

30 percent of average wages and black levels only displaced 8.6 percent, pension parity was achieved

in 1993 with a universal wage replacement rate of 15.5 percent (Van der Berg 1997: 488). Figure

6.3 shows the outcome of this process of equalisation for the OPG.

100
90
80
70
60
Percentage

50
40
30
20
10
0
1947 1960 1970 1991 1993
Coloureds Indians Blacks

Figure 6.3: OPG Benefit Levels as a Percentage of White Level (Source: Kruger 1992; Van den Berg 1997)

The period also witnessed the introduction of two further social grants. The state created the

FCG following the promulgation of the 1983 Child Care Act (RSA 1983). This grant was aimed

at supporting foster parents who were caring for a child committed to their care by way of a court

order (SALC 2002: 714). This was followed by the introduction of the Care Dependency Grant

CDG81 in terms of the Social Assistance Act of 1992, supporting the caregivers of disabled children
81
Lund describes the administrative inefficiency that characterised this grant at that time by stating that ‘in 1996 it

was still impossible to determine the number of CDGs. . . The CDG was applied for through the health department, but

120
(RSA 1992). Even with attempts to limit the fiscal impact of full equalisation and the elimination

of racial barriers in terms of the OPG, the DG, the FCG, and the CDG (the implementation of the

SMG meanwhile remained highly discriminatory), the effect was to rapidly increase overall social

assistance spending. Spending on the OPG thus rose from 0.59 percent of GDP in 1970 (which was

down from 0.8 percent in 1960) to 1.82 percent by 1993 (Van den Berg 1997: 487-488).

In spite of this development in relation to the deracialisation of some programmes, other funda-

mental social assistance norms that emerged during earlier periods remained wholly intact. Even

as programmes were expanded to include more groups, overall coverage remained limited through

not providing for impoverished people of working age, benefits remained unconditional and the ori-

entation continued to be individualist. In fact, the process of deracialisation during late apartheid

served to further embed these norms by beginning to extend them without alteration to cover the

entire population.

Social Assistance at the End of Apartheid

The preceding analysis demonstrates that the pre-1994 development of the South African social

assistance system was inexorably tied to the overall process of state formation within a segregationist

and eventual apartheid political economy. Table 6.1 provides a broad summary of the evolution of

the system.
delivered through the judicial services (magistrates’ courts), and the records were held at these courts, individually,

and never centralised. The welfare department simply did not know how many grants were being awarded’ (2008:

19).

121
 Period Note

1921-1948 Introduction and fragmented expansion

1921-1936: Introduction

1936-1948: Fragmented and selective expansion

1948-1961 Contraction

1961-1994 Expansion and reduced inequality

1961-1980: Expansion

1980-1994: Deracialisation

Table 6.1: Evolution of the pre-1994 South African Social Assistance System (Source: Adapted from Driabe et al.

1995: 5)

When apartheid ended, social assistance provision had thus become

an outflow of [South Africa’s] apartheid history, which allowed the tender plant of the welfare

state to be artificially protected by racial barriers, and later accounted for the surprisingly rapid

expansion of social security [through deracialisation] as the balance of social and political power

changed (Van der Berg 1997: 500-501).

By the 1990s, many of ‘the social pensions and grants which were set up to protect the white

population [had] gradually expanded their eligibility rules to include all South Africans’ (Lund 1993:

22). Despite this incorporation of other groups, the system nevertheless continued to reflect the

underlying norms of the one originally created to protect the white population under the segrega-

tionist and apartheid paradigm. As a first step towards assessing the general landscape of social

assistance at this time, table 6.2 indicates the social assistance programmes in operation by 1994.

The striking conclusion is that the vast majority of the programmes in existence today were already

established in the decades preceding the transition to democracy.

122
 Name Year Created Target Current Status

State Maintenance 1921 Impoverished Children (0- Defunct

Grant 18/22)

Care Dependency Grant 1992 Disabled Children (0-18) Operational

Foster Child Grant 1983 Foster Children (0-18) Operational

Disability Grant 1936 Impoverished Disabled (18-65) Operational

Older Persons Grant 1929 Impoverished Elderly Operational

(60+/65+)

Table 6.2: South African Social Assistance in 1994

In addition to bequeathing these programmes to the incoming government, the norm of only

providing limited coverage through excluding the working age poor was also firmly entrenched

during previous periods. Reconstructing South African lifecycle risks for 1994 indicates that it

indeed mirrors the one currently in operation (with only slight differences in age eligibility) because

it too excluded able-bodied impoverished people who were too old to receive the SMG,82 and too

young for the OPG.

Birth Time Death

Education Phase (age Accumulation Phase Dependent Phase (age

0-18/22) Risks: (age 18/22-60/65) Risks: 60+/65+) Risks:

Poverty/Old Age
Poverty/Education Poverty/Unemployment Risk:
Funding Risk: Risk:

 ? 
Health/Disability Health/Disability Risk: Health/Disability


Risk: Risk:

 

Figure 6.4: South African Lifecycle Risks in 1994 (Source: Adapted from Smith 2011)

Additionally, the core norms of unconditionally distributing benefit payments and assessing the
82
While the SMG was legally available to all groups at this point, it practically remained a highly discriminatory

programme that covered only a very small percentage of potential black beneficiaries

123
needs of potential beneficiaries through individualist means testing were also well-established during

the preceding decades. Finally, despite the significant progress that had been made in deracialising

the welfare state, the SMG remained a highly discriminatory social assistance programme by the end

of apartheid. When democracy arrived in April 1994, South Africa had thus already constructed an

elaborate social assistance system dating back some 70 years. This tumultuous historical process led

to the deep entrenchment of particular social assistance norms (limited coverage, unconditionality

and an individualist orientation) which would subsequently prove to have powerful path dependent

effects upon the development of post-1994 social assistance policies.

6.2.2 The Legal Codification of Social Assistance Norms in South Africa

The transition to democracy during the early 1990s represented a critical juncture with the potential

to break the path dependent institutional stickiness that had developed through the entrenchment

of productive social assistance policies and norms during previous decades. While the South African

Constitution is often lauded as a highly transformative document that severed institutional ties with

pre-democratic South Africa, the area of social assistance represents a clear case of institutional

continuity (Sunstein 2001).

Instead of transforming them, the Constitution served to further embed productive norms into

the political economy of social assistance by translating them into the country’s supreme law. This

process formalised institutional norms into concrete legal rules, enhancing their ultimate effect upon

policy preference formation. This subsection reviews the constitutional provisions and policy White

Paper related to the provision of social assistance in an attempt to illustrate the ways in which they

served to embed the norms that were carried over from the pre-1994 period into the institutional

framework of contemporary South Africa.

124
The Constitution (1996)

The supreme source of institutional norms in South Africa is the Constitution, which was promul-

gated in 1996. Up until the adoption of this final Constitution, the country had functioned according

to the provisions of a 1993 interim Constitution that arose from the negotiations between the NP

and the African National Congress (ANC) to end apartheid. Negotiations and deliberations aimed

at drawing up the text for the final Constitution continued after the overwhelming electoral victory

of the ANC on 27 April 1994 and the subsequent inauguration of Nelson Mandela as President. The

task of finalising the Constitution thus fell upon the Government of National Unity (GNU),83 which

adopted the final text on 11 October 1996. After being duly certified by the Constitutional Court,

it was signed by President Mandela on 10 December and officially proclaimed on 18 December 1996.

The Constitution came into full effect on 4 February 1997.

The South African Constitution is generally regarded as one of the most progressive in existence,

and has even been described as ‘the most admirable Constitution in the history of the world’

(Sunstein 2001: 261). This is because it ‘gives extensive recognition to and entrenches not only

so-called first generation (civil and political rights), [but] also so-called second and third generation

fundamental rights (socio-economic and environmental rights, respectively)’ (Olivier 2002: 119-120).

This is especially true with regards to the Bill of Rights, which bestows extensive rights in areas

ranging from labour relations, the environment, land, housing, healthcare, food and water, and

with particular regard to children, education, language and cultural rights, the rights of cultural,

religious and linguistic communities, and just administrative action (Olivier 2002: 119-120; RSA

1996a: chapter two, sections 23-33). A further example of the entrenchment of a socio-economic
83
The interim Constitution provided for the establishment of a GNU, whereby any party with 20 or more seats in

Parliament following the 1994 election could claim one or more Cabinet portfolios in government. The main purpose

of the arrangement was to adopt the final Constitution in an inclusive manner. The NP announced its withdrawal

from the GNU as soon as the final Constitution was effected on 4 February 1997, although the arrangement only

officially lapsed at the end of the first Parliament’s term in 1999.

125
right is the right to social assistance.

Section 27 in the Bill of Rights of the Constitution discusses a range of rights, including the

declaration in subsection (1) (c) that ‘everyone has the right to have access to. . . social security,

including, if they are unable to support themselves and their dependents, appropriate social assis-

tance’ (RSA 1996a). This framing of access to social assistance as a constitutional right amounts to a

codification and powerful reinforcement of the pre-established institutional norm of unconditionality

in the delivery of benefits.

Reading this section in conjunction with section 2 of the Bill of Rights — which states that ‘this

Constitution is the supreme law of the Republic; law or conduct inconsistent with it is invalid, and

the obligations imposed by it must be fulfilled’ –– justifies the conclusion that ‘the fundamental

right to access to social [assistance] is enforceable, because section 2 explicitly states that duties

imposed by the Constitution must be performed’ (RSA 1996a; Olivier 2002: 122). This is further

‘fortified by the constitutional provision that the state must respect, protect, promote and fulfil the

rights in the Bill of Rights’ (Olivier 2002: 122). The unambiguous language in the Constitution

therefore effectively prohibits the implementation of conditional social assistance programmes, as

withholding welfare payments on the basis of a failure to comply with conditions would be deemed

unconstitutional (Interview 11).

However, in addition to establishing a firm rights-based approach to social assistance, section

21 (1) (c) cited above also laid the initial groundwork for entrenching the norm of providing only

limited coverage by declaring that everyone has the right to social assistance ‘if they are unable

to support themselves and their dependents’ (RSA 1996a). This had the effect of ‘locking in’ the

principle of limited coverage (Bond 2014: 7). The Constitution furthermore introduces the principle

of the ‘progressive realisation’ of certain rights. Section 27 (2) specifically elaborates on the right to

social assistance by requiring that ‘the state must take reasonable legislative and other measures,

within its available resources, to achieve the progressive realisation of [this] right’ (RSA 1996).

126
 In this context, progressive realisation is thus regarded as a ‘qualifying subsection’ to the right

to social assistance (SAHRC, 2009: iii). The United Nations Committee on Economic, Social and

Cultural Rights (CESCR) defines the concept as constituting

a recognition of the fact that full realisation of all economic, social and cultural rights will

generally not be able to be achieved in a short period of time. . . It is on the one hand a necessary

flexibility device, reflecting the realities of the real world and the difficulties involved for any

country in ensuring full realisation of economic, social and cultural rights. On the other hand, the

phrase must be read in the light of the overall objective. . . which is to establish clear obligations

for states in respect of the full realisation of the rights in question (1990: para. 9).

The concept is based on the recognition that socio-economic rights would generally not be realised

in a short period of time, and it thus introduces ‘an element of some flexibility in terms of the

obligations of states’ (Chenwi, 2010: 19).

The Constitution therefore indicates that, alongside the right of access, ‘financial viability’ is

to be regarded as a key paradigm informing the development of social assistance in South Africa

(Olivier 2002: 122). While the Constitution places an obligation on the state to ensure access to

social assistance, it simultaneously allows a degree of elasticity in relation to three key aspects:

‘the progressive realisation of the right, the taking of reasonable measures and the availability of

resources’ (Olivier 2002: 122). When read in conjunction with section 7 of the Bill of Rights, which

declares that ‘the rights in the Bill of Rights are subject to the limitations contained. . . elsewhere

in the Bill’, it becomes clear that the availability of resources is to be a key factor in determining

whether the state is deemed to have taken ‘reasonable measures’.

In effect, ‘resource constraints could be a basis for the state justifying its rate of progress

(or lack thereof) in achieving the full realisation of social [assistance] rights’ (Olivier 2002: 122).

In this way, the concept effectively becomes a limitation that is placed upon the right to access

social assistance. The significance of institutionally incorporating progressive realisation as the

second pillar of constitutional social assistance rights is to legally entrench, legitimate and provide

127
constitutional justification for the already extant institutional norm of limiting access to social

assistance programmes to those deemed ‘deserving’.

The Bill of Rights’ extensive incorporation of both negative and positive rights84 further results

in the inclusion of strong provisions on legal and social equality. Section 9 effectively amounts to

a definition of individual equality that is more detailed than the Universal Declaration of Human

Rights. It asserts that

everyone is equal before the law and has the right to equal protection and benefit of the law. . . The

state may not unfairly discriminate directly or indirectly against anyone on one or more grounds,

including race, gender, sex, pregnancy, marital status, ethnic or social origin, colour, sexual

orientation, age, disability, religion, conscience, belief, culture, language and birth (RSA 1996a).

Within the context of an extremely socially and culturally diverse country such as South Africa,

this focus on promoting individual equality effectively discourages the conceptualisation and defi-

nition of social assistance beneficiaries in terms of family structure, as this would almost inevitably

involve normative definitions involving issues of marital status, religion, conscience, belief and cul-

ture.85 These provisions thus make it clear that the Constitution encourages the institutional norm

of assessing the needs of beneficiaries on an individualist basis.

The popular narrative asserts that ‘the South African Constitution is the world’s leading ex-

ample of a transformative Constitution. A great deal of the document is an effort to eliminate

apartheid “root and branch”’ (Sunstein 2001: 224-225). While there is undeniable truth to this

general statement, the effort to eliminate institutional norms originating in the apartheid and pre-

apartheid era ‘root and branch’ did not fully extend to the area of social assistance. Even though

complete deracialisation was to be achieved (completing a process that had already been underway
84
Negative rights generally include first generation civil and political rights and largely oblige inaction (prevents

infringement upon certain rights). Positive rights usually include second and third generation rights obliging an active

pursuit of the fulfilment of these rights (Sunstein 2001: 221).

85
See Lund 2008: 65 for a discussion of these ‘deep level problems’.

128
for some time), the fundamental systemic social assistance norms survived wholly intact in the new

Constitution.

In the same vein, the Constitutional Court has argued that the Constitution ‘retains from the

past only what is acceptable and represents a radical and decisive break from that part of the past

which is unacceptable’ (ZACC 1996). In terms of social assistance, racial discrimination was deemed

to be a wholly unacceptable part of the past. However, the Bill of Rights then implicitly deems as

‘acceptable from the past’ the three core norms underwriting the social assistance system in South

Africa: a structure of limited coverage, featuring an unconditional approach and an individualist

orientation. The Constitution effectively codified legal and institutional norms that allowed for

the further entrenchment of a productive social assistance system in the political economy of post-

apartheid South Africa.

The White Paper for Social Welfare (1997)

The adoption of the Constitution was followed by the publication of the White Paper for Social

Welfare (WPSW) in August 1997 (RSA 1997b). The DSD describes it as ‘the primary policy

document [serving] as the foundation for social welfare in the post-1994 era’ (DSD 2014). It is

meant to further express the social welfare provisions introduced in the Constitution by articulating

the ‘principles, guidelines, recommendations, proposed policies and programmes for. . . social welfare

in South Africa’ (Potts 2012: 76). The way in which the WPSW builds on the abovementioned

constitutional provisions further reinforces the institutional norms associated with a productive

social assistance system.

In addition to the need for completing the process of deracialisation through phasing out ‘all

disparities in social welfare programmes’, the document echoes the Constitution’s effective entrench-

ment of an unconditional approach through stating that

the government will take steps to ensure the progressive achievement of. . . appropriate social

assistance for those unable to support themselves and their dependents. . . to give effect to the

129
 constitutional right to social security (WPSW 1997: 21; 45).

In this regard, the WPSW takes ‘the lead from the [newly introduced] constitutional right of access

to social security’, acknowledging the salience of this rights-based approach and the related norm

of unconditionality (Lund, 2008: 13).

The WPSW further builds on the constitutional caveats of limiting social assistance to those who

are ‘unable to support themselves and their dependents’ and financial viability through ‘progressive

realisation’ by introducing the paradigm of ‘developmental social welfare’ to serve as the overall

principle guiding policy formulation (RSA 1997b). The concept is

geared towards providing citizens with an ‘opportunity to play an active role in promoting their

own well-being’. The priority on individual self-activation, under the guise of ‘empowerment’

discourse, was combined with a view of social security and social services as ‘investments which

lead to tangible economic gains’. Public spending expansion was seen as dependent on economic

growth in a context where ‘the high expectations of many people for the new democratic govern-

ment to deliver welfare services and programmes to address pressing needs cannot be fully met

in the short-term’. Market forces, productivism and individual responsibility came therefore to

provide strict constraints to redistributive and decommodifying policy interventions (Barchiesi

2005: 343).

The priority assigned to providing only limited coverage through a residual system that excludes

the ‘non-deserving’ poor (unemployed able-bodied people of working age) is forthrightly expressed

in chapter 8 of the WPSW: ‘whilst welfare programmes should be available to all South Africans,

the focus must be on the poor, those who are vulnerable and those who have special needs’ (RSA

1997b: 8.4). This approach, based on providing limited coverage, is furthermore explicitly framed

in macroeconomic terms through the admission that ‘fiscal constraints’ make it impossible ‘for the

welfare function to grow in real terms in the medium term’ (RSA 1997b: 8.11). The resulting

‘residualist orientation accompanied its emphasis on reducing dependence on state social assistance

and increased self-reliance’ (Barchiesi 2005: 344).

130
 The WPSW explicitly strives for the construction of a productive social assistance through

framing the state as a ‘provider of last resort’, advocating for ‘more stringent and appropriate

means testing and eligibility requirements’ and programmes intended to ‘divert people from the

welfare system’ (RSA 1997b: 45; 22). The document

ultimately and ‘programmatically’ departs from approaches based on universalism of access and

decommodification of provision, by promoting further selectivity and privatization. To reinforce

the point, wage labour discipline is presented as a stark alternative to welfare ‘dependency’

(Barchiesi 2005: 345).

According to the WPSW’s logic, social assistance should be oriented to the economy’s needs

— not to the needs of society (Bond 2014: 4). This rationale leaves no space for social assistance

programmes to support ‘the [able-bodied] 35 to 40 percent of the working age population who

cannot find work or even engage in informal entrepreneurial activity’ (Bond 2014: 4). The WPSW

unambiguously sets out an institutional path towards constructing a productive social assistance

system in South Africa.

6.2.3 Social Assistance Norms in South Africa

This section has examined the way in which the social assistance norms associated with South

Africa’s productive system came to be deeply institutionalised in the country’s contemporary polit-

ical economy. It illustrates that the rudimentary origins of the current system can be traced back

to the policies that were initially enacted during the period of segregationist and apartheid state

construction over a period of eight decades. The principles of limited coverage, an unconditional

approach and an individualist orientation all originated through this historical process.

The discussion also highlighted the fact that these norms were not discarded during the tran-

sition to democracy in the early 1990s. Instead, these principles remained unaltered as they were

legally codified in the 1996 Constitution and expanded upon in the WPSW of 1997. The historical

institutional context facing policy actors hoping to define the contours of social assistance in demo-

131
cratic South Africa was thus clear: they were operating within a policy environment where distinct

institutional norms were not only historically deeply entrenched, but had also been constitutionally

embedded. The effect of this high level of entrenchment was to significantly reduce the policy op-

tions available to actors through the mechanism of path dependence as they looked to define social

assistance in a democratic South Africa from 1994 onwards.

6.3 State Structure

In addition to the historical development and legal codification of particular norms, the institutional

context facing policy actors was further shaped by contemporary South Africa’s highly concentrated

state structure. This section thus addresses the second independent variable through examining the

way in which the centralisation and concentration of political power in post-apartheid South Africa

influenced the institutional context and, ultimately, the capabilities and preferences of actors. It

takes a broad approach to defining state structure, focusing both on the formal state institutions

shaping policymaking capacity and the way in which political power was practically distributed.

The initial section details the formal institutional and organisational structure of the state,

highlighting the way in which social assistance policymaking capacity was wholly centralised in the

national DSD. This is followed by an examination of the distribution of political power in post-

apartheid South Africa, a discussion which serves to highlight the extreme concentration of power

in the hands of the ANC. Synthesising these two elements culminates in the conclusion that in terms

of the institutional capability to formulate and enact social assistance policy, South Africa features

a highly concentrated state structure. The capacity to determine social assistance policy is solely

vested in the national government, while said national government has been, and continues to be,

politically entirely dominated by the impressive electoral majorities of the ANC.

132
6.3.1 Concentrated Institutional Structure

The South African Constitution elucidates the organisational structure of the contemporary South

African state. The country is a constitutional democracy with a three-tier system of national

government and features an independent judiciary (GCIS 2012a: 286). Legislative authority is

vested in the national Parliament, reflecting the fact that South Africa is governed according to a

parliamentary system. Parliament itself is firstly composed of the National Assembly (NA), which

constitutes the lower house.

According to section 46 of the Constitution, between 350 and 400 representatives are elected to

the NA every five years through a closed party-list proportional representation system, where half

of the members are proportionally elected from the nine provincial party lists and the other half are

chosen from national lists (RSA 1996a). The upper house of Parliament is the National Council of

Provinces (NCOP). The function of the NCOP is to ensure that provincial interests are represented

during the national legislative process (Parliament 2014). Section 60 of the Constitution stipulates

that the NCOP ‘is composed of a single delegation from each province consisting of ten delegates’,

thereby giving equal representation to all provinces, irrespective of population size (RSA 1996a).

Section 91 of the Constitution further specifies that executive authority is vested in the Cabinet,

which is composed of the President, Deputy President, Ministers and Deputy Ministers (RSA 1996a).

The President appoints and assigns the powers of the Deputy President and Ministers, and may

dismiss them. The Deputy President and Ministers (save for two) are selected from among the

members of the NA. The President furthermore appoints a member of Cabinet to be the leader

of government business in the NA. In turn, the President is elected from among the members of

the NA (RSA 1996a: 86). Once elected President, ‘a person ceases to be a member of the NA

and, within five days, must assume office by swearing or affirming faithfulness to the Republic and

obedience to the Constitution’ (RSA 1996a: 87). The President serves a five year term of office,

and ‘no person may hold office as President for more than two terms’ (RSA 1996a: 88).

133
 The third tier of national government is the judiciary, with judicial authority being vested in

the courts. Courts are wholly independent and subject only to the Constitution and the law, which

they are required to apply impartially and without fear, favour or prejudice. Organs of state are

constitutionally required to assist and protect the courts (RSA 1996a: 165). The highest court in

the land is the Constitutional Court, followed by the Supreme Court of Appeal, the High Court of

South Africa and the Magistrates’ Courts (RSA 1996a: 166). The country features a hybrid legal

system, primarily consisting of a Roman-Dutch civil law system, a British system of common law,

as well as elements of African customary law (SCA 2014).

In addition to the national layer, the country also features provincial and local governmental

spheres. The nine provinces have their own provincial legislatures, with members being elected

through a closed party-list proportional representation system and which consist of between 30

and 80 members (RSA 1996a: 105). Provincial executive power is located in the office of the

Premier and their Executive Committee. Managed according to chapter seven of the Constitution,

the local government level includes 226 local municipalities and 44 district municipalities (RSA

1996a). The executive and legislative authority of a municipality is vested in its Municipal Council.

Municipalities have the right ‘to govern, on their own initiative, the local government affairs of its

community’ (RSA 1996a: 151).

Importantly, schedules four and five of the Constitution specify the areas of concurrent national,

provincial and local legislative competence. Although the Constitution thus nominally contains

certain weak federal elements, the country ‘is largely a unitary state’ (Ile 2007: 2). In fact, the

South African case closely mirrors Ile’s definition of a unitary state as

a type of government that seeks to concentrate power at the centre for various reasons. It may

de-concentrate power to other subunits of government to enable it to achieve its objectives.

This means that such governmental subunits are mere extensions of the central government or

agencies of the central government (2007: 15).

This statement is particularly true in the case of social assistance policies. According to these formal

134
institutional arrangements, welfare (including social assistance) policymaking is strictly a national

competency in South Africa.

Only the national government has the constitutional authority to design social assistance pro-

grammes (Interview 1; Interview 2). The Constitutional Court has further independently affirmed

‘that social assistance is a national government competence’ and that only a national government

entity has ‘the competence to administer social assistance’ (Interview 2). SASSA was subsequently

set up in the wake of this ruling as a subsidiary of the national DSD to implement all social assis-

tance programmes, thereby further reflecting the extreme degree of centralisation of social assistance

policymaking and provision in South Africa. The result is that subnational political units have no

control or authority over the formulation or implementation of social assistance policies, with all

such authority being vested in the national DSD.

6.3.2 Political Power Concentration

This highly concentrated policymaking structure, derived from the formal institutional configu-

ration, is further reinforced through the reality of political power distribution in South Africa.

Figure 6.5 illustrates that ever since the onset of democracy, the country has been a single-party

dominated state, where ‘multiple parties compete for power, but only one party wins consecutive

elections’ (Wieczorek 2012: 29).86

Under the leadership of Nelson Mandela, the ANC won its first general election in 1994 with

62.65 percent of the vote (AED 2014). This increased to 66.35 percent in 1999 when Thabo Mbeki

became President, followed by a showing of 69.69 percent in 2004 (enough support to achieve the
86
It is worth briefly considering the underlying reasons for this political power concentration in the hands of the

ANC. The defining role played by the country’s apartheid history has resulted in a situation where politics is largely

defined by racial nationalist discourse. As the party of liberation, the effect of the ANC’s nationalist legitimacy has

been to entrench its political support throughout the period under examination. This resulted in the postponement

of class mobilisation and the associated potential proliferation of political power, allowing the ANC to exclusively

direct policy design in a top-down manner.

135
 100
90
80
70
Percentage 60
50
40
30
20
10
0
1994 1999 2004 2009 2014
African National Congress Democratic Alliance Other Opposition

Figure 6.5: Electoral Support, General Elections (1994-2014) (Source: AED 2014)

two-thirds majority required for constitutional change) (AED 2014). The party got 65.90 percent

when Jacob Zuma became President in 2009, followed by 62.15 percent in 2014 (AED 2014). The

period has witnessed a small consolidation of the opposition vote, with the Democratic Alliance

(DA)87 being the primary beneficiary. The party managed to increase its share of the national vote

from 1.73 percent in 1994 to 22.23 percent in 2014 (AED 2014).

The same one-party logic is visible in terms of provincial and local government, although op-

position parties have been marginally more successful in these spheres. The ANC controlled seven

out of nine provinces after the 1994 elections, eight out of nine following the 1999 election, all

nine provinces after the 2004 elections, and again eight out of nine following the 2009 and 2014

general elections (IEC 2014). The party has also consistently governed more than 90 percent of

the country’s municipalities since 1994 (IEC 2014). The overall conclusion suggested by the data

is clear: in contemporary South Africa, political power is highly concentrated in one party and po-

litical ‘constraints on the ANC are very weak’ (Giliomee, Myburgh & Schlemmer 2001: 163). The

combination of one-party dominance and the strong unitary features of the state outlined above
87
The Democratic Alliance was known as the Democratic Party prior to the 2004 general election.

136
means that social assistance policymaking authority is thus highly concentrated in South Africa.

6.4 The Institutional Setting

The preceding sections of this chapter have served to outline the South African institutional con-

text. The analysis indicates that social assistance norms regarding coverage, conditionality and

orientation were historically deeply entrenched, preserved and constitutionally codified during the

transition to democracy; and that both the formal institutional composition of the state, as well

as the de facto distribution of political power, means that South Africa features a highly concen-

trated state structure — particularly in terms of social assistance policymaking capacity. Table 6.3

consequently summarises the South African case in terms of the study’s independent variables.

Case (IV1) Institutionalisation of Norms (IV2) Concentration of State Struc-

ture

South Africa High High

Table 6.3: Independent Variables – South Africa

6.5 Institutions, Actors and Social Assistance in South Africa

It is important to pause here and emphasise that this concludes the first part of the two-step ACI

analytical procedure.88 As discussed in chapter two, this framework

proceeds from the assumption that social phenomena are to be explained as the outcome of inter-

actions among intentional actors. . . but that these interactions are structured, and the outcomes

shaped, by the characteristics of the institutional settings within which they occur (Scharpf 1997:

1).

The preceding sections thus set the scene for an examination of the way in which the South African

institutional setting shaped the distinct preferences of policy actors, and how these preferences were
88
This section further develops and expands upon some initial insights previously published in Schreiber 2014.

137
ultimately translated into concrete social assistance policy outputs.

This is possible because ‘once we know the institutional setting of interaction, we know a good

deal about the actors involved, about their options, and about their perceptions and preferences’

(Scharpf 1997: 41). Analysing the way in which actors embedded within the previously specified

institutional setting went about the process of policy formulation initially entails articulating the

relevant actor constellation and deducing their institutionally-derived preferences, followed by an

empirical examination of the way in which interactions among members of this policy community

ultimately determined the shape of South Africa’s productive social assistance policy regime.

The shape of the South African social assistance policy community was itself determined by

the country’s concentrated state structure. Due to this institutional configuration — where only

the national government had the constitutional authority to formulate social assistance policy —

the primary venue for policymaking since 1994 has been the national DSD. The Department has

been headed by six different Ministers since then, all from the ANC. This setting witnessed the

emergence of a distinct group of actor sets comprising the social assistance policy community and

driving the policy development process.

The first such set was national state actors,89 composed of bureaucrats from the national DSD,

national politicians and state legal advisers (Schreiber 2014: 269). Although this was undoubtedly

a highly diverse group of individual actors, they are treated as a collective due to the fact that

they were ‘institutionally constituted’ according to the same set of ‘pre-existing rules and that they

depend on [this same set of] rules for their continuing existence and operation’ (Scharpf 1997: 39).

This allows them to ‘coordinate their choices within a common frame of reference that is constituted

by institutional rules’ (Scharpf 1997: 39). These actors were all operating directly under the rules

of the institution that is the national government sphere.

89
For stylistic reasons, the remainder of this chapter will interchangeably refer to this group as ‘state actors’, but

strictly in reference to state actors at the national level.

138
 The second composite group of actors that featured in the policy development process were civil

society actors. This actor set was composed of ‘independent experts...economists, representatives

from civil society organisations and academic researchers’ (Schreiber 2014: 269). This was quite

clearly also a highly diverse set of individuals, but they are treated as a composite group due to the

fact that they too were governed by a common set of pre-existing rules upon which they depended

for their continuing existence and operation; this group of actors all operated under the rules which

guided external interaction with the national government.

Interactions within this policy community translated institutionally-derived preferences into

concrete social assistance policy outputs in South Africa. The nature of these interactions was

moulded by the rules governing engagements between national state actors and civil society actors.

It is important to note once more that only national state actors ultimately had the authority to

formulate policy. No politician from the provincial or local government sphere, or activist lobbying

the national DSD, could directly determine policy outputs (Interview 2). Instead of having direct

influence, the principal interaction avenue that was open to civil society actors came in the shape

of central government commissions.90

Such commissions were created by the national DSD on an ad hoc basis as a formal venue for

civil society actors to provide input on social policy formulation. This mode of interaction however

featured highly

unequal power relations, with state actors having the upper hand. This is because civil society

actors were dependent on the state actors for their legitimacy. The committees were created by

the very actors who were ultimately to consider their reform suggestions. Power resided with

state actors, because any particular reform recommendation made by civil society actors could

not be implemented without their approval. This observation serves to emphasise the role of

power relations (Schreiber 2014: 268).

90
The discussion in section 6.2 makes it clear that establishing advisory commissions composed of external ‘experts’

was a deeply engrained practise in South Africa, perhaps itself constituting an institution of sorts.

139
As a result, the characteristic mode of interaction in the South African case was hierarchical direction

(Scharpf 1997: 171-194).

The role of the institutional setting was however not limited to shaping the interactions within

the policy community; it also decisively influenced the very preferences held by actors (Scharpf

1997: 1; 21). The preceding description of the institutional framework allows us to glean a great

deal of information about the preferences held by these two actor sets. At the most fundamental

level, the analysis proceeds from the assumption that state actors were in all cases motivated to

attain or retain power91 (Davis, Hinich & Ordeshook 1970: 438). In the South African context,

the overwhelming electoral majorities of the ANC however meant that these actors faced ‘little

immediate electoral incentive’ (Seekings 2008: 35). They were therefore insulated to a significant

extent from the pressures of electoral competition, with very little fear that they would be removed

from power. In policy terms, this meant that there was no need for ANC actors to distinguish

themselves from the competition, as there was no real competition.

In terms of their specific preferences, constitutional provisions firstly prioritised the process of

deracialisation and equalisation of welfare programmes which had commenced during the previous

decade. The language contained in the Bill of Rights was clearly incompatible with racially discrim-

inatory policies, which meant that ‘there was a constitutional imperative that it could not only be

provided for one racial group and not for the others’ (Interview 11). As outlined above, the decla-

rations contained in section 27 of the Bill established a preference for an individualist, rights-based

approach to social assistance (RSA 1996a; Interview 11). State actors were further influenced by

the provisions related to ‘fiscal constraints’ and ‘progressive realisation within available resources’,

as outlined in section 27 (2) of the Constitution and the WPSW (RSA 1996a; RSA 1997b: 8.11).

This had the effect of creating an additional preference for fiscal restraint (Bond 2014: 10; Seekings
91
This well-established inference ‘assumes that the candidate’s objective is to maximise his plurality. Although the

rewards [actors] seek vary, it is important to note that winning, at the very least, is instrumental for realising most

such goals’ (Davis et al. 1970: 438).

140
2007: 384). The preferences of state actors thus arose in an institutional context characterized by

a strict rights-based approach, fiscal restraint and individualism — premised on continued political

and policymaking power concentration. This structure encouraged a top-down process of policy

preservation predicated on the political insulation of national state actors.

In sharp contrast to this, the preferences of civil society actors were largely determined by

functional factors. These actors repeatedly expressed a desire to ‘solve the problem’ as they saw

it (Lund 2008: 25). Positioned outside the institutional constraints of the central government,

the preferences of this actor set were ‘directly influenced’ by the perceived features, and functional

shortcomings, of the social assistance system the country inherited from its apartheid past (Schreiber

2014: 271). The result was a thoroughgoing preference for functional effectiveness, with civil society

actors agitating for the creation of a system that could above all be deemed ‘appropriate’ and

‘effective’ (Woolard & Leibbrandt 2010: 11; Schreiber 2014: 274).

The fact that national state actors and civil society actors occupied different positions within the

institutional setting meant that they had different answers to the question: what should the post-

apartheid social assistance system look like? State actors were primarily concerned with preserving

the deeply embedded norms of limited coverage, unconditionality and an individualist orientation.

In partial contrast, civil society actors were largely concerned with implementing functionally ap-

propriate interventions. These divergent preferences, combined with the institutional rules guiding

interactions within this policy community, set the scene for the interactions which would ultimately

lead to the development of the country’s contemporary social assistance system.

6.6 The Policy Development Process

Having specified the institutional setting, actor constellation, modes of interaction and the distinc-

tive preferences held by different groups of actors, this subsection proceeds to empirically trace

the way in which interactions between actor sets translated these preferences into concrete policy

141
outputs. The discussion highlights the way in which the high level of institutionalisation of norms

related to social assistance, combined with a highly concentrated state structure, led to the en-

trenchment of a productive social assistance system in South Africa. The analysis is subdivided

into three chronological parts: an initial period (1994-1999) which saw the first round of policy

formulation, a second period that witnessed largely failed attempts by civil society actors to under-

take comprehensive reforms (1999-2009), as well as a final section (2009-present) characterised by

systemic expansion and consolidation.

6.6.1 Initial Policy Design (1994-1999)

The first step in the policymaking process was taken shortly after the April 1994 elections, with the

appointment of the Lund Committee on Child and Family Support (hereafter ‘Lund Committee’

and ‘the Committee’) by the Minister of Welfare and Population Development92 in December 1995.

The Committee was headed by Francie Lund, a leading expert in social security research, and

was further composed93 of economists, representatives from civil society organisations, organised

labour and Non-Governmental Organisations (NGOs) (Lund 2008: 22-23). Its aim was to make
92
This is the previous name of the DSD.
93
In addition to being chaired by Lund, the Committee consisted of economists Pieter le Roux and Servaas van der

Berg; Laura Kganyago from the Rural Women’s Movement; Jackie Loffell of the National Committee for the Rights of

Children; Marilyn Setlalentoa from the National Welfare, Social Services and Development Forum; Jean Triegaardt of

the Joint University Committee for Social Work; Ndivhuho Sekhoba from the Maintenance Action Group; academic

Debbie Budlender; Marj Brown of the Black Sash; Marion Stewart from the Programme for Unemployed Women with

Children; and Este Lohrentz of the national Department, who was there in the role of secretariat (Lund 2008: 22-23).

142
recommendations for the reform of the state’s family support system94 (Patel 2011: 369).

The Committee’s terms of reference ‘can be read to signify the acceptance by the welfare ministry

of the discourse of targeting and limited assistance’ (Lund 2008: 25). While the appointment of

the Committee signalled a certain level of trust in the expertise of external experts, the Committee

‘crucially lacked any institutional and policymaking power’ (Schreiber 2014: 269). The outcome

of interactions between these external civil society actors and the national state actors that had

appointed the Committee were to provide the first concrete social policy outputs in post-apartheid

South Africa.

The Committee convened for the first time on 1 December 1995, with the first full meeting taking

place on 9 February 1996 (Lund 2008: 19). Committee members largely interpreted the heart of

their mandate as ‘solving the problem of the SMG’ (Lund 2008: 25). As outlined in section 6.2.1,

racial parity had largely been achieved by 1993 in terms of individual grants, with the SMG being

the primary exception. For the Lund Committee, fulfilling their mandate by ‘solving the problem of

the SMG’ meant that the content of their reform recommendations would be determined exclusively

by functional interpretations of what they concluded to be the SMG’s shortcomings.

By this time, the grant had been an entrenched feature of the social assistance system for more

than 70 years. Legally, it was available to a parent or guardian95 living with a child younger than

18 years (this age limit was extended to 22 if the child was in secondary or tertiary education)

(Haarmann 1998: 144; Lund 2008: 62). The requirements for qualification included that the parent
94
The precise mandate called on the Committee to undertake a critical appraisal of the existing system of state

support, in all departments, to children and families; to investigate the possibility of increasing parental financial

support through the private maintenance system; to explore alternative policy options in relation to social security

for children and families as well as other anti-poverty, economic empowerment and capacity-building strategies; to

develop approaches for effective targeting of programmes for children and families; and to present a report giving

findings and recommendations (Lund 2008: 24).

95
For most of its life span the SMG was only available to female caregivers and it was only in 1992 that it was

extended to include males.

143
or guardian had been deserted by the spouse for more than six months; was widowed, separated

or unmarried; had a spouse who had been in a drug treatment centre, prison or similar facility for

more than six months; or had a spouse who had been declared unfit to work for more than six

months or was receiving a social grant themselves (Lund 2008: 15). The SMG was unconditional.

It also featured a complicated two-pronged means test, with the actual cash transfer also con-

sisting of two components. Firstly, qualification for the monthly flat rate child payment of R135

(US$24) per child was limited to those applicants with monthly earnings below R1 118 (US$202)

for a single income and below R2 236 (US$404) for a dual income (when one partner received a

disability grant) (Haarmann 1998: 81). The child payment was traditionally available for up to four

children, but in 1992 this was reduced to two (Lund 2008: 15). The second part was a caregiver

allowance with a maximum pay-out of R430 (US$78) per month (Haarmann 1998: 81).

However, this component featured another, extremely rigid means test. For every R1 (US$0.18)

earned above R258 (US$47) per month, 50 cents (US$0.09) was deducted from the maximum of R430

(US$78). This continued until the minimum pay-out level of R90 (US$16) was reached (Haarmann

1998: 144). This meant that the total value of the grant for a caregiver with a single child varied

from R225 (US$40) to R565 (US$102) per month. Qualification for the caregiver component was

further predicated upon an applicant being a single parent with a child falling within the appropriate

age range. An applicant was not eligible if they received private maintenance and it had to be proven

that an attempt was made to obtain private maintenance from the partner or other parent of the

child through the private system.

In 1995/96, South Africa’s welfare budget totalled R13.4 billion (US$3.1 billion) (RSA 1995b:

4.10). The SMG’s share of the expenditure on social grants was R1.2 billion (US$279 million)

— approximately 9 percent of the total welfare budget (Haarmann 1998: 83). It represented 84

percent of total social assistance spending directed at child and family care, indicating its undisputed

position as the most significant intervention in that regard (Lund 2008: 14). Calculations for the

144
projected costs of maintaining the existing SMG system but extending it to the entire eligible

population through deracialisation and equalisation ranged from R12 billion (US$2.7 billion) to

R13.7 billion (US$3.2 billion) (Lund 2008: 18; Haarmann 1998: 88). The latter total would have

exceeded the country’s entire welfare budget.

A vital de facto feature of the SMG was that it continued to be tremendously skewed in racial

terms. Indeed, ‘black children [were] almost entirely outside the system, for the few who [did]

receive grants, the average grant size [was] much smaller than for the other races’ (Dlamini &

Simkins 1992: 70). Additionally, the system of providing the SMG had become so fractured that

the impression existed that ‘Pretoria really did not have control over what was being implemented

with regard to this particular grant’ (Lund 2008: 16). Dysfunction, uncertainty and ambiguity

prevailed. The implication of the priority assigned to providing the SMG principally to whites,

Indians and coloureds while severely neglecting blacks, combined with the administrative chaos,

was that although black children were by far the poorest and most vulnerable, very few of them

actually received the SMG. This was in contrast to coloured and Indian children, who were less

impoverished but constituted the biggest beneficiary pool. Some white children also qualified and

got the grant, but due to much lower poverty levels among this group the demand was simply not

very high.

Statistics from 1993 (the last year for which racially disaggregated welfare spending data is

available) indicates that 5 percent of coloured children, 4 percent of Indian children, 1.4 percent of

white children and only 0.2 percent of black children were in receipt of the SMG (Lund 2008 as cited

in Woolard & Leibbrandt 2010: 8). This was despite the fact that out of all impoverished people

in South Africa in 1996 (defined as less than R250 (US$45) per month) 91 percent were black, 7

percent were coloured, 1 percent were Indian and 1 percent were white (Leibbrandt, Woolard &

Woolard 2007: 15).

When this fact is interpreted in conjunction with figure 6.6 below, it meant that even though

145
the vast majority (91 percent) of poor individuals in South Africa were black, this group was only

receiving about 1.8 percent of all funds allocated through the SMG. There was thus a complete

mismatch between those in need of child-directed social assistance and those actually receiving it.

This is vividly illustrated in figures 6.6 and 6.7, with figure 6.8 showing that the situation persisted

for multiple decades. These features meant that the SMG was by no means a functionally effective

programme.

70

60

50
Percentage

40

30

20

10

0
Black Coloured Indian White

Figure 6.6: Poverty Headcount Ratio by Population Group at R250 (US$45) per month (1996) (Source: Leibbrandt

et al. 2007)

Beyond the desire to remove these racially discriminatory elements which were severely limit-

ing the grant’s ability to combat poverty among the majority of children, the Committee further

concluded that the SMG’s effects were being diluted through operational assumptions that were

inappropriate in the South African context (Woolard & Leibbrandt 2010: 11). As discussed in

section 6.2.1, it was originally designed in the 1920s to protect white family life (Lund 2008: 15).

The focus was very much on single mothers who were widowed or had been abandoned by their

spouses and excluded all other potential family scenarios. The inference of this approach was that

the SMG assumed a nuclear family configuration where both a father and mother were present to be

146
 50

45

40

35

30
Percentage

25

20

15

10

0
Black Coloured Indian White

Figure 6.7: Percentage of SMGs received by Population Group (1993) (Source: Author’s calculations based on

figures provided by Lund 2008 as cited in Woolard & Leibbrandt 2010: 8)

the norm, with only those ‘fringe’ cases involving widowhood or disability requiring state support.

Moreover, it clearly reflected a belief in the prevalence of near-full employment where access to

paid work was typical, as evidenced by the exclusion of (paternal) unemployment and low wages as

valid grounds for qualification. It assumed that unemployment and poverty was not a widespread

problem and that for a (white) family to be caught in such a position was somewhat irregular. It

thus clearly reflected the assumptions that — initially only within the white community — families

largely consisted of a married father and mother living together and raising children; unemployment

and poverty was not a widespread problem; and that support was largely needed when men were

unable to provide for their families due to abandonment or disability.

Comparing these underlying assumptions to the reality of the situation throughout all communi-

ties in South Africa during the 1990s, the Committee concluded that they were entirely inappropriate

and severely limiting the SMG’s effectiveness (Lund 2008: 80). The majority of households living in

poverty often contained three or four generations, both in rural and urban areas. Poorer households

thus tended to be much larger than richer ones (RSA 1996b). The legacy of the migrant labour

147
 60

Number of SMGs per 1 000 children

50

40

30

20

10

0
1960 1965 1970 1975 1980 1985 1990
Year

Total White Coloured Indian Black

Figure 6.8: Number of SMGs per 1 000 children aged 0-17 (Source: Van der Berg 1996 as cited in RSA 1996b and

Haarmann 1998)

system was also evident in these households, as the ‘middle’ generation was often missing due to

the father or mother working in a different area or having died (RSA 1996b).

This problem was especially pronounced in the case of fathers, although the precise extent was

unknown.96 Large numbers of men established dual households: one in the traditional rural area and

one in the urban centre where these labourers went to work, resulting in the eventual establishment

of two permanent households (RSA 1996b). Subsequently, in 1996 the proportion of children with

absent fathers who were still alive stood at 42 percent (SAIRR 2011: 2). Moreover,

a frequent occurrence [was] that a mother has a child with one partner, out of a formal union.

When she has a child later with another man, he will not agree to care for another man’s child.

She leaves the child or children with her own mother, or aunt, or other traditional guardian

(RSA 1996b).

The outcome was that by 1995 less than half of all black children under seven years old lived
96
There are various technical problems related to the process of capturing data related to this subject in surveys,

including reasons for non-disclosure.

148
with both parents, and fully 20 percent of South African children were not living with either of their

parents (Budlender 1996; SALDRU 1993). A very significant proportion of children were thus not

being continuously parented by either or both of their parents (RSA 1996b). Household boundaries

were extremely fluid, with family members and children themselves being moved around in search

of better opportunities. While it is possible that caregivers other than parents may be able to

provide quality care, the fact is that there was ‘a sequence of different caregivers, and generations

of children [were] brought up with no parental role models’ (RSA 1996b).

Another prominent shortcoming of the SMG’s assumptions about family structure was a cultural

one. The fact that high numbers of children were born outside formal partnerships was extremely

problematic with regards to the grant’s supposition that only nuclear family structures were legit-

imate. The existence of different cultures and customs where betrothal is not necessarily an event

but a process during which a child may be born; the birth of children culturally considered to be

entirely legitimate into formal polygamous relationships; and the side-by-side existence of ‘Western’

and traditional values and systems of paternal obligations within the South African context meant

that the entire concept of ‘illegitimacy’ was problematic (RSA 1996b).

The members of the Lund Committee were thus convinced that the SMG’s exclusive focus on

nuclear family structures stigmatised other household forms by not incorporating such structures

into the benefit, limiting its overall effectiveness (RSA 1996b). They also deemed it to be inappro-

priate due to the fact that it regarded problems of unemployment and poverty to be fairly limited.

Statistics from the time reveal that in 1994 the official unemployment rate stood at 20 percent, one

of the highest figures in the world (World Bank 2013a). Similarly, the data reveals that in 1993 the

overall poverty rate was 24.3 percent (World Bank 2013b). Overall, economic and social deprivation

was clearly widespread. The outcome was that the Committee not only wanted to deracialise the

system of child support; they also desired a move away from the misguided assumptions that formed

the basis of the SMG and led to its ineffective operation (Lund 2008: 24).

149
 The first set of recommendations produced by the Lund Committee clearly reflected this. The

proposals contained in what came to be known as the Itala Agreement followed from the Itala think

tank, which was held at a game reserve in Kwazulu-Natal during May 1996. In addition to the mem-

bers of the Committee, this gathering brought together a wide range of civil society stakeholders97

and would prove to be a ‘defining event in the Committee’s work and in the policy reform’ (Lund

2008: 32). The aim of the think tank was to ‘have a concentrated time of learning and discussion,

the outcome of which would be decisions about the main policy options and recommendations’

(Lund 2008: 32).

The Itala Agreement was thus a pure articulation of the functional reform preferences held by

these civil society actors. The Agreement called for a child benefit available to all children below a

defined age to replace the SMG (Lund 2008: 130). It also contained a proposal to pay the new child

benefit to the ‘primary caregiver’, as well as statements calling for ‘no means testing because all

the logic goes the other way’ (Lund 2008: 130). This was further underscored through the stated

desire to link the grant ‘with participation in some form of health activity’ (Lund 2008: 130). The

document also called for the continued implementation of the FCG and the CDG.

A mere two weeks after the Itala think tank, the government formally introduced the Growth,

Employment and Redistribution (GEAR) framework to guide its macroeconomic policy. GEAR

was a succinct statement of state actors’ preference for fiscal restraint, as it essentially insisted that

it was largely up to the private sector to lead economic development in South Africa. Deep cuts

in government spending were advocated, with the implication that social service delivery budgets

needed to be reprioritised in order to address the claims of the poor. At the same time, those social
97
These stakeholders included the members of the Committee, as well as ‘the Development Bank’s social policy

analyst Benny Mokaba; former Black Sash pensions worker Marj Brown, who was then co-opted on to the Committee,

paediatrician Neil McKerrow, for his specialist knowledge and activism around the AIDS epidemic and children;

economist John Kruger, for his knowledge of [the SMG] and welfare in general; and economist Charles Simkins, with

his broad knowledge of economic and social policy’ (Lund 2008: 32-33).

150
services that could be undertaken more effectively by the private sector or could not be provided at

all, such as social assistance grants to impoverished children, were to be scaled down or eliminated

(Visser 2004: 9). The consequence of GEAR’s policy orientation was that economic concerns were

to crowd out social welfare concerns in cases where they were not aligned (Hölscher 2008: 116).

The formalisation of the state’s preference for fiscal restraint through the introduction of GEAR

on 14 June 1996 thus directly clashed with the Lund Committee’s desire to eliminate means testing.

In effect, the Committee’s proposal challenged the preference of state actors to apply fiscal restraint

(as they feared that such an approach would be significantly more expensive). Lund recounts that

the Deputy Minister of Finance directly told her that the political leadership would not ‘entertain’

any policy recommendations which did not ‘reform within the existing envelope’ (Lund 2008: 30).

A policy recommendation that did not take fiscal restraint seriously ‘would itself not be taken

seriously’ (Lund 2008: 90).

The Committee consequently recognised that their interactions with these powerful state actors

were exclusively premised on hierarchical direction, which created an array of possible veto points

due to the fact that their recommendations would have to ‘travel a political road’ (Lund 2008: 90).

Indeed, Committee members perceived themselves to be involved in ‘a battle to preserve the R1.2

billion allocation to welfare’ (Lund 2008: 86). This led them to emphasise

the functionality of the welfare system to economic growth within the GEAR framework [in order

to] respond to the need of rescuing (emphasis added) the most advanced features of the existing

system, like the [OPG] and children grants, from pressures inside the government, especially the

Treasury, demanding more substantial cutbacks (Barchiesi 2005: 346).

This distress strikingly illustrates just how unevenly power was distributed during the process, as

it was anticipated that state actors had the ability to block the entire reform initiative — and

scrap the practice of state child support altogether — if their preference for fiscal restraint was

meaningfully challenged.

The outcome was the Committee’s calculation that they would have to ‘strategically. . . work

151
within fiscal constraints’ (Lund 2008: 30). This situation led to a vivid confrontation between the

different actors during a meeting involving Committee members and government lawyers. Amidst

appeals by the Committee that the desire for scrapping the means test was the ‘result of a considered

and purposeful policy choice’, the reply of legal advisors was that ‘this is just going to be another

handout’ (Lund 2008: 87). In an expression of the hierarchical structure of interactions, state actors

proceeded to veto the proposal to scrap means testing and the Committee had to ‘[retreat]. . . into

a diminished scale of provision’ (Lund 2008: 87). The result was that the final report of the Lund

Committee, contrary to the preference expressed in the Itala Agreement, called for means testing.

The Committee’s final report was submitted to Cabinet on 5 March 1997 (Lund 2008: 128). Its

proposals argued for the replacement of the SMG by the CSG, which was to be

paid to the primary caregiver of a child according to a simple means test; it should be payable

from birth for a limited number of years, with the number of years used as a cost-containment

mechanism; a condition for receiving the grant should be the proper registration of the birth of

the child, as well as positive health-related activities (Lund 2008: 131).

It also recommended that the FCG and the CDG remain in place (Lund 2008: 131). The Lund

Committee’s tenure ended in March 1997 and the process of finalising the CSG’s implementation

details was taken forward by the Child Support Grant Task Team (CSGTT), which continued to

include some of the original members of the Committee (Lund 2008: 128). But the publication of

the Lund Committee’s final report was not the end of the policymaking road, as state actors again

deemed certain elements of these recommendations to be inconsistent with their preferences.

Principal among these was the call to create linkages between the CSG and the healthcare

system. Although the Lund Committee was careful to point out that they were not in favour of

conditionalities where beneficiaries could be cut-off from the grant, they did propose that the CSG

be linked to the healthcare system as an incentive (Lund 2008: 68; Interview 7). This was to be

achieved through suggesting that the primary caregiver needs to be in possession of a Road to

Health Card. This card was already being used by 75 percent of caregivers as a way to monitor

152
for signs of under-nutrition in children between zero and five years old (RSA 1999b). A further

possibility was to promote child immunisation by taking the child to a healthcare centre around the

age of 30 months (Lund 2008: 69).

In a reflection of their preference for functional efficiency, these proposals were intended to ‘draw

people administratively closer to the system’ (Interview 7). However, despite generally supporting

the CSG in Cabinet, the Minister of Health vetoed these suggestions; she was unwilling to make

any commitment for the cooperation of the health services in implementing the programme, thereby

rejecting even these modest proposals for linking the CSG to health (Lund 2008: 70). The reason

for the vetoing of this suggestion was state actors’ perception that such measures would leave them

open to possible constitutional challenge, reflecting their preference for taking a strict rights-based

and unconditional approach to social assistance provision (Interview 11).

The regulations which were finally passed as part of the Welfare Laws Amendment Act in 1997

to enable the introduction of the CSG and replace the SMG were thus far removed from the initial

functional preferences expressed by civil society actors in the Itala Agreement (RSA 1997c). Despite

the fact that the CSGTT experienced some very significant challenges during the initial process of

implementing the CSG,98 its fundamental features had become clearly discernible by the end of

the Mandela administration’s tenure in 1999. The CSG was to be a means tested grant, initially

available to children between the ages of zero and seven (up to their seventh birthday). Instead of

being linked to the healthcare system through ‘soft’ conditionalities, the CSG was to be entirely

unconditional. It would also take a strictly individualistic approach to the issue of orientation,

premised on the notion of the primary caregiver.

By 1999, the South African social assistance system thus provided support to impoverished

children between the ages of zero and seven; disabled children between the ages of zero and 18; foster

children between the ages of zero and 18; disabled adults; and impoverished elderly people above the

age of 60/65. Table 6.4 summarises the coverage of this productive policy regime which continued
98
See Lund 2008: 72-75 for a description of these initial challenges.

153
to be premised on limited provision (by excluding all but the ‘deserving poor’), unconditionality

and an individualist orientation to defining beneficiaries.

Name Year Created Target

Child Support Grant 1997 Impoverished Children (0-7)

Care Dependency Grant 1992 Disabled Children (0-18)

Foster Child Grant 1983 Foster Children (0-18)

Disability Grant 1936 Impoverished Disabled (18-65)

Older Persons Grant 1929 Impoverished Elderly

(60+/65+)

Table 6.4: South African Social Assistance in 1999

6.6.2 Attempts at Comprehensive Reform (1999-2009)

The phasing out of the SMG and the implementation of the CSG was shortly followed by the

establishment of an interdepartmental task team convened by the national DSD in order to review

the social security system in the country. It identified a number of inefficiencies in the system, and

subsequently recommended the establishment of another national committee composed of external

experts to ‘investigate a move to a comprehensive and integrated social security structure’ (RSA

2002: 2). This led to the creation of the Committee of Inquiry into a Comprehensive System of

Social Security for South Africa (hereafter ‘Taylor Committee’ and ‘the Committee’), chaired by

Professor Vivienne Taylor from the University of Cape Town’s Department of Social Development.

The Taylor Committee was to carry out the most comprehensive review of social security provi-

sion ever undertaken in South Africa. It was composed of an astounding array of leading national

and international experts,99 which meant that the Committee’s proposals amounted to a compre-

hensive statement on the preferences of these external civil society actors. With regards to social

assistance policies, the Committee’s terms of reference called for them to evaluate ‘the entire social
99
See RSA 2002: v-viii for a full list of participants.

154
assistance mechanism including all grants, their funding mechanisms, and the efficiency with which

they achieve their goals’ (RSA 2002: 2). After two years of work, the Committee submitted its final

report in March 2002.

It contained a number of recommendations on reforming the social assistance system. Principal

among these was the call for the introduction of a universal BIG in the amount of R100 (US$10)

per month ‘as a means of providing social security to all and alleviating poverty’ (Brockerhoff 2013:

25). The Committee’s functional rationale was based on the belief that the BIG would overcome the

fact that the UIF covered less than 40 percent of the total labour force and a mere 6 percent of the

unemployed population. It was therefore essentially proposed as an effective solution that enabled

the inclusion of the unemployed and those working within the informal economy in the country’s

social assistance system (Brockerhoff 2013: 25). The report made it clear that the recommendation

to implement a BIG unambiguously reflected the preference of civil society actors for designing the

most functionally effective system.

Their analysis indicated that the BIG had ‘the potential, more than any other possible so-

cial protection intervention, to reduce poverty and promote human development and sustainable

livelihoods’ (RSA 2002: 62). It pointed towards the benefits of such an approach, including a forti-

fication of the poor’s ability to manage risk, the fostering of multiplier effects related to improved

self reliance, as well as strengthening overall societal cohesiveness (RSA 2002: 62). The BIG was

thus directly aimed at plugging ‘the coverage gaps within South Africa’s social [assistance] system’,

thereby making it ‘a general social assistance grant for all South Africans’ (RSA 2002: 326-327).

The Committee recommended that the BIG be financed through progressive tax reforms (RSA

2002: 327).

The recommendations proposed that the BIG be phased in over a period of 13 years, from 2002

to 2015 (RSA 2002: 65). The first step in this process involved changing the name of the CSG

in order for it to be called an ‘income support grant’, thereby reflecting the desire to establish

155
universalism (RSA 2002: 343). This remodelled CSG was then to be expanded to cover all children

up to the age of 18, before it was transformed into a BIG that made social assistance available to

destitute people of all ages. After extensively examining the effect of means testing, the Committee

also called on the state to ‘simplify the means tests where they obstruct equity, administrative

justice and are costly to implement’ (RSA 2002: 64).

But members of the Committee were well aware that national state actors did not necessarily

share their preferences, and that they operated in an environment where interactions were premised

on hierarchical direction. In recognition of this fact, the Committee’s report noted that these

national state actors could instead prefer to maintain the status quo

based on the view that...income poverty interventions cannot be accommodated due to inflexible

fiscal constraints...There will probably be considerable support for this option, particularly from

those mainly concerned about possible financial implications of any proposed interventions. The

Committee, however, is of the view that this would be an ultimately short-sighted position (RSA

2002: 63).

It turned out that these fears were well-founded. The Committee’s reform suggestions amounted

to an attempt at fundamentally transforming the deeply embedded institutional norm of providing

only limited social assistance coverage in South Africa. This directly clashed with the preferences

of national state actors.

Given their position within the institutional context, these state actors continued to have very

little incentive to radically reform the country’s productive approach to social assistance provision.

Their political power was completely consolidated, having won more than 66 percent of the vote in

the most recent national election (AED 2014). They also perceived the suggestions of implementing

a universal BIG and removing the means test as being completely at odds with their preference

for fiscal restraint. Instead of adopting these measures aimed at relatively greater redistributive

decommodification, the ANC’s 2002 national policy conference ‘placed a major emphasis on the

creation of short-term employment in the form of public works programmes for the sake of “pride

156
and self-reliance of communities”’ (Barchiesi 2006: 16).

In his 2003 State of the Nation Address, President Mbeki

reaffirmed that the task of his government is to ‘reduce the number of people dependent on

social welfare, increasing the numbers that rely for their livelihood on normal participation in

the economy’, while Minister of Finance, Trevor Manuel, lambasted the BIG as an ‘unsustainable’

and ‘populist’ idea. ANC ideologue and government spokesperson Joel Netshitenzhe argued that

the best form of social inclusion for South Africans remains to ‘enjoy the opportunity, the dignity

and the rewards of work’ (Barchiesi 2006: 16).

In a further reflection of the institutionally-derived power configuration, the Taylor Committee’s

proposals were ‘essentially forestalled by the intensifying hostility of the government and the ANC’

during the remainder of the Mbeki presidency (Barchiesi 2006: 15).

The only element of the recommendation that was ultimately adopted was the gradual extension

of the CSG to children older than the age of seven — to nine years old in 2003; 11 in 2004; 14

in 2005; 15 in 2008; and 18 in 2009 (DSD et al. 2012: 2). This continued to reflect the state’s

preference for a rights-based approach, as the country’s legal framework defined a child as ‘a person

under the age of 18 years’ (RSA 2005). Not providing the CSG to children older than seven thus left

the state vulnerable to constitutional challenge. This was the only part of the recommendations to

be adopted because it was the only reform suggestion which aligned with the preferences of national

state actors. The rest of the Taylor Committee’s social assistance policy suggestions were all vetoed.

State actors saw this preference for a rights-based approach and the concomitant fear of Consti-

tutional Court challenge strengthened in 2006, when the government was sued for gender and age

discrimination in terms of the OPG. Up until the point of the court challenge, the OPG had been

provided, since 1937, to women from the age of 60, and for men from the age of 65. In Roberts

and Others versus the Minister of Social Development and Others, the Constitutional Court ruled

that this age discrimination violated section 27 (1) (c) of the Constitution on the right to social

assistance, and consequently struck is down (Seekings 2008: 49). As admitted by the DSD, this

157
served to further emphasise the preference for narrowly following the rights-based approach of the

Constitution (at least when it came to the ‘deserving poor’), as ‘Constitutional Court rulings have

made clear’ (Seekings 2008: 37).

This period additionally marked the start of the government’s embrace of public works pro-

grammes, as the EPWP was also formally adopted in 2003. This occurred ‘despite the fact that the

Taylor Committee had rejected the idea of public works programmes as ill-suited for tackling South

Africa’s unemployment problem’ (Brockerhoff 2013: 27). State actors thus explicitly rejected the

Taylor Committee’s proposals for universalism in social assistance policy in favour of intensifying

wage-centred commodification through public works schemes. This

story of the struggles and debates on the BIG in post-apartheid South Africa reveals to what

extent the government’s wage-centred social policy has marginalised fiscal transfers and social

grants [which are independent from] individual employment, despite the many signals on the

growing inadequacies of wages to keep citizens out of poverty (Barchiesi 2006: 17).

By the end of the two Mbeki and brief Motlanthe100 administrations in 2009, the country

had thus strongly reaffirmed its commitment to a productive social assistance regime. Despite the

challenge mounted to the underlying norms of this system by external civil society actors in the shape

of the Taylor Committee, interactions based on hierarchical direction with powerful national state

actors and their conflicting preferences meant the even stronger entrenchment of limited coverage,

unconditionality and an individualist orientation.

100
Following the ANC’s ouster of Mbeki at its Polokwane conference, Kgalema Motlanthe was South African Presi-

dent from 25 September 2008 to 9 May 2009 as a ‘placeholder’ for Jacob Zuma.

158
 Name Year Created Target

Child Support Grant 1997 Impoverished Children (0-18)

Care Dependency Grant 1992 Disabled Children (0-18)

Foster Child Grant 1983 Foster Children (0-18)

Disability Grant 1936 Impoverished Disabled (18-65)

Older Persons Grant 1929 Impoverished Elderly (60+)

Table 6.5: South African Social Assistance in 2009

6.6.3 Consolidation (2009-present)

The period since Jacob Zuma assumed the presidency in 2009 has been characterised by the consol-

idation of this system. It has featured no further significant practical policy changes, and has been

primarily marked by a continued growth in take-up rates and enhancing administrative efficiency,

particularly in terms of the role of SASSA as the implementing agent. One interesting development

that did however take place during this period is related to a somewhat schizophrenic ideational

shift by the DSD on the issue of conditionalities.

Regulations introduced on 31 December 2009 stipulated that beneficiaries of the CSG must

ensure that children between the ages of seven and 18 are enrolled at and attending school (RSA

2009). It further called for beneficiaries to provide ‘proof of school or an educational institutional

enrolment and attendance’ within one month of their initial participation in the programme, as well

as the submission of their child’s signed school report card to the DSD every six months. In the

case of a CSG beneficiary’s child not attending school, the regulations required that ‘appropriate

steps’ be taken to ensure that the child is enrolled at and attending school (RSA 2009).

As noted at the time by the Deputy-Director General of the DSD, these ‘provisions are not of

a punitive nature. Where a child does not attend school, the grant is not going to be stopped’

(PMG 2010). Despite the language contained in the regulations clearly appearing to approximate

conditionality, he also insisted that ‘condition is too strong a word, but [this is] rather a provision

159
that provides for an incentive that caregivers ensure that their children attend school’ (PMG 2010).

The Minister of Social Development at the time, Edna Molewa, further emphasised this by stating

that ‘proof of school enrolment is not a condition for the Child Support Grant’.

In a turn of phrase that echoes the sentiments of one of the early architects of the Brazilian

system,101 the Deputy Minister (and current Minister) Bathabile Dlamini argued that this provision

would ‘help us towards ensuring...that we use education as an instrument for fighting poverty’

(PMG 2010). The DSD had clearly warmed to the idea of educational conditionalities. But, as

Dlamini herself noted in response to a later parliamentary question, they soon found that they

were prohibited from actually implementing them because ‘the Constitution provides for the right

of access to social [assistance] without necessarily providing for a condition of school attendance’

(GCIS 2012b). The result was that the CSG remained unconditional, while the 2009 regulations

were not enforced (GCIS 2012b; DSD 2012b).

In 2010, the Zuma administration launched yet another central government commission, the Na-

tional Planning Commission (NPC). It was tasked with taking a ‘broad, cross-cutting, independent

and critical view of South Africa’ in order to produce clear recommendations for government (NPC

2014). This mandate culminated in the publication of the National Development Plan (NDP) in

November 2011, which sought to offer a ‘vision for 2030’ in ‘the process of charting a new course for

[the] country’ (NPC 2011: 1). In sharp contrast to both the Lund Committee in 1995 and the Taylor

Committee in 2002, the NPC was primarily composed of national state actors (chaired by Deputy

President Cyril Ramaphosa and Minister in the Presidency Jeff Radebe), while Vivienne Taylor

(the former chair of the Taylor Committee) was the only social policy expert on the Commission.

The NPC thus had much less political independence than either of its predecessors. Its mandate

was clearly also much broader. This came to the fore vividly in terms of the NDP’s recommendations

regarding social assistance policies. It noted the functional problem that

there [is a] critical gap [in the South African social assistance system, through] the lack of
101
See section 7.5.1.

160
 protection for many working-age people. . . For those who are willing and able to work, but who

are locked out of the economy, there is no meaningful level of social protection (NPC, 2011: 331).

But despite reaching this conclusion, neither the BIG nor any other of the Taylor Committee’s

proposed changes to the social assistance system are even mentioned in the document’s 430 pages.

Instead, it again emphasised the government’s commitment to the EPWP due to its view that

it ‘provides people with...the dignity of being productive, rather than dependent...[which allows]

unemployed people to become a productive part of the economy’ (NPC 2011: 334). The recom-

mendations contained in the NDP, which the ANC government regards to be its ‘blueprint’ for the

country’s future, thus clearly reflected the preferences of national state actors. While it is safe to

assume that certain members of the NPC (as well as external experts providing inputs) lobbied for

undertaking reforms that mirrored their own functionalist preferences, the policy recommendations

contained in the NDP make it clear that those suggestions were again vetoed by powerful national

state actors. The above quotation on the EPWP in fact neatly captures the essence of the country’s

productive approach to social assistance provision.

The system has thus undergone little real change during the Zuma administration, with minor

adjustments to benefit levels announced in the annual budget being the only practical modifica-

tions currently being made (Gordhan 2014). As indicated in table 6.6, the outcome is the policies

currently being implemented in South Africa: a productive social assistance system that continues

to be structured around the provision of limited coverage, unconditionality and an individualist

orientation.

161
 Grant Name Target Group Age Bracket Monthly Value Beneficiaries

Child Support Impoverished Children 0-18 R300 (US$30) 10 898 923

(max 5)

Foster Child Foster Children 0-18 R800 (US$80) 478 781

Care Dependency Disabled Children 0-18 R1 270 (US$127) 119 575

Disability Impoverished 18-59 R1 270 (US$127) 1 208 301

Disabled Adults

Older Persons Impoverished 60+ R1 270/R1 290 2 938 693

Elderly Persons (US$127/US$129)

Total - - - 15 644 273

Table 6.6: South African Social Assistance in 2014 (Sources: RSA 2014; SASSA 2013)

6.7 Conclusion

This chapter has undertaken a thorough review of the development of social assistance policies in

South Africa. The analysis was carried out according to the ACI analytical framework, proceeding in

two phases. The first involved a detailed specification of the institutional context. The discussion

was initially guided by the study’s first independent variable, namely the institutionalisation of

specific social assistance norms. It showed that norms regarding limited coverage, unconditionality

and an individualist orientation were historically deeply entrenched and constitutionally codified

during the transition to democracy in the early 1990s. This was followed by an examination of state

structure as the second independent variable. Following an exposition of the formal institutional

policymaking structure and the practical distribution of political power, this section concluded that

South Africa features a highly concentrated state structure with social assistance policymaking

being entirely centralised.

This specification of the institutional configuration set the scene for the second analytical step,

which involved an examination of the ways in which the institutional setting introduced during

162
the first part of the chapter shaped both the preferences held by different groups of actors, as well

as their interactions. Finally, a discussion on the process whereby interactions between actor sets

within the social assistance policy community ultimately shaped policy outputs empirically traced

the development of social assistance policies in South Africa. The investigation indicates that policy

outputs were profoundly shaped — through the mechanism of path dependence and preference for-

mation — by the institutional configuration, which featured high levels of norm institutionalisation

and a highly concentrated state structure. It showed that, despite the existence of different possible

choices on multiple occasions, the state repeatedly preferred not to embark on alternative policy

paths.

Following the transition to democracy, the functional recommendations of numerous govern-

ment committees indeed repeatedly called for a shift towards precisely such alternative paths. But

through interactions premised on hierarchical direction, their proposals were consistently rejected by

powerful national state actors facing little incentive for reform by virtue of their operation within

a context of high norm institutionalisation and a concentrated state structure. In this way, the

analysis is thus able to account for the implementation of a productive social assistance system in

post-apartheid South Africa.

163
Chapter 7

The Development of Social Assistance

in Brazil

7.1 Introduction

This chapter extends the analytical endeavour to the Brazilian case study, with its core aim being

to account for the emergence of a protective social assistance system in the country. This entails

undertaking a thorough examination of the policy development process, guided by the analytical

approach of ACI. Figure 7.1 illustrates the analytical procedure to be carried out for the Brazilian

case study. This two-step process firstly entails a detailed explication of the institutional configu-

ration according to the study’s two independent variables. This information is subsequently used

to introduce actor preferences, followed by an analysis of the ways in which strategic interactions

between these actors transformed institutionally-derived preferences into tangible policy outputs.

Figure 5.2 outlines the analytical structure of the chapter.

Examining the institutional context initially involves addressing the first independent variable

by discussing the historical emergence and institutionalisation of social assistance norms in Brazil.

The discussion highlights the fact that such programmes were not deeply entrenched by the time of

164
 Institutional Setting
(1) Norms weakly Institutionalised
(2) Diffuse State Structure

Problems Actors Constellation Modes of Interaction Protective

- Preferences - Subnational State - Negotiated System
- Capabilities Actors Agreement
- Federal State
Actors
- Civil Society Actors

Policy Environment

Figure 7.1: Actor-Centred Institutionalism – Brazil (Source: Adapted from Scharpf 1997: 44)

IV1:
Institutionalisation of
Social Assistance
(Low)
Constitutional- Preferences of Interactions Policy
Legal Policy Actors between Actor Sets Outputs
Framework (Protective
IV2: Concentration of System)
State Structure (Low)

Figure 7.2: Explanatory Framework – Brazil

the country’s transition to democracy, producing relatively weak path dependence effects and pro-

viding policymakers with a comparatively clean institutional slate. This is followed by a subsection

analysing Brazilian state structure as the second independent variable. Through an examination

that is focused upon both the formal policymaking structure and the practical distribution of po-

litical power, the discussion shows that Brazil featured a diffuse social assistance policymaking

structure. The result is that policymaking capacity was decentralised.

The final section carries out the second step in the ACI procedure by investigating the ways

165
in which interactions premised on negotiated agreement between the relevant actor sets within

the Brazilian social assistance policy community led to the emergence of the policy outputs that

constitute the contemporary social assistance system in that country. This is achieved through

empirically tracing the process of policy development. Through an examination of the decisive role

played by institutional structure, the chapter is thereby able to account for the emergence of a

protective social assistance system in Brazil following democratisation.

7.2 Weak Institutionalisation of Social Assistance Norms

The initial part of this section undertakes an overview of the provision of social assistance in Brazil

prior to democratisation in the late 1980s. The key insight produced by the analysis is that spe-

cific institutional norms regarding social assistance were only weakly entrenched in the Brazilian

political economy. Indeed, prior to the 1970s, non-contributory social assistance programmes were

wholly absent in the country, and it was only during the final decades leading up to democratisation

that such policies slowly began to emerge. This historical examination is followed by an analysis

of the legal norms constitutionally codified during the democratic transition. The section is there-

fore aimed at addressing the first independent variable by analysing the historical emergence and

institutionalisation of social assistance in Brazil.

7.2.1 Social Assistance before 1990

By the time that Fernando Collor became Brazil’s first directly elected democratic President since

the military seized power in 1964, the country had followed a very different historical trajectory from

South Africa in terms of the development of its social welfare regime. In keeping with the general

trend throughout Latin America, Brazilian social welfare was historically designed according to

Bismarckian principles (Schwarzer & Querino 2002; Amann & Barrientos 2014). In effect, this meant

that social security provision was overwhelmingly dominated by the construction of contributory

166
social insurance schemes, while social assistance provision was practically non-existent (Schwarzer

& Querino 2002: 3).

The initial part of the examination reveals the historical lack of development of social assistance

in Brazil where, prior to the 1970s, social assistance did not feature on the policy agenda at all.

This is followed by a discussion on the introduction of the first de facto social assistance scheme by

the military dictatorship, the Renda Mensal Vitalı́cia (RMV – Lifetime Monthly Income), which

targeted impoverished people over the age of 70. The subsection thus traces the historical process

in order to illustrate the way in which social assistance norms were only weakly institutionalised

by the time of the transition to democracy. In turn, this state of affairs played a decisive role

in shaping the institutional setting which ultimately led to the construction of the contemporary

Brazilian system.

Welfare without Social Assistance

As is the case in much of the colonial world, a dearth of historical data on the provision of social

security in Brazil prior to the arrival of European colonists does not necessarily entail a complete

absence of strong informal institutions. But it does mean that this review must commence with the

arrival of Pedro Álvares Cabral and his fellow Portuguese settlers in 1500. In general, the colonial

and imperial periods102 in Brazil were characterised by the way in which the provision of care for the

poor was considered a matter of Christian charity, particularly through the introduction and growth

of charitable hospitals known as Santas Casas de Misericórdia (Hospitals of Mercy) (Schwarzer &

Quirino 2002: 2).

In this case it was largely the Roman Catholic Church that provided ad hoc poor relief, with the

first organised welfare scheme during the 17th and 18th centuries being an insurance plan granting
102
The colonial period lasted from 1500 until 1808, when the Portuguese royal court fled from Napoleon’s invasion

and moved the imperial government to Rio de Janeiro. This inaugurated the imperial period, which featured the

declaration of independence by Dom Pedro I in 1822 and lasted until the republican coup of 1889.

167
pensions to the dependent widows and children of deceased navy officers (MPS 2009: 7). A decree

issued by Dom Pedro I in 1821 is recognised as the ‘first legal text that registered the issue of social

welfare in Brazil’ (MPS 2009: 7). But concern with assisting the poor remained almost entirely

absent during the transition from a society based on labour coercion and slavery (which lasted until

1888) to a system of waged-based employment (Schwarzer & Querino 2002: 2).

Where ad hoc arrangements did exist, they were frequently ‘reduced to a scheme of local poverty

management with patronising characteristics’ (Schwarzer & Querino 2002: 2). The rupture with

the Old Republic and the emergence during the 1930s of the Estado Novo (New State), with its

more centralised power structure, allowed for the implementation of national (instead of regional

and ad hoc) policies (Driabe, Castro & Azeredo 1995: 3). The first modern social protection

law institutionalising social insurance was the Eloy Chaves Law, introduced in 1923 (MPS 2009:

7; McGuire 2010: 159). This law initially granted contributory pension and health coverage to

railroad workers and certain other private sector employees.

The following decade saw this insurance system evolve into a ‘very fragmented, company-based,

low coverage set of isolated programmes, operating under a full capitalisation regime’ (Beltrão,

Pinheiro & Oliveira 2002: 3). The fledgling system was eventually consolidated into sector-wide

institutes during the first Vargas period of the Estado Novo (1930-1945) (Beltrão et al. 2002: 3;

Schwarzer & Querino 2002: 2). These moves coincided with initial efforts to undertake industrial-

isation, as well as with the tentative emergence of the labour movement (Beltrão et al. 2002: 3).

Although this movement was often repressed, members of the new urban working classes neverthe-

less joined trade unions and pushed the federal government throughout the 1920s, 1930s and 1940s

for social security provision. The resulting programmes were all based on contributory insurance

principles (Schwarzer & Querino 2002: 3).

Schwarzer and Quierino further conclude that

the concentric expansion of social security coverage under the populist governments of those

decades had much to do with the political power enjoyed by a particular social/professional

168
 group or with the potential threat it may have represented (2002: 3).

In a similar vein, Hunter notes that during the corporatism of the Vargas period social policy’s foun-

dation rested upon occupational status, and was segmented accordingly (2014: 18). By providing

benefits exclusively to the civilian and military bureaucracy, as well as to the most well-organised

and strategic sectors of the working class, the state revealed its motive: to pre-empt the development

of an autonomous and militant working class (Hunter 2014: 18).

By the end of the 1940s, this strategy resulted in there being ten times more insured mem-

bers than in 1934 (MPS 2009: 8). Despite this expansion, social insurance provision remained

qualitatively and quantitatively extremely unequal throughout the period — while social assistance

continued to be entirely absent (MPS 2009: 8). In what can be regarded as a striking indication

of the low priority assigned to social security provision in general, vast resources from the social

insurance institutes were used to finance President Juscelino Kubitschek’s vision of ‘fifty years of

prosperity in five’ during the 1950s. Through the ‘biggest real estate investment in their existence’,

these institutes effectively financed a significant portion of the costs related to the construction of

the new capital city, Brası́lia, from 1956 to 1960. This was done without any ‘guarantee of necessary

remuneration’ (MPS 2009: 9).

By the end of the so-called Second Republic in 1964, Brazil had thus constructed the very

rudimentary foundations of a welfare state. Unlike in South Africa, Brazil’s approach was entirely

based upon a Birmarckian system reliant on contributory social insurance. This meant that the

weakest social groups who lacked political power, including rural workers, the indigenous population,

as well as urban informal and domestic workers, were effectively not considered for social security

coverage (Schwarzer & Querino 2002: 3). The result was that social assistance was totally absent,

with no support available for people who were locked-out from the formal, contributory system. This

also meant that no specific institutional norms had emerged in terms of social assistance provision

during this period.

169
Military Dictatorship and a Tentative First Step: The RMV

It was only following the 1964 coup d’état establishing the military dictatorship that a comprehensive

Lei Orgânica da Seguridade Social (LOSS – Organic Social Security Law) was adopted (FRB 1964).

After 14 years of debate in Congress, this law finally unified the costing and benefit schemes across

the various social insurance institutes (Beltrão et al. 2002: 4). This was followed by the creation

of the Instituto Nacional de Previdência Social (INPS – National Social Security Institute) in

1966 to unify and consolidate the management of social insurance by incorporating the revenues,

expenditures, assets and liabilities of the respective institutes (MPS 2009: 10; Beltrão et al. 2002:

4). Although the majority of urban workers were theoretically covered by the system by the mid-

1960s, in practice coverage was still below 50 percent even for the wage-earning, employed population

(Beltrão et al. 2002: 4).

The focus throughout the 1970s was on the further administrative consolidation of welfare

provision with the founding and expansion103 of the Sistema Nacional de Previdência e Assistência

Social (SNAPS – National Social Welfare and Assistance System), which was aimed at integrating

the provision and maintenance of benefits, rendering services, funding activities and programmes,

as well as administrative and financial management (MPS 2009: 9). Steps to incorporate the rural

population into the social security system also commenced during this decade with the introduction

of the Programa de Assistência ao Trabalhador Rural (FUNRURAL – Rural Workers’ Assistance

Programme) in 1971. Effective insurance coverage for this group was achieved by 1974, with a
103
The period saw the further introduction of the Instituto Nacional de Assistência Médica da Previdência Social

(INAMPS – National Institute of Medical Assistance of Social Welfare), the Instituto de Administração Financeira da

Previdência e Assistencia Social (IAPAS – Institute of Financial Administration of Social Welfare and Assistance), the

Central de Medicamentos (CEME – Centre for Medication), the Empresa de Processamento de Dados da Previdência

Social (DATAPREV – Social Welfare Data Processing Company), the Fundação Nacional do Bem-Estar do Menor

(FUNABEM – Foundation of Minors’ Welfare) and the Legião Brasileira de Assistência (LBA – Brazilian Legion of

Assistance) (MPS 2009: 9-10).

170
benefit equal to half the minimum wage being made available to the rural disabled and workers

older than 65 (Beltrão et al. 2002: 6). The inclusion of this relatively marginal political group

signalled an important directional shift that would subsequently lead to the establishment of the

first programme resembling a social assistance intervention.

The establishment of FUNRURAL was thus shortly followed by the introduction of the RMV

in 1974 (FRB 1974). Assessing the nature of the programme is somewhat problematic, as it was

technically classified as a social insurance programme aimed at impoverished people over the age of

70104 (both rural and urban) who had made at least 12 contributions to the social security system

during their lifetimes (FRB 1974; MDS 2014a; Beltrão et al. 2002: 6). Beneficiaries had to earn

less than 60 percent of the minimum wage, and the RMV’s value was equal to half the minimum

wage (FRB 1974). The requirement for prior contributions was however rarely enforced, resulting

in the RMV being implemented in a way more akin to a social assistance benefit (Interview 12).

Following administrative adjustments in response to the intense fiscal and popular pressures of

the 1980s, Brazil’s social welfare system was further elaborated under the technocratic authoritarian

regime installed in 1964. These reforms defined the core of state intervention, the centralised appa-

ratus to support this intervention was established, the required funds and resources were aligned,

the principles and operational procedures were defined, as well as the rules of social inclusion and

exclusion (Driabte et al. 1995: 4). The initial establishment of a system entirely premised on social

insurance during the 1930s thus culminated in the consolidation of the Brazilian welfare state by

the mid-1970s.

Instead of granting social protection to all citizens, this arrangement was characterised by the

nearly exclusive extension of social protection to urban workers formally participating in the labour

market, as well as to segments of the corporatively organised middle class. It featured highly

fragmented access which was ‘associated positively with systems of force, bargaining, and privileges,
104
This high age threshold was very exclusionary as, by 1980, average life expectancy in Brazil was still only 63 years

(World Bank 2014b).

171
and negatively with low levels of universality and uniformity of social benefits’ (Driabe et al. 1995:

1). The military regime elaborated a system premised almost solely on social insurance, with the

only programme even resembling a social assistance scheme being the belatedly-introduced RMV.

The result was that specific norms regarding the operation of social assistance continued to be

largely absent.

Social Assistance at the End of Military Rule

This historical discussion highlights the way in which welfare provision in Brazil emerged within

its own particular historical political economic context. It was initially closely tied to the process

of state formation following the end of the Old Republic and the emergence of the Estado Novo in

the 1930s. The initial introduction of a purely Bismarckian insurance-based system was expanded

during the subsequent periods, particularly during the phase of military rule. By the late 1980s,

Brazil had very little experience with social assistance provision. A broad overview of the (limited)

development of the system prior to 1990 is introduced in table 7.1.

Period Note

1974-1990 Introduction

1974: Introduction of the RMV

1977-1985: Administrative adjustment

1985-1990: Progressive reformulation attempts

Table 7.1: Evolution of the pre-1990 Brazilian Social Assistance System (Source: Adapted from Driabe et al. 1995:

5)

The lack of social assistance therefore stands out when assessing the social welfare landscape

at this time. Table 7.2 shows that social assistance consisted exclusively of the RMV (which itself

was legally established as a contributory insurance scheme). This means that, in sharp contrast

to South Africa, no programme currently in existence in Brazil originated in the period prior to

172
democratisation. The fact is therefore that social assistance was not historically entrenched in

Brazil. Apart from the principle hesitantly embodied by the RMV that the impoverished elderly

and disabled were entitled to some modicum of de facto non-contributory support, social assistance

norms were not historically institutionalised. Constructing Brazilian lifecycle risks for 1990 in figure

7.3 confirms that there were serious coverage gaps in the system at that time.

Name Year Created Target Current Status

Renda Mensal Vitalı́cia 1974 Impoverished Elderly (70+) Defunct

Table 7.2: Brazilian Social Assistance in 1990

Birth Time Death

Education Phase (age Accumulation Phase Dependent Phase (age

0-18) Risks: (age 18-70) Risks: 70+) Risks:
Poverty/Education Poverty/Unemployment Poverty/Old Age
Funding Risk: Risk: Risk:

? ? 
Health/Disability Health/Disability Risk: Health/Disability


Risk: Risk:

 

Figure 7.3: Brazilian Lifecycle Risks in 1990 (Source: Adapted from Smith 2011)

When full democracy was established in 1990, Brazil had largely failed to construct a social

assistance system. Instead, through its own particular historical process primarily propelled by the

political economy of the Estado Novo and the military dictatorship, the country had built a welfare

system almost entirely predicated on social insurance. The general absence of social assistance

norms patently meant that there was subsequently much less potential for the manifestation of

path dependent institutional stickiness.

173
7.2.2 The Legal Codification of Social Assistance Norms in Brazil

This state of affairs went on to have an important impact on the subsequent development of social

assistance policies in a democratic Brazil. In a similar vein to South Africa, the 1988 Constitution

was to lay the basis for the construction of a new state. But instead of being constrained by a set

of deeply institutionalised social assistance norms, the drafters of the Brazilian Constitution had a

relatively blank slate to work with (Interview 12). This subsection reviews the relevant provisions

of the Constitution and the LOAS in order to illustrate the way in which they reflected this shallow

institutionalisation of norms, placing significantly less constraints on social assistance policymakers

than their South African counterparts.

The Constitution (1988)

Following a ‘lost decade’ in the 1980s marked by recession, stagflation and civil unrest featuring

demands by Brazilians to directly elect their leaders,105 the transition to democracy got underway

in earnest in 1985. The first move was the indirect election of Tancredo Neves as President by the

Electoral College, an institution that was then still dominated by the military (Zaverucha 1997:

5). In a dramatic turn of events adding further uncertainty to an already volatile situation, Neves

collapsed and died the night before his inauguration. He was consequently replaced by his running

mate and Vice-President, José Sarney. Sarney thereby became ‘the President of a government

tasked with leading the transition from authoritarianism to democracy’ (Zaverucha 1997: 7).

The government subsequently passed a constitutional amendment calling for elections for a

National Constituent Assembly (NCA). One of the leaders of the resistance movement, Ulysses

Guimarães, was elected to preside over the NCA, which was granted full powers to draft and enact

a new democratic Constitution. The NCA was in session from February 1987 until October 1988,

while the military was ‘able to negotiate the terms of its disengagement from the government with
105
Principally through the Diretas Já movement.

174
civilians’ (Zaverucha 1997: 7). During these 20 months, ‘Congress and Brası́lia was the centre of

Brazilian life’ (Souza 2005: 81). The Constitution was eventually ratified and adopted on 5 October

1988 by the NCA.

Similar to its South African correlative, the Brazilian Constitution is generally quite detailed

and articulates an array of social and economic rights (Souza 2005: 82). But it is again instructive

to examine the specific language when it comes to social assistance provision. Unlike the South

African Constitution, the Brazilian document draws a clear distinction between social insurance106

and social assistance. Title 8 of the Constitution deals with the ‘social order’, with section III being

solely dedicated to social insurance provision (FRB 1988). This is followed by section IV, where the

framework for social assistance is introduced. Instead of enshrining access to social assistance as a

fundamental right, article 203 simply states that ‘social assistance shall be rendered to whomever

may need it’, followed by a list of ‘objectives’ (FRB 1988).

The only social assistance right directly entrenched is in article 203 (V), which guarantees a

monthly benefit equal to one minimum wage to those elderly and handicapped who are able to

prove their incapability of providing for themselves or having support provided for by their families

(FRB 1988). The inclusion of this section in the Constitution represents an ‘evolution of the RMV’

and was eventually to lead to the creation of the BPC (Interview 12). It is no coincidence that

the only de facto social assistance programme that existed prior to the transition to democracy in

Brazil was directly carried over into the new Constitution. This serves to further affirm the power

of pre-existing institutional norms in shaping subsequent policy outputs.

Apart from the section guaranteeing the BPC, the Brazilian Constitution thus does not enshrine

the provision of social assistance payments as a fundamental right (Interview 12). The implication

is that the government has much greater discretion in designing the parameters of such programmes,

especially in terms of linking them to conditionalities. The Constitution does not guarantee general

access to social assistance, and leaves space for the denial of benefit payments through conditional-
106
Under the heading of ‘social security’ in section III of title 8.

175
ities (Interview 12).

The document also introduced significant measures to potentially insulate welfare spending from

discretionary budget cuts. Article 195 calls for the establishment of a separate social welfare budget

that is ‘not part of the budget of the Union’ (FRB 1988).107 As set out in section II, the social

welfare budget pertains to ‘all direct and indirect administration entities or bodies connected with

social security, as well as funds and foundations instituted and maintained by the government’ (FRB

1988). In short, ‘government actions in the area of social assistance shall be implemented with funds

from the social welfare budget’ (FRB 1988). These funds were to be collected from payrolls and

earmarked taxes on the gross turnover (Contribuição para o Financiamento da Seguridade Social

– COFINS) and net profits (Contribuição Social sobre o Lucro Lı́quido – CSLL) of enterprises, in

addition to other minor sources (Schwarzer & Querino 2002: 4).

The process laid out in the Constitution resulted in the Brazilian budget being very rigid.108 It

directly mandates a vast array of expenditures and earmarks tax revenues to specific uses, desig-

nating them as mandatory. The effect is to insulate a very large part of the budget from scrutiny

and potential cuts. In fact, estimates indicate that over 90 percent of the annual budget is subject

to such rigidities (Blöndal, Goretti & Kristensen 2003: 101-102). This situation is further compli-

mented by the abovementioned practice of tax earmarking spending on the social sectors (Blöndal,

Goretti & Kristensen 2003: 101-102). Most fundamentally, the Brazilian Constitution did not con-
107
Instead of a single integrated budget, the Constitution calls for the creation of three distinct budgets: ‘the fiscal

budget regarding the powers of the union, their funds, bodies and entities of the direct and indirect administration,

including foundations instituted and maintained by the government; the investment budget of companies in which the

union directly or indirectly holds the majority of the voting capital; [and] the social welfare budget, comprising all

direct and indirect administration entities or bodies connected with social security, as well as funds and foundations

instituted and maintained by the government’ (FRB 1988).

108
Part of the reason for this rigidity is itself historically rooted, as ‘during Brazil’s high inflation years, it was

essentially meaningless to have a specific appropriation in the budget in nominal terms. By linking expenditures with

a revenue source, it was possible to “insure” the expenditures against the effects of inflation, as the revenues would

rise in line with inflation’ (Blöndal et al.: 2003: 102).

176
tain any reference to the notion of ‘progressive realisation’ of rights and the implied primacy of

fiscal restraint.

The diversification of the financing of contributions and taxes was in fact one of the guiding

social policy principles of the Brazilian Constitution, constructed with the goal of ‘increasing the

stability of social policy funding’ (Schwarzer & Querino 2002: 4). The effect of this was to make it

harder for fiscally conservative actors to constitutionally justify reduced scales of provision, thereby

providing still greater policy development space. The Constitution had built-in mechanisms which

provided the potential to shelter social welfare spending. In sum, Brazilian policymakers were

not as severely constrained as their South African counterparts by constitutional notions of fiscal

discipline.

A final significant feature of the Constitution was the way in which it defined the family as

‘the foundation of society’, which was to ‘enjoy special protection from the state’ (FRB 1988).

Article 203 lists a range of ‘objectives’ for social assistance, including an explicit reference to ‘the

protection of the family’ (FRB 1988). Eligibility for the constitutionally guaranteed benefit (which

was to become the BPC) was predicated upon proof that the handicapped and elderly could not have

support ‘provided by their family’ (FRB 1988). This language clearly encouraged the conclusion

that ‘the family unit (rather than the individual) was the appropriate entity to receive the benefit

and should in turn bear responsibility for meeting programme requirements’ (Lindert 2006: 68).

The Brazilian Constitution thus outlines an institutional framework which differs in important

respects from the South African one. These differences primarily result from the fact that social

assistance norms were historically only weakly institutionalised in the Brazilian political economy

prior to the transition to democracy. This reduced the potential for path dependence effects,

resulting in the ‘representatives of the population that were writing the Constitution at that time

on the social rights chapter, and specifically on social assistance, [having] a blank paper that they

could fill in’ (Interview 12). The outcome was, with the exception of the principles introduced by

177
the RMV, the construction of an entirely new set of institutional norms.109

Organic Social Assistance Law (1993)

The enactment of the Constitution was followed by the publication of the LOAS in 1993 (although it

was only fully implemented from 1995 onwards) (Schwarzer & Querino 2002: 16). The LOAS further

elaborated on the principles introduced by the Constitution, meaning that it ‘regulates this aspect

of the Constitution and establishes standards and criteria for the organisation of social assistance’

(MDS 2009: 4). The way in which the provisions contained in the LOAS built upon relevant

constitutional provisions served to strengthen the emerging norms associated with a protective

approach to social assistance.

In this vein, it elaborated on article 203 (V) of the Constitution regarding the guarantee of a

monthly benefit equal to the minimum wage for impoverished handicapped and elderly persons by

mandating the creation of the BPC (FRB 1993: art. 20). In language very similar to that of the

Constitution, it called for the implementation of a

continuous benefit [in the form of] a guarantee of one monthly minimum wage for the handi-

capped and the elderly older than 70110 years or more who have no means to provide for their

own maintenance or have it provided by their family (FRB 1993: art. 20).

However, apart from the BPC, the LOAS did not directly guarantee any other specific social assis-

tance benefits, thereby continuing to leave the door open for the potential enactment of condition-

alities with regards to other programmes.

109
While this section conveys the point that the lack of pre-existing norms and their path dependence effects created

greater institutional space, this naturally raises an important question about the process whereby the new Constitution

was drafted and associated new norms created. See Hochstetler (1997) for an examination of the role played by social

movements in the construction of new norms and Souza (2005) for an analysis which regards them as a result of the

need to legitimate democracy.

110
This has subsequently been adjusted to 65 years or older.

178
 Instead, it articulated a set of general guiding principles. In addition to once more noting the

primacy of the ‘protection of the family’ and ‘respect for...the family’, the LOAS strongly emphasised

the ‘supremacy of meeting social needs above the requirements of economic profitability’ (FRB 1993:

art. 4). The combination of this approach with the Constitution’s relative insulation of social welfare

budgets meant that the provisions contained in these documents sharply diverged from the guidelines

introduced by the Constitution and WPSW in South Africa. The Brazilian documents de-link social

assistance from general economic considerations to a greater extent, thereby creating space for the

introduction of programmes aimed at relatively greater universalism and decommodification. They

ultimately illuminate an institutional path towards a more protective social assistance system in

Brazil.

7.2.3 Social Assistance Norms in Brazil

The preceding discussion has investigated the historical development of social assistance norms in

Brazil. It shows that these norms were weakly institutionalised, with only the principle of providing

some support to the elderly and disabled (through the RMV) emerging in the decades leading

up to the transition to democracy. This means that, in contrast to South Africa, norms related to

coverage, conditionality and orientation did not originate through a deep historical process. Instead,

the delegates of the NCA charged with writing the Constitution had a significant amount of leeway

in codifying an emerging set of principles.

The result was that the historical institutional context facing policy actors in Brazil allowed them

to operate within a policy environment where norms were only weakly institutionalised. Although

the Constitution favoured certain outcomes — the protection of the family primary among them

–– it provided a much greater degree of flexibility by not enshrining general social assistance as a

fundamental right, and by constructing mechanisms to insulate social welfare spending from the

government’s fiscal and investment budgets. The effect of this low level of institutionalisation was

to provide policy actors with a larger set of potential policy options in their efforts to, for the first

179
time in the country’s history, construct a social assistance system in Brazil from 1990 onwards.

7.3 State Structure

The second aspect of the institutional framework shaping the policy context in Brazil was the

country’s diffuse state structure. The following section examines this further institutional variable

by discussing the role played by the diffusion of political power in influencing the institutional

context and ultimately channelling policy outputs. Although the discussion initially centres on

the legal distribution of policymaking capacity as defined by formal aspects of state structure, the

analysis is also sensitive to the practical distribution of political power.

The first section therefore highlights the fact that, in terms of formal institutional and organ-

isational arrangements, social assistance policymaking capacity is highly decentralised in Brazil.

The subsequent discussion elucidates the distribution of political power in democratic Brazil and

serves to emphasise the high degree of power diffusion. Taken together, the analysis concludes that

in both legal and practical terms, Brazil features a diffuse state structure in the realm of social

assistance policymaking. The capacity to formulate and enact such policies is widely distributed

and not limited to the federal government sphere, while no political grouping has been able to attain

a dominant power position.

7.3.1 Diffuse Institutional Structure

The Brazilian Constitution outlines the formal organisational structure of the state. The country

is a federal presidential republic composed of three ‘independent and harmonious’ branches: the

executive, legislature and judiciary (FRB 1988: art 2). Reflective of this is the fact that the

President is both head of state and the head of government. Executive powers are exercised by the

President and the Ministers of the federal government. The President has the exclusive power to

initiate legislation, appoint and dismiss Ministers, and to preside over the upper management of

180
the federal administration (FRB 1988: art 84).

The President and Vice-President are simultaneously directly elected on the first Sunday of

October of the year prior to the termination of the sitting President’s term. The President’s mandate

runs for four years and commences on 1 January of the year following his or her election (FRB 1988:

art 77; art 82). There is a two-term limit to the presidency, but a former two-term President may be

re-elected provided that they did not stand as a candidate during the four-year period immediately

after their second term in office. Voting is also compulsory for literate citizens between the ages of

18 and 70.111

The legislative branch is the National Congress, which is composed of the Chamber of Deputies

(lower house) and the Senate (upper house). The Chamber of Deputies represents the people of

each state, and is currently composed of 513 Deputies.112 Members are elected every four years

according to a system of proportional representation. The role of the Senate, meanwhile, is to

represent the 26 states and the federal district. Each state is represented by three Senators, who

are directly elected for eight year terms. The configuration of the electoral cycle means that either

one-third or two-thirds of the Senate are elected every four years.

An independent judiciary is the final branch of the Brazilian state. As articulated in articles 101

to 103-B of the Constitution, the highest court in the country — equivalent to the Constitutional

Court in South Africa — is the Supremo Tribunal Federal (Supreme Federal Court). This is followed

by the Superior Tribunal de Justiça (Superior Court of Justice) as the highest appellate court

for non-constitutional matters, other superior courts, the Tribunais Regionais Federais (Regional

Federal Courts) and the Conselho Nacional de Justiça (National Justice Council).

In addition to these three branches, Brazil’s federal structure means that the country features
111
Voting is non-compulsory for youths between the ages of 16 and 18, people older than 70, and illiterate citizens.

Voter turnout is usually around the order of 80 percent. Failure to comply with the responsibility to vote is subject

to a fine.
112
Each state is guaranteed a minimum of eight and a maximum of 70 members, while the number of seats per state

is calculated according to figures from the national census.

181
three tiers of government: the federal (national) level,113 states (including the federal district) and

municipalities (FRB 1998: art 1). The country is comprised of 26 states plus the federal district,

which are further subdivided into a total of 5 561 municipalities (Souza 2005: 78). Each state

has its own Constitution, while states and municipalities have their own legislative and executive

institutions. Elections for the President, governors and for congressional and state representatives

take place simultaneously every four years. This is followed by mayoral and Municipal Council

elections two years later (Souza 2005: 85).

A crucial part of the Brazilian state structure is the decentralised distribution of power between

these three spheres. The country features a system of symmetrical federalism, with ‘each constituent

unit having the same powers as those granted to constituent units in the United States and Mexico’

(Souza 2005: 84). This extends to the issue of municipal autonomy, as the 1988 Constitution incor-

porated municipalities alongside states as part of the federation and ‘shifted considerable political

power and tax resources from the federal government to the states and municipalities’ (Selcher

1998: 25). The adoption of the current Constitution entailed dramatic changes to the resources

made available to subnational governments, particularly municipalities.

As a result, the share of public revenue being directed towards local governments increased from

18.2 percent to 22.8 percent (3.5 percent of GDP) (Souza 2002a: 3). This has resulted in a situation

where Brazilian municipalities are financially well-off compared to municipalities in other parts of

the developing world. The period between the promulgation of the 1988 Constitution and 1998

featured an increase in local revenue of 197 percent, with local revenue representing 1.6 percent

of GDP in the same year — a higher amount than federal constitutional transfers (Souza 2002a:

3-13). This is reflective of the fact that when the Constitution was adopted, ‘one of the main

objectives...was to leave subnational levels plenty of room to determine where and how to spend

the resources transferred to them’ (Souza 2002a: 3-13).

In all, ‘there is a broad consensus among analysts that Brazil is the most decentralised country in
113
The federal government sphere is legally known as the União (Union).

182
the developing world’ (Souza 2002b: 24). This means that Brazilian federalism is characterised by

various power centres, features financial interdependence among governmental units, and contains

multiple routes for designing and implementing policy (Souza 2002b: 45). The adoption of the 1988

Constitution resulted in the emergence of competing power centres with access to decision making

and policy implementation mechanisms, particularly in terms of social policies (Souza 2002b: 45).

These multiple power centres114 ‘compete both among themselves and with the federal executive’

(Souza 2005: 87).

This is particularly relevant in analysing the development of social assistance programmes.

Article 22 of the Constitution lists the exclusive competencies of the national level, followed by

article 23’s explication of the areas of concurrent powers (FRB 1988). It notes that

the Union, the states, the federal district and the municipalities, in common, have the power:

II –– to provide for health and public assistance, for the protection and safeguarding of hand-

icapped persons; X –– to fight the causes of poverty and the factors leading to substandard

living conditions, promoting the social integration of the underprivileged sections of the popula-

tion...The Union, the states and the federal district have the power to legislate concurrently on:

XII — social security, protection and defence of health; XIV –– protection and social integration

of handicapped persons; XV — protection of childhood and youth (FRB 1988: art. 23-24).

In short, the federal, state and municipal levels thus have a shared competence for implementing

‘health and social welfare [policies, as well as] combating poverty and social marginalisation’ (Souza

2005: 86). Additionally, preschool and primary education are areas of ‘mainly local’ constitutional

authority (Souza 2005: 86). The result is that, in Brazil, the subnational spheres formally enjoy

much greater power to design and implement social assistance policies affecting these issue areas.
114
Despite the decentralised nature of the Brazilian state, it is important to note that the federal level nevertheless

‘holds the largest number of exclusive powers’ (Souza 2005: 85).

183
7.3.2 Political Power Fragmentation

This constitutional diffusion of policymaking authority is further complemented by the practical

distribution of political power in Brazil. Figure 7.4 introduces the results of presidential elections

between 1994 and 2014.115 It illustrates that these national elections are characterised by a high

degree of competition between the candidates from the two primary national political parties, the

Partido dos Trabalhadores (PT – Workers’ Party) and the Partido da Social Democracia Brasileira

(PSDB – Brazilian Social Democracy Party).

100
90
80
70
Percentage

60
50
40
30
20
10
0
1994 1998 2002 2006 2010 2014
Partido dos Trabalhadores
Partido da Social Democracia Brasileira
Other Opposition

Figure 7.4: Electoral Support, Presidential Elections (1994-2014) (Source: TSE 2014)

Following the tumultuous years of the Collor and Franco presidencies, the politics around pres-

idential elections settled into a pattern of fierce contestation between the PT and PSDB, with

potential third-party spoilers being an ever-present reality. The success of the economic stabilisa-

tion programme known as the Plano Real (Real Plan) initially propelled Fernando Henrique Cardoso

to the presidency in 1994. Cardoso won the election by a margin of 27.3 percent, the largest in
115
The personalistic nature of politics in Brazil means that the value of thinking in terms of competition between

parties is limited. It is however useful for the purposes of this section by illustrating the high levels of competition.

184
Brazilian history (TSE 2014).

Subsequent to a constitutional amendment allowing him a second term, Cardoso was re-elected

in 1998 with 52.1 percent of the vote (TSE 2014). However, in the 2002 election, the PT swept to

victory under Lula, garnering 46.4 percent in the first round and winning comfortably in the runoff

against the PSDB candidate (TSE 2014). Lula was elected for a second term in 2006 in an initially

close race with the PSDB (he won the first round with 48.6 to 41.6 percent). Lula was subsequently

succeeded in 2010 by Dilma Rousseff, who won re-election in the 2014 race — a campaign that

featured the closest runoff election in the country’s history (TSE 2014).

The pattern of intense competition is even more pronounced in the case of legislative elections.

Figure 7.5 indicates that the number of parties represented in the Chamber of Deputies has never

been less than 17, with no party ever having managed to garner more than 20 percent of the vote

(TSE 2014). Additionally, every governing coalition has consisted of between four and nine parties

in the Chamber of Deputies.116 The intensity of political fragmentation under Brazil’s multiparty

presidential democracy has been characterised as a situation of ‘permanent minority presidentialism’

(Mainwaring 1992: 1, also see Figueiredo & Limongi 2000; Geddes & Neto 1992). The result is that

all Brazilian presidents must engage in a process of coalition-building in order to attain a majority

in Congress (Mainwaring 1992).

The picture of intense competition is further confirmed at the state and municipal levels.117 Dur-

ing the 2014 gubernatorial elections, candidates from nine different parties were elected as governors

for the 26 states and the federal district (TSE 2014). The pattern is also clear in the municipal
116
While this section focuses on the dynamics within the Chamber of Deputies, the same fractious pattern is visible

in the Federal Senate. A total of 15 parties are currently represented, with eight of them being part of the governing

coalition.
117
In contrast to South Africa, where the overwhelming racial nationalist legitimacy of the ANC has postponed the

pluralisation effects of class mobilisation, Brazil has seen the emergence of a political discourse centred on class which

has served as an underlying reason for the diffusion of political power. The result of this has been to further fuel the

process of bottom-up policy development characteristic of the Brazilian case.

185
 30

25

Number of Parties 20

15

10

0
1994 1998 2002 2006 2010 2014
Total Governing Coalition

Figure 7.5: Number of Parties Represented in the Chamber of Deputies (1994-2014) (Sources: TSE 2014; Figueiredo

2007: 190; Pereira 2010: 1.)

sphere where, in the 2012 municipal elections, mayors were elected from 26 different political parties

(TSE 2014). Out of the 5 561 municipalities in the country, the Partido do Movimento Democático

Brasileiro (PMDB – Brazilian Democratic Movement Party) won the most mayoralties: 1 014; fol-

lowed by the PSDB with 699 and the PT with 623 (TSE 2014). This means that no party controls

more than 19 percent of Brazilian municipalities.

The implications in relation to the topic of this study are clear: contemporary Brazilian pol-

itics, both formally and in terms of practical power distribution, are marked by a high level of

competition and the diffusion of authority. When this fact is combined with the country’s decidedly

federal structure, it becomes clear that the framework for social assistance policymaking is highly

decentralised in Brazil.

186
7.4 The Institutional Setting

The first sections of this chapter have introduced the Brazilian institutional context. The discussion

has illustrated that, compared to South Africa, social assistance norms were historically only weakly

entrenched; and that both in terms of the formal design of the state and through the practical

distribution of political power, Brazil features a diffuse state structure, especially in the area of social

assistance policymaking. Based on this preceding analysis, table 7.3 introduces the independent

variables for the Brazilian case.

Case (IV1) Institutionalisation of Norms (IV2) Concentration of State Struc-

ture

Brazil Low Low

Table 7.3: Independent Variables – Brazil

7.5 Institutions, Actors and Social Assistance in Brazil

The above examination of the institutional context amounts to the first step in the two-part ACI

analytical procedure, setting the scene for an examination of the impact that this institutional

setting had on the construction of actor preferences and the ways in which these preferences were

ultimately transformed into social assistance policy outputs. This initially involves an introduction

of the relevant policy community alongside their institutionally-derived preferences, as well as the

predominant mode of interaction, followed by an empirical analysis of the policy development process

that led to the construction of a protective social assistance system in Brazil.

The preceding discussion highlights the fact that the country’s diffuse state structure meant

that policymaking capacity was not limited to the federal government sphere. The authority to

develop social assistance policy in post-authoritarian Brazil was widely dispersed throughout the

state, involving policy actors from municipalities, states and the federal government in a process

187
of decentralised policy innovation. The country thus featured a more expansive policy community

than the South African case. Brazil’s diffuse state structure meant that state actors were effectively

divided into two subgroups, namely subnational state actors and federal state actors.118

The first subgroup included a wide array of municipal mayors, state governors and their associ-

ated advisers,119 while the second set of state actors was composed of policymakers at the federal

level, including Presidents, Ministers of Social Development, Health and Education, legal advisers

and federal bureaucrats. The final group constituting the social assistance policy community was

civil society actors, which included journalists, social activists, economists, historians and other

academics (Interview 16).

The shape of interactions within this actor constellation was institutionally determined. In

contrast to the case of South Africa, where the highly concentrated state structure meant that almost

all relevant policy interactions were conducted through ad hoc central government committees, the

Brazilian case featured less centralised and more regular avenues of interaction. Due to the fact

that subnational actors had the authority to independently design social assistance interventions, a

range of interactions took place at the local and regional level, largely between local politicians and

civil society actors. At the same time, a further set of interactions took place at the federal level

involving actors from all three groups.

Interactions were often conducted by way of the newly-established social assistance councils

which included representatives from civil society120 — through the Conselho Nacional de Assistência

Social (CNAS – National Social Assistance Council) at the federal level and its subnational corre-

lates on the state- and municipal levels.121 Even when policies were introduced through pieces of
118
The distinction between subnational and federal state actors is consistent with the approach adopted by Coêlho

(2012) in his review on the diffusion of the PBE in Brazil.

119
By 2001, this subnational group included actors from at least seven states and between 80 and 200 cities (Aguiar

& Araujo 2002; Villatoro 2005).

120
Côrtes notes that the ‘representatives who sit on the councils are seen as leaders who can influence policies’ (2013:

137).
121
See section 7.6.

188
federal legislation, the decentralised social assistance policymaking structure and the neocorporatist

nature of social assistance councils meant that interactions were largely characterised by instances

of negotiated agreement (Scharpf 1997: 116-150).

The institutional setting also played an important role in shaping the preferences held by policy

actors (Scharpf 1997: 1; 21). As in the case of South Africa, the analysis proceeds from the

fundamental assumption that both subnational and federal state actors were motivated to retain

or attain political power (Davis et al. 1970: 438). Whereas in the South African case this was an

almost trivial statement because of the ANC’s deeply entrenched power position, the assumption

becomes a much more potent force for shaping preferences in the case of politically competitive

Brazil.

In contrast to South Africa, where state actors faced ‘little immediate electoral incentive’ when

designing the post-apartheid social assistance system, Brazilian policymakers at all levels were

constantly subject to intense political competition (Seekings 2008: 35). The fact that legislative

turnover in Brazil has consistently exceeded 50 percent during each election since 1990 means that

state actors held a preference to design and implement social assistance policies that would reach as

many people in their jurisdiction as effectively as possible in order to distinguish themselves from

the competition and to be identified as ‘reformers’ (Samuels 2000: 481; Keefer & Khemani 2003a;

Keefer & Khemani 2003b). The goal was to create social policies which were popularly perceived

as effective in order to reap electoral rewards.

The incentive to utilise social assistance provision as a vehicle for attaining political credit was

further enhanced by the flexibility introduced by the lack of historical norms, as expressed in the

country’s constitutional-legal framework. The institutional structure also meant that even though

these groups were sometimes differentially positioned within the state and potentially competing

with each other, federal and subnational actors held broadly similar policy preferences driven by

the intensity of political competition. Both federal and subnational actors consequently wanted to

189
be able to take credit for designing social assistance policies that were effective and reached a wide

swath of voters. The preferences of state actors thus arose in an institutional context characterized

by greater policy and fiscal flexibility, and premised on political power diffusion. This structure

encouraged bottom-up policy experimentation and innovation fueled by the realities of political

competition.

The preferences held by civil society actors122 in Brazil mirrored those of their counterparts in

South Africa. The group often included actors previously involved in the Diretas Já movement,

and their institutional position outside the narrow constraints of the state meant that this group

of actors were naturally not subject to the imperatives of political competition. Instead, they

were focused on creating functionally effective policies available ‘to all citizens entitled to social

assistance’ (Côrtes 2013: 136). The historical lack of social assistance programmes in Brazil often

meant that this concern with functional effectiveness implied lobbying for more expansive coverage,

particularly through the movement that was to develop around the notion of a universal BIG in the

country.

The institutional setting thus gave rise to a somewhat different set of preferences in the Brazilian

case. Beyond the need to comply with the requirements of the new Constitution, state actors were

concerned with maximising the potential electoral returns of effective social assistance policies, while

civil society actors wanted to continue the process started during the struggle for democracy by

creating a significantly expanded system. The institutional setting guiding their preferences and

interactions would go on to decisively influence the development of Brazil’s contemporary social

assistance system.
122
Côrtes (2013) provides a fascinating account of the dynamics within these social assistance councils and the way

in which their preferences were conveyed.

190
7.6 The Policy Development Process

Following the specification of the institutional setting, actor constellation, modes of interaction and

actor preferences, this subsection empirically traces the policy development process for the Brazilian

case. Its principal aim is to illustrate the route whereby the weak institutionalisation of social

assistance norms combined with a diffuse state structure directed the construction of a protective

system in Brazil. The chronological discussion traces the role of these two independent variables in

four phases: an initial period (1990-1997) characterised by the emergence of the BPC and the first

conditional programmes at subnational level, a second period which saw the first actions to provide

federal support and eventually federal coordination for the major social assistance initiatives (1997-

2003), a third period characterised by the unification and initial universalisation of social assistance

(2003-2006) and a final period marked by the expansion and consolidation (2006-present) that has

entrenched the features of the current system.

7.6.1 Initial Policy Design (1990-1997)

The tumultuous fledgling years of Brazil’s new democracy were consumed by issues of economic

instability and runaway inflation. Federal politics centred on the fight against inflation, with seven

different economic stabilisation plans being implemented in the period between 1986 and 1994.123

It was only with the adoption of the Plano Real in July 1994 that stability started to return to

the Brazilian economy and inflation was tamed. The architect of the Plano Real was Fernando

Henrique Cardoso, who was finance minister during the presidency of Itamar Franco. Cardoso was

subsequently elected President in October 1994.

The first steps on the road towards constructing the contemporary Brazilian system of social

assistance followed shortly after the economic stabilisation brought about by the Plano Real and
123
These were the Plano Cruzado, Plano Cruzado II, Plano Bresser, Polı́tica Feijão com Arroz, Plano Verão, Plano

Collor and, finally, the Plano Real.

191
the implementation of the LOAS in 1995 (Schwarzer & Querino 2002: 16). The LOAS had a

particularly profound impact on the creation of the system, as it ‘deeply changed the structure of

social assistance in Brazil’ (Schwarzer & Querino 2002: 17). Its implementation led to the creation

of a Secretaria de Estado da Assistência Social (SEAS – Secretariat for Social Assistance) within

the federal MPS in 1995 with the aim of coordinating the decentralised policy design and execution

process.

This was soon followed by the construction of a national social assistance policy, the issuance of

basic operational norms, as well as the creation of intergovernmental administrative commissions

tasked with promoting ‘dialogue between the three government levels, [determining] the scope of

each level, [settling] financial issues, and [implementing] a participative council scheme at the three

levels’ (Schwarzer & Querino 2002: 17). Civil society groups were represented at each governmental

level through these participatory social assistance councils.

The LOAS required every municipality, state as well as the federal government to create such

an advisory social assistance council with the purpose of ‘providing richer policies, securing par-

ticipation, and making policies more transparent. The principle guiding the councils is to reach

consensus (emphasis added) between the different actors’ (Legido-Quigley 2009: 4). It also called

for each administrative unit to draw up an official social assistance policy document stating its

main objectives and principles, as well as the creation of an executing agency (a social assistance

secretary) at each government level (Schwarzer & Querino 2002: 17). Table 7.4 indicates that rapid

progress was made in these regards, with 88.6 percent of municipalities having a social assistance

council and 82.5 percent having a formal social assistance policy by 2000 (Schwarzer & Querino

2002: 18).

The first actual policy changes took place with regards to the Constitution’s call in article 203

(V) for the creation of a guaranteed monthly benefit equal to one minimum wage to the impoverished

handicapped of all ages and the impoverished elderly (FRB 1988). This new benefit — which later

192
Year Total Municipalities Municipalities with Social As- Municipalities with Social As-

sistance Council sistance Policy

1998 5 506 3 927 2 165

1999 5 507 4 840 4 482

2000 5 507 4 878 4 543

Table 7.4: Municipal Social Assistance Competency (Sources: Schwarzer & Querino 2002: 18)

became known as the BPC — was to be regulated by the LOAS itself in order to replace the RMV

(FRB 1993; Schwarzer & Querino 2002: 16; Interview 12). It called for the BPC to be coordinated

by the Ministério da Previdência Social (MPS – Ministry of Social Security) which, through the

INSS, was already responsible for implementing social insurance pensions in Brazil. This meant

that the non-contributory BPC had to be incorporated alongside these contributory programmes

(Legido-Quigley 2009: 3).

The programme stipulations laid out in the legal framework were very detailed: the federal

government was required to create a social assistance programme directed at disabled people of

any age and to those above the age of 65 whose household per capita income was less than one

quarter of the minimum wage per month. It was constitutionally mandated that this programme be

unconditional and assess the needs of beneficiaries based on their family structure (FRB 1988; FRB

1993). Given the fact that the BPC’s core institutional norms were defined in such great detail by

the Constitution and LOAS, policymakers did not grapple with issues of coverage, conditionality

and orientation.

The institutional framework was explicit: the BPC was to be unconditional; its orientation was

to be familial; and it was to cover impoverished disabled people of all ages and the impoverished

elderly by providing a monthly benefit equal in value to one minimum wage. This did not leave

much room for policy contestation. Instead, the main issues under consideration during the policy

design process were related to the BPC’s operationalisation. Throughout 1995, negotiations around

193
this were coordinated at the federal level by the SEAS and involved actors from all three spheres of

government and civil society through the newly-established administrative commissions and social

assistance councils (Schwarzer & Querino 2002: 18).

The main issue dealt with during these negotiations was the question of the benefit payment

mechanism. Given the fact that the INSS was already experienced in delivering the benefit payments

for contributory pensions, and despite discussions about the possibility of adopting a system where

the payments would be contracted out to local authorities, a decision was reached to task the INSS

with delivering the BPC payments across the country (Legido-Quigley 2009: 4). The BPC benefits

would consequently be paid out by the INSS via the same channels (commercial banks and post

offices) and on the same dates as contributory pensions (Legido-Quigley 2009: 4). Even though

the BPC has thus been a federal programme since its inception, the brief process that took place

to determine its operation hinted at what was to come: a decentralised procedure of negotiated

agreement between different policy actors.

The BPC officially became operational on 2 January 1996, marking the first implementation of a

genuine non-contributory social assistance intervention in Brazilian history. At the same time, new

applications for the RMV were no longer accepted after 1995, while benefits which were granted

up to 1995 continue to be paid.124 In a further reflection of the power of institutional norms, the

operationalisation of the BPC got off to a somewhat rocky start primarily due to the fact that INSS

officials were not used to working with a non-contributory benefit. As a result, ‘the mentality within

the INSS was that you had a right to a pension if you had previously contributed to the system’

(Legido-Quigley 2009: 4). Despite this initial difficulty, by 1998 the BPC was being paid out on a

non-contributory basis to a total of 844 632 beneficiaries every month (Schwarzer & Querino 2002:

18).

At the same time that the BPC was being phased in, there were two closely related policy
124
In 2014, a total of 183 531 beneficiaries were technically still considered to be receiving the RMV (MPS 2014:

11).

194
stirrings taking place elsewhere which would go on to have a decisive impact on shaping the pro-

tective social assistance system in Brazil. The first was the intensification of a theoretical debate

around the potential introduction of a BIG, while the second was the practical creation of the first

(localised) conditional CTs aimed at the impoverished population not covered by the fledgling BPC

(children, families and unemployed adults). Along with the implementation of the BPC, examining

the way in which these developments were institutionally channelled provides a persuasive account

of the contemporary features of Brazilian social assistance.

Early discussions around the potential for introducing a BIG originated among civil society

actors in the 1970s, but it was only during the 1990s that the issue was taken up in policy circles.

The political champion of this agenda was Senator Eduardo Suplicy from the PT, who introduced a

draft bill to the Senate in 1991 calling for the establishment of a social assistance scheme that would

guarantee a minimum income to all adults with incomes below a specified threshold (Suplicy 2002:

2; Lindert et al. 2007: 10). The bill was approved by the Senate, but the Chamber of Deputies failed

to vote on it. Despite the failure to adopt this particular bill, the debate around it placed the issue

of expanded social assistance squarely on the political agenda. Through a process of intense policy

experimentation, competition and proliferation, the idea eventually found some practical expression

in the introduction of the first subnational social assistance programmes in January 1995.

The first one to be created, known as the Programa Bolsa Escola (PBE – School Grant Pro-

gramme), became operational on 3 January 1995 in the federal district (Lindert et al. 2007: 12).

As the first programme to be implemented, the features introduced by the PBE would subsequently

go on to have a tremendous impact on shaping the eventual characteristics of the PBF programme.

After running on the platform of the PBE, the programme was introduced by incoming PT governor

Cristovam Buarque immediately upon assuming office (Interview 16). The initial impetus behind

the PBE originated through interactions with functionally-oriented civil society actors during Buar-

que’s time as rector of the University of Brası́lia in the late 1980s, where he engaged in debates on

195
challenges facing the country with academics (including sociologist and future President Cardoso)

and other civil society actors through the Centre for Studies of Contemporary Brazil (Buarque 2013:

11).

Along with journalist Gilberto Dimenstein, Buarque went on to further expound on the idea of

the PBE in two books.125 Buarque and his civil society collaborators emphasised the idea that ‘the

principal vector of development was not the economy, but was, indeed, quality education’ (Buarque

2013: 10). As such, they began to view social assistance provision as a tool to enhance educational

outcomes through conditionality. The ideational foundation was simple: if there were families with

children who are unable to go to school due to poverty, then they should be paid on the condition

that the children must study (Buarque 2013: 14). The aim was to use the poor’s ‘need for income

as a way to induce them to place their children in school’ (Buarque 2013: 14).

The notion of the PBE was first introduced to the general public in August 1994 during the

election campaign for the governorship of the federal district. The implementation of the programme

was initiated immediately after Buarque won the election. The PBE was set to become the first

conditional CT in Brazil. It would pay the equivalent of one minimum wage to families with

children between the ages of seven and 14, and would be conditional upon school attendance. The

programme was managed by the Secretary of Education and was well and truly a local programme,

as it was entirely funded by the federal district (Buarque 2013: 20).

A few months prior to the initiation of the PBE in the federal district in January 1995, the

then-mayor of the city of Campinas in São Paulo state, José Roberto Magalhães Texeira, had

shown an interest in the ideas previously published by Buarque, and invited him to discuss them
125
The first was entitled Revolução das Prioridades – da Modernidade Técnica à Modernidade Ética (‘The Revolution

of Priorities – from Technical Modernity to Ethical Modernity’) and A Segunda Abolição – uma proposta para a

erradicação da pobreza no Brasil (‘Abolishing Poverty: A Proposal for Brazil’). The case for a conditional CT was

further made in a series of academic and newspaper articles published in the Folha de São Paulo by economist José

Márcio Camargo (Lindert et al. 2007: 11).

196
in person (Buarque 2013: 16). Shortly after witnessing the introduction of the PBE, Teixeira (who

was from the rival PSDB) initiated his own programme126 in Campinas on 3 March 1995. Although

the Campinas programme closely emulated the PBE model, it was managed by the Secretary for

Welfare (instead of Education) and featured somewhat softer conditionality (Buarque 2013: 20).

By the end of 1998, 20 000 families in the federal district were receiving the PBE, while Campinas

had around 7 000 beneficiaries (Buarque 2013: 22).

What followed these initial experiments amounts to a remarkable tale of policy diffusion. The

PBE immediately proved to be immensely popular, both with beneficiaries and in the media. The

press played an important role in disseminating information across the country through ringing

endorsements that appeared in some of Brazil’s leading titles, including O Globo (19 November

1995; 20 December 1996); Veja (10 July 1996; 10 October 1996); Policarpo Júnior (8 October

1997); Sandra Brasil (8 October 1997); Jornal de Brası́lia (19 September 1996); and Estado de São

Paulo (5 February 1996) (Buarque 2013: 51-59).

Headlines included phrases like ‘The great silent revolution: Brazil surmounts the phase of

pompous solutions and begins to resolve its education problem with simple ideas’; ‘Help and even

kindness out of the horror – there are three solutions for the drama of childhood lost in the street:

school, school, and school’; ‘House and school – federal district government helps poor children

– and their parents as well’; ‘The example of Brası́lia’; ‘Complementing family income reduces

school dropouts in the federal district – program guarantees R$100 a month to needy families that

keep all their children in school’; and ‘UN gives prizes to federal district educational project –

Bolsa Escola pays one minimum wage to the students who do not skip school’ (Buarque 2013: 51-

59). The programme also won the prestigious Getúlio Vargas Foundation Public Management and

Citizenship Award in 1996.

The PBE furthermore received international recognition, with a profile appearing in Time mag-
126
This intervention was known as the Programa de Garantia de Renda Familiar Mı́nima (PGRFM – Familial

Minimum Income Guarantee Programme).

197
azine as early as December 1995. Nobel Laureate for Economics Gary Becker praised it in a 1997

Business Week article, while the programme won the 1996 UNICEF Children and Peace Prize

(Buarque 2013: 60-62). As a result of its ubiquitous profile, it became

one of the programmes that [was] most consulted by the country’s other governors. Every day

the federal district Secretary of Education received dozens of telephone calls, letters, and visits

of those interested in setting up the project (Buarque 2013: 59).

In this way, the programme effectively became a model which was replicated across the states and

municipalities of Brazil. By 2001, more than 100 municipalities and at least seven states covering

200 000 families were operating programmes based on the PBE (Lindert et al. 2007: 11; Buarque

2013: 32).

Without any involvement from the federal government, municipalities and states governed by

political parties from across the ideological spectrum that were ‘deeply diverse in terms of popu-

lation, poverty levels and economic weight...decided to fund the same programme, an event that

occurred in a relatively short period of time’ (Interview 12; Coêlho 2012: 57). The driving force

behind this tremendous horizontal policy diffusion was political competition channelled by Brazil’s

highly competitive and decentralised institutional structure, as politicians quickly learnt through

election results that resoundingly endorsed the popularity of the programmes (Fonseca & Montali

1996; Caccia Bava et al. 1998; Lindert et al. 2007; Coêlho 2012).

Heading into the 1998 presidential election year, the foundation for the construction of the

Brazilian social assistance system had thus been established. It provided comprehensive federal

support to impoverished disabled people of all ages and the impoverished elderly via the BPC, while

a disparate set of subnational programmes were providing coverage through conditional programmes

to families with school-aged children in a majority of municipalities and states. Already at this time,

the system had thus incorporated conditionalities and assessed the needs of potential beneficiaries

on a familial basis. Table 7.5 provides a snapshot into this evolving system.127
127
There was no unified, national PBE at this time, and this table only indicates that there were a range of such

198
Name Year Created Target Beneficiaries

Bolsa Escola (subnational) 1995 Families with school-aged children (gener-

ally 0-14)

Benefı́cio de Prestação Continuada (federal) 1995 Disabled (all ages)

Elderly (67+)

Table 7.5: Brazilian Social Assistance in 1997 (Sources: CEF 2014; MPS 2013: 11)

7.6.2 Federal Coordination (1997-2003)

Despite initial scepticism from bureaucrats surrounding the cost implications of federal social assis-

tance provision, it would soon simply become impossible for the government in Brası́lia to ignore

the continued proliferation, positive impact evaluations128 and growing electoral popularity of the

subnational PBE programmes (Interview 16; Buarque 2013: 62). Recognition of the model’s success

had in fact caused it to begin spreading beyond Brazil’s borders, as the Mexican federal govern-

ment had set up the Progresa (Progress) conditional CT in 1997 explicitly based upon the federal

district’s PBE (Interview 16).129

Driven by the realities of political competition, the federal government decided to get in on the

act. The PSDB administration identified an initial opportunity to do so by leveraging its financial

resources to induce cash-strapped municipalities into sharing political credit for the programmes

with the national sphere. The federal government thus identified and initiated negotiations with

financially constrained subnational units about the possibility of creating some room for federal
programmes being implemented by states and municipalities across the country. The figure for the BPC includes

remaining RMV beneficiaries.

128
Evaluations indicating positive impacts were carried out as early as 1997 by organisations including the World

Bank, the Inter-American Development Bank, the UNDP, ILO, UNESCO and UNICEF.
129
In Mexico, the programme was subsequently renamed Oportunidades (Opportunity), and is currently known as

Prospera (Prosper). President Ernesto Zedillo had sent a delegation (which included his wife) to the federal district

in 1996 to learn about the PBE before setting up the Mexican scheme (Interview 16).

199
involvement in the financing of local PBEs.

The process resulted in the introduction of the Programa de Garantia de Renda Mı́nima (PGRM

– Programme for a Guaranteed Minimum Income) in 1998. The bill creating this scheme was

introduced to Congress by Nelson Marchezan from the PSDB. The PGRM was managed by the

federal Ministry of Education, and provided transfers to municipalities which were implementing

their own conditional CTs but lacked the necessary resources to sustain the programmes (Lindert

et al. 2007: 12). Municipalities were however still required to contribute at least 50 percent of

programme funds in order to be eligible for federal support.

Through this process of negotiated agreement, the PGRM was ultimately extended to cover 1

350 of the poorest municipalities in the country (Coêlho 2012: 57). Although it was not a social

assistance intervention per se, the PRGM represented the first foray by the federal government

into the provision of social assistance to groups not eligible for coverage under the BPC. It thus

enabled municipalities to continue implementing such programmes by providing financial support.

In turn, municipalities ceded some of their political authority to the federal government. As such,

the PGRM was not only an important factor in further promoting and sustaining subnational CTs,

but also became a gateway for the design and implementation of future federal social assistance

initiatives (Lindert et al. 2007: 12).

Meanwhile, Cardoso was re-elected in the 1998 presidential election and assumed office in Jan-

uary 1999 to commence his second term — with the diffusion of subnational PBE programmes

continuing unabated across the country. Despite its financial involvement, this reality meant that

the federal government risked foregoing the opportunity to be definitively associated with what had

become the most popular social policy initiative in Brazilian history. It had to act. Much of the

impetus for federal action continued to come from Congressman Marchezan (Cardoso’s colleague

in the PSDB) through the hosting of a national seminar on minimum income linked to education

in the Chamber of Deputies during November 2000. The outcome of these efforts was to convince

200
the Cardoso administration of the (political) necessity of establishing a nationwide, federal PBE

(Buarque 2013: 36).

The final two years of the Cardoso presidency subsequently saw the introduction and rapid

expansion of federal CTs. Federal politicians had become thoroughly convinced by the potential

of CTs to generate political support, and ‘every Minister thought: “this is a good idea, I want

one too”. It started to come out everywhere’ (Interview 13). Fittingly, the first to be introduced

was a federal PBE programme. On 11 April 2001 Cardoso created, through executive decree,130 a

national version of the programme that was originally implemented in the federal district six years

earlier to replace the PGRM (Lindert et al. 2007: 13). The federal PBE thereby consolidated the

scattered array of subnational programmes into a single federal scheme.

Managed by the Ministry of Education, the programme targeted families with children between

the ages of seven and 16 years old and paid a monthly sum of R$15 (US$7) per child to families with

per capita income of less than R$90 (US$43) conditional upon a minimum school attendance rate

of 85 percent per child (Lindert et al. 2007: 13). The operation of the programme was therefore

identical to its subnational predecessors. The primary change was an organisational one: although

still implemented in a decentralised fashion by individual municipalities, the PBE had now become

a national programme funded exclusively by the federal government. It had thereby gained much

firmer legal footing, even as political credit for the PBE continued to be shared between local and

federal politicians. From a small local programme targeting 20 000 families in the federal district
130
The widespread use of executive decrees in Brazil’s presidential system is a topic of some controversy among

analysts (Ames 1995; 2001; Limongi & Figueiredo 1995; Figueiredo & Limongi 2000; Reich 2002). The Constitution

grants Presidents the power to enact a measure for a period of 30 days through executive decree, during which time

Congress is supposed to vote on its permanence. But in a vast majority of cases Congress fails to vote on executive

decrees (Reich 2002: 10). When this happens, the decree expires. However, the practice of simply reissuing the same

decree for months or even years on end — thereby enabling the continued implementation of the particular provision

— has become a deeply embedded presidential practice which also manifested itself in the creation of conditional CTs

in Brazil.

201
in 1995, the federal PBE ‘was present in almost all cities in the country’ and reached a total of 4.8

million families by 2001 (Coêlho 2012: 58; Buarque 2013: 37).

The federal executive quickly moved to consolidate this political momentum by additionally

creating the Cadastro Único (Central Registry)131 with the aim of collecting information on the

living conditions of all impoverished families in the country. This would allow the government to

design further programmes targeting this population. The first subsequent step was taken on 6

September 2001, when Cardoso issued a presidential decree creating the Bolsa Alimentação (BA –

Food Grant). Explicitly modelled on the PBE, this CT was aimed at families with children between

the ages of zero and seven years old and was conditional upon complying with a schedule of pre- and

post-natal medical visits, ensuring that all children were vaccinated according to an official schedule,

participation in nutritional education seminars, as well as growth monitoring. It was managed by

the Ministry of Health and also paid $15 (US$7) per child to families with a per capita monthly

income of less than R$90 (US$43). Once eligibility for the programme expired at age seven, families

with eligible children transitioned onto the federal PBE (Lindert et al. 2007: 13).

The federal government’s newfound zeal for CTs was also expressed through the creation of the

Auxı́lio Gás (AG – Gas Voucher) programme in January 2002 as a ‘compensatory measure for the

phasing out of cooking gas subsidies’ (Lindert et al. 2007: 13). This scheme was administered by

the Ministry of Mines and Energy, and paid a transfer of R$15 (US$7) in bi-monthly instalments

to families with monthly per capita income equal to less than half the minimum wage (Lindert et

al. 2007: 13).

By the end of Cardoso’s second term, profound changes had consequently been made to the

Brazilian social assistance regime. Although still incomplete, the system now not only provided

coverage to the disabled and elderly through the BPC, but also extended federal coverage to impov-
131
The creation of the Cadastro Único proved to be an administrative masterstroke. The database currently contains

information on the living conditions of 78 million of the poorest Brazilians and has become an operational centrepiece

in the country’s social assistance efforts (ILO 2014: 2).

202
erished families with children between the ages of zero and 16. It continued to define beneficiaries

in familial terms, and strongly incorporated elements of conditionality. Table 7.6 illustrates the

composition of the system at that time.132

Name Year Created Target Beneficiaries

Bolsa Alimentação 2001 Families with children aged 0-7

Bolsa Escola 1995 (subnational) Families with children aged 7-16

2001 (federal)

Benefı́cio de Prestação Continuada 1995 Disabled (all ages)

Elderly (67+)

Table 7.6: Brazilian Social Assistance in 2003 (Sources: CEF 2014; MPS 2013: 11)

7.6.3 Unification and Universalisation (2003-2006)

On one level, the 2002 presidential election brought profound change to Brazilian politics. After two

full terms under Cardoso’s PDSB government, Brazilians elected Lula and his PT in a second-round

runoff during his fourth campaign for President. Lula and the PT had first run for President in

1989, and had for many years openly advocated for the implementation of socialism in the country.

But the successful adoption of market reforms during the Cardoso years had ‘rendered the party’s

socialist project unviable’, forcing it closer to the centre of the ideological spectrum and ultimately

enabling it to achieve electoral victory (Hunter 2007: 440).

The pressures of vote maximisation produced by a highly competitive political arena, combined

with the need to demonstrate results in office following a series of successful subnational elections

in the 1990s, increased the weight of pragmatists in the party and led the PT to undertake a

credible shift closer to the political centre (Samuels 2004; Hunter 2010). This undertaking was most

profoundly expressed prior to the election in Lula’s famous Carta aos Brasileiros (‘Letter to the

Brazilian People’) where he promised to honour the country’s debts and uphold current economic
132
This list does not include the niche AG programme. The figure for the BPC includes remaining RMV beneficiaries.

203
policy. The same pattern of continuity is visible in the area of social assistance, where Lula’s election

did not imply a profound break with previous policy. Instead, successive PT administrations would

go on to do a masterful job of building on the prior foundational reforms implemented by Cardoso

(Faulbaum 2013).

But the Lula administration would have to learn a hard lesson about institutional stickiness

first. During his inaugural address on 1 January 2003, the new President announced the creation

of the Fome Zero (FZ – Zero Hunger) programme as his government’s flagship programme. It

amounted to a multi-dimensional umbrella scheme composed of over 60 different interventions aimed

at eradicating hunger in the country (Lindert et al. 2007: 13). Lula also announced the related

creation of the Ministro de Estado Extraordinário de Segurança Alimentar e Combate à Fome

(MESA – Special Ministry for Food Security) to coordinate the introduction of a pilot project known

as the Cartão Alimentação (CA – Food Card), where money was to be transferred to impoverished

beneficiaries to enable them to make food purchases (Bither-Terry 2013: 1; Lindert et al. 2007:

13).

The FZ intervention soon proved to be a dismal failure, largely because it disregarded existing

institutions and ‘failed to engage constructively with policy legacies from the Fernando Henrique

Cardoso government’ (Bither-Terry 2013: 1; Interview 13). At the same time, a new set of supple-

mentary survey questions in the 2004 household survey had found that CTs were having positive

impacts on the poor. It made a big difference in terms of the poverty headcount ratio and, perhaps

more significantly in the Brazilian context, was having a ‘huge impact’ in terms of reducing inequal-

ity (Interview 13). By the end of Lula’s first year in office, he was ‘[beating] a hasty retreat from FZ

and invested in social programmes similar to Cardoso’s, increasing their resources’ (McCann 2008:

40). The administration was admitting policy failure as far as FZ was concerned: it abolished the

special ministry and ended the CA pilot project.

From 2004 onwards, the Lula government shifted its attention away from the institutionally

204
incompatible interventions of FZ towards what was to become the PBF. Even though voices in some

quarters of the PT, including Cristovam Buarque who was now federal Minister of Education, had

been lobbying Lula since 2002 to further expand CTs across the country, it took the failure of FZ to

finally convince the President that these interventions were more politically effective (Buarque 2013:

37; Interview 13). Lula consequently requested officials from the Ministry of Social Assistance133

and the Instituto de Pesquisa Econômica Aplicada (IPEA – Institute of Applied Economic Research)

to outline a proposal for an integrated CT programme (Lindert et al. 2007: 14).

On 9 January 2004, Lula formally created the PBF through an executive decree (FRB 2004c).

The programme unified the existing CTs (PBE, BA, and CA) into one massive conditional federal

programme. The language used in the operational justifications for the reforms that resulted in

the PBF included references to ‘leveraging synergies’, ‘rationalisation and consolidation’, ‘improved

efficiency and targeting’, ‘strengthening monitoring and evaluation’, ‘promoting vertical integration’,

and ‘leveraging opportunities to systematise complementarities’ (Lindert et al. 2007: 14).

More decisive than this rationale was the political justification. In the face of intensifying

criticism and the failed FZ approach, the leftist Lula government was eager to claim political credit

for what were proving to be extreme popular social programmes (Interview 13). Indeed,

after the much-ballyhooed Fome Zero programme failed to get off the ground in 2003-4, Lula was

receiving low marks in this area. He set his sights on reversing this situation in the second half

of his administration. Staking his prospects on the Bolsa Famı́lia was a cost-effective strategy

(Hunter & Power 2007: 19).

Motivated by the realities of political competition, the administration proceeded to build on the

policy legacies inherited from the Cardoso years with the aim of leveraging the PBF into political

support.

It did this while being fully cognisant of the realities of the country’s diffuse state structure, as

the negotiated agreements that were reached with subnational actors in the creation of the PBF
133
This ministry was merged with MESA in January 2004 to form the MDS, headed by Minister Patrus Ananias.

205
allowed them to continue sharing in some of the political credit for the creation of the programme:

by sharing the day-to-day operation of Bolsa Famı́lia with Brazil’s 5 500 municipal mayors, the

President allowed local elites to reap some of the benefits of this hugely popular programme

(Hunter & Power 2007: 18).

This outcome was crucial, as it enabled the Lula government to take credit for the creation of

the PBF while simultaneously ensuring continued political buy-in from subnational actors in the

context of Brazil’s diffuse state structure.

The other relevant development that took place during this period was related to the debate

surrounding the BIG in Brazil. The discussion, still championed by the PT’s Suplicy, had already

played a role in spurring the expansion of social assistance provision. The BIG had been part of

Lula’s PT election campaigns throughout the 1990s, and his election finally led to a vote in Congress

on a BIG bill authored by Suplicy (Suplicy 2006: 43). Both houses of Congress approved the bill

during 2003, and it was signed into law by Lula on 8 January 2004 (FRB 2004a). Brazil had thereby

become the first country in the world to adopt a law calling for the creation of a universal basic

income.

However, the language contained in the bill made it clear that it was the prerogative of the

executive to determine the pace of its implementation (FRB 2004a). The BIG law was a very

flexible piece of legislation, as ‘the amount given and its realisation will be gradual and given under

the criterion of the national executive, which gives priority to the neediest until everybody can

receive it’ (Suplicy 2006: 53). In practise, the PBF was regarded by many as the first step on the

road towards implementing a BIG by gradually extending its coverage (Lindert et al. 2007: 14).

In the face of mounting criticism regarding its initially slow progress on social policy issues

and the failed FZ programme, the Lula government thus responded to the political pressure by

claiming the PBF as its own, and then framing the PBF as the first step in implementing a BIG

across Brazil134 (Hunter & Power 2007: 19; Suplicy 2006: 10; Lindert et al. 2007: 14). The
134
Despite the fact that the law was politically framed in this way, some analysts question whether the PBF will

206
influence of the BIG law was made particularly visible through the introduction of the PBF’s basic

benefit (a benefit that did not exist for any of its predecessors), which enabled individuals living

in families classified as ‘extremely poor’ to qualify for a per capita payment of R$50 (US$23) per

month regardless of their demographic composition and whether they had any children (Lindert et

al. 2007: 16). This step had the effect of introducing, for the first time, universal coverage that

also included the able-bodied poor of working age in the Brazilian social assistance system.

The final two years of the first Lula administration were dominated by the execution of the

administrative actions required to fully operationalise the PBF. The programme underwent a tran-

sition year in 2004 during which its foundations were established. This included the creation of

the MDS in January of that year, followed by the issuance of operational instructions. The year

2005 ‘represented a period of consolidation and maturation for the PBF’ during which its ‘core

architecture’ was strengthened (Lindert et al. 2007: 14). ‘Massive efforts’ were also undertaken to

enhance the accuracy of the Cadastro Único and monitoring conditionalities (Lindert et al. 2007:

14). By the end of 2004, the programme was already reaching 59 percent of eligible families; this

grew to 77 percent by 2005 and almost 100 percent by 2006 for a total of 11.1 million families, or

44 million individuals (Hunter & Power 2007: 19; Lindert 2006: 67).

The 2006 general election was the next decisive event in the construction of the Brazilian social

assistance system. Despite his party being embroiled in the Mensalão (‘big monthly payment’)

corruption scandal, Lula remarkably won the runoff against the PSDB’s Geraldo Alckmin with

more than 60 percent of the votes (TSE 2014). The election results represented a clear testament

to the political success of expanding the conditional, familial PBF programme through a process

of negotiated agreement (Hunter & Power 2007; Zucco 2008; Zucco & Power 2013). Lula’s support

base had undergone a pronounced shift, as voters from wealthier regions of the country abandoned

him due to the corruption scandal, while voters from relatively more impoverished regions of the
ultimately result in the implementation of a BIG; see Britto 2011.

207
country (who had not supported him in 2002) now voted for him en masse135 (Hunter & Power

2007; Zucco 2008; Zucco & Power 2013).

The evidence suggests that

the voters who carried Lula to victory appear to have been strongly influenced by the govern-

ment’s social policies, especially the PBF. . . The poorest, least educated citizens in Brazil have

seen material improvements in their lives since 2003. In 2006, they made clear that they are avail-

able to be mobilized by politicians who provide them with poverty-reducing, equity-enhancing

benefits. . . In sum, the social policy story is arguably the single most plausible explanation for

Lula’s reelection (Hunter & Power 2007: 20-25).

The 2006 election results thus vindicated Lula’s gamble on the PBF and testified ‘to the wisdom of

[his] acceleration of social policy in the second half of his first term’ (Hunter & Power 2007: 19).

The signature achievement of Lula’s first term was the unification and universalisation of social

assistance in Brazil through the introduction of the PBF. It was the culmination of a decade long

bottom-up policy development process where the norms of universal provision, conditionality and

a familial orientation were continuously being entrenched. Table 7.7 provides an overview of what

the system looked like by the time that Lula was re-elected for a second term as President.

Name Year Created Target Beneficiaries

Bolsa Famı́lia Basic 2004 Extremely poor (all ages)

Bolsa Famı́lia Variable 1 1995 (subnational) Poor families with children aged 0-16 (up

2004 (federal) to 3 children)

Benefı́cio de Prestação Continuada 1995 Disabled (all ages)

Elderly (67+)

Table 7.7: Brazilian Social Assistance in 2006 (Sources: CEF 2014; MPS 2013: 11)

135
Even though this interpretation was criticised by Bohn (2011), the data introduced by Hunter and Power (2007),

as well as by Zucco and Power (2013), amount to highly compelling evidence that Lula’s voter base did indeed swing

from the wealthy southwest to the impoverished northeast between 2002 and 2006.

208
7.6.4 Expansion and Consolidation (2006-present)

The adoption of the BIG law and the political popularity of the PBF would go on to play a decisive

role in spurring further moves towards universal social assistance provision in the country. With

the ever present reality of political competition lurking in the background, the PT government had

become thoroughly convinced that expanding the PBF to cover the entire impoverished population

— based on conditionality and a familial orientation — was a winning electoral strategy (Hunter &

Power 2007).

In addition to continuing to implement the BPC and strengthening the administrative mecha-

nisms of all programmes, Lula’s second term from 1 January 2007 to 31 December 2010 was mainly

characterised by the PBF’s continued expansion in coverage and benefit values. The programme

effectively had three sets of eligibility criteria between 2003 and 2010: R$50 and R$100 from 2003

to 2006; R$60 and R$120 from 2006 to 2009; as well as R$70 and R$140 since 2009 to the present

(Osório & Ferreira de Souza 2013: 1). The benefit levels were adjusted four times during this

period: in 2007, 2008, 2009 and 2011. The biggest expansion in the programme’s coverage was the

introduction of the adolescent (variable 2) benefit in December 2007. This adjustment expanded

coverage to families earning between R$60 (US$25) and R$120 (US$50) per capita with children up

to the age of 18 (Osório & Ferreira de Souza 2013: 2). The maximum eligibility age for the variable

benefit thus increased from 16 to 18. Table 7.8 summarises these expansionary changes.

209
Year Changes and Adjustments Benefit Design (at year end)

2003 October Extreme Poverty: up to R$50

Creation of the PBF, with two levels of eligibility that referred, but Basic: R$50

were not bound, to the amounts of 1/4 and 1/2 the minimum wage Children: R$15 to R$45

(R$200 in early 2003) of per capita household income, with a basic Poverty: from R$50 to R$100

benefit given only to extremely poor families, and a variable benefit, Children: R$15 to R$45

given per child aged 0-15 years, for a maximum of three children.

2006 April Extreme Poverty: up to R$60

The eligibility levels are adjusted for the first time, with no change to Basic: R$50

benefit design. Children: R$15 to R$45

Poverty: from R$60 to R$120

Children: R$15 to R$45

2007 July Extreme Poverty: up to R$60

The benefits are readjusted. Basic: R$58

December Children: R$18 to R$45

The benefit design is altered for the first time, with the creation of a Adolescents: R$30 to R$60

benefit for up to two 16- and 17-year-old adolescents. Poverty: from R$60 to R$120

Children: R$18 to R$45

Adolescents: R$30 to R$60

2008 June Extreme Poverty: up to R$60

The benefits are readjusted. Basic: R$62

Children: R$20 to R$60

Adolescents: R$30 to R$60

Poverty: from R$60 to R$120

Children: R$20 to R$60

2009 April Extreme Poverty: up to R$70

The levels are readjusted to R$69 and R$137. Basic: R$68

July Children: R$22 to R$66

The eligibility levels are again readjusted to the amounts that would Adolescents: R$33 to R$66

remain in effect until the end of 2012. The benefits are also read- Poverty: from R$70 to R$140

justed in July. Children: R$22 to R$66

Adolescents: R$33 to R$66

210
 2011 March Extreme Poverty: up to R$70

The benefits are readjusted, and the benefit design undergoes a sec- Basic: R$70

ond change, expanding the limit from three to five children. Children: R$32 to R$160

Adolescents: R$38 to R$76

Poverty: from R$70 to R$140

Children: R$32 to R$160

Adolescents: R$38 to R$76

2012 May With children aged 0-16 years

The PBSM’s per capita transfer is introduced, aimed at households Extreme Poverty: up to R$70

which, even after receiving the PBF benefit, had remained extremely Basic: R$70

poor. Children: R$32 to R$160

November Adolescents: R$38 to R$76

The age range of children eligible to participate in the PBSM is rede- PBSM: remaining per capita gap

fined as 0-15 years of age. Without children 0-16 years

Extreme Poverty: up to R$70

Table 7.8: The Evolution of the Bolsa Famı́lia from 2003 to 2012 (Sources: Osório & Ferreira de Souza 2013: 2)

By the end of Lula’s second term in 2010, the PBF was covering 12.7 million families (52

million individuals) (ECLAC 2013). The social assistance system theoretically provided an absolute

monthly minimum income of R$70 (US$30) to every impoverished person in the country, while

children, the disabled and elderly were entitled to an additional monthly amount (up to three

children per family between zero and 18 received an additional amount of R$32 (US$14) or R$38

(US$17) through the PBF, while the disabled and elderly received R$677 (US$294) via the BPC).

Thanks in large part to these policies, the number of people living in poverty had fallen by 20 million

under Lula; from 49.5 million (28.5 percent) to 29 million (16 percent) (The Economist 2010). At

the same time, the number of Brazilians who were too poor to properly feed themselves had fallen

from 17 percent in 2003 to 8.8 percent by 2008 (The Economist 2010).

It was within this context that the 2010 general elections took place. Lula’s successor as PT

211
presidential candidate and his former chief of staff, Dilma Rousseff (popularly known simply as

Dilma), came up against the PSDB’s José Serra. In an illustration of the extent to which social

assistance had become politically entrenched, Serra promised that he would maintain the PBF if

elected (R7 Notı́cias 2010). But the PT’s association with such programmes allowed Dilma to

aggressively campaign on the promise of further expanding social assitance (R7 Notı́cias 2010). The

strategy again paid-off, as Dilma was elected by a margin of 56 percent to 44 percent in the runoff

with Serra (TSE 2014).

Reflecting the priority assigned to the continued expansion of social assistance provision, one of

the first moves made by the Dilma government was to increase the benefit values and the number

of children eligible per family for the child-focused part of the PBF (variable 1 benefit) from three

to a maximum of five (Osório & Ferreira de Souza 2013: 1). This was followed by the introduction

of the PBSM in June 2011, which amounted to the biggest coverage expansion since the creation

of the PBF in 2004. This plan built on the institutional legacy inherited from the Lula years by

introducing the residual benefit to the PBF. The plan explicitly targeted the ‘approximately 16.2

million Brazilian identified by the 2010 census who are still in a situation of extreme poverty, that

is, receiving a monthly income below R$70’ (FRB 2011b: 6).

The programme also featured an ‘active search’ component aimed at proactively looking for and

registering all of the remaining extremely poor families that had not been located yet (FRB 2011b:

6). Dilma made it clear that the PBSM’s aim of finally eradicating extreme poverty in Brazil by

ensuring that no person lived on less than R$70 (US$30) per month by 2014 was her government’s

‘key priority’ (BBC 2011). In order to achieve this, the PBF’s residual benefit would forthwith cover

the remaining poverty gap for each beneficiary family (FRB 2011b). For example, if a family’s total

income after the basic and variable PBF transfers still amounted to only R$50 per capita, then the

state would pay an additional R$20 per person to ensure that the household was brought above the

extreme poverty line.

212
 Dilma’s first term was focused on implementing the PBSM. While it continued to incorporate

conditionality and a clear familial orientation, the PBF under the PBSM also served to further

emphasise what had become the universalistic nature of the Brazilian system. And the electorate

was about to reaffirm once more that the expansion of this system was politically advantageous.

Even as a wave of popular middle class protests swept over Brazil and the economy weakened in

2013, Dilma was claiming that the government had almost met the target of eliminating extreme

poverty (Boadle 2013). By late 2014 and in the heat of the next election campaign, the Minister of

Social Development, Tereza Campello, stated that the government had complied ‘with all the goals

of the PBSM’136 (MDS 2014b).

Following a tumultuous campaign which again prominently featured the PT candidate primarily

running on the government’s record of social assistance provision, Dilma was re-elected in an tight

runoff against the PSDB’s Aécio Neves (TSE 2014). She embarked on her second term as President

in a Brazil that had undergone profound changes in terms of social assistance provision. From a

baseline of practically zero, the previous two decades had witnessed the construction of a protective

system based on universality, conditionality and a familial orientation. Table 7.9 illustrates the

component programmes of this system.

136
There is some controversy surrounding this claim, see Carneiro 2014 and Magalhães 2014.

213
Name Year Created Target Beneficiaries

Bolsa Famı́lia Basic 2004 Extremely poor (all ages)

Bolsa Famı́lia Variable 1 1995 (subnational) Poor families with children aged 0-16 (up

2004 (federal) to 5 children)

Bolsa Famı́lia Variable 2 2007 Poor families with adolescents aged 16-18

(up to 2 adolescents)

Bolsa Famı́lia Residual 2012 All families with per capita income still

below R$70

Benefı́cio de Prestação Continuada 1995 Disabled (all ages)

Elderly (67+)

Table 7.9: Brazilian Social Assistance in 2014 (Sources: CEF 2014; MPS 2013: 11)

7.7 Conclusion

The chapter has conducted a thorough overview of the development of social assistance policies in

Brazil. It has illustrated the way in which a productive system emerged within an institutional

context characterised by the weak institutionalisation of norms and a diffuse state structure. The

analysis proceeded according to the ACI procedure, which entailed two sequential steps. The first

involved the explication of the Brazilian institutional structure according to the two independent

variables. The discussion initially highlighted the fact that institutional norms relating to coverage,

conditionality and orientation were only weakly entrenched, owing to the absence of social assistance

provision in Brazil prior to the transition to democracy. This left the constitutional architects with

a relatively blank slate, which they subsequently filled with a flexible set of emerging norms.

This was followed by an examination of state structure, which indicated that formal policy-

making capacity and the practical distribution of political power were both highly dispersed. The

specification of the institutional setting subsequently enabled the second analytical step, involving

an examination of the role played by this configuration in shaping the preferences and interac-

214
tion structures of actors within the policy community. The final section empirically reviewed the

actual process of social assistance policy development in Brazil through a focus on interactions

between diverse sets of policy actors, culminating in an approach premised on universal coverage,

conditionality and a familial orientation.

The analysis shows that even when efforts were made to undertake alternative policy paths,

such as with the Lula government’s unilateral introduction of the FZ programme at the potential

expense of social assistance interventions, the incentives produced by an institutional structure

premised on weak norm institutionalisation and a diffuse state structure ultimately undermined such

attempts. The chapter thus illustrates that even as it was initiated by the transition to democracy

and influenced by functional and ideational factors, the process leading to the construction of a

protective social assistance system in Brazil was ultimately decisively channelled and shaped by the

country’s particular institutional configuration.

215
Chapter 8

Explaining Variation

8.1 Introduction

This chapter explicitly returns to the study’s research question by interpreting the preceding case

studies. As such, it aims to account for the variation on the dependent variable (the creation of

a productive system in South Africa versus a protective approach in Brazil) by pointing to the

variation on the independent variables (institutionalisation of norms and state structure) as the

primary explanatory factor. In short, it argues that differences on the institutional dimension most

effectively explain the fact that South Africa has adopted a productive approach to social assistance,

while Brazil has constructed a protective system.

It achieves this by initially investigating the veracity of the set of competing claims introduced

in section 3.5. The analysis shows that while functional and ideational factors were tautologically

relevant to the policy development process, these explanations fail to comprehensively account

for the differing outcomes. This is followed by a return to the study’s central argument that

differences in institutional features offer the most plausible explanation. In a powerful confirmation

of the institutional hypothesis, the discussion demonstrates the way in which (often similar) policy

problems and ideas were institutionally channelled into varying outcomes across the cases and

216
ultimately led to the construction of a productive social assistance system in South Africa, and a

protective regime in Brazil.

8.2 Alternative Explanations

This section subjects the empirical discussion on the development of social assistance policies in

South Africa and Brazil to a functional and ideational interpretation. These two dimensions have

emerged in the literature as the principle explanatory vectors in accounting for the recent design

and implementation of such policies in the developing world (Graham 2002; Fiszbein et al. 2009;

Hanlon et al. 2010; UNRISD 2010; Liesering & Barrientos 2013; Mothiane 2014). As the analysis

will show, these interpretations are however inadequate in accounting for the emergence of different

policy regimes across the cases, as measured along the dimensions of coverage, conditionality and

orientation. Regarding the recent introduction of social assistance provision in the developing world

purely as a product of broadly similar ‘objective’ policy problems and/or ideas is an oversimplifica-

tion which severely underestimates the decisive role played by local institutions in shaping varying

policy outputs.

8.2.1 The Functional Hypothesis

A functional account of the rise of social assistance in South Africa and Brazil provides a useful

exogenous compliment to the inherent endogeneity of the institutionalist approach. Proponents

of this approach assert that the implementation of CTs represents a response to ‘economic crisis,

structural adjustment and global integration, where the limitations of residual, ad hoc safety nets

to address the social consequences of neoliberal policies became painfully apparent’ (UNRISD 2010:

135). The introduction of these programmes is therefore seen as a functional response to the policy

problem of persistently high poverty (and inequality) rates. It is argued that the interventions

‘reflect a situation of high poverty and inequality’ that have ‘increasingly led governments...to

217
examine whether social protection in general — and cash transfer programmes in particular — can

address some of the...challenges’ (Fiszbein et al. 2009: 9; Garcia & Moore 2012: 1).

If the design features of social assistance interventions purely represent ‘appropriate’ responses

to exogenous policy problems, then it can be expected that cases featuring similar problems would

exhibit similar responses. The veracity of the functional claim thus depends upon demonstrating

that governments adopted similar policies in similar contexts. Beyond the somewhat tautological

claim that poverty needs to be present for CTs to be justified, a brief examination of the features of

the South African and Brazilian cases illustrates the highly limited applicability of this explanation

in accounting for their specific features in terms of coverage, conditionality and orientation.

A functional explanation for the variations in coverage (limited in South Africa, universal in

Brazil) would need to demonstrate that broader coverage was required to alleviate poverty in Brazil

than in South Africa. Specifically, such an account would need to explain why it is that able-bodied

unemployed adults are covered in Brazil but not in South Africa. Straightforward proxies for testing

this are provided by data on poverty and working-age unemployment rates: perhaps poverty and

unemployment rates were simply higher in Brazil than in South Africa when the systems were

designed, necessitating the incorporation of this impoverished societal segment there. Figure 8.1

compares the two cases on these metrics.

218
 45

40

35

30
Percentage
25

20

15

10

0
Poverty (US$1.25 per day - Poverty (US$2 per day - Unemployment Rate
PPP) PPP)
South Africa Brazil

Figure 8.1: Poverty and Unemployment Rates (1995) (Source: World Bank 2015a; 2015b; 2015c)

The data indicate that poverty and unemployment rates were higher in South Africa than in

Brazil at the beginning of the period under examination (and have remained that way throughout).

It is safe to deduce from these figures that able-bodied unemployed adults were probably in some-

what greater objective need in South Africa than in Brazil. According to the functionalist view,

this would have necessitated broader coverage in South Africa than in Brazil — the exact opposite

of what actually transpired. There is thus no functional justification for the provision of universal

coverage in Brazil and only limited coverage in South Africa.

A functional argument for the implementation of conditionalities in Brazil but not in South

Africa is equally unsatisfactory. An explanation for the variation in conditionality according to this

approach again rests upon the assumption that there was some greater objective need in Brazil

for conditions than in South Africa. Given the fact that Brazilian conditionality primarily focuses

on improving health indicators for young children through vaccinations and school attendance for

adolescents, the data would need to show that Brazil fared significantly worse than South Africa on

these measures in order for the argument to be plausible. Using the best available cross-country data,

219
vaccination rates137 are introduced in figure 8.2 as a proxy for testing the functional interpretation

in relation to health conditionalities.

100
90
80
70
Percentage

60
50
40
30
20
10
0
BCG (1995) Hep B (2000) DTP (1995) Polio (1995) Tetanus Measles
(2000) (1995)
South Africa Brazil

Figure 8.2: Vaccination Coverage for Children 0-7 (Source: WHO 2014; World Bank 2015d)

In another refutation of the hypothesis, the data clearly shows that vaccination rates were actu-

ally somewhat higher in Brazil. As such, the policy problem appears to have been relatively more

acute in South Africa, making health conditionalities for young children seem more likely there

than in Brazil. But the actual policy outputs were again contrary to this functionalist expectation.

Owing to a severe dearth of cross-country data, assessing the validity of this explanation for educa-

tional conditionalities presents more of a challenge.138 The only reliable figure capable of serving as

a proxy for school attendance is introduced in figure 8.3, where the net primary school enrolment

rates for South Africa and Brazil in 2000 are compared.

While it is hard to draw firm conclusions, the limited available data make it seem likely that
137
The indicated vaccinations all come from the official schedule used by the PBF.
138
In addition to a simple lack of data for actual daily school attendance, there are significant problems in creating

accurate comparisons due to the potential presence of myriad complicating factors such as dropout rates and grade

repetition. In a reflection of this situation, the data introduced here is thus of a very rudimentary nature.

220
 100

90

80

70

60
Percentage

50

40

30

20

10

0
South Africa Brazil

Figure 8.3: Net Primary School Enrolment Rates (2000) (Source: UNESCO 2011; World Bank 2015e)

enrolment figures in primary and secondary education was not very different across the cases (UIS

2015). But this still says nothing about daily school attendance rates, which is a different metric

from the once-off yearly act of school enrolment. While it is therefore difficult to directly compare

the cases, the latest research on the South African education system makes it clear that the country

features a very high school dropout rate, making it likely that educational conditionalities could

have been functionally justified there as well (Spaull 2015). Taken together with the data on

vaccination rates, the functionalist interpretation is thus unable to conclusively explain the use of

conditionalities in Brazil but not in South Africa.

What about the functional hypothesis’ applicability to the issue of orientation, where the South

African system assesses the need of beneficiaries in individualist terms while Brazil has adopted a

familial approach? Liesering and Barrientos make a functionalist case for focusing on families by

arguing that they

are increasingly the unit of intervention as regards social protection and social assistance. This

focus on households arises from the practice of social protection. In sub-Saharan Africa, for

example, children are a majority of the population. In practical terms, reaching children for

221
 the purpose of ensuring their rights to social protection involves working through their parents,

guardians or carers (2013: 60).

According to this logic, a higher share of children in the population would serve as an important

functional justification for a familial orientation. However, despite the fact that children compose

a greater population share in South Africa than in Brazil139 (as indicated in figure 8.4), it is the

latter rather than the former that has adopted a strict focus on families. The statistics on household

composition shown in figure 8.4 does nevertheless appear to provide some functional basis for South

Africa’s individualist ‘follow the child’ approach. This is due to the fact that a mere 36 percent

of children live in families with a nuclear structure where both parents are present (IPUMS 2015).

This is very different from Brazil, where nuclear families are the norm — with 76 percent of children

living in a family with both parents (IPUMS 2015). Additionally, 70 percent of households in South

Africa include members of the extended family or other adults (IPUMS 2015).

The upshot is that family structures are significantly more diverse in South Africa than in Brazil,

bestowing at least some explanatory credibility on a functional interpretation where policymakers

are concerned with defining child beneficiaries in the most effective way. This is however only

applicable to child-focused programmes like the CSG and certain components of the PBF, and

does not provide a conclusive answer as to why systemic orientation also differs in terms of social

assistance benefits that are exclusive to adults (such as parts of the BPC in Brazil, and the OPG

and DG in South Africa).

At the most general level, it is clearly a truism that some level of poverty and inequality needs to

be present in order for social assistance to become functionally ‘necessary’. However, as the analysis

indicates, it is a gross oversimplification to assume that programme parameters are designed purely

as a mechanistic response to policy problems. In addition to the fact that the functional hypothesis

does not say anything about timing (as these problems persisted for many decades prior to the recent

implementation of social assistance schemes), it also fails to provide a convincing account of why
139
This statistic is for children between the ages of zero and 14 years.

222
 80

70

60

50
Percentage
40

30

20

10

0
Children as Children living Children living Children living Probability of
share of total with 0 biological with 1 biological with 2 biological household
population parents parent parents including
extended family
South Africa Brazil

Figure 8.4: Family Structures (Source: CIA World Factbook 2015a; 2015b; IPUMS 2015)

the South African and Brazilian systems took on different shapes. While functional considerations

thus certainly played some role as part of the policy context, they are unable to explain why South

Africa implemented a productive social assistance system whereas Brazil went the protective route.

8.2.2 The Ideational Hypothesis

Another potential exogenous explanation for the differences between the cases is that ideational

bases were different in South Africa and Brazil. Perhaps the adoption of a productive system in

South Africa simply reflects the salience of ‘neoliberal’ ideology, while Brazil’s protective system

is a function of a more ‘social democratic’ orientation? The ideational hypothesis fundamentally

posits that, in explaining the features of social assistance, ‘it is important to pay attention to the

understandings of poverty which support it’ (Liesering & Barrientos 2013: 59).

As a result,

the ideology of the state is an important factor. . . Neoliberal economic policy encourages the

reduction of social grant programmes through reduced state spending and requires more action

or responsibility on the part of beneficiaries. By contrast, social democratic policies may encour-

223
 age the state to increase commitment to social spending and to increase the capacity of grant

programmes through (in some cases) more universal grant policies’ (Mothiane 2014: 4).

If it is true that ‘differences in political attitudes about redistribution’ can account for variations in

the structure of social assistance, then we would expect to see significant variations in beliefs about

poverty between the two cases under examination (Graham 2002: 2). These beliefs may include

differences in the ideology of the governing parties, as well as in society at large.

The two parties primarily responsible for the construction of the systems, the ANC and the PT,

exhibit a number of striking ideological similarities. Measures of party family classification indicate

that their ideological underpinnings represent a case of ‘most similar’ orientations (Lipset & Rokkan

1967). Specifically, the ANC and PT are parties ‘that mobilised in similar historical circumstances

or with the intention of representing similar interests’ (Gallagher, Laver & Mair 2001: 202). Where

the ANC arose as a leftist party in alliance with labour unions and communists within the context

of the anti-apartheid struggle, the PT was established by trade unionists and socialist intellectuals

in the struggle against the military dictatorship.

In addition to therefore having emerged to represent broadly similar interests, the two par-

ties have actively affirmed their similar ideational beliefs through membership in the transnational

Progressive Alliance. The organisation’s stated aim it to be a global network of ‘progressive, demo-

cratic, social democratic, socialist and labour movements’ (PA 2015). The ANC and the PT were

thus born into similar historical circumstances to represent similar interests, and have voluntarily

affirmed their ideological similarities as mass-based centre-left parties.

But it is also apparent that the ideational orientations of the South African and Brazilian

publics need to be taken into account. If there were vast differences in societal beliefs about the

role of welfare and redistribution, then the ideational hypothesis may still be plausible. In terms of

explaining variations in coverage, the data would have to specifically indicate that concern about

poverty as well as support for redistribution and universal social assistance was greater in Brazil

224
than in South Africa. Data from the World Values Survey provide a useful tool for assessing this

argument by comparing the basic ideational orientations of South Africa and Brazil towards the

public provision of social welfare. Figure 8.5 compares the percentage of people who strongly agreed

with each statement about redistribution.

70

60

50
Percentage

40

30

20

10

0
It is humiliating People who Incomes should Government Reducing
to receive money don't work be made more should tax the extreme poverty
without having become lazy equal rich and is a top priority
to work for it subsidise the
poor
South Africa Brazil

Figure 8.5: Public Attitudes towards Redistribution (Source: WVS 2005-2008)

The results show that there is little variation between the two cases, with the exception of beliefs

about the importance of government taxing the rich and subsidising the poor — a statement that

South Africans agree with much more strongly than Brazilians. In a further set of surveys, 76

percent of Brazilians indicated that people are poor because ‘society is unjust’ while 70.5 percent

agreed with the statement that ‘the poor have very little chance to escape from poverty’ (Fiszbein

et al. 2009: 10). In South Africa, recent research has similarly found that 82.3 percent of people

strongly agree that ‘most people on social grants desperately need the help’, while 65.9 percent

strongly agree with the notion that ‘the government should spend more money on social grants for

the poor, even if it leads to higher taxes’ (Noble, Ntshongwana & Surender 2008: 13-14).

On the specific issue of whether impoverished adults who are unable to find work should be

entitled to social assistance, fully 74.4 percent of South African respondents agreed with the principle

225
that ‘people who can’t get work deserve help in the form of social grants’ (Noble et al. 2008:

10). Given these fairly similar ideational beliefs about public welfare provision, it thus seems

highly unlikely that policymakers were simply responding to societal beliefs about social assistance

provision when they were designing the systems in their respective countries. This is particularly

true in terms of systemic coverage, where strong societal support for extending provision to the

able-bodied unemployed was only translated into policy in Brazil, but not in South Africa. The

ideational hypothesis is therefore unable to account for the differences in coverage.

The empirical evidence further suggests that the absence or presence of conditionalities was also

not ultimately determined by ideational factors. As discussed in the case studies, both cases featured

actors calling for the introduction of conditionalities on ideational grounds. This includes the current

South African Minister of Social Development, who has expressed her desire to incorporate them

in an attempt to ‘take children out of intergenerational poverty’ — language very similar to that

employed in Brazil (PMG 2010). Despite the fact that the most recent amendments to the relevant

legislation allude to this, conditionalities were never imposed in South Africa but have been a part

of the Brazilian approach since the very beginning. An ideational explanation alone cannot account

for this variation between two cases both featuring calls for conditionalities amid broad ideological

similarities.

It further appears to be unlikely that different ideational beliefs about the appropriate mode of

assessing the needs of beneficiaries are able to convincingly account for South Africa’s individualist

orientation, while Brazil has defined beneficiaries in familial terms. As indicated in figure 8.6 below

and in spite of the relatively more precarious position of families in South Africa (as described above),

fundamental beliefs about the importance of ‘promoting family life and strengthening families’ are

similar across the cases (RSA 2012: 8).

It would thus appear that the claim that South Africa’s productive system emerged in contrast

to Brazil’s protective approach purely because South African politics was somehow ‘more neoliberal’

226
 120

100

80
Percentage
60

40

20

0
Family is very important in Greater emphasis on I trust my family
my life families would be a good completely
thing
South Africa Brazil

Figure 8.6: Public Attitudes towards Families (Source: WVS 2005-2008)

and Brazil’s ‘more social democratic’ amounts to another oversimplification (Mothiane 2014). While

the case studies indicate that ideational factors, just like functional elements, were certainly present

during the policy development process, they alone are plainly unable to satisfactorily explain why

South Africa opted for limited coverage, unconditionality and an individualist orientation while

Brazil pursued universal coverage, conditionality and a familial orientation. Instead, the specifics

of policy problems and ideas about how to solve them were themselves shaped and channelled by

the specific configuration of each country’s institutional setup.

8.3 The Institutional Explanation

It is the contention of this study that variation on the institutional dimension is the single most

important factor in explaining the differences between the cases. However, it does this without

outright denying that exogenous elements were also present; the existence of high levels of poverty

and inequality as well as ideational underpinnings surely were relevant factors during the policy

development process. But these elements were institutionally informed and channelled, meaning

227
that the specifics of institutional configuration amount to the fundamental reason for divergent

policy outcomes in cases featuring such broadly similar policy problems and ideational orientations.

At the most general level, the case studies make it clear that the transition to more inclusive

democratic institutions triggered much of the momentum for reform. But the specific shapes of the

reform efforts, and the features of the systems they eventually gave rise to, were determined by the

underlying institutional structure of each case.

8.3.1 Coverage

This is clear when examining the variation in coverage. The discussion on the South African

case highlights the way in which excluding impoverished able-bodied adults from social assistance

provision initially emerged as a norm during the process of segregationist state formation. The

transition to democracy did not break with this fundamental feature, as notions of ‘progressive

realisation’ and ‘fiscal restraint’ made their way into the constitutional-legal framework of the

newly emerging state, providing policymakers with incentives to effectively maintain the status quo

in terms of coverage (even as social assistance was deracialised). Contrary to the Brazilian case,

there were no built-in mechanisms to protect spending on these programmes, which meant that

economic concerns were to crowd out social welfare concerns in cases where they were not aligned

(Hölscher 2008: 116).

The role played by South Africa’s concentrated state structure is also abundantly clear. At no

stage did the ANC face any credible threat to its overwhelmingly dominant power position. Even in

cases where opposition parties have managed to win regional elections in a few local municipalities

or provinces, they quickly found that they had no policymaking capacity whatsoever to design

their own social assistance programmes. The ANC plainly ‘faces little immediate [electoral or

institutional] incentive to expand the welfare state. South Africa is, in this respect, very different

from many other countries, including Brazil...where there have been, and are, strong...pressures to

expand the social welfare system’ (Seekings 2008: 35).

228
 The established historical practice of only providing limited coverage (due to the deep entrench-

ment of institutional norms) and the lack of political incentives (due to a highly concentrated state

structure) meant that the functional and ideational arguments made by actors serving on the Lund

Committee (1996), Taylor Committee (2002) or potentially even the NPC (2010) for expanding

coverage to include impoverished people between the ages of 18 and 60 years were doomed to fail.

The government in fact went in the exact opposite direction by excluding this group, rejecting calls

for a BIG, and instead implementing the EPWP. The decision to implement this programme in lieu

of providing social assistance coverage was only possible because of the country’s particular insti-

tutional configuration: a concentrated state structure that did not create adequate incentives for

expanding provision, combined with a deeply embedded historical commitment to only providing

limited coverage.

The opposite happened in Brazil, where coverage has been systematically expanded to the point

where social assistance has in practise become nearly universally available to those in extreme

poverty. The lack of historically established social assistance practises meant that policymakers

were not beholden to any given pattern of provision. In what may initially appear to be a paradox,

the near-complete absence of such programmes prior to democratisation subsequently made it easier

to pursue universal coverage. This was because (historically constructed) concerns regarding ‘pro-

gressive realisation’ and ‘fiscal restraint’ were not incorporated into the Brazilian constitutional-legal

framework. Instead, the case’s ‘blank slate’ allowed constitutional architects to construct mecha-

nisms with the potential to ‘increase the stability of social policy funding’ (Interview 12; Schwarzer

& Querino 2002: 4). As the LOAS itself made clear, the result was the ‘supremacy of meeting social

needs above the requirements of economic profitability’ (FRB 1993: art. 4).

Brazil’s diffuse state structure further played a decisive role in driving the universalisation of

social assistance provision. In addition to the creation of the BPC, the combination of the fact that

subnational governments had the authority to independently design policy interventions and the

229
intensity of political competition at all levels in Brazil fuelled a gradual but persistent bottom-up

expansion in coverage. Municipalities and states outright copied the earliest examples of municipal

CTs during the mid-1990s in an attempt to profit from the evident popularity of the programmes.

The federal PSDB government subsequently sought to associate itself with the policies through the

1998 introduction of the PGRM and the creation of the first federal interventions in 2001. This

was followed by the adoption of the BIG law, federal consolidation and full universalisation into

the PBF under the PT in 2004, as well as the 2012 introduction of the residual benefit to eradicate

extreme poverty through social assistance provision.

The lack of historical precedents regarding coverage (due to the weak institutionalisation of

norms) and the presence of clear political incentives (produced by a diffuse state structure) led to

the adoption of a universalistic approach to social assistance provision in Brazil. This institutional

configuration channelled the functional and ideational considerations of policymakers, who had both

more room and greater incentives to implement policies that produced significant political payoffs

among as many beneficiaries as possible. The result was that universal coverage in Brazil, as well

as limited coverage in South Africa, was not principally the result of functional factors or ideology.

The variation was fundamentally directed by the particularities of each case’s institutional features.

8.3.2 Conditionality

A similar pattern is evident in accounting for variation in terms of conditionalities. The empiri-

cal discussion makes it clear that an unconditional approach to social assistance had already be-

come a deeply entrenched institutional norm in South Africa prior to the transition to democracy.

In a powerful expression of path dependence, this norm was transferred wholeheartedly into the

constitutional-legal structure of the post-apartheid state. The Constitution explicitly prohibited

the implementation of conditions, a feature which was reinforced in the WPSW and subsequent

legislation.

A top-down process of policy formulation determined by the country’s highly concentrated state

230
structure ultimately led to the entrenchment of an unconditional approach. As the case analysis

makes clear, there were nevertheless calls from civil society actors to introduce some linkages to the

CSG on functional grounds, while the current Minister of Social Development herself and policy

actors close to her have all expressed some ideational preference for introducing conditionalities

(Lund 2008: 68; Interview 7; PMG 2010). But none of these efforts were successful, because

the institutional structure did not allow for it. The lack of conditionalities in South Africa did

therefore not result because policy problems or ideological orientations were fundamentally different

from those in Brazil. Instead, faced with little competition or political incentive (due to a highly

concentrated state structure) to impose conditionalities that went against established practices (due

to the deep entrenchment of institutional norms), policymakers in South Africa designed a system

based on an unconditional approach.

The Brazilian usage of conditionalities for the child-focused components of the PBF stands in

complete contrast to this. Norms about conditionality were not entrenched in the decades prior to

democratisation, as the social assistance system remained largely undeveloped. This in turn meant

that conditionality for children was not prohibited in the constitutional-legal framework. The

Constitution rather left open the possibility of incorporating conditionality into these programmes,

while the LOAS and derivative legislation further affirmed this approach.

Conditionalities came to be a feature of the Brazilian system through a bottom-up process of

policy development channelled by the country’s diffuse state structure, which enabled local govern-

mental units to independently design policies within a highly competitive political context. The case

study makes it clear that subnational actors, although also influenced by functional considerations

and ideological orientations, were fundamentally encouraged by this diffuse state structure and the

weak institutionalisation of norms to adopt a conditional approach. Motivated by the skyrocketing

popularity of early conditional programmes, policymakers embarked on a remarkable process of

policy diffusion that culminated in the eventual establishment of the contemporary system.

231
 Conditionality in Brazil did thus not result because policy problems or ideational beliefs were

fundamentally different from those in the South African case. Rather, policymakers responded to

an institutional setting characterised by competition and clear political incentives (due to a diffuse

state structure) that did not constitutionally prohibit conditionalities (due to the weak historical

development of institutional norms) by incorporating a conditional approach which had proven to

be politically advantageous.

8.3.3 Orientation

The same analytical pattern is visible when it comes to accounting for the fact that South Africa

defined potential beneficiaries in individualist terms, while Brazil adopted a familial orientation.

The process tracing in terms of the South African case again shows how deeply entrenched the

practise of means testing potential beneficiaries on an individualist basis had become during the

formative pre-democracy decades of the system. Examining the constitutional-legal framework

makes it clear that this norm had also been transferred into the contemporary institutional context

by placing relatively greater emphasis on individual equality. This encouraged the preservation of

the norm of defining beneficiaries individually and independent of family structures.

The country’s concentrated state structure once more served to prevent any meaningful chal-

lenges to this norm from arising. The few regional governments under opposition control were pre-

cluded by this structure from designing any potential alternative arrangements, while the ANC’s

strongly fortified national political position did not incentivise it to challenge the status quo. While

the earlier discussion shows that there is indeed some functional justification for defining beneficia-

ries in individualist terms in the South African case, the decision to do so was not fundamentally

directed by functional considerations. Instead, the engrained practise of assessing beneficiaries in

individualist terms (due to the deep entrenchment of institutional norms) and the lack of any polit-

ical rationale for challenging this established method (due to a highly concentrated state structure)

led to the effective maintenance of the status quo in terms of orientation.

232
 The Brazilian case meanwhile highlights the way in which the lack of norms regarding orientation

enabled the adoption of a familial approach. The near-total lack of social assistance provision

prior to the 1990s meant that path dependence and associated institutional stickiness was not

made manifest during the construction of the contemporary constitutional-legal framework. Early

policymakers took advantage of this policy space by adopting the position that the family unit was

‘the appropriate entity to receive [social assistance] benefits and...bear responsibility for meeting

programme requirements’ (Lindert 2006: 68).

The country’s diffuse state structure subsequently served to embed the practise of defining

beneficiaries in familial terms. The policy autonomy of different governmental spheres combined

with the ferocity of competition produced incentives to copy and spread the operational features

of programmes that had proven to be hugely popular right from their onset in the mid-1990s. The

bottom-up policy development process meant that each round of programme proliferation further

entrenched the practise of assessing the needs of beneficiaries in familial terms, culminating in the

eventual adoption of the approach for all federal social assistance schemes.

The relative dearth of historical precedents for defining social assistance beneficiaries (due to

the weak institutionalisation of norms) and the presence of clear political incentives to implement

programme features that were popularly perceived as being effective (due to the country’s diffuse

state structure) led to the enactment of a familial orientation to social assistance provision in

Brazil. As was the case with coverage and conditionality, this particular institutional configuration

channelled functional and ideational factors in a direction that eventually led to the adoption of an

approach which differed greatly from the one embraced in the South African case.

8.3.4 Productive versus Protective

Synthesising the above findings allows for a return to the dependent variables introduced during the

descriptive overview of social assistance provision in South Africa and Brazil. It will be recalled that

the features of the South African case — limited coverage, unconditionality and an individualist

233
orientation — were categorised as constituting a productive regime. This regime centres on relatively

less decommodification and emphasises that the able-bodied poor are to rely exclusively on the

labour market for overcoming poverty. In South Africa, the adoption of the EPWP instead of

a BIG law or universal social assistance is a clear marker of this approach. It also incorporates

other ‘liberal’ features, such as not making individual choice conditional and defining beneficiaries

in individualist terms.

This stands in contrast to the Brazilian case, where universal provision, conditionality and a

familial orientation led to its classification as a protective system. This course involves a ‘social

democratic’ element through relatively greater decommodification based on universal provision that

emphasises the role of social assistance in overcoming poverty. In the case of Brazil, the adoption

of a BIG law and the explicit framing of the PBF’s role in eradicating extreme poverty (especially

following the introduction of the PBSM) are indicators of this. This regime furthermore contains

a ‘conservative’ streak through its incorporation of conditionality and the practise of denoting

beneficiaries in familial terms.

The case analysis shows that the features constituting these regimes did not mechanistically arise

as a result of differing policy problems or profoundly different ideational orientations. In addition to

problems and ideological beliefs being broadly similar across the cases, testing these two competing

claims in fact demonstrates a number of instances where policy choices were in direct conflict with

what would have been expected if functional and ideational factors were paramount. Despite the

allure of these exogenous interpretations, it is instead endogenous variation on the institutional level

that is able to most convincingly account for the fact that South Africa and Brazil have implemented

qualitatively different social assistance regimes.

The deep historical entrenchment of norms related to coverage, conditionality and orientation

combined with a highly concentrated state structure produced powerful path dependence effects

that resulted in the (now deracialised) maintenance of a productive system in South Africa. In

234
contrast, the weak historical rooting of institutional norms regarding coverage, conditionality and

orientation in combination with a diffuse state structure created an environment largely free from

path dependent institutional stickiness that both enabled and incentivised political actors to enact

reforms which produced the greatest political payoffs and eventually resulted in the implementation

of a protective social assistance system in Brazil.

8.4 Conclusion

This chapter has utilised the empirical discussions introduced during the case studies in order to

assess the veracity of the functional, ideational and institutional hypotheses. It initially assessed

whether the presence of different policy problems and ensuing functional responses across the cases

were able to satisfactorily account for the cross-case variation. The examination showed that in ad-

dition to the cases featuring broadly similar problems, a number of areas produced policy responses

which were actually in direct conflict with the logic of the functional hypothesis. A similar pattern

emerged when the claims of the ideational interpretation were tested, as ideological beliefs and

societal views regarding coverage, conditionality and orientation were not found to be substantially

different.

This was followed by an assessment of the study’s institutionalist claim. The discussion showed

that variations on the institutional dimension are able to most convincingly account for the vari-

ations on the dependent variable. The features of South Africa’s contemporary productive social

assistance regime, predicated on limited coverage, unconditionality and an individualist orientation,

fundamentally arose out of the fact that norms which were historically deeply embedded were trans-

ferred into the institutional structure of the post-apartheid state, while a highly concentrated state

structure further incentivised the preservation of these practices. In contrast, the analysis indicated

that the emergence of Brazil’s contemporary protective system based on universal coverage, condi-

tionality and a familial orientation was directed by an institutional framework that featured weak

235
entrenchment of norms and provided much greater space for policy innovation, as well as a diffuse

state structure which incentivised and rewarded policies popularly perceived as being effective.

236
Chapter 9

Concluding Remarks

9.1 Summary of Findings

The remarkable proliferation of redistributive social assistance initiatives across the developing world

shows no signs of abating, with prominent proposals for the United Nations Sustainable Develop-

ment Goals (SDGs) regarding them as a ‘development priority in the post-2015 development agenda’

(ECA, ILO, UNCTAD, UNDESA & UNICEF 2012). In addition to expanding the scholarship on

emerging welfare states, social assistance in South Africa and Brazil, as well as historical institu-

tionalism, the urgent global relevance of this theme serves to further underscore the significance of

the resultant findings.

Discussions around this topic have largely centred on examining the social and economic effects

produced through these programmes. But there has been a glaring lack of emphasis on exploring the

factors which shape the emergence of such policies in the first place. This study’s research question

addressed this omission head-on by launching an inquiry into the variation between the South

African and Brazilian models of redistributive social assistance provision. Following a specification

of the conceptual and analytical framework, a review of the literature that clearly illustrated the

relevance and novelty of the project, as well as a discussion on the inquiry’s research design, the

237
first step in this process involved a descriptive exercise that sought to clearly define the variation

across the cases.

This involved a highly detailed technical evaluation of the social assistance systems of South

Africa and Brazil, addressing operational features such as coverage, the respective legal and institu-

tional frameworks, conditionality, eligibility, orientation, appeals mechanisms, delivery, and specify-

ing the individual programmes. The analysis ultimately showed that there were clear differences in

the policy regimes of the two countries regarding coverage, conditionality and systemic orientation.

These differences were subsequently classified into a set of distinct typologies which constituted the

study’s dependent variables, with the South African system being regarded as protective and the

Brazilian regime as productive.

The case study chapters subsequently conducted historical institutional ACI analyses of the

development of social assistance policies in South Africa and Brazil. This was achieved through

initially examining the historical emergence of particular norms, combined with a specification of

each case’s state structure. In the case of South Africa, the analysis found that norms regarding

coverage, conditionality and orientation had become deeply entrenched in the decades prior to

democratisation. In combination with the country’s highly concentrated state structure, these

norms went on to have powerful path dependence effects in shaping the contemporary productive

system. In contrast, the Brazilian case clearly showed the way in which historical norms were absent

in the period leading up to democratisation. This feature combined with the country’s diffuse state

structure to drive a policy development process that ultimately resulted in the implementation of

a protective system.

The case studies were followed by a comparative chapter which returned to the initial set of

hypotheses proffered as potential answers to the research question. It conducted an assessment of

the two most commonly discussed explanations, premised on exogenous functional and ideational

factors, and found them both wanting. The variation across the cases in terms of the dependent

238
variables could simply not be satisfactorily explained by these two interpretations. Instead, the

latter part of the analysis integrated the insights from the case studies in order to show that

the differences between the systems in fact primarily resulted from variation on the institutional

dimension. Whereas South Africa featured deeply entrenched historical norms regarding coverage,

conditionality and orientation, as well as a highly concentrated state structure, Brazil clearly lacked

such norms and operated under a diffuse state structure. These differences amount to qualitatively

distinct institutional frameworks which ultimately channelled policy outputs in different directions.

These findings are of great significance given the prominence that has been afforded to this

field by both scholars and practitioners. On the theoretical level, the analysis extends the histor-

ical institutionalist ACI analytical framework into an exciting new area. Empirically, it makes a

tremendous contribution by offering the first answer to the question of why there is a clear pat-

tern of divergence in the social assistance regimes implemented across the developing world; the

differences result from distinct local institutional configurations. This finding in turn feeds into

the broader literature on the rise of emerging welfare states incorporating different distributional

regimes. Finally, the practical significance of the study is to provide development and policy prac-

titioners with powerful evidence that local institutional configurations should feature prominently

in any future efforts to implement social assistance programmes elsewhere. Taken together, the

findings thus entail a theoretical extension of historical institutionalism; produce novel empirical

answers to a popular scholarly field of inquiry; and practically inform future social assistance policy

development.

239
9.2 Generalizability

The significance of these findings are further enhanced through briefly examining their relevance

in the case of another prominent emerging welfare state: Mexico.140 While it is important to

again emphasise that the universe of potential further cases is fairly limited due to the fact that

much of the global proliferation of social assistance programmes has subsequently been driven by a

somewhat distinct process often involving financing and guidance from international development

organisations, it is revealing to briefly consider the Mexican case in order to ascertain whether the

processes at work resemble those which took place in either of the cases examined in this study.

A cursory review of the Mexican case indeed illustrates that it strongly mirrors the develop-

mental processes which were at work in Brazil. Like Brazil, Mexico followed a pattern of welfare

state building throughout the twentieth century which was very much predicated upon the delivery

of social insurance. It was also only during the 1970s that Mexico first introduced programmes

resembling social assistance interventions (Lopez 2011: 295). But the first true step towards the

elaboration of the current social assistance system operating in Mexico, which covers upwards of 25

percent of the total population, was taken with the 1997 introduction of the Progresa CT.

Initially modelled on the Brazilian PBE and supported by the World Bank, this programme

rapidly grew to the point where it currently serves more than 5.8 million families across the country

(Lopez 2011: 297). Now renamed Prospera, it is child-focused and incorporates educational and

health conditionalities with a strict familial orientation. This programme is complemented on

the other side of the spectrum by the 70 y más (70 and Over) old age pension (Willmore 2014).

This intervention, which currently supports 3.5 million elderly Mexicans, emerged in a decidedly

bottom-up fashion. It was initially introduced by the governor of the federal district in 2001 and

subsequently spread throughout all states of the country before recently being fully federalised.
140
The prominence of the Mexican case is confirmed through the fact that it has been the subject of more impact

evaluations than any other.

240
 In 2009 the Mexican government additionally moved towards the universalisation of the country’s

social assistance system through the creation of the Programa de Apoyo Alimentario (Food Support

Programme) (UMS 2014). This nutritional aid programme targets impoverished families who do

not qualify for Prospera, including those without children (UMS 2014). It pays an amount of $310

(US$20) per month, along with in-kind transfers, to more than 26 million households (Barrientos

et al. 2010). The similarities between the Brazilian and Mexican systems thus include a conceptual

approach to coverage premised on universalism,141 the incorporation of conditionalities for child-

focused programmes, as well as a familial orientation. In short, Mexico appears to have also adopted

a protective policy regime.

The findings produced by this study suggests that this commonality is the result of similari-

ties in terms of institutional configuration. It is indeed the case that Mexico featured a similarly

weak entrenchment of social assistance norms prior to the consolidation of democracy during the

2000s (Bruhn 2009). The contemporary Mexican state structure also strongly resembles Brazilian

federalism, with a high level of formal policymaking decentralisation and a diffusion of political

power.142

This correlation indicates that the development of the Mexican system may well have been

shaped by institutional features similar to those which were present in the Brazilian case. This

further enhances the significance of the findings, as it suggests that similar processes may have been

at work in many other cases — and that domestic institutional configurations will continue to exert

their influence as the proliferation of social assistance interventions continues unabated across the

countries of the developing world.

141
Although, in practise, the Brazilian system currently provides broader coverage.
142
The current era of greater competition was inaugurated by the liberalising reforms of the 1990s and the 2000

electoral victory of Vicente Fox, leading to the first change of government in 71 years.

241
9.3 Opportunities for Future Research

This also makes it clear that the research agenda on emerging welfare states will benefit from simi-

larly detailed inquiries into the social assistance policy regimes adopted in those other cases where

these programmes have also come to occupy centrally important positions. The perceived success

of these policies in cases where they were domestically designed has further spurred international

development organisations, spearheaded by the World Bank, into developing and financing similar

programmes in a range of countries that are too poor to autonomously afford them. This presents a

further opportunity for researchers interested in examining the processes whereby these programmes

have proliferated in recent decades. The study thus ultimately lays the groundwork for future in-

quiries in a field that is set to assume even greater academic and practical prominence as an ever

increasing portion of the world’s population gains access to social protection.

These findings additionally expand avenues for potential future research by conclusively estab-

lishing the practise of conducting comparative political economy analyses of emerging welfare states.

As discussed in chapter two, this is a practise that has become widely accepted when assessing the

constellation of welfare regimes extant in the developed world. The typologies constructed and

findings produced by this study however provides a powerful justification for undertaking similarly

extensive analyses with regards to the countries of the global South. The recent spread of social pro-

tection initiatives throughout the developing world means that scholars currently have a tremendous

opportunity to gain insights into the different distributional regimes that exist in these countries.

In terms of the two cases under examination, the greatest opportunity lies in expanding the

scope of the research. The many similarities across the cases means that there are a whole host of

other social policy arenas — including social insurance, labour protection, and healthcare — that

are deserving of similarly focused comparative efforts. While the topic of this study addresses a

vitally important aspect of the overall distributional regimes of these two cases, the other elements

comprising such a regime are certainly also worthy of further study. Although the social assistance

242
programmes examined here are arguably the most significant redistributive interventions, a potential

goal of future comparative research efforts would thus be to produce a complete picture of the nature

and origin of related policies in the important emerging welfare states of South Africa and Brazil.

9.4 Recommendations for Policy Practitioners

A number of relevant practical recommendations also result from these findings. At the most general

level, the study serves to affirm the decisive importance of ‘institutional fit’ in the implementation of

social protection programmes. The fact that two cases characterised by such tremendous similarity

and facing broadly similar problems nevertheless ended up with greatly varying policy approaches

serves as a clear testament to the decisive importance of local institutions in shaping successful

and scalable interventions. The additional fact that both South Africa and Brazil domestically

and autonomously designed their social assistance regimes without the guidance of international

development organisations means that their cases represent something akin to ‘natural experiments’.

And the results produced by both ‘experiments’ make it clear that large scale social protection

programmes are first and foremost shaped by local institutional configurations.

As can be observed from the case analyses carried out in earlier chapters, one of the primary

concerns for policy practitioners interested in implementing functionally effective social protection

schemes relate to the maximisation of the ability of such programmes to address particular policy

problems. The foundational idea is to create a system which would operate as effectively as possible

within the bounds of its given resources. In turn, a significant portion of achieving this aim hinges

on the scalability of such programmes. Indeed, during the previous two decades, international

development organisations have often initially launched such interventions on the basis of pilot

projects with the aim of scaling them up only after evaluations have proven such scalability to be

feasible.

This research suggests that practitioners would be well served by efforts to investigate and

243
understand the local institutional context prior to launching such projects. This would aid in

avoiding situations where, for example, exclusive responsibility for policy design and implementation

is assigned to the national government sphere in a context characterised by a diffuse state structure.

This is unlikely to be a sustainable approach. Likewise, preceding research aimed at establishing

the operational norms of any potentially pre-existing social protection practices (both formal and

informal) would allow for the implementation of interventions which do not challenge entrenched

implementation conventions.

Apart from such specific examples, the ultimate practical contribution of the study is to compre-

hensively reorient perspectives on the emergence of social welfare policies in the developing world.

Far from operating within a vacuum which yearns only for the enlightened advice of policy experts,

the continued rise of emerging welfare states is just as much shaped by domestic institutional struc-

tures as their counterparts in the developed world were. Incorporating this reality would go a long

way towards enabling an approach to development policy which takes seriously the particularities

of institutional configuration on the road towards designing interventions which are both effective

and sustainable.

244
 Bibliography

Abramovay, M., Andrade, C. & Waiselfisz, J. G. 1998. Bolsa-Escola, melhoria educacional e redução

da pobreza. Brası́lia: UNICEF, UNESCO.

Abu Sharkh, M. & Gough, I. 2009. ‘Global Welfare Regimes: A Cluster Analysis’ in Stanford Uni-

versity Center for Democracy, Development and the Rule of Law Working Paper, 91: 1-31.

Acemoglu, D. & Robinson, J. A. 2006. Economic Origins of Dictatorship and Democracy. Cambridge:

Cambridge University Press.

Adato, M. & Hoddinott, J. 2009. ‘Conditional Cash Transfer Programs: A “Magic Bullet” for Reduc-

ing Poverty?’ in Von Braun, J., Hill, R. E. & Pandya-Lorch, R. (eds.), The Poorest and Hungry:

Assessments, Analyses, and Actions. Washington: International Food Policy Research Institute.

AED (African Elections Database). 2014. ‘Elections in South Africa’ on African Elections Database

[online] Available at: http://africanelections.tripod.com/za.html [Accessed 11 September 2014].

Aguëro, J. M., Carter, M. R. & Woolard, I. 2006. ‘The Impact of Unconditional Cash Transfers on

Nutrition: The South African Child Support Grant’ in SALDRU Working Paper Series, 6 (8):

1-32.

Aguiar, M. & Arujo, C. H. 2002. Bolsa Escola: educação para enfrentar a pobreza. Brası́lia: UNESCO.

Ahmed, S. S. 2005. ‘Delivery Mechanisms of Cash Transfer Programs to the Poor in Bangladesh’ in

World Bank Social Protection Discussion Paper Series, 0520: 1-73.

245
Akresh, R., De Walque, D. & Kazianga, H. 2013. ‘Cash Transfers and Child Schooling: Evidence

from a Randomized Evaluation of the Role of Conditionality’ in The World Bank Policy Research

Working Paper, 6340.

Almond, G. 1990. A Discipline Divided: Schools and Sects in Political Science. Newbury Park: Sage

Publications.

Amann, E. & Barrientos, A. 2014. ‘Is there a new Brazilian model of development? Main findings

from the IRIBA research programme’ in International Research Initiative on Brazil and Africa

Working Paper, 13: 1-25.

Amaral, C. & Ramos, S. 1999. Programas de Renda Mı́nima e Bolsa-Escola: Panorama Atual e

Perspectivas. Rio de Janeiro: UNICEF.

Ambler, K. 2011. ‘Bargaining with Grandma: The Impact of the South African Pension on Household

Decision Making’ in University of Michigan Population Studies Center Research Report, 11-741:

1-65.

Ames, B. 1995. ‘Electoral Rules, Constituency Pressures, and Pork Barrel: Bases of Voting in the

Brazilian Congress’ in Journal of Politics, 57: 324-343.

—— 2001. The Deadlock of Democracy in Brazil. Ann Arbor: University of Michigan Press.

Arts, W. & Gelissen, J. 2002. ‘Three worlds of welfare capitalism or more? A state-of-the-art report’

in Journal of European Social Policy, 12 (2): 137-158.

Baird, S., Ferreira, H. G., Özler, B. & Woolcock, M. 2013. Relative Effectiveness of Conditional

and Unconditional Cash Transfers for Schooling Outcomes in Developing Countries: A Systemic

Review. Oslo: The Campbell Collaboration.

Baird, S., McIntosh, C. & Özler, B. 2011. ‘Cash or Condition? Evidence from a Cash Transfer

Experiment’ in Quarterly Journal of Economics, 126 (4): 1709-1753.

246
Baldwin, P. 1996. ‘Can we define a European Welfare State Model?’ in Greve, B. (ed.), Comparative

Welfare Systems: The Scandinavian Model in a Period of Change, 29-44. London: MacMillan

Publishers.

Baloyra, E. 1987. Comparing New Democracies: Transition and Consolidation in Mediterranean

Europe and the Southern Cone. Boulder: Westview Press.

Barchiesi, F. 2005. Social Citizenship and the Transformations of Wage Labour in the making of

Post-Apartheid South Africa, 1994-2001. Johannesburg: Doctoral Dissertation, University of the

Witwatersrand.

—— 2006. ‘The Debate on the Basic Income Grant in South Africa: Social Citizenship, Wage Labour

and the Reconstruction of Working-Class Politics’ presented at Harold Wolpe Memorial Trust’s

Tenth Anniversary Colloquium. Cape Town, 21-23 September 2006.

Barrientos, A. 2013. Social Assistance in Developing Countries. Cambridge: Cambridge University

Press.

Barrientos, A., Niño-Zarazúa, M. & Maitrot, M. 2010. Social Assistance in Developing Countries

Database, Version 5.0. Manchester: Brooks World Poverty Institute and Chronic Poverty Research

Centre.

Bates, R. H. 1981. Markets and States in Tropical Africa: The Political Basis of Agricultural Policies.

Berkeley: University of California Press.

—— 1989. Beyond the Miracle of the Market: The Political Economy of Agrarian Development in

Rural Kenya. Cambridge: Cambridge University Press.

BBC. 2011. ‘Brazil launches scheme to lift millions out of poverty’ on BBC News [online] Available

at: http://www.bbc.com/news/world-latin-america-13626951 [Accessed 18 January 2015].

247
Becker, K. F. 2004. Fact Finding Study: The Informal Economy. Stockholm: Swedish International

Development Cooperation Agency.

Beltrão, K. I., Pinheiro, S. S. & Oliveira, F. E. 2002. ‘Population and Social Security in Brazil: An

Analysis with Emphasis on Constitutional Changes’ in Instituto de Pesquisa Econômica Aplicada

Working Paper, 862: 1-44.

Berger, D. L. 1983. White Poverty and Government Policy in South Africa, 1892-1934. Philadelphia:

Doctoral Dissertation, Temple University.

Bither-Terry, R. 2013. Zero Hunger: The Politics of Anti-Hunger Policy in Brazil. Chapel Hill:

Doctoral Dissertation, University of North Carolina at Chapel Hill.

Blekesaune, M. & Quadagno, J. 2003. ‘Public Attitudes toward Welfare State Policies: A Compara-

tive Analysis of 24 Nations’ in European Sociological Review, 19 (5): 415-427.

Blöndal, J. R., Goretti, C. & Kristensen, J. K. 2003. ‘Budgeting in Brazil’ in OECD Journal on

Budgeting, 3 (1): 97-131.

Bloomberg, L. D. & Volpe, M. 2012. Completing your Qualitative Dissertation: A Road Map from

Beginning to End. Thousand Oaks: Sage Publications.

Boadle. 2013. ‘Brazil’s Rousseff says extreme poverty almost eradicated’ on Reuters [online] Avail-

able at: http://www.reuters.com/article/2013/02/19/us-brazil-poverty-idUSBRE91I14F20130219

[Accessed 18 January 2015].

Boessen, A. C. M. 2008. The politics of European Union health policy-making. Maastricht: Doctoral

Dissertation, University of Maastricht.

Bohn, S. R. 2011. ‘Social Policy and Vote in Brazil: Bolsa Famı́lia and the Shifts in Lula’s Electoral

Base’ in Latin American Research Review, 46 (1): 54-79.

248
Bond, P. 2014. “‘Talk left, walk right” in South African Social Policy’ presented at University of

KwaZulu-Natal Centre for Civil Society Seminar. Durban, March 2014.

Bonoli, G. 1997. ‘Classifying Welfare States: A Two-Dimensional Approach’ in Journal of Social

Policy, 26 (3): 351-372.

Bonoli, G. & Natali, D. 2012. The Politics of the New Welfare State. Oxford: Oxford University

Press.

Bratton, M. & Van de Walle, N. 1997. Democratic Experiments in Africa: Regime Transition in

Comparative Perspective. Cambridge: Cambridge University Press.

Britto, T. 2011. ‘Conditional Cash Transfers and the Basic Income Grant in Brazil: Will They Ever

Merge?’ in International Policy Centre for Inclusive Growth One Pager, 124.

Brockerhoff, S. 2013. ‘A Review of the Development of Social Security Policy in South Africa’ in

Studies in Poverty and Inequality Institute Working Paper, 6: 1-46.

Bromberger, N. 1982. ‘Government Policies affecting the Distribution of Income’ in Schrire, R. (ed.),

South Africa: Public Policy Perspective, 165-203. Johannesburg: Juta and Company.

Bruhn, K. 2009. ‘Mexico: Democracy and the Future’ in Insights on Law and Society, 9 (3): 1-30.

Buarque, C. 2013. Bolsa-Escola: History, Theory and Utopia. Brası́lia: Thesaurus Editora.

Budlender, D. 1996. ‘The Circumstances of Young Children 0 to 7 in South Africa’ report prepared

for the Lund Committee on Child and Family Support.

Caccia Bava, S., Da Silva Telles, V., Bejerano, S. R., Araújo, C. H. & De Sousa, N. H. B. 1998.

Programa de Renda Mı́nima no Brasil: Impactos e Potencialidades. São Paulo: POLIS.

Cardoso, E. & Souza, A. P. 2004. ‘The Impact of Cash Transfers on Child Labour and School

Attendance in Brazil’ in Vanderbilt University Working Paper, 04 (07): 1-51.

249
Carnegie Commission. 1932. The Poor White Problem in South Africa: Report of the Carnegie

Commission: Joint Findings and Recommendations. Stellenbosch: Ecclesia Publishers.

Carneiro, J. D. 2014. ‘IBGE: redução da desigualdade no Brasil estaciona nos nı́veis de 2011’ on

BBC Brasil [online] Available at: http://www.bbc.co.uk/portuguese/noticias/2014/09/140918

desigualdade ibge brasil pnad rb [Accessed 26 January 2015].

Case, A. & Deaton, A. 1998. ‘Large Cash Transfers to the Elderly in South Africa’ in The Economic

Journal, 108: 1330-1361.

CEF (Caixa Econômica Federal). 2014. ‘Bolsa Famı́lia’ on Caixa [online] Available at: http://

www.caixa.gov.br/Voce/Social/Transferencia/bolsa familia/como receber.as [Accessed 19 Febru-

ary 2014].

CESCR. 1990. ‘General Comment No. 3: The Nature of States Parties’ Obligations’ in United Nations

Document, E/1991/23. Geneva: United Nations.

Chenwi, L. 2010. ‘Monitoring the Progressive Realisation of Socio-Economic Rights: Lessons from

the United Nations Committee on Economic, Social and Cultural Rights and the South African

Constitution’ in Studies in Poverty and Inequality Institute Research Paper.

CIA World Factbook. 2015a. ‘South Africa’ on CIA World Factbook [online] Available at: https:

//www.cia.gov/library/publications/the-world-factbook/geos/sf.html [Accessed 4 February 2015].

CIA World Factbook. 2015b. ‘Brazil’ on CIA World Factbook [online] Available at: https://www.

cia.gov/library/publications/the-world-factbook/geos/br.html [Accessed 4 February 2015].

Cnaan, R. A. 1992. ‘Three Worlds of Welfare Capitalism’ in Acta Sociologica, 35 (1): 69-71.

Cochrane, A. 1993. ‘Comparative Approaches and Social Policy’ in Cochrane, A. & Clarke, J. (eds.),

Comparing Welfare States: Britain in International Context. London: Sage Publications.

250
Coêlho, D. B. 2012. ‘Political Competition and the Diffusion of Conditional Cash Transfers in Brazil’

in Brazilian Political Science Review, 6 (2): 56-87.

Collie, J. 1924. Memorandum on the subject of Old Age Pension Scheme for the Union of South

Africa, together with a brief description of various existing schemes. Pretoria: Government

Gazette.

Côrtes, S. V. 2013. ‘Brazil’s National Council for Social Assistance and the policy community sup-

porting social assistance as a right’ in Vaitsman, J., Riberio, J. M. & Lobato, L. (eds.), Policy

Analysis in Brazil, 135-149. Bristol: Policy Press.

Creswell, J. W. 2007. Qualitative Inquiry and Research Design. Thousand Oaks: Sage Publications.

Das, J., Do, Q. & Özler, B. 2005. ‘Reassessing Conditional Cash Transfer Programs’ in The World

Bank Research Observer, 20 (1): 57-80.

Davie, G. 2005. Poverty Knowledge in South Africa: The Everyday Life of Social Science Expertise

in the Twentieth Century. Ann Arbor: Doctoral Dissertation, University of Michigan.

Davis, M., Hinich, J. & Ordeshook, P. C. 1970. ‘An Expository Development of a Mathematical

Model of the Electoral Process’ in The American Political Science Review, 64 (2): 426-448.

Denzin, N. K. & Lincoln, Y. S. 2005. The Sage Handbook of Qualitative Research. Thousand Oaks:

Sage Publications.

De Janvry, A., Finan, F., Sadoulet, E. & Vakis, R. 2006. ‘Can conditional cash transfer programs

serve as safety nets in keeping children at school and from working when exposed to shocks?’ in

Journal of Development Economics, 79: 349-373.

De Janvry, A. & Sandoulet, E. 2005. ‘Conditional cash transfer programs for child human capital

development: Lessons derived from experience in Mexico and Brazil’ presented at GRADE 25th

Anniversary Conference. Lima, 15-17 November 2005.

251
Dekker, A. H. 2008. ‘Mind the gap: suggestions for bridging the divide between formal and informal

social security’ in Law, Democracy and Development, 12 (1): 117-131.

Dethier, J. J. 2007. ‘Social Security: What Can Developing Countries Learn from Developed Coun-

tries?’ in 2020 Focus Brief on the World’s Poor and Hungry People. Washington: International

Food Policy Research Institute.

Diamond, L. 1999. Developing Democracy: Toward Consolidation. Baltimore: Johns Hopkins Uni-

versity Press.

Dlamini, T. & Simkins, C. 1992. ‘The Problem of Supporting Poor Children in South Africa’ in

Burman, S. & Preston-Whyte, E. (eds.), Questionable Issue: Illegitimacy in South Africa, 16-42.

Cape Town: Oxford University Press.

Downing, B. M. 1992. The Military Revolution and Political Change: Origins of Democracy and

Autocracy in Early Modern Europe. Princeton: Princeton University Press.

Doyle, M. 1986. Empires. Ithaca: Cornell University Press.

DPW (Department of Public Works). 2012. Expanded Public Works Programme Phase II Training

Framework. Pretoria: Republic of South Africa.

—— 2012. ‘Budget Vote Speech by the Minister of Social Development, Ms Bathabile Dlamini, at the

National Council of Provinces (NCOP), Cape Town’ on Department of Social Development [on-

line] Available at: http://www.dsd.gov.za/index.php?option=com content&task=view&id=400&

Itemid=82 [Accessed 26 January 2015].

—— 2014. ‘Legislative Mandate’ on Department of Social Development [online] Available

at: http://www.dsd.gov.za/index.php?option=com content&task=view&id=30&Itemid=56 [Ac-

cessed 2 June 2014].

252
DSD (Department of Social Development), SASSA (South African Social Security Agency) and

UNICEF (United Nations Children’s Fund). 2012. The South African Child Support Grant Impact

Assessment: Evidence from a Survey of Children, Adolescents and their Households. Pretoria:

UNICEF South Africa.

Duflo, E. 2003. ‘Grandmothers and Granddaughters: Old-Age Pensions and Intrahousehold Alloca-

tion in South Africa’ in The World Bank Economic Review, 17 (1): 1-25.

Duncan, D. 1993. ‘The origins of the “welfare state” in pre-apartheid South Africa’ in Societies of

Southern Africa in the 19th and 20th Centuries, 19 (45): 106-119.

ECA (Economic Commission for Africa), ILO (International Labour Organisation), UNCTAD

(United Nations Conference on Trade and Development), UNDESA (United Nations Department

of Economic and Social Affairs) & UNICEF (United Nations Children’s Fund). 2012. ‘Social pro-

tection: A development priority in the post-2015 UN development agenda’ in UN System Task

Team on the post-2015 UN Development Agenda Thematic Think Piece.

ECLAC (Economic Commission for Latin America and the Caribbean). 2013. ‘Non-contributory

social protection programmes database’ on Economic Commission for Latin America and the

Caribbean [online] Available at: http://dds.cepal.org/bdptc/programa/?id=6 [Accessed 16 Jan-

uary 2015].

Ekiert, G. 1996. The State Against Society: Political Crises and their Aftermath in East Central

Europe. Princeton: Princeton University Press.

Ellis, F. 2012. “‘We Are All Poor Here”: Economic Difference, Social Divisiveness and Targeting

Cash Transfers in Sub-Saharan Africa’ in Journal of Development Studies, 48 (2): 201-214.

Esping-Andersen, G. 1990. The Three Worlds of Welfare Capitalism. Oxford: Blackwell Publishers.

253
Esping-Andersen, G. 1993. ‘The Comparative Macro-Sociology of Welfare States’ in Moreno, L. (ed.),

Social Exchange and Welfare Development, 123-136. Madrid: Consejo Superior de Investigaciones

Cientı́ficas.

Esping-Andersen, G. 1994. ‘Welfare States and the Economy’ in Smelser, N. J. & Swedberg, R. (eds.),

The Handbook of Economic Sociology, 711-732. Princeton: Princeton University Press.

Esping-Andersen, G. 1996. ‘Welfare States without Work: The Impasse of Labour Shedding and

Familialism in Continental European Social Policy’ in Esping-Andersen, G. (ed.), Welfare States

in Transition, 66-87. London: Sage Publications.

Esping-Andersen, G. 1997. ‘Hybrid or Unique? The Japanese Welfare State between Europe and

America’ in Journal of European Social Policy, 7 (3): 179-189.

Esping-Andersen, G. 1999. Social Foundations of Post-Industrial Economies. Oxford: Oxford Uni-

versity Press.

Fajth, G. & Vinay, C. 2010. ‘Conditional Cash Transfers: A Global Perspective’ in Millennium

Development Goals Insights, 1: 1-6.

Faulbaum, L. H. V. 2013. Bolsa Famı́lia: A Case Study of Political Stability. London: Master’s

Thesis, King’s College London.

Feagin, J., Orum, A. & Sjoberg, G. 1991. A Case for Case Study. Chapel Hill: University of North

Carolina Press.

Feenstra, R. C., Inklaar, R. & Timmer, M. P. 2013. ‘The Next Generation of the Penn World Table’

on Penn World Table [online] Available at: www.ggdc.net/pwt [Accessed 26 September 2014].

Fernald, L. C. H., Gertler, P. J. & Neufeld, L. M. 2008. ‘Role of cash in conditional cash transfer

programmes for child health, growth, and development: an analysis of Mexico’s Oportunidades’

in The Lancet, 371: 828-837.

254
FIAP (International Federation of Pension Fund Administrators). 2011. Non-contributory pension

programs in FIAP countries. Santiago: FIAP.

Figueiredo, A. C. 2007. ‘Government Coalitions in Brazilian Democracy’ in Brazilian Political Science

Review, 1 (2): 182-216.

Figueiredo, A. C. & Limongi, F. 2000. ‘Presidential Power, Legislative Organisation, and Party

Behaviour in Brazil’ in Comparative Politics, 32 (2): 151-170.

Fiszbein, A., Schady, N., Ferreira, F. H. G., Grosh, M., Keleher, N., Olinto, P. & Skoufi, E. 2009.

Conditional Cash Transfers: Reducing Present and Future Poverty. Washington: World Bank.

Flora, P. 1983. The Growth of Mass Democracies and Welfare States. Frankfurt am Main: Campus

Verlag.

—— 1986. Growth to Limits: The Western European Welfare States since World War II. Berlin: De

Gruyter Publishers.

Flora, P. & Heidenheimer, A. J. 1981. The Development of Welfare States in Europe and America.

New Brunswick: Transaction Books.

Fonseca, A. & Montali, L. 1996. ‘O Programe de Renda Mı́nima de Campinas: Uma Estratégia de

Combate á Pobreza Urbana’ in XX Encontro Anual da ANPOCS.

FRB (Federative Republic of Brazil). 1943. Consolidação das Leis do Trabalho, 5.452/1943. Rio de

Janeiro: Diário Oficial da União.

—— 1964. Lei Orgânica da Seguridade Social, 4.591/1964. Brası́lia: Diário Oficial da União.

—— 1974. Institui amparo previdenciário para maiores de setenta anos de idade e para inválidos, e

dá outras providências, 6.179/1974. Brası́lia: Diário Oficial da União.

—— 1988. Constitution of the Federative Republic of Brazil. Brası́lia: Diário Oficial da União.

255
—— 1993. Lei Orgânica de Assistência Social, 8.742/1993. Brası́lia: Diário Oficial da União.

—— 2004a. Institui a renda básica de cidadania e dá outras providências, 10.835/2004. Brası́lia:

Diário Oficial da União.

—— 2004b. Polı́tica Nacional de Assistência Social. Brası́lia: Government Printer.

—— 2004c. Cria o Programa Bolsa Famı́lia e dá outras providências, 10.836/2004. Brası́lia: Diário

Oficial da União.

—— 2007. Regulamenta o benefı́cio de prestação continuada da assistência social devido à pessoa

com deficiência e ao idoso, Decreto 6.214/2007. Brası́lia: Government Printer.

—– 2011a. Altera a Lei 8.742, de 7 de dezembro de 1993, que dispõe sobre a organização da As-

sistência Social 12.435/2011. Brası́lia: Diário Oficial da União.

—— 2011b. Plano Brasil Sem Miséria. Brası́lia: Government Printer.

—— 2013. 10 Years of Bolsa Famı́lia. Brası́lia: Government Printer.

Friedberg, A. L. 2000. In the Shadow of the Garrison State: America’s Anti-Statism and its Cold

War Grand Strategy. Princeton: Princeton University Press.

Friedman, S. & De Villiers, R. 1996. Comparing Brazil and South Africa: Two Transitional States

in Political and Economic Perspective. Johannesburg: Center for Policy Studies.

Gallagher, M., Laver, M. & Mair, P. 2001. Representative Government in Modern Europe. Boston:

McGraw Hill.

Garcia, M. & Moore, M. T. 2012. The Cash Dividend. Washington: World Bank.

GCIS (Government Communication and Information System). 2012a. South Africa Yearbook

2012/13. Pretoria: GCIS.

256
—— 2012b. ‘Dept investigating Bolsa-Familia-type grant’ on SANews [online] Available at: http:

//www.sanews.gov.za/south-africa/dept-investigating-bolsa-familia-type-grant [Accessed 26 Jan-

uary 2015].

Geddes, B & Neto, A. R. 1992. ‘Institutional Sources of Corruption in Brazil’ in Third World Quar-

terly, 13 (4): 641-661.

Gerring, J. 2004. ‘What is a case study and what is it good for?’ in American Political Science

Review, 98 (2): 341-354.

Gertler, P. 2004. ‘Do Conditional Cash Transfers Improve Child Health? Evidence from Progresa’s

Control Randomized Experiment’ in The American Economic Review, 94 (2): 336-341.

Giliomee, H. 2003. The Afrikaners: Biography of a People. Cape Town: Tafelberg Publishers.

Giliomee, H., Myburgh, J. & Schlemmer, L. 2001. ‘Dominant Party Rule, Opposition Parties and

Minorities in South Africa’ in Democratization, 8 (1): 161-182.

Glewwe, P. & Kassouf, A. L. 2012. ‘The Impact of the Bolsa Escola/Famı́lia Conditional Cash

Transfer Program on Enrollment, Grade Promotion and Dropout Rates in Brazil’ in Journal of

Development Economics, 2: 505-517.

Goldstone, J. A. 1991. Revolution and Rebellion in the Early Modern World. Berkeley: University of

California Press.

Goodwin, J. 2001. No Other Way Out: States and Revolutionary Movements, 1945-1991. Cambridge:

Cambridge University Press.

Gordhan, P. 2014. 2014 Budget Speech. Pretoria: National Treasury.

Gough, I. 2004a. ‘Welfare regimes in development contexts: a global and regional analysis’ in Gough,

I. & Wood, G. (eds.), Insecurity and Welfare Regimes in Asia, Africa and Latin America. Cam-

bridge: Cambridge University Press.

257
Gough, I. 2004b. ‘Social policy regimes in the developing world’ in Kennet, P. (ed.), A Handbook of

Comparative Social Policy, 239-260. Northampton: Edward Elgar Publishing.

Gough, I. 2013. ‘Social policy regimes in the developing world’ in Kennet, P. (ed.), A Handbook of

Comparative Social Policy, 217-238. Northampton: Edward Elgar Publishing.

Gough, I. & Wood, G. 2004. Insecurity and Welfare Regimes in Asia, Africa and Latin America.

Cambridge: Cambridge University Press.

Gould, A. 1999. Origins of Liberal Dominance: State, Church, and Party in Nineteenth Century

Europe. Ann Arbor: University of Michigan Press.

Gourevitch, P. 1986. Politics in Hard Times. Ithaca: Cornell University Press.

Graham, C. 2002. ‘Public Attitudes Matter: A Conceptual Frame for Accounting for Political Econ-

omy in Safety Nets and Social Assistance Policies’ in World Bank Social Protection Discussion

Paper Series, 0233: 1-37.

Haarmann, D. 1998. From State Maintenance Grants to a New Child Support System: Building a

Policy for Poverty Alleviation with Special Reference to the Financial, Social, and Developmental

Impacts. Cape Town: Doctoral Dissertation, University of the Western Cape.

Hacker, J. 1998. ‘The Historical Logic of National Health Insurance: Structure and Sequence in the

Development of British, Canadian, and U. S. Medical Policy’ in Studies in American Political

Development, 12: 57-130.

Haddad, M. A. 2008. ‘Bolsa Famı́lia and the Needy: Is Allocation Contributing to Equity in Brazil?’

in Journal of International Development, 20: 654-669.

Haggard, S. & Kaufman, R. R. 1995. The Political Economy of Democratic Transitions. Princeton:

Princeton University Press.

258
—— 2008. Development, Democracy and Welfare States. Princeton: Princeton University Press.

Hall, A. 2008. ‘Brazil’s Bolsa Famı́lia: A Double-Edged Sword?’ in Development and Change, 39 (5):

799-822.

Hall, K. 2007. ‘Where to Draw the Line? Targeting and leakage in the Child Support Grant’ presented

at First Annual Conference on the Economics of Social Protection. Magaliesberg, 15 June 2007.

Hall, P. 1986. Governing the Economy: The Politics of State Intervention in Britain and France.

New York: Oxford University Press.

Hanlon, J., Barrientos, A. & Hulme, D. 2010. Just Give Money to the Poor. Sterling: Kumarian

Press.

Hansen, J. M. 1991. Gaining Access: Congress and the Farm Lobby, 1919-1981. Chicago: University

of Chicago Press.

Hattam, V. C. 1993. Labor Visions and State Power: The Origins of Business Unionism in the

United States. Princeton: Princeton University Press.

Heclo, H. 1978. ‘Issue networks and the executive establishment’ in King, A. (ed.), The new American

political system. Washington: American Enterprise Institute for Public Policy Research.

Hicks, A. 1991. ‘The Three Worlds of Welfare Capitalism’ in Contemporary Sociology, 20 (3): 399-401.

Hochstetler, K. 1997. ‘Democratizing Pressures from Below? Social Movements in new Brazil-

ian Democracy’ presented at Latin American Studies Association XX International Congress.

Guadalajara, 17-19 April 1997.

Hölscher, D. 2008. ‘The Emporer’s new Clothes: South Africa’s Attempted Transition to Develop-

mental Social Welfare and Social Work’ in International Journal of Social Welfare, 17: 114-123.

259
Hossain, N. 2010. ‘School Exclusion as Social Exclusion: the Practices and Effects of a Conditional

Cash Transfer Programme for the Poor in Bangladesh’ in Journal of Development Studies, 46 (7):

1264-1282.

Huber, E. & Stephen, J. D. 2001. Development and Crisis of the Welfare State. Chicago: University

of Chicago Press.

Hunter, W. 2007. ‘The Normalization of an Anomaly: The Workers’ Party in Brazil’ in World

Politics, 59 (3): 440-475.

Hunter, W. & Power, T. J. 2007. ‘Rewarding Lula: Executive Power, Social Policy, and the Brazilian

Elections of 2006’ in Latin American Politics and Society, 49 (1): 1-30.

Hunter, W. 2010. The Transformation of the Workers’ Party in Brazil. Cambridge: Cambridge

University Press.

—— 2014. ‘Making Citizens: Brazilian Social Policy from Getúlio to Lula’ in Journal of Politics in

Latin America, 6 (3): 15-37.

Huntington, S. 1968. Political Order in Changing Societies. New Haven: Yale University Press.

IEC (Independent Electoral Commission). 2014. ‘Election results’ on Independent Electoral Commis-

sion [online] Available at: http://www.elections.org.za/content/Elections/Election-results/ [Ac-

cessed 11 September 2014].

Ikenberry, J. G. 1988. ‘Conclusion: An Institutional Approach to American Foreign Economic Policy’

in Ikenberry, J. G., Lake, D. A. & Mastanduno, M. (eds.), The State and American Foreign

Economic Policy, 222-223. Ithaca, Cornell University Press.

—— 2001. After Victory: Institutions, Strategic Restraint, and the Rebuilding of Order after Major

Wars. Princeton: Princeton University Press.

260
Ile, I. U. 2007. A Public Administration Approach to Managing Intergovernmental Relations Systems

in the Governance of the State: A Case Review of Nigeria and South Africa. Pretoria: Doctoral

Dissertation, University of Pretoria.

Iliffe, J. 1987. The African Poor – A History. Cambridge: Cambridge University Press.

ILO (International Labour Organisation). 2000. ‘Employment and Social Protection in the Informal

Sector’ presented at 277th Session of the International Labour Organisation. Geneva, March 2000.

—— 2014. ‘Cadastro Único – Operating a registry through a national public bank’ in Social Protection

in Action: Building Social Protection Floors, 1: 1-4.

Im, H. B. 1987. ‘The Rise of Bureaucratic Authoritarianism in South Korea’ in World Politics, 39:

231-257.

Immergut, E. 1992. Health Politics: Interests and Institutions in Western Europe. Cambridge: Cam-

bridge University Press.

IPUMS (Integrated Public Use Microdata Series). 2015. ‘Living Arrangements’ on Integrated Public

Use Microdata Series [online] Available at: https://international.ipums.org/international-action/

variables/group [Accessed 4 February 2015].

Jaccoud, L., Hadjab, P. D. E. & Chaibub, J. R. 2010. ‘The Consolidation of Social Assistance in

Brazil and its Challenges, 1988-2008’ in Institute for Applied Economic Research Working Paper,

76: 1-33.

Jones, N. 2014. ‘Social Transfers: Strengthening Girls’ and Women’s Potential as Protagonists in

Development’ in Policy in Focus, 27: 24-26.

Jordan, G. & Richardson, J. J. 1979. Governing under Pressure. Oxford: Martin and Robertson

Publishing.

261
Kakwani, N., Soares, F. V. & Son, H. H. 2005. ‘Conditional Cash Transfers in African Countries’ in

UNDP International Poverty Centre Working Paper, 9: 1-89.

Karl, T. R. 1997. The Paradox of Plenty: Oil Booms and Petro-States. Berkeley: University of

California Press.

Katzenstein, P. 1978. Between Power and Plenty. Madison: University of Wisconsin Press.

Katznelson, I., Zolberg, A. R. 1986. Working-Class Formation: Nineteenth-Century Patterns in

Western Europe and the United States. Princeton: Princeton University Press.

Keefer, P. & Khemani, S. 2003a. ‘Democracy, Public Expenditures, and the Poor’ in World Bank

Policy Working Paper, 3164.

—— 2003b. ‘Why do the Poor receive Poor Services?’ in Economic and Political Weekly, 39 (9):

935-943.

Klasen, S. & Woolard, I. 2008. ‘Surviving Unemployment without State Support: Unemployment

and Household Formation in South Africa’ in Journal of African Economies, 18 (1): 1-51.

Klausen, S. 2004. Race, Maternity and the Politics of Birth Control in South Africa, 1910-1939.

Basingstoke: Palgrave Macmillan Publishers.

Kohl, J. 1993. ‘Der Wohlfahrtsstaat in vergleichender Perspektive’ in Zeitschrift für Sozialreform,

39: 67-82.

Krasner, S. D. 1978. Defending the National Interest: Raw Materials Investments and U. S. Foreign

Policy. Princeton: Princeton University Press.

—— 1983. International Regimes. Ithaca: Cornell University Press.

Krikler, J. 2005. The Rand Revolt: The 1922 Insurrection and Racial Killing in South Africa. Cape

Town: Jonathan Ball Publishers.

262
Kruger, J. J. 1992. State Provision of Social Security: Some Theoretical, Comparative and His-

torical Perspectives with Reference to South Africa. Stellenbosch: Master’s Thesis, University of

Stellenbosch.

Kryder, D. 2000. Divided Arsenal: Race and the American State during World War II. Cambridge:

Cambridge University Press.

Lagarde, M., Haines, A. & Palmer, N. 2007. ‘Conditional Cash Transfers for Improving Uptake of

Health Interventions in Low- and Middle-Income Countries: A Systematic Review’ in The Journal

of the American Medical Association, 298 (16): 1900-1910.

Legido-Quigley, H. 2009. ‘The Brazilian Social Assistance Pensions (BPC)’ in Non-Contributory

Pensions and Poverty Prevention Working Paper, 3: 1-13.

Leibbrandt, M., Woolard, I. & Woolard, C. 2007. ‘Poverty and Inequality Dynamics in South Africa:

Post-apartheid Developments in the Light of the Long-Run Legacy’ presented at IPC-DRCLAS

Workshop. Brası́lia, 11-13 January 2007.

Leroy, J. L., Ruel, M. & Verhofstadt, E. 2009. ‘The impact of conditional cash transfer programmes

on child nutrition: a review of evidence using a programme theory framework’ in Journal of

Development Effectiveness, 1 (2): 103-129.

Lessenich, S. & Ostner, I. 1998. Welten des Wohlfahrtskapitalismus: Der Sozialstaat in vergleichender

Perspektive. Frankfurt am Main: Campus Verlag.

Levi, M. 1988. Of Rule and Revenue. Berkeley: University of California Press.

Lichtman, M. 2013. Qualitative Research in Education. Thousand Oaks: Sage Publications.

Lieberman, E. S. 2003. Race and Regionalism in the Politics of Taxation in Brazil and South Africa.

Cambridge: Cambridge University Press.

263
—— 2009. Boundaries of Contagion. Princeton: Princeton University Press.

Liesering, L. & Barrientos, A. 2013. ‘Social citizenship for the global poor? The worldwide spread of

social assistance’ in International Journal of Social Welfare, 22 (1): 50-67.

Lim, S. S., Dandona, L., Hoisington, J. A., James, S. L., Hogan, M. C. & Gakidou, E. 2010. ‘In-

dia’s Janani Suraksha Yojana, a conditional cash transfer programme to increase births in health

facilities: an impact evaluation’ in The Lancet, 375: 2009-2023.

Limongi, F. & Figueiredo, A. C. 1995. ‘Partidos politicos na Camara dos Deputados: 1989-1994’ in

Dados, 38: 497-527.

Lindert, K. 2006. ‘Brazil: Bolsa Famı́lia Program – Scaling-up Cash Transfers for the Poor’ in

Emerging Good Practice in Managing for Development Results, First Edition. Washington: OECD

and World Bank.

Lindert, K., Linder, A., Hobbs, J. & De la Brière, B. 2007. ‘The Nuts and Bolts of Brazil’s Bolsa

Famı́lia Program: Implementing Conditional Cash Transfers in a Decentralized Context’ in World

Bank Social Protection Discussion Paper, 0709: 1-143.

Lipset, S. M. & Rokkan, S. 1967. Party Systems and Voter Alignments: Cross-National Perspectives.

New York: Free Press.

Lopez, B. L. G. 2011. ‘A Social Protection Floor - Mexico’ in Sharing Innovative Experiences: Suc-

cessful Social Protection Floor Experiences. New York: United Nations Development Programme.

Luna, F. V. & Klein, H. S. 2014. The Economic and Social History of Brazil since 1889. New York:

Cambridge University Press.

Lund, F. J. 1993. ‘State Social Benefits in South Africa’ in International Social Security Review, 46

(1): 5-25.

264
Lund, F. J. 2008. Changing Social Policy: The Child Support Grant in South Africa. Cape Town:

HSRC Press.

Lustick, I. 1993. Unsettled States, Disputed Lands: Britain and Ireland, France and Algeria, Israel

and the West Bank-Gaza. Ithaca: Cornell University Press.

Magalhães, J. C. 2014. ‘IBGE afirma que não revisará números sobre miséria’ on

Folha de São Paulo [online] Available at: http://www1.folha.uol.com.br/poder/2014/11/

1544326-ibge-afirma-que-nao-revisara-numeros-sobre-miseria.shtml [Accessed 26 January 2015].

Mahoney, J. 2000. ‘Knowledge Accumulation in Comparative-Historical Analysis: The Case of

Democracy and Authoritarianism’ presented at The Conference on Comparative-Historical Anal-

ysis. Providence, 28-29 April 2000.

Mahoney, J. 2001. The Legacies of Liberalism: Path Dependence and Political Regimes in Central

America. Baltimore: Johns Hopkins University Press.

Mainwaring, S. 1992. ‘Dilemmas of Multiparty Presidential Democracy: The Case of Brazil’ in Kellogg

Institute Working Paper, 174: 1-52.

Maioni, A. 1998. Parting at the Crossroads: The Emergence of Health Insurance in the United States

and Canada. Princeton: Princeton University Press.

March, J. & Olsen, J. P. 1984. ‘The New Institutionalism: Organizational Factors in Political Life’

in American Political Science Review, 78: 734-749.

—— 1989. Rediscovering Institutions. New York: Free Press.

Marshall, T. H. 1950. Citizenship and Social Class and other Essays. Cambridge: Cambridge Uni-

versity Press.

—— 1963. Sociology at the Crossroads and other Essays. London: Heinemann.

265
—— 1965. Social Policy. London: Hutchinson and Company.

—— 1981. The Right to Welfare and other Essays. London: Heinemann.

Marx, A. 1998. Making Race and Nation: A Comparison of South Africa, the United States, and

Brazil. Cambridge: Cambridge University Press.

Maylam, P. 1990. ‘The Rise and Decline of Urban Apartheid in South Africa’ in African Affairs, 89

(354): 57-84.

Mayntz, R. & Scharpf, F. W. 1995. ‘Der Ansatz des akteurzentrierten Institutionalismus’ in Mayntz,

R. & Scharpf, F. W. (eds.), Gesellschaftliche Selbstregelung und politische Steuerung, 39-72. Frank-

furt am Main: Campus Verlag.

McAdam, D. 1982. Political Process and the Development of Black Insurgency, 1930-1970. Chicago:

University of Chicago Press.

McAdam, D., Tarrow, S. & Tilly, C. 2001. Dynamics of Contention. Cambridge: Cambridge Univer-

sity Press.

McCann, B. 2008. The Throes of Democracy: Brazil since 1989. Halifax: Fernwood Publishing.

McCord, A. G. 2012. Public Works and Social Protection in Sub-Saharan Africa: Do Public Works

Work for the Poor? Tokyo: United Nations University Press.

McGuire, J. W. 2010. Wealth, Health and Democracy in East Asia and Latin America. Cambridge:

Cambridge University Press.

MDS (Ministério do Desenvolvimento Social e Combate à Fome). 2009. Lei Orgânica de Assistência

Social Anotada. Brası́lia: Ministério do Desenvolvimento Social e Combate à Fome.

266
—— 2014a. ‘Renda Mensal Vitalı́cia’ on Ministério do Desenvolvimento Social e Combate

à Fome [online] Available at: http://www.mds.gov.br/assistenciasocial/beneficiosassistenciais/

renda-mensal-vitalicia-rmv [Accessed 10 August 2014].

—— 2014b. ‘Tereza Campello: “Cumprimos todas as metas do Brasil Sem Miséria”’ on Portal Brasil

Sem Miséria [online] Available at: http://www.brasilsemmiseria.gov.br/noticias/ultimas-noticias/

2014/dezembro/tereza-campello-201ccumprimos-todas-as-metas-do-brasil-sem-miseria201d [Ac-

cessed 18 January 2015].

Medeiros, M., Diniz, D. & Squinca, F. 2006. ‘Cash Benefits to Disabled Persons in Brazil: An

Analysis of the BPC – Continuous Cash Benefit Programme’ in UNDP International Poverty

Centre Working Paper, 16: 1-25.

Meth, C. 2011. ‘Employer of Last Resort? South Africa’s Expanded Public Works Programme

(EPWP)’ in Southern Africa Labour and Development Research Unit Working Paper, 58: 1-49.

Midgley, J. 2004. ‘Social development and social welfare: implications for social policy’ in Kennett, P.

(ed.), A Handbook of Comparative Social Policy, 217-238. Northampton: Edward Elgar Publishing.

Miller, R. B. 1993. ‘Science and society in the early career of H. F. Verwoerd’ in Journal of Southern

African Studies, 19 (4): 634-661.

Mommsen, W. J. 1981. The Emergence of the Welfare State in Britain and Germany. London: Croom

Helm Publications.

Monama, T. 2013. ‘State to take over Payment of Grants’ on IOL News [online] Available at:

http://www.iol.co.za/news/politics/state-to-take-over-payment-of-grants-1.1572270 [Accessed 4

February 2014].

Moore, B. 1966. Social Origins of Dictatorship and Democracy. Boston: Beacon Press.

267
Moraes, A. & Santana, S. R. 1997. Avaliação do Programa Bolsa-Escola do GDF. Brası́lia: UNICEF,

UNESCO.

Mothiane, M. 2014. Conditionality in Social Grants: The Cases of Brazil and South Africa. Johan-

nesburg: Master’s Thesis, University of the Witwatersrand.

Mpedi, L. G. 2014. ‘Job Creation Policies for South African Youth – The Role of Safety Nets’ in

Policy in Focus, 28: 32-34.

MPS (Ministério da Previdência Social). 2009. Overview of Brazilian Social Welfare. Brası́lia: Min-

istério da Previdência Social.

—— 2013. Social Security Statistical Bulletin. Brası́lia: Ministério da Previdência Social.

—— 2014. Social Security Statistical Bulletin. Brası́lia: Ministério da Previdência Social.

Neuman, W. L. 2005. Social Research Methods: Qualitative and Quantitative Approaches. Boston:

Pearson.

North, D. C. Institutions, Institutional Change and Economic Performance. Cambridge: Cambridge

University Press.

NPC (National Planning Commission). 2011. National Development Plan: Vision for 2030. Pretoria:

Republic of South Africa.

—— 2014. ‘About’ on National Planning Commission online] Available at: http://www.npconline.

co.za/ [Accessed 24 September 2014].

Offe, C. 1991. ‘The Three Worlds of Welfare Capitalism’ in American Journal of Sociology, 96 (6):

1 555-1557.

Olivier, M. 2002. ‘Constitutional Perspectives on the Enforcement of Socio-Economic Rights: Recent

South African Experiences’ in Victoria University of New Zealand Law Review, 33 (1): 117-151.

268
Osório, R. G. & Ferreira de Souza, P. H .G. 2013. ‘Bolsa Famı́lia after Brasil Carinhoso: An Analysis

of the Potential for Reducing Extreme Poverty’ in International Policy Centre for Inclusive Growth

Research Brief, 41: 1-8.

PA (Progressive Alliance). 2015. ‘Progressive Alliance: A Network of Progressive Forces for the

21st Century’ on Progressive Alliance [online] Available at: http://progressive-alliance.info/

basic-document/ [Accessed 13 February 2015].

Pancaldi, F. 2012. ‘The Turn to Real Actors: Achievements and Challenges for Actor-Centered

Institutionalism in the Comparative Political Economy’ presented at XXVI Convegno SISP. Rome,

13-15 September.

Parliament. 2014. ‘National Council of Provinces’ on Parliament of the Republic of South Africa

[online] Available at: http://www.parliament.gov.za/live/content.php?Category ID=27 [Accessed

11 September 2014].

Patel, L. 2011. ‘Child Support Grants – South Africa’ in Sharing Innovative Experiences: Successful

Social Protection Floor Experiences. New York: United Nations Development Programme.

Pereira, C. 2010. ‘Brazil’s Executive-Legislative Relations under the Dilma Coalition Government’

in Brookings Institution Working Paper.

Pestieau, P. 2006. The Welfare State in the European Union: Economic and Social Perspectives.

Oxford: Oxford University Press.

Peters, G. B. 2012. Institutional Theory in Political Science: The New Institutionalism. London:

Continuum Books.

Pfeifer, M. 2010. ‘Comparing Unemployment Protection and Social Assistance in 14 European Coun-

tries: Four Worlds of Protection for People of Working Age’ in International Journal of Social

Welfare, 21 (1): 13-25.

269
Pierson, P. 1994. Dismantling the Welfare State? Reagan, Thatcher, and the Politics of Retrench-

ment. Cambridge: Cambridge University Press.

—— 2000a. ‘Increasing Returns, Path Dependence, and the Study of Politics’ in The American

Political Science Review, 94 (2): 251-267.

—— 2000b. ‘Three Worlds of Welfare State Research’ in Comparative Political Studies, 33 (6/7):

791-821.

—— 2001. The New Politics of the Welfare State. Oxford: Oxford University Press.

Pierson, P. & Skocpol, T. 2002. ‘Historical Institutionalism in Contemporary Political Science’ in

Katznelson, I. & Milner, H. V. (eds.), Political Science: State of the Discipline, 693-721. New

York: W. W. Norton.

PMG (Parliamentary Monitoring Group). 2010. ‘Minister on changes to Social

Assistance Act Regulations & Child Support Grant extension’ on Parliamen-

tary Monitoring Group [online] Available at: http://www.pmg.org.za/report/

20100209-minister-changes-social-assistance-act-regulations-child-support-gran [Accessed 26

January 2015].

Polanyi, K. 1944. The Great Transformation. Boston: Beacon Press.

Pollak, H. 1981. ‘State Social Pensions, Grants and Social Welfare’ in Van der Horst, S. T. & Reid,

J. (eds.), Race Discrimination in South Africa: A Review, 152-186. Cape Town: David Philip.

Potts, R. 2012. ‘Social Welfare in South Africa: Curing or Causing Poverty?’ in Penn State University

Journal of International Affairs, 1 (2): 74-92.

R7 Notı́cias. 2010. ‘Veja as principais propostas dos presidenciáveis Dilma, Marina e Serra

para cinco áreas’ on R7 Notı́cias [online] Available at: http://noticias.r7.com/brasil/noticias/

270
 veja-as-principais-propostas-dos-presidenciaveis-dilma-marina-e-serra-para-cinco-areas-20100707.

html [Accessed 18 January 2015].

Rawlings, L. B. & Rubio, G. M. 2005. ‘Evaluating the Impact of Conditional Cash Transfer Programs’

in The World Bank Research Observer, 20 (1): 29-55.

Reich, G. 2002. ‘Executive Decree Authority in Brazil: How Reactive Legislators Influence Policy’

in Legislative Studies Quarterly, 27 (1): 5-31.

Rhodes, R. A. W. 1986. The National World of Local Government. Oxford: Allen and Unwin.

Richards, A. & Waterbury, J. 1990. A Political Economy of the Middle East: State, Class, and

Economic Development. Boulder: Westview Press.

Ritchie, J. & Lewis, J. 2003. Qualitative Research Practise. London: Sage Publications.

Robertson, L., Mushati, P., Eaton, J. W., Dumba, L., Mavise, G., Makoni, J., Schumacher, C.,

Crea, T., Monasch, R., Sherr, L., Garnett, J. P., Nyamukapa, C. & Gregson, S. 2013. ‘Effects

of unconditional and conditional cash transfers on child health and development in Zimbabwe: a

cluster-randomised trail’ in The Lancet, 381 (9874): 1283-1292.

RSA (Republic of South Africa). 1970. Bantu Homelands Citizenship Act. Pretoria: Government

Gazette.

—— 1983. Child Care Act. Pretoria: Government Gazette.

—– 1995a. Labour Relations Act. Pretoria: Government Gazette.

—— 1995b. Budget Review. Pretoria: Government Gazette.

—— 1996a. Constitution of the Republic of South Africa. Pretoria: Government Gazette.

—— 1996b. Report of the Lund Committee on Child and Family Support [online] Available at: http:

//www.info.gov.za/otherdocs/1996/lund.htm [Accessed 10 February 2013].

271
—— 1997a. Basic Conditions of Employment Act. Pretoria: Government Gazette.

—— 1997b. White Paper for Social Welfare. Pretoria: Government Gazette.

—— 1997c. Welfare Laws Amendment Act. Pretoria: Government Gazette.

—— 1998. Employment Equity Act. Pretoria: Government Gazette.

—— 1999a. Public Finance Management Act. Pretoria: Government Gazette.

—— 1999b. Preliminary Findings from the Health and Demographic Survey. Pretoria: Department

of Health.

—— 2000a. Promotion of Access to Information Act. Pretoria: Government Gazette.

—— 2000b. Promotion of Administrative Justice Act. Pretoria: Government Gazette.

—— 2000c. Promotion of Equality and Prevention of Unfair Discrimination Act. Pretoria: Govern-

ment Gazette.

—— 2002. Report of the Committee of Inquiry into a Comprehensive System of Social Security for

South Africa. Pretoria: Department of Social Development.

—— 2004a. Social Assistance Act. Pretoria: Government Gazette.

—— 2004b. South African Social Security Agency Act. Pretoria: Government Gazette.

—— 2005. Children’s Act. Pretoria: Government Gazette.

—— 2009. Regulations Relating to the Application for and Payment of Social Assistance and the

Requirements or Conditions in Respect of Eligibility for Social Assistance. Pretoria: Government

Printer.

—— 2010. Social Assistance Amendment Act. Pretoria: Government Gazette.

272
—— 2012. White Paper on Families in South Africa. Pretoria: Government Gazette.

—— 2014. South African Government Services [online] Available at: http://www.services.gov.za/

services/content/Home [Accessed 29 January 2014].

Rudra, N. 2007. ‘Welfare States in Developing Countries: Unique or Universal?’ in The Journal of

Politics, 69 (2): 378-396.

—— 2008. Globalization and the Race to the Bottom in Developing Countries. Cambridge: Cambridge

University Press.

Rueschemeyer, D. A., Stephens, E. H. & Stephens, J. D. 1992. Capitalist Development and Democ-

racy. Chicago: University of Chicago Press.

SAHRC (South African Human Rights Commission). 2009. The 7th Economic and Social Rights

Report: Millennium Development Goals and the Progressive Realisation of Economic and Social

Rights in South Africa. Johannesburg: South African Human Rights Commission.

SAIRR (South African Institute of Race Relations). 2011. First Steps to Healing the South African

Family. Johannesburg: South African Institute of Race Relations.

SALDRU (Southern Africa Labour and Development Research Unit). 1993. Project for Statistics on

Living Standards and Development, [Computer File]. S0037. Cape Town, South Africa: School of

Economics, University of Cape Town [Producer] 1993; Pretoria, South Africa: South African Data

Archive, National Research Foundation [Distributor], 1995.

SALC (South African Law Commission). 2002. Review of the Child Care Act. Pretoria: South African

Law Commission.

Samson, M., Lee, U., Ndebele, A., MacQuene, K., Van Niekerk, I., Gandhi, V., Harigaya, T. &

Abrahams, C. 2004. The Social and Economic Impact of South Africa’s Social Security System.

Cape Town: Economic Policy Research Institute.

273
Samuels, D. 2000. ‘Ambition and Competition: Explaining Legislative Turnover in Brazil’ in Legisla-

tive Studies Quarterly, 25 (3): 481-497.

—— 2004. ‘From Socialism to Social Democracy: Party Organization and the Transformation of the

Workers’ Party in Brazil’ in Comparative Political Studies, 37 (9): 999-1024.

Santos, L. M. P., Paes-Sousa, R., Miazagi, E., Silva, T. F. & Da Fonceca, A. M. M. 2011. ‘The

Brazilian Experience with Conditional Cash Transfers: A Successful Way to Reduce Inequity

and to Improve Health’ presented at World Conference on Social Determinants of Health. Rio de

Janeiro, 19-21 October.

SASSA (South African Social Security Agency). 2012. Presentation to University of Stellenbosch.

Stellenbosch: 18 October 2012.

—— 2013. Statistical Report: Issue No 12 of 2013 – 31 December 2013. Pretoria: Republic of South

Africa.

—— 2014. ‘Our Vision, Mission and Values’ on South African Social Security Agency [online] Avail-

able at: http://www.sassa.gov.za/index.php/about-us/our-vision-mission-and-values [Accessed 3

February 2014].

SCA (Supreme Court of Appeal). 2014. ‘The Legal System’ on Supreme Court of Appeal of South

Africa [online] Available at: http://www.justice.gov.za/sca/historysca.htm [Accessed 11 Septem-

ber 2014].

Schady, N. & Araujo, M. C. 2006. ‘Cash Transfers, conditions, school enrolment, and child work:

Evidence from a randomized experiment in Ecuador’ in World Bank Policy Research Working

Paper, 3920: 1-31.

Scharpf, F. W. 1991. ‘Games Real Actors Could Play: The Challenge of Complexity’ in Journal of

Theoretical Politics, 3 (3): 277-304.

274
—— 1997. Games Real Actors Play: Actor-Centred Institutionalism in Policy Research. Boulder:

Westview Press.

—— 2000. ‘Institutions in Comparative Policy Research’ in Comparative Political Studies, 33 (6/7):

762-790.

Schreiber, L. A. 2014. ‘Institutions and Policy Change: The Development of the Child Support Grant

in South Africa’ in Politikon: South African Journal of Political Studies, 41 (2): 267-288.

Schwarzer, H. & Querino, A. C. 2002. ‘Non-contributory pensions in Brazil: The impact on poverty

reduction’ in International Labour Office Extension of Social Security Working Paper, 11: 1-49.

Scruggs, L. & Allan, J. 2006. ‘Welfare-state decommodification in 18 OECD countries: a replication

and revision’ in Journal of European Social Policy, 16 (1): 55-72.

Sedlacek, G., Gustafsson-Wright, E., Ilahi, N. & Lannon, M. 2001. ‘An Assessment of the Bolsa

Escola Programs’ in World Bank Report, 20208-BR: 1-54.

Seekings, J. 2000. ‘The Origins of Social Citizenship in Pre-Apartheid South Africa’ in South African

Journal of Philosophy, 19 (4): 386-404.

—— 2006. ‘The Carnegie Commission and the Backlash Against Welfare-State Building in South

Africa, 1931-1937’ in Centre for Social Science Research Working Paper, 159: 1-36.

—— 2007. “‘Not a Single White Person Should Be Allowed to Go Under”: Swartgevaar and the

Origin of South Africa’s Welfare State, 1924-1929’ in The Journal of African History, 48: 375-394.

—— 2008. ‘Deserving Individuals and Groups: The Postapartheid State’s Justification of the Shape

of South Africa’s System of Social Assistance’ in Transformation (68): 28-52.

—— 2014. ‘Welfare State-Building in Africa in Comparative Perspective’ presented at Seminar on

African Welfare States. Geneva, September 2014.

275
Seekings, J. & Nattrass, N. 2005. Class, Race, and Inequality in South Africa. New Haven: Yale

University Press.

Seidman, G. 1994. Manufacturing Militance: Workers’ Movements in Brazil and South Africa, 1970-

1985. Berkeley: University of California Press.

Selcher, W. A. 1998. ‘The Politics of Decentralized Federalism, National Diversification, and Region-

alism in Brazil’ in Journal of Interamerican Studies and World Affairs, 40 (4): 25-50.

Shepsle, K. 1986. ‘Institutional Equilibrium and Equilibrium Institutions’ in Weisberg, H.(ed.), Po-

litical Science: The Science of Politics, 51-81. New York: Agathon Press.

Simmons, B. A. 1994. Who Adjusts? Domestic Sources of Foreign Economic Policy during the

Interwar Years, 1923-1939. Princeton: Princeton University Press.

Skocpol, T. 1979. States and Social Revolutions: A Comparative Analysis of France, Russia, and

China. Cambridge: Cambridge University Press.

—— 1992. Protecting Soldiers and Mothers: The Political Origins of Social Policy in the United

States. Cambrdige: Belknap Press of Harvard University Press.

Skowronek, S. 1997. The Politics President Make: Leadership from John Adams to Bill Clinton.

Cambridge: Belknap Press of Harvard University Press.

Smith, S. 2011. ‘Towards Integrated and Comprehensive Social Security in South Africa’ in Critique

of the Social Security and Retirement Reform.

Soares, F. V., Ribas, R. P. & Osório, R. 2010. ‘Evaluating the Impact of Brazil’s Bolsa Famı́lia:

Cash Transfer Programs in Comparative Perspective’ in Latin American Research Review, 45 (2):

173-190.

276
Soares, F. V. & Silva, E. 2010. ‘Conditional cash transfer programmes and gender vulnerabilities:

Case studies of Brazil, Chile and Colombia’ in International Policy Centre for Inclusive Growth

Working Paper, 69: 1-39.

Soares, F. V., Soares, S., Medeiros, M. & Osório, R. G. 2006. ‘Cash Transfer Programmes in Brazil:

Impacts on Inequality and Poverty’ in International Poverty Centre Working Paper, 21: 1-30.

Souza, C. 2002a. ‘Brazil’s System of Local Government, Local Finance, and Intergovernmental Re-

lations’ on Mimeo [online] Available at: http://info.worldbank.org/etools/docs/library/229986/

Souza%20Brasil%27s%20system%20of%20local%20government.pdf [Accessed 21 November 2014].

—— 2002b. ‘Brazil: The Prospects of a Center-Constraining Federation in a Fragmented Polity’ in

Publius: The Journal of Federalism, 32 (2): 23-48.

—— 2005. ‘Federal Republic of Brazil’ in Kincaid, J. & Tarr, G. A. (eds.), Constitutional Origins,

Structure, and Change in Federal Countries. Montreal: McGill-Queen’s University Press.

Spaull, N. 2015. ‘How to raise the “real” matric pass rate’ on Africa Check [online] Available at:

http://africacheck.org/2015/01/13/how-to-raise-the-real-matric-pass-rate/ [Accessed 4 February

2015].

Spicker, P. 1984. Stigma and Social Welfare. Mountain View: Croom Helm Publications.

Stake, R. E. 1995. The Art of Case Study Research. Thousand Oaks: Sage Publications.

Stats SA (Statistics South Africa). 2013. Quarterly Labour Force Survey: Quarter 3, 2013. Pretoria:

Statistics South Africa.

Steinmo, S. 1996. Taxation and Democracy: Swedish, British, and American Approaches to Financing

the Modern State. New Haven: Yale University Press.

—— 2001. ‘The New Institutionalism’ in Clarke, P. B. & Foweraker, J. (eds.), Encyclopedia of

Democratic Thought, 570-573. London: Routledge.

277
—— 2008. ‘Historical institutionalism’ in Della Porta, D. & Keating, M. (eds.), Approaches and

Methodologies in the Social Sciences: A Pluralist Perspective, 118-138. Cambridge: Cambridge

University Press.

Streeck, W. 1992. Social Institutions and Economic Performance. Newbury Park: Sage Publications.

Sunstein, C. R. 2001. Designing Democracy: What Constitutions Do. Oxford: Oxford University

Press.

Suplicy, E. M. 2002. ‘Legitimizing Basic Income in Developing Countries: Brazil’ presented at World

Summit on Sustainable Development and Basic Income Grant. Johannesburg, August 2002.

—— 2006. Citizen’s Basic Income: The Answer is Blowing in the Wind. Brası́lia: Programa Brasil.

—— 2008. ‘From the Family Scholarship Program towards the Citizen’s Basic Income in Brazil’

presented at XII International Congress of BIEN – Basic Income Earth Network. Dublin, 20-21

June 2008.

Tarrow, S. 1998. Power in Movement: Social Movements and Contentious Politics. Cambridge:

Cambridge University Press.

The Economist. 2008. ‘Brazil: Happy families’ on The Economist [online] Available at: http://www.

economist.com/node/10650663 [Accessed 10 March 2015].

—— 2010. ‘Brazil’s Presidential Campaign: In Lula’s Footsteps’ on The Economist [online] Available

at: http://www.economist.com/node/16486525 [Accessed 18 January 2015].

Thelen, K. 1993. ‘European Labor in Transition: Sweden and Germany Compared’ in World Politics,

46 (1): 23-49.

—— 1994. ‘Beyond Corporatism: Toward a New Framework for the Study of Labor in Advanced

Capitalism’ in Comparative Politics, 27: 107-124.

278
Thelen, K. & Steinmo, S. 1992. ‘Historical institutionalism in comparative politics’ in Steinmo, S.,

Thelen, K. & Longstreth, F. (eds.), Structuring politics: Historical institutionalism in comparative

analysis, 1-32. Cambridge: Cambridge University Press.

Thomas, G. 2011. ‘A Typology for the Case Study in Social Science following a Review of Definition,

Discourse, and Structure’ in Qualitative Inquiry, 17 (6): 511-521.

Titmuss, R. 1958. Essays on the Welfare State. London: Allen and Unwin.

—— 1974. Social Policy. London: Allen and Unwin.

TSE (Tribunal Superior Eleitoral). 2014. ‘Eleições’ on Tribunal Superior Eleitoral [online] Available

at: http://www.tse.jus.br/eleicoes [Accessed 21 November 2014].

UIS (UNESCO Institute for Statistics). 2015. ‘Data Centre’ on UNESCO Institute for Statistics

[online] Available at: http://data.uis.unesco.org/ [Accessed 4 February 2015].

UKDFID (United Kingdom Department for International Development). 2005. Social Transfers and

Chronic Poverty: Emerging Evidence and the Challenge Ahead. London: United Kingdom Depart-

ment for International Development.

—— 2011. Cash Transfers: Literature Review. London: UKDFID.

Ulriksen, M. S. 2014. ‘18 Years Old? You Are on your Own: Safety Nets and Youth Unemployment

in South Africa’ in Policy in Focus, 28: 30-32.

UMS (United Mexican States). 2014. Acuerdo por el que se emiten las Reglas de Operación del

Programa de Apoyo Alimentario, para el ejercicio fiscal 2014. Mexico City: Diario Oficial de la

Federación.

UNESCO (United Nations Educational, Scientific and Cultural Organisation). 2011. ‘State of

Education in Brazil’ on United Nations Educational, Scientific and Cultural Organisation

279
 [online] Available at: http://portal.unesco.org/geography/es/files/13662/12960781625TOM -

Brazil%27s Ed System EN.pdf/TOM%2B-%2BBrazil%27s%2BEd%2BSystem EN.pdf [Accessed

4 February 2015].

UNICEF (United Nations Children’s Fund). 2008. Quantitative Analysis of the Impact of the Child

Support Grant. Cape Town: UNICEF.

UNRISD (United Nations Research Institute for Social Development). 2010. Combating Poverty

and Inequality: Structural Change, Social Policy and Politics. Geneva: United Nations Research

Institute for Social Development.

USA (Union of South Africa). 1911. Mines and Works Act. Pretoria: Government Gazette.

—— 1913. Children’s Protection Act. Pretoria: Government Gazette.

—— 1921. Children’s Protection Amendment Act. Pretoria: Government Gazette.

—— 1922. Final Report of the Unemployment Commission. Pretoria: Government Printer.

—— 1924. Yearbook of the Union of South Africa, Volume 9. Pretoria: Government Printer.

—— 1925. Report of the Economic and Wage Commission. Pretoria: Government Printer.

—— 1927. First Report of the Commission on Old Age Pensions and National Insurance. Cape Town:

Government Printer.

—— 1928. Old Age Pension Act. Pretoria: Government Gazette.

—— 1929. Third Report of the Commission on Old Age Pensions and National Insurance. Cape

Town: Government Printer.

—— 1937a. Old Age Pension Amendment Act. Pretoria: Government Gazette.

—— 1937b. Children’s Act. Pretoria: Government Gazette.

280
—— 1944. Report of the Social Security Committee and Report No. 2 of the Social and Economic

Planning Council entitled: Social Security, Social Services and the National Income. Pretoria:

Government Printer.

—— 1946. Disability Grants Act. Pretoria: Government Gazette.

Van der Berg, S. 1989a. ‘Long Term Economic and Development Prospects in South Africa’ in African

Affairs, 88 (351): 187-203.

—— 1989b. Social Reform and Economic Development. Stellenbosch: Mimeo.

—— 1997. ‘South African social security under apartheid and beyond’ in Development Southern

Africa, 14 (4): 481-503.

Van Niekerk, R. 2003. ‘The Evolution of Health and Welfare Policies in South Africa: Inherited

Institutions, Fiscal Restraint, and the Deracialisation of Social Policy in the Post-Apartheid Era’

in The Journal of African American History, 88 (4): 361-376.

Villatoro, P. 2005. ‘Conditional Cash Transfer Programs: Experiences from Latin America’ in CEPAL

Review, 86: 83-96.

Visser, W. 2004. ‘Shifting RDP into GEAR: The ANC Government’s Dilemma in Providing an

Equitable System of Social Security for the “New” South Africa’ presented at 40th ITH Linzer

Konferenz. 16-19 September 2004.

Watts, J. 2013. ‘Brazil’s bolsa familia scheme marks a decade of pioneering poverty relief’ on The

Guardian [online] Available at: http://www.theguardian.com/global-development/2013/dec/17/

brazil-bolsa-familia-decade-anniversary-poverty-relief [Accessed 10 March 2015].

Weigand, C. & Grosh, M. 2008. ‘Levels and Patterns of Safety Net Spending in Developing and

Transition Countries’ in World Bank Social Protection Discussion Paper, 0817: 1-14.

281
Weingast, B. 1996. ‘Political Institutions: Rational Choice Perspectives’ in Goodin, R. E. (ed.), A

New Handbook of Political Science, 167-190. Oxford: Oxford University Press.

WHO (World Health Organisation). 2014. ‘Immunization Coverage Data’ on World Health Organisa-

tion [online] Available at: http://www.who.int/immunization/monitoring surveillance/data/en/

[Accessed 3 February 2015].

Wickham-Crowley, T. 1992. Guerillas and Revolutions in Latin America: A Comparative Study of

Insurgents and Regimes since 1956. Princeton: Princeton University Press.

Wieczorek, R. 2012. ‘ANC Dominance and Democratic Consolidation in South Africa’ in Penn State

Journal of International Affairs, 1 (2): 27-38.

Wilensky, H. L. 1975. The Welfare State and Equality: Structural and Ideological Roots of Public

Expenditures. Berkeley: University of California Press.

Williams, M. J. 2007. ‘The Social and Economic Impacts of South Africa’s Child Support Grant’ in

Economic Policy Research Institute Working Paper, 40: 1-47.

Willmore, L. 2014. ‘Towards universal pension coverage in Mexico’ in Pension Watch Briefing, 13.

Woolard, I. 2003. ‘Social Assistance Grants, Poverty and Economic Growth in South Africa’ presented

at DPRU/TIPS Conference. Johannesburg, 8-10 September 2003.

Woolard, I., Harttgen, K. & Klasen, S. 2010. ‘The Evolution and Impact of Social Security in

South Africa’ presented at Promoting Resilience through Social Protection in Sub-Saharan Africa

Conference. Dakar, 28-30 June 2010.

Woolard, I. & Leibbrandt, M. 2010. ‘The Evolution and Impact of Unconditional Cash Transfers in

South Africa’ in Southern Africa Labour and Development Research Unit Working Paper, 51.

282
World Bank. 2013a. ‘Unemployment, Total (% of total labor force)’ on World Bank [online] Avail-

able at: http://data.worldbank.org/indicator/SL.UEM.TOTL.ZS?page=3 [Accessed 11 February

2013].

—— 2013b. ‘Poverty Headcount Ratio at $1.25 a Day (PPP) (% of population)’ on World Bank

[online] Available at: http://data.worldbank.org/indicator/SI.POV.DDAY?page=3 [Accessed 11

February 2013].

—— 2014a. ‘Conditional Cash Transfers – Country Overviews and Project Information’ on World

Bank [online] Available at: http://go.worldbank.org/SUHLHT69C0 [Accessed 27 February 2014].

—— 2014b. ‘Life expectancy at birth, total (years)’ on World Bank [online] Available at: http:

//data.worldbank.org/indicator/SP.DYN.LE00.IN?page=6 [Accessed 10 October 2014].

—— 2015a. ‘Poverty headcount ratio at $1.25 a day (PPP) (% of population)’ on World Bank [online]

Available at: http://data.worldbank.org/indicator/SI.POV.DDAY?page=3 [Accessed 3 February

2015].

—— 2015b. ‘Poverty headcount ratio at $2 a day (PPP) (% of population)’ on World Bank [online]

Available at: http://data.worldbank.org/indicator/SI.POV.2DAY?page=3 [Accessed 3 February

2015].

—— 2015c. ‘Unemployment, total (% of total labor force) (modeled ILO estimate)’ on World Bank

[online] Available at: http://data.worldbank.org/indicator/SL.UEM.TOTL.ZS?page=3 [Accessed

3 February 2015].

—— 2015d. ‘Immunization, measles (% of children ages 12-23 months)’ on World Bank [online]

Available at: http://data.worldbank.org/indicator/SH.IMM.MEAS?page=3 [Accessed 3 February

2015].

283
—— 2015e. ‘School enrolment, primary (% net)’ on World Bank [online] Available at: http://data.

worldbank.org/indicator/SE.PRM.NENR?page=2 [Accessed 4 February 2015].

—— 2015f. ‘Social Safety Net Projects’ on World Bank [online] Available at: http://www.worldbank.

org/projects/search?lang=en&searchTerm=&themecode exact=54 [Accessed 30 March 2015].

WVS (World Values Survey). 2005-2008. ‘Wave 5 Official Aggregate v.20140429’ on

World Values Survey Association [online] Available at: http://www.worldvaluessurvey.org/

WVSDocumentationWV5.jsp [Accessed 11 February 2015].

Yashar, D. 1997. Demanding Democracy: Reform and Reaction in Costa Rica and Guatamala, 1870s-

1950s. Stanford: Stanford University Press. Yin, R. K. 2003. Case Study Research: Design and

Methods. Thousand Oaks: Sage Publications.

ZACC (Constitutional Court of South Africa). 1996. Shaballala and Others v. Attorney-General of

the Transvaal and Another, 1996 South Africa 725 (C.C).

Zaverucha, J. 1997. ‘The 1988 Brazilian Constitution and its Authoritarian Legacy: Formalizing

Democracy while Gutting its Essence’ presented at Meeting of the Latin American Studies Asso-

ciation. Guadalajara, 17-19 April 1997.

Zucco, C. 2008. ‘The President’s “New” Constituency: Lula and the Pragmatic Vote in Brazil’s 2006

Presidential Elections’ in Journal of Latin American Studies, 40 (29-49).

Zucco, C. & Power, T. J. 2013. ‘Bolsa Famı́lia and the Shift in Lula’s Electoral Base, 2002-2006: A

Reply to Bohn’ in Latin American Research Review, 48 (2): 3-24.

Zysman, J. 1994. ‘How Institutions Create Historically Rooted Trajectories of Growth’ in Industrial

and Corporate Change, 3 (1): 243-283.

284
Appendix A: Interviews – South

Africa

Interview Position Date

Number

1 Provincial Minister of Social Development 6 November 2012

2 SASSA senior manager 12 November 2012

3 Former Lund Committee member 12 November 2012

4 Professor of Economics 15 November 2012

5 SASSA executive official 16 November 2012

6 National Deputy Minister 20 November 2012

7 Former Lund Committee chairperson; 20 November 2012

8 Member of Parliament (Social Development portfolio) 21 November 2012

9 Professor of Social Policy 6 December 2012

10 Senior Research Fellow in Social Policy 6 December 2012

11 Deputy Director-General in the Department of Social Development 10 December 2012

285
Appendix B: Interviews – Brazil

Interview Position Date

Number

12 Research Coordinator at the International Policy Centre for Inclusive Growth 25 October 2013

13 Director of Social Policy at the Instituto de Pesquisa Econômica Aplicada 6 November 2013

14 Social Assistance Analyst at the Instituto de Pesquisa Econômica Aplicada 20 November 2013

15 Social Security Analyst at the Instituto de Pesquisa Econômica Aplicada 27 November 2013

16 Former State Governor 29 November 2013

17 President of the Instituto de Pesquisa Econômica Aplicada 29 November 2013

286
Appendix C: Eligibility Requirements

– South Africa

Name Means Test Further Requirements

(Annual)

Child Support Single: R34 800 The applicant must be the child’s primary caregiver (parent, grandparent

Grant (US$3 480) or a child over the age of 16 heading a household). If the applicant is not

Married: R69 the child’s parent, proof must be provided that they are the child’s primary

600 (US$6 960) caregiver through an affidavit from a police official, a social worker’s report,

an affidavit from the biological parent or a letter from the principal of the

school attended by the child. The applicant must be a South African citizen

or permanent resident. The child must be under the age of 18 years; not be

cared for in a state institution; live with the primary caregiver (who is not

being paid to look after the child); and both the applicant and child must live

in South Africa.

Foster Child N/A The applicant must be taking care of a foster child placed in their custody

Grant by a court as a result of being either orphaned, abandoned, at risk, abused or

neglected. The applicant must be a South African citizen, permanent resident

or refugee; both the applicant and the child must live in South Africa; the

foster must have been legally placed in the care of the applicant and must

remain in their care; and the child must be younger than 18 years.

287
Care Depen- Single: R151 The grant is aimed at supporting the primary caregivers of children with se-

dency Grant 200 (US$15 120) vere disabilities in need of fulltime and special care. The applicant must be

Married: R302 a parent, primary caregiver or a foster parent appointed by the court; be a

400 (US$ 30 South African citizen or permanent resident. The child must be younger than

240) 18 years; not be cared for permanently in a state institution; and must have a

The means test severe disability and be in need of fulltime and special care — a state medical

does not apply officer assesses the child before the grant is approved. Both the applicant and

to foster par- the child must live in South Africa.

ents.

Disability Grant Single: R49 200 The Disability Grant is aimed at beneficiaries with a physical or mental dis-

(US$ 4 920); ability that makes them unfit to work for a period of longer than six months.

assets <R831 Beneficiaries qualify for a permanent disability grant if their disability will

600 (US$83 160) continue for more than a year and a temporary disability grant if their dis-

Married: R98 ability will last for a continuous period of not less than six months and not

840 (US$9 884); more than 12 months. A permanent disability grant does not mean that

assets <R1 663 they will receive the grant for life, but that it will continue for longer than

200 (US$166 12 months. An applicant must be a South African citizen, permanent resi-

320) dent or refugee and living in South Africa at the time of application; be be-

tween 18 and 59 years of age; not be cared for in a state institution; possess

a valid ID document; and must undergo a medical examination where a doc-

tor appointed by the state assess the degree of their disability — the doctor

completes a medical report and forwards it to SASSA. The report is valid for

three months from the date of the medical assessment.

Grant-in-Aid N/A The Grant-in-Aid is directed at people already receiving another social grant,

but who are unable to care for themselves. The benefit is intended to enable

them to pay for the person who takes fulltime care of them. In order to qual-

ify, the applicant must already be receiving a Disability Grant, War Veterans

Grant or a Older Persons Grant; not be able to look after themselves owing

to a physical or mental disability and be in need of fulltime care provided by

another person; and must not be cared for in an institution that received a

subsidy from the government for their care or housing.

288
War Veterans Single: R49 920 The War Veterans Grant is aimed at former soldier who fought in the First

Grant (US$4 992); World War (1914-1918), the Second World War (1939-1945), the Zulu Upris-

assets <R831 ing (1906) or the Korean War (1950-1953) and are unable to support them-

600 (US$83 160) selves. An applicant must live in South Africa; be older than 60 years of age

Married: R99 or be disabled; not receive any other social grant; and must not be cared for

840 (US$9 984); in a state institution.

assets <R1 663

200 (US$166

320)

Older Persons Single: R49 200 An applicant must older than 60 years of age; be a South African citizen or

Grant (US$4 920); permanent resident; live in South Africa; not receive any other social grant;

assets <R831 and must not be cared for in a state institution.

600 (US$83 160)

Married: R99

840 (US$9 984);

assets <R1 663

200 (US$166

320)

Required Documents

In addition to completing an application form, applicants are required to submit a set of documents

at their nearest SASSA office as part of the application process. The requirements are slightly differ-

ent for each grant, and can also depend on the specifics of the applicant’s situation. The following

list of documents may be required: ID; birth certificates for children; proof of any maintenance

received for children; if they are a refugee, their status permit and refugee ID is required in addi-

tion to the child’s birth certificate or ID from the country of origin; proof of earnings; a marriage

certificate; if an applicant is divorced, the court order indicating that they have custody of a child;

if one or both parents are dead or missing, the death certificate of the deceased or proof that the

289
parent is missing (such as a missing person’s report from the police); a medical assessment report

that confirms the applicant or child’s disability; salary slips, bank statements for the previous three

months, pension slips, and any other proof of income.

If the applicant does not have an ID or the child’s birth certificate, they must complete an

affidavit on a standard SASSA form in the presence of a Commissioner of Oaths who is not a

SASSA official and submit a sworn statement signed by a reputable person (such as a councillor,

traditional leader, social worker, minister of religion or school principal) who knows the applicant

and child. They are furthermore required to submit proof of application for an ID and/or birth

certificate at the Department of Home Affairs; a temporary ID issued by the Department of Home

Affairs (if applicable); a baptismal certificate if available; a road to health clinic card if available;

and a school report if available.

290
Appendix D: Eligibility Requirements

– Brazil

Name Family Means Test (per person Further Requirements

per month)

Bolsa Famı́lia Basic Up to R$70 (US$30) N/A

Bolsa Famı́lia Variable 1 R$70.01 (US$30) to R$140 (US$61) The benefit is aimed at household with children

between the ages of 0 and 16, as well as pregnant

or lactating woman.

Bolsa Famı́lia Variable 2 R$70.01 (US$30) to R$140 (US$61) This benefit is aimed at households with children

between the ages of 16 and 18.

Bolsa Famı́lia Residual <R$70 (US$30) after other PBF N/A

benefits

Benefı́cio de Prestação R$169 (US$73) The BPC’s disability component is aimed at dis-

Continuada abled individuals of all ages. In order to qualify

(Disability) for this benefit, an applicant must be disabled

to the extent that they are unable to work and

lead an independent life. The applicant’s per

capita family income must also be less than 25

percent of the legal minimum wage (currently

R$169 (US$73) per month).

291
Benefı́cio de Prestação R$169 (US$73) The old age element of the BPC is directed at

Continuada impoverished individuals older than 65 years.

(Old Age) An applicant’s per capita family income must be

less than 25 percent of the legal minimum wage

(currently R$169 (US$73) per month) in order to

qualify.

Required Documents

Determining eligibility for the PBF is ultimately the responsibility of the MDS, as directed through

the Cadastro Único. Potential beneficiaries are required to enrol onto the database of the Cadastro

Único prior to formally applying for the PBF. Applications are then made at local municipal offices,

where prospective beneficiaries self-report their family incomes and submit birth certificates and civil

identification documents. This data is then cross-referenced against the information contained in the

Cadastro Único, which determines benefit eligibility according to income and family composition.

This is followed by additional cross checks by the MDS in order to further validate self-reported

income.

In the case of the BPC, applicants must submit a declaration issued by an authority (judge,

local police chief or municipal administrator) containing information about the per capita income

of the candidate’s family. These declarations are submitted to the local municipality, where a social

worker receives and forwards them to the regional office of the INSS. The INSS then analyses the

documents and issues an authorisation to grant the benefit. Applicants for the BPC’s disability

component must additionally submit a medical assessment report.

292
Abstract

Recent decades have witnessed the profound proliferation of social assistance cash transfer pro-

grammes throughout numerous countries of the developing world. While scholars have subsequently

paid a great deal of attention to the largely positive socioeconomic effects generated through the

expansion of these redistributive interventions, the literature has neglected the fact of tremendous

empirical variation regarding the ways in which different countries have designed their social as-

sistance policy regimes. The smattering of studies which have taken up the question have usually

been content to regard fundamental differences in these patterns of welfare provision as the result of

functional and ideational factors on the basis of conventional wisdom, rather than through rigorous

analysis.

In the first study of its kind, this inquiry explicitly addresses this gap in the literature by

resolutely focusing on the systemic variation in social assistance provision in the two benchmark

emerging welfare states of South Africa and Brazil. Its central aim is to account for the disparity

between the policy regimes instituted by the two countries. The fact that South Africa and Brazil

have adopted very different approaches regarding issues of coverage, conditionality and orientation

results in the construction of distinct typologies which sees the South African system classified

as ‘productive’ and the Brazilian approach as ‘protective’. The description of this empirically

observable divergence is followed by an account of its manifestation through the application of

a historical institutionalist framework, operationalised by employing the analytical approach of

Actor-Centred Institutionalism.

293
 The resulting investigation into the comparative political economy of social assistance provision

reveals that, while conventionally accepted functional and ideational hypotheses fail to explain the

divergence across the cases, variation on the institutional dimension is able to convincingly account

for the fact that South Africa and Brazil have adopted qualitatively different policy approaches. The

study thereby provides the first explanatory account of the contemporary domestic emergence of

programmes which have become centrally important in combating poverty and inequality through-

out the countries of the developing world. In addition to making a significant and entirely novel

explanatory contribution to the knowledge on the subject, the findings attest to the decisive prac-

tical importance of the role played by local institutional structures in shaping context appropriate

policy interventions.

294
Zusammenfassung

In den vergangenen Jahrzehnten hat eine tiefgreifende Verbreitung von Geldtransferprogramme in

zahlreichen Entwicklungsländern stattgefunden. Obwohl Sozial-wissenschaftler seither großes Au-

genmerk auf die überwiegend positiven sozioökonomischen Effekte, welche durch die Ausweitung

derartiger redistributiver Interventionen erfolgten, gelegt haben, wurde in der vorliegenden Liter-

atur vielfach vernachlässigt, einen Nachweis über die umfangreichen empirischen Variationen in

Bezug auf die Ausgestaltung von sozialpolitischen Regimen unterschiedlicher Länder zu erbringen.

Studien, die sich bislang der Fragestellung angenommen haben, begnügen sich in der Regel damit

fundamentale Unterschiede in den Mustern von Wohlfahrtsmaßnahmen - auf konventionellem Wis-

sen basierend - als Resultat von funktionalen und Ideengeleiteten Faktoren anzusehen, anstatt eine

wirklich tiefergehende Analyse vorzunehmen.

Die vorliegende Studie adressiert daher als erste seiner Art die bestehende Lücke in der wis-

senschaftlichen Literatur indem es einen deutlichen Fokus auf die systemische Variation von sozialen

Hilfsprogrammen in den beiden Bezugsländern Südafrika und Brasilien legt. Zentrales Ziel ist

es dabei die unterschiedlichen institutionalisierten Politikregime der beiden ausgewählten Länder

aufzuzeigen. Die Tatsache, dass Südafrika und Brasilien unterschiedliche Ansätze hinsichtlich der

Reichweite, der Konditionalität und der Orientierung ihrer sozialen Hilfsprogramme gewählt haben,

resultiert in der Konstruktion verschiedener Typologien, derer zufolge das südafrikanische System

als ‘produktiv’ und das brasilianische System als ‘protektiv’ angesehen werden kann. Nach einer

Beschreibung dieser empirisch nachweisbaren Unterschiede erfolgt in der vorliegenden Arbeit eine

295
Darstellung ihrer jeweiligen Ausgestaltung mittels der Anwendung eines historisch-institutionellen

Theoriegerüstes, welches wiederum durch den analytischen Ansatz des Akteurszentriertem Institu-

tionalismus operationalisiert wird.

Die resultierende Untersuchung im Bereich der Vergleichenden Politischen Ökonomie von sozialen

Hilfsprogrammen macht deutlich, dass, während konventionelle funktionale und Ideengeleitete Hy-

pothesen sich als unzureichend erweisen unterschiedliche Fälle zu erkennen, Variationen in den

institutionellen Dimensionen es ermöglichen, überzeugend darzulegen, dass Südafrika und Brasilien

qualitativ unterschiedliche Politikansätze gewählt haben. Die vorliegende Studie liefert demgemäß

die erste tiefgreifende Untersuchung in Bezug auf das Aufkommen zeitgenössischer länderspezifischer

Programme, welche von zentraler Bedeutung im Kampf gegen Armut und Ungleichheit in Entwick-

lungsländern sind. Abgesehen von einer signifikanten und vollständig neuen Untersuchung zum Wis-

sen zu dem Thema, zeigen die Ergebnisse auf, dass lokale institutionelle Strukturen von zentraler

Bedeutung sind in Bezug auf die Ausgestaltung von Kontextangemessenen Politikinterventionen

sind.

296
Publications during Course of

Research

Peer Reviewed

Schreiber, L. A. 2014. ‘Institutions and Policy Change: The Development of the Child Support

Grant in South Africa’ in Politikon: South African Journal of Political Studies, 41 (2): 267-288.

Reports

Schreiber, L. A. 2013. ‘The Expansion of Social Assistance in Brazil and South Africa’ on Friedrich

Naumann Foundation.

Conference Proceedings

Schreiber, L. A. 2013. ‘Institutions and Policy Change: The Development of the Child Support

Grant in South Africa’ presented at 11th Annual International Conference on Politics and

International Affairs. Athens, Greece: 17-20 June 2013.

Schreiber, L. A. 2013. ‘Social Assistance in South Africa’ presented at 1st Berlin Forum on Global

Politics. Berlin, Germany: 18-19 April 2013.

Schreiber, L. A. 2013. ‘Institutions and Policy Change’ presented at 9th Neoinstitutionalism

Workshop. Warsaw, Poland: 14-15 March 2013.

297
Journal Referee

Journal of Development Studies (Ad Hoc; July 2014-present)

Newspaper Articles and Opinion Pieces

Schreiber, L. A. 2015. ‘Turkye se eie Zuma’ in Rapport, 7 June, p. 8-9 [print].

Schreiber, L. A. 2015. ‘Britse kiesers gee les vir SA’ in Rapport, 17 May [print].

Schreiber, L. A. 2015. ‘Ons is almal só: Nie oukei nie’ in Rapport, 26 April, p. 4-5 [print].

Schreiber, L. A. 2015. ‘Small Yet BIG: The Basic Income Guarantee’ on The Huffington Post, 23

April.

Schreiber, L. A. 2015. ‘Time to burst our deluded notions of exceptionalism’ in Daily Dispatch, 21

April [print].

Schreiber, L. A. 2015. ‘South Africa, we are not okay’ on Politicsweb, 19 April.

Schreiber, L. A. 2015. ‘Moenie wurg as ekonomie lê’ in Rapport, 23 February, p. 10 [print].

Schreiber, L. A. 2015. “n Griekse boodskap aan Europa’ in Rapport, 1 February, p. 14-15 [print].

Schreiber, L. A. 2014. ‘Betraying Mandela’ on Politicsweb, 5 December.

Schreiber, L. A. 2014. ‘Armoede neem Muur se plek in’ in Rapport, 9 November [print].

Schreiber, L. A. 2014. ‘Ons kan leer by Brasilië’ in Rapport, 12 October, p. 12-13 [print].

Schreiber, L. A. 2014. ‘Sal SA ooit die Skotse riel dans?’ in Rapport, 14 September, p. 12-13 [print].

Schreiber, L. A. 2014. ‘Linkses en regses saam in ‘n kamp’ in Beeld, 17 July [print].

Schreiber, L. A. 2014. ‘The European Emperor is Naked’ on Politicsweb, 9 June.

Schreiber, L. A. 2014. ‘Die Europese keiser is kaal’ in Rapport, 1 June, p. 22-23 [print].

Schreiber, L. A. 2014. ‘Haatlike simbool nie verdedigbaar’ in Beeld, 25 February, p. 11-12 [print].

Schreiber, L. A. 2014. ‘Tyd dat die middelklas begin saamstaan’ in Rapport, 4 January, p. 5 [print].

Schreiber, L. A. 2013. ‘Why is the voice of middle class protest absent from South Africa?’ on

Politicsweb, 14 August.

298
Schreiber, L. A. 2013. ‘Welfaart laat hul kla’ in Rapport, 7 June, p. 3-4 [print].

Schreiber, L. A. 2013. ‘Protests in Brazil are the result of progress’ in Business Day, 2 July [print].

Schreiber, L. A. 2013. ‘The NDP and policy framing’ on Politicsweb, 12 March.

Schreiber, L. A. 2013. ‘Time to think BIG again?’ on Politicsweb, 26 February.

Schreiber, L. A. 2012. ‘Why a “Lula moment” is not possible in South Africa’ on Politicsweb, 8

October.

Schreiber, L. A. 2012. ‘Learning from German reforms’ on Politicsweb, 14 June.

Schreiber, L. A. 2012. ‘An important experiment in the Western Cape’ on Politicsweb, 15 May.

Schreiber, L. A. 2012. ‘Reducing citizens to subjects’ on Politicsweb, 20 April.

299
Biography

Leon Schreiber is a Ph.D. Candidate in Political Science at the Freie Universität Berlin and a

Visiting Student Research Collaborator at Princeton University. Originally hailing from the rural

Namaqualand region of South Africa, Schreiber went on to earn a Certificate in Business Man-

agement from the University of South Africa, as well as a Bachelor’s degree, a Bachelor’s degree

with Honours and a Master’s degree in International Studies and Political Science from Stellen-

bosch University. His research interests include social policy, poverty, and the comparative political

economy of development, while his Ph.D. dissertation comparatively examines the development of

social welfare policies in South Africa and Brazil since democratization.

Schreiber’s work has been recognized through numerous grants and awards from Stellenbosch

University, the European Commission, the German Academic Exchange Service, the St. Gallen

Symposium Wings of Excellence Award, Princeton University, and the Berlin Consortium for Ger-

manic Studies. He is a columnist for one of South Africa’s largest Sunday newspapers, and regularly

appears as a political analyst on a weekly actuality television programme. Schreiber has previously

worked in various capacities at a wide range of social impact organisations across four different

continents, including think tanks, universities, local government structures, political parties and

international development agencies.

300