From Recognition to Representation: CollectiveRights and Democratic Citizenship in the Philippines

by

Nina McMurry

B.A., Stanford University (2010)

Submitted to the Department of Political Sciencein partial fulfillment of the requirements for the degree of

Doctor of Philosophy

at the

MASSACHUSETTS INSTITUTE OF TECHNOLOGY

September 2020

c○ Massachusetts Institute of Technology 2020. All rights reserved.

Author . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Department of Political Science

September 8, 2020

Certified by. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Lily Tsai

Ford Professor of Political ScienceThesis Supervisor

Accepted by . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Fotini Christia

Professor of Political ScienceChair, Graduate Program Committee

2

From Recognition to Representation: Collective Rights and

Democratic Citizenship in the Philippines

by

Nina McMurry

Submitted to the Department of Political Scienceon September 8, 2020, in partial fulfillment of the

requirements for the degree ofDoctor of Philosophy

Abstract

How does the recognition of self-determination rights for indigenous and tribal com-munities affect governance in modern democratic states? Nearly half of UN memberstates recognize indigenous groups in their constitutions, many devolving control overland and local governance functions. A dominant perspective in political science,rooted in the concept of the nation-state, implies these policies, by empowering non-state authorities and crystallizing sub-national identities, are likely to have negativeunintended consequences. Yet few studies have investigated these predictions directly.

This study examines the effects of collective recognition for indigenous commu-nities on state consolidation and democratic representation. Rather than weakeningstates and undermining democratic accountability, I argue that given underlying con-ditions of state weakness, collective recognition can encourage the incorporation ofmarginalized populations by enabling more effective claim-making through formaldemocratic politics. I evaluate empirical implications of this theory in the Philip-pines, which has one of the most robust frameworks for indigenous recognition inSoutheast Asia. Drawing on more than two years of fieldwork in the country, I com-bine analysis of administrative data, original survey data and survey experiments,and in-depth qualitative interviews with indigenous leaders and policymakers.

I find that recognition through the granting of collective land titles leads to in-creased indigenous self-identification, but also to greater attachments to nationalidentity and multiple indicators of state integration. In addition, I find evidence thatrecognition, rather than simply entrenching political elites, increases community elec-toral mobilization directed toward obtaining public goods from the state. This worknot only speaks to contemporary debates surrounding indigenous rights, but alsohas broader implications for our understandings of post-colonial state consolidation,ethnic and identity politics, and collective participation in democratic systems.

Thesis Supervisor: Lily TsaiTitle: Ford Professor of Political Science

3

4

Acknowledgments

I am tremendously grateful for the opportunity to write and submit this dissertationand to the many people who made it possible.

First and foremost, I am indebted to the members of my dissertation committee:Lily Tsai, Evan Lieberman, Roger Petersen, and Danny Hidalgo. Lily, thank youfor supporting me not only as my dissertation committee chair but as a mentorthroughout my time at MIT. Your scholarly work, your dedication to your studentsand to the broader MIT community, and your pioneering efforts to bridge the gapbetween social science and policy are a constant source of inspiration. Thank you forsticking with me through many different iterations of this project, for helping me todistill concrete questions from initially inchoate thoughts, and for being a constantvoice of encouragement through many challenges and setbacks. Evan, thank youfor always being willing to give timely and honest feedback, for helping me stay ontrack, for drawing connections I would not otherwise have seen, and for pushing meto address the broader implications of my work. Roger, thank you for knowing whento pull me out of the weeds and make me consider big picture questions and whento pull me into the weeds and make me consider the implications of my claims forreal politics and everyday human interactions. Danny, thank you for the invaluablefeedback you have provided on so many aspects of this project, from the empiricalanalysis to the theoretical logic to positioning within various literatures, and for yourbroader contributions to my methodological training.

I am also grateful for support and feedback I received from many other MIT polit-ical science faculty members. In particular, I thank Rich Nielsen, Chappell Lawson,Regina Bateson, Fotini Christia, and Ariel White for detailed feedback and guidanceat key stages of the process.

I could not have asked for a more supportive group of fellow graduate students inthe department. Thank you to my officemates Ben Morse, Tesalia Rizzo, Leah Rosen-zweig, and Guillermo Toral for the many spontaneous conversations that contributedgreatly to my work and to my overall well-being. Thank you to my amazing co-hort: Ben Armstrong, Parrish Bergquist, Tugba Bozcaga, Loreto Cox, Alex Copulsky,Elizabeth Dekeyser, Michael Freedman, Mayumi Fukishima, Meg Goldberg, DaniloLimoeiro, Philip Martin, Brittany Montgomery, Cullen Nutt, Kelly Pasolli, Tim Mc-Donnell, and Weihuang Wong. I don’t know how I would have made it through thefirst few years of graduate school without your support, encouragement, and friend-ship. Thank you to the MIT GOV/LAB community, who took the time to providefeedback on drafts and half-baked ideas alike. Thank you also to Elissa Berwick,Dan de Kadt, Gabriel Nahmias, Marika Landau-Wells and Yiqing Xu, and (outsideof MIT) to Stephanie Schwartz and Michael Rubin for helpful feedback and advice.Thank you to Chad Hazlett for encouraging me to pursue the PhD route and forreassuring me along the way.

Many people in the Philippines contributed tremendously to this project. I amespecially grateful to the staff of Legal Network for Truthful Elections. Thank you toRona Ann Caritos for taking a chance on me and for your patience and leadership,

5

and to Rich Apollo Mamhot for countless hours spent conducting and arranginginterviews, training and managing field staff, dealing with many unforeseen logisticalchallenges, and for the many conversations in airports, on boats, and on buses thatshaped every aspect of this project. Brizza Rosales, Rosabel Montemar, and Renzodel Rosario also provided invaluable assistance. Adit Butoy offered a sounding boardfrom the very beginning and helped me in so many ways, from facilitating meetingsto giving me places to stay in multiple cities to taking me out for karaoke. Thank youto the Commission on Elections and the National Commission on Indigenous Peoples,especially to April Mabanes and Erwin Caliba. Thank you also to Pura Sumangil,Ester Alkonga, Bing van Tooren, Dante de los Angeles, Johnmart Salunday, RebeccaBumahit, and Mary Anne Pudduan. Finally, I thank the many interviewees andsurvey respondents who took the time to share their experiences and perspectives. Ilook forward to future trips and to many more conversations as this work continuesto evolve.

Last but not least, I am grateful to my family. Thank you to my parents, Dianeand Kevin, for your unwavering belief in me and to my brother Joel for the manyon-demand pep talks. Thank you to my husband James for moving across the worldto support my career (and for agreeing to do it a second time), for tolerating months-long stretches apart, for forcing me to go outside every once in a while, and for yourconstant love and support.

All errors are my own.

6

Contents

1 Introduction 17

1.1 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

1.2 Indigenous recognition, national identity, and the state . . . . . . . . 26

1.3 Indigenous recognition, democratic representation, and accountability 28

1.4 State consolidation, democratic accountability, and systemic legitimacy 31

1.5 Summary of the argument, findings, and contributions . . . . . . . . 33

1.6 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

1.6.1 Qualitative interviews . . . . . . . . . . . . . . . . . . . . . . 36

1.6.2 Community survey . . . . . . . . . . . . . . . . . . . . . . . . 37

1.6.3 Administrative data . . . . . . . . . . . . . . . . . . . . . . . 38

1.7 Plan of the dissertation . . . . . . . . . . . . . . . . . . . . . . . . . . 39

2 Recognition as Integration 41

2.1 The statebuilder’s dilemma . . . . . . . . . . . . . . . . . . . . . . . . 43

2.2 Recognition or assimilation? . . . . . . . . . . . . . . . . . . . . . . . 45

2.2.1 The state . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

2.2.2 Traditional elites . . . . . . . . . . . . . . . . . . . . . . . . . 49

2.2.3 Communities . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

2.3 Implications for representation . . . . . . . . . . . . . . . . . . . . . . 54

2.4 Predictions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

7

3 Indigenous recognition in the Philippines 61

3.1 From assimilation to recognition . . . . . . . . . . . . . . . . . . . . . 62

3.2 Recognition in practice . . . . . . . . . . . . . . . . . . . . . . . . . . 68

3.2.1 Connection to the theoretical framework . . . . . . . . . . . . 68

3.2.2 What I hope to illustrate . . . . . . . . . . . . . . . . . . . . . 71

3.2.3 Local context and interview methodology . . . . . . . . . . . . 72

3.2.4 Unmediated interaction with the state threatens traditional elites 78

3.2.5 Egalitarian governance limits claim-making . . . . . . . . . . . 81

3.2.6 Collective claim-making increases efficacy . . . . . . . . . . . . 85

3.2.7 Collective recognition increases collective claim-making capac-

ity and creates incentives to engage with the state . . . . . . . 90

3.2.8 Systemic legitimacy can affect cooperation with the state . . . 95

3.2.9 Community accountability mechanisms limit elite rent-seeking 98

3.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

4 Recognition and integration 103

4.1 Observational Data Analysis . . . . . . . . . . . . . . . . . . . . . . . 105

4.1.1 Research Design . . . . . . . . . . . . . . . . . . . . . . . . . . 105

4.1.2 Empirical strategy . . . . . . . . . . . . . . . . . . . . . . . . 108

4.1.3 Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

4.1.4 Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

4.1.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113

4.1.6 Additional Robustness Checks . . . . . . . . . . . . . . . . . . 118

4.1.7 Additional Alternative Explanations . . . . . . . . . . . . . . 120

4.2 Survey Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

4.2.1 Survey Data and Priming Experiment . . . . . . . . . . . . . 123

4.2.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125

4.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

8

5 From recognition to representation 131

5.1 Collective recognition and voting behavior: nationwide . . . . . . . . 134

5.2 Collective recognition and voting behavior: survey results . . . . . . . 142

5.2.1 Collective recognition and voter coordination . . . . . . . . . . 142

5.2.2 Collective recognition and leader influence on vote choice . . . 143

5.3 Collective recognition beyond land titling . . . . . . . . . . . . . . . . 153

5.4 Collective recognition and state public goods provision . . . . . . . . 161

5.4.1 Discussion and conclusion . . . . . . . . . . . . . . . . . . . . 165

6 Voice, not exit 167

6.1 Implications for indigenous rights in the Philippines and beyond . . . 168

6.2 Implications for political science literature . . . . . . . . . . . . . . . 170

6.2.1 Ethnic and identity politics . . . . . . . . . . . . . . . . . . . 171

6.2.2 Traditional authority . . . . . . . . . . . . . . . . . . . . . . . 176

6.2.3 State-building and democracy . . . . . . . . . . . . . . . . . . 182

6.2.4 Collective participation, identity politics, and liberal democracy 185

6.3 Directions for future research . . . . . . . . . . . . . . . . . . . . . . 188

A Survey and Interview Details 193

A.1 Survey Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . 193

A.1.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193

A.1.2 Sample selection . . . . . . . . . . . . . . . . . . . . . . . . . 194

A.1.3 Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . 196

A.1.4 Ethics approval and permissions . . . . . . . . . . . . . . . . . 196

A.2 Elite interview guide . . . . . . . . . . . . . . . . . . . . . . . . . . . 198

B Empirical Appendix 203

B.1 Supplementary material for Chapter 4 . . . . . . . . . . . . . . . . . 203

B.1.1 Summary statistics for observational analysis . . . . . . . . . . 203

9

B.1.2 Identity robustness checks . . . . . . . . . . . . . . . . . . . . 204

B.1.3 Compliance robustness checks . . . . . . . . . . . . . . . . . . 210

B.1.4 Instrumenting for land titling . . . . . . . . . . . . . . . . . . 219

B.1.5 Alternative explanations . . . . . . . . . . . . . . . . . . . . . 221

B.1.6 Survey experiment details . . . . . . . . . . . . . . . . . . . . 221

B.2 Supplementary material for Chapter 5 . . . . . . . . . . . . . . . . . 230

B.2.1 Additional analysis of survey outcomes . . . . . . . . . . . . . 230

B.2.2 Conjoint experiment details and robustness checks . . . . . . . 233

B.2.3 COMELEC polling places intervention . . . . . . . . . . . . . 233

B.2.4 Public goods robustness checks . . . . . . . . . . . . . . . . . 242

10

List of Figures

1-1 Qualitative interview sites . . . . . . . . . . . . . . . . . . . . . . . . 37

2-1 Governance strategies differentiated by modes of state-society interaction. 47

3-1 Certificates of Ancestral Domain Title (CADTs) and eligible areas as

of October 2016 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

3-2 In-depth interview municipalities in Oriental Mindoro, Occidental Min-

doro, and Palawan . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

4-1 Cumulative number of Certificates of Ancestral Domain Title (CADTs)

approved by year, as of March 2018 . . . . . . . . . . . . . . . . . . . 107

5-1 Experimental effects by candidate attribute . . . . . . . . . . . . . . 149

5-2 Effects of experimental primes on perceived influence of indigenous

leaders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151

5-3 Experimental effects by candidate attribute, untitled barangays only . 152

5-4 Average effects of candidate attributes on vote choice in titled and

untitled barangays . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154

5-5 Effects of titling on local public goods provision . . . . . . . . . . . . 164

A-1 Target survey sample . . . . . . . . . . . . . . . . . . . . . . . . . . . 195

B-1 Baseline levels of indigenous population and birth registration (eligible

universe) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215

11

B-2 Birth registration rates by age in 2000 . . . . . . . . . . . . . . . . . 218

B-3 Recognition Prime (Culture) . . . . . . . . . . . . . . . . . . . . . . . 224

B-4 Recognition Prime (Material) . . . . . . . . . . . . . . . . . . . . . . 225

B-5 Identity showcards . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226

B-6 Identity rankings by experimental condition . . . . . . . . . . . . . . 227

B-7 Effects of priming on individual components of state attitudes index . 228

B-8 Voting Game (Conjoint) Candidate Attribute Images . . . . . . . . . 234

B-9 Average Marginal Component Effect estimates in the full sample. . . 234

B-10 Effects of priming treatments on perceived influence of indigenous leaders.236

B-11 Experimental effects by candidate attribute, titled barangays only . . 237

B-12 Marginal means by candidate attribute and experimental condition . 239

B-13 Marginal means by candidate attribute and titling status . . . . . . . 240

12

List of Tables

2.1 Predictions of Effects of Collective Recognition . . . . . . . . . . . . . 57

3.1 Interview locations and tribes by titling status . . . . . . . . . . . . . 74

4.1 Land Titling (Continuous) and Indigenous Identification (Differences-

in-Differences) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

4.2 Land Titling (Continuous) and Birth Registration (Two-Way FE) . . 117

4.3 Land Titling (Continuous) and Legibility (Two-Way FE) . . . . . . . 119

4.4 Priming Treatment and Top Ranking of Identity Attributes . . . . . . 127

4.5 Priming Effects on State Attitudes . . . . . . . . . . . . . . . . . . . 128

5.1 Land Titling (Continuous) and Average 2016 Precinct-Level Vote Mar-

gin (Barangay Level) . . . . . . . . . . . . . . . . . . . . . . . . . . . 136

5.2 Land Titling (Continuous) and Average 2016 Precinct-Level Incumbent

Vote Share (Barangay Level) . . . . . . . . . . . . . . . . . . . . . . . 139

5.3 Land Titling and 2016 Mayoral Vote Margin (Municipal Level) . . . . 141

5.4 Land Titling and Voter Coordination . . . . . . . . . . . . . . . . . . 144

5.5 SPP/AVC Pilot, Barangay Turnout, Margin, and Incumbent Voteshare

(Cross-Sectional) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159

5.6 SPP/AVC Pilot, Barangay Turnout, Margin, and Incumbent Voteshare

(Difference-in-Differences) . . . . . . . . . . . . . . . . . . . . . . . . 160

13

5.7 SPP/AVC Pilot, Barangay Turnout, Margin, and Incumbent Voteshare

(Two-way Fixed Effects) . . . . . . . . . . . . . . . . . . . . . . . . . 161

A.1 Covariate Balance (Targeted vs. Actual Sample) . . . . . . . . . . . . 197

A.2 Sample Demographics . . . . . . . . . . . . . . . . . . . . . . . . . . 197

B.1 Pre-Treatment Covariate Balance Among Eligible Barangays (Titled

in 2018 vs. Untitled in 2018) . . . . . . . . . . . . . . . . . . . . . . . 204

B.2 Pre-Treatment Covariate Balance Among Titled Barangays (Titled Be-

fore 2010 vs. Titled After 2010) . . . . . . . . . . . . . . . . . . . . . 205

B.3 Summary Statistics (Study Barangays) . . . . . . . . . . . . . . . . . 206

B.4 Land Titling (Binary) and Indigenous Identification (Differences-in-

Differences) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206

B.5 Land Titling and Population Change 2000-2010 (Differences-in-Differences)207

B.6 Land Titling and Recent Migration 2000-2010 (Differences-in-Differences)207

B.7 Land Titling and Migration Among Indigenous Peoples 2000-2010 (Differences-

in-Differences) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208

B.8 Land Titling (Continuous) and Indigenous Identification (Differences-

in-Differences) - IP Migration Controls . . . . . . . . . . . . . . . . . 208

B.9 Land Titling and Indigenous Identification (Differences-in-Differences) 209

B.10 Land Titling and Indigenous Identification (Conley SE) . . . . . . . . 209

B.11 Land Titling (Binary) and Birth Registration (Two-Way FE) . . . . . 212

B.12 Land Titling and Birth Registration (Two-Way FE), IP Pop Interaction213

B.13 Land Titling (Continuous) and Birth Registration (Two-Way FE), IP

Majority Interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . 214

B.14 Land Titling (Continuous) and Birth Registration (Two-Way FE) -

Excluding 90% Baseline Reg. . . . . . . . . . . . . . . . . . . . . . . 215

14

B.15 Land Titling (Continuous) and Legibility (Two-Way FE), IP Pop In-

teraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216

B.16 Land Titling (Continuous) and Lagged Birth Registration . . . . . . . 217

B.17 Land Titling (Continuous) and Lagged Legibility . . . . . . . . . . . 217

B.18 Land Titling and Birth Registration (Differences-in-Differences) . . . 217

B.19 Land Titling and Birth Registration (Conley SE) . . . . . . . . . . . 218

B.20 Land Titling and Indigenous Identity (Instrumental Variables Estimates)219

B.21 Land Titling (Continuous) and Indigenous Identification - Land Value

Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222

B.22 Land Titling (Continuous) and Birth Registration - Land Value Inter-

actions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222

B.23 Priming Treatment and Top Ranking of Identity Attributes . . . . . . 227

B.24 Prime Treatment and Ranking of Identity Attributes . . . . . . . . . 229

B.25 Recognition Primes and Belief that Government Conducted Survey . 229

B.26 Land Titling and Indigenous Leader Influence on Vote Choice (Index

Components) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231

B.27 Land Titling and Vote Coordination (Index Components) . . . . . . . 232

B.28 Candidate Conjoint Results by Experimental Prime . . . . . . . . . . 235

B.29 Candidate Conjoint Results by Land Titling Status . . . . . . . . . . 238

B.30 Separate Polling Places Pilot - Balance Table (Genetic Matching) . . 241

B.31 Land Titling (Continuous) and Lagged Highway Access . . . . . . . . 242

B.32 Land Titling and Highway Access (Conditional Logit) . . . . . . . . . 243

15

16

Chapter 1

Introduction

States around the world recognize collective self-determination rights for indigenous

and tribal communities, defining distinct identity groups and in many cases devolving

substantial powers over land and the administration of justice to indigenous institu-

tions. A recent study found that nearly half of UN member states recognized indige-

nous governance in their constitutions (Holzinger et al. 2018), while dozens more do so

through statutory provisions.1 Another found that indigenous peoples, in many cases

through tenure rights afforded by governments, collectively manage or hold rights to

territory covering more than a quarter of the Earth’s land area (Garnett et al. 2018).

While this may appear to be a holdover from the pre-colonial or colonial peri-

ods, likely to fade as modern state institutions consolidate, the trend appears to be

moving in the opposite direction. Many states have recognized self-governance rights

for indigenous groups only in recent decades (Yashar 2005, Cuskelly 2011, Chartock

2013).2 These developments find support in current international law, and are lauded

by many as positive advances in human rights, the preservation of cultural heritage,

1Holzinger et al. (2018) count 94 countries that contain at least one “indigenous provision” in theirconstitutions. This includes provisions in three non mutually-exclusive categories: indigenous groupacknowledgment or special rights (73 countries), recognition of traditional political institutions (48countries), and recognition of customary law (53 countries).

2Similarly, on the African continent, scholars identified a resurgence of state recognition of tra-ditional institutions beginning in the 1990s (Herbst 2000, Englebert 2002, Baldwin 2011)

17

and even environmental protection.3 But they also raise important questions about

prospects for nation building and state consolidation. In particular, by recogniz-

ing multiple identities and forms of governance within their borders, do states risk

undermining national unity and rendering themselves less effective and legitimate?

This concern is not merely theoretical. As indigenous groups began advocating

for recognition within the United Nations and other international forums in the 1960s

and 1970s, many state governments registered strong objections to proposals to ex-

tend collective self-governance rights to these groups. These objections came from

states across the spectrum of industrialization and post-colonial status. For some, the

very idea of collective rights for sub-national communities was seen as antithetical to

principles of national sovereignty and universal human rights.4 Others expressed more

practical concerns about the implications of indigenous self-determination for their

ability to govern. During a 1988 convention on the issue, for example, a representative

of the Venezuelan government argued that recognizing collective self-determination

rights would “service to set up micro-States within States as sanctuaries to shelter

subversion, guerilla warfare, drug trafficking and common delinquency” (Rodríguez-

Piñero 2006). As late as 2006, the ratification of the UN Declaration on the Rights of

Indigenous Peoples (UNDRIP) was delayed due to opposition from a group of African

states concerned that recognizing collective self-determination rights for indigenous

communities would undermine national unity and threaten their territorial integrity

(Charters and Stavenhagen 2009).

The view that granting collective self-governance rights to sub-national commu-

3Most prominently, the Indigenous and Tribal Peoples Convention of 1989 (ILO-Convention 169)and United Nations Declaration on the Rights of Indigenous People (UNDRIP), adopted by the UNGeneral Assembly in 2007, establish universal rights to collective self-determination for indigenouspeoples.

4In a debate on the issue in 1996, a representative from the French government argued that“collective rights did not exist in international human rights law.” The representative of Japanargued that “there had to be equality before the law in a State and this [collective recognition] poseda problem with regard to accepting separate ‘legal systems’ (UN Commission on Human Rights1996).

18

nities poses a threat to states is consistent with a dominant strand in the political

science literature on state-building. Scholars have long considered a unified national

identity and system of government to be prerequisites for the development of effective

modern states (Zolberg 1967, Weber 1976, Anderson 1983, Gellner 1983). The persis-

tence of societal fragmentation —both in terms of ethnic diversity and the existence of

competing authority structures —is linked to state weakness, corruption, poor public

goods provision, and internal conflict (Alesina, Baqir and Easterly 1999, Miguel 2004,

Collier 2010). Furthermore, in democratic settings, traditional elites (who may be em-

powered by the granting of collective self-governance rights) are often conceived of as

obstacles to electoral accountability (Acemoglu, Reed and Robinson 2014, De Kadt

and Larreguy 2015, Mamdani 1996, Koter 2013, Ntsebeza 2005, Anderson, Francois

and Kotwal 2012, Mershon 2015, Boone 2014).

In light of concerns expressed by states about collective recognition and the near-

consensus among scholars about the negative effects of societal diversity, recent moves

by states to recognize and to expand indigenous self-governance rights appear puz-

zling. Why are states taking action that they have previously resisted and that,

according to the scholarly conventional wisdom, will render them weak and ineffec-

tive? One possibility is that states are responding to evolving human rights norms

favoring indigenous self-determination. Yet many countries with strong indigenous

recognition policies regularly flout international human rights norms in other areas.

Another is that leaders are driven by electoral incentives: devolving control to indige-

nous elites may create vote brokers that can deliver easy votes to the ruling party

(Baldwin 2011, De Kadt and Larreguy 2015). This may be true in some cases, but it

has limited explanatory power outside of dominant party systems. A third is that col-

lective recognition is merely a contemporary version of indirect rule, a strategy used

by colonial powers and empires to “govern on the cheap” and extract resources with-

out having to implement costly nation-building strategies (Herbst 2000, Boone 2003,

19

Gerring et al. 2011, Acemoglu et al. 2014). But while indirect rule typically involved

the direct delegation of extractive functions (e.g. tax collection), the recognition of

self-governance rights for indigenous communities often entails a direct loss of state

revenue through tax exemption. In addition, modern independent states have gen-

erally proven more ambitious than colonial states and many have in fact attempted

assimilative nation-building (Migdal 1988, Scott 1998, Miguel 2004, McGovern 2012,

Bandyopadhyay and Green 2013).

Regardless of why states do it, decisions to devolve authority to non-state gov-

ernance structures and to recognize sub-national groupings are assumed to produce

sub-optimal governance outcomes. Rather than theorizing the motivation of states,

this project seeks to answer what is in some ways a more first-order question: is

this assumption correct? Is it actually the case that states, in choosing to recognize

sub-national collectives and non-state sources of authority, are tying their hands and

making it more difficult to govern in the way most modern states aspire to? Relat-

edly, does the recognition of indigenous self-determination rights in fact undermine

the quality of democratic representation as existing theories would predict?

Despite the ubiquity of these policies and the extent to which they exist in tension

with dominant views in the literature, few scholars have addressed this question ex-

plicitly. Holzinger et al. (2018) and Behr (2018) examine cross-national relationships

between various forms of recognition for traditional institutions, country character-

istics, and development outcomes. However, as these authors and others note, the

decision of a state to enact policies of recognition is almost certainly a function of the

underlying power of these traditional institutions, making it difficult to disentangle

their effects in comparison to a valid counterfactual (Herbst 2000, Acemoglu et al.

2014). Understanding the effects of collective recognition policies has important im-

plications both for policy and for our theoretical understanding of the relationship be-

tween nationalism and state-building. If policies granting collective self-determination

20

rights to sub-national communities do indeed weaken national identity and under-

mine state authority, we require an explanation for states’ adoption of apparently

self-undermining policies. If they do not, this suggests a need to revisit foundational

theories linking cultural and institutional heterogeneity to state weakness.

In this project, I study the effects of collective recognition for indigenous commu-

nities, focusing primarily on sub-national variation within a single country case: the

Philippines. The Philippines has one of the most robust frameworks for indigenous

recognition, both within Southeast Asia and more broadly, making it a qualitatively

important case and a relatively fair test case for understanding the effects of indige-

nous recognition as it is envisioned in international law (Inguanzo 2014). I offer a new

theory that challenges the conventional wisdom about the effects of collective recogni-

tion on both state consolidation and democratic representation. In short, I argue that

given underlying conditions of state weakness and relative to other potential gover-

nance strategies, collective recognition of marginalized groups can both advance the

consolidation of state authority and improve the representation of these groups within

democratic institutions. Drawing on more than two years of fieldwork in the country,

I combine observational analysis of historical and contemporary administrative data,

original survey data, including survey experiments, and in-depth qualitative inter-

views with indigenous leaders and policymakers to illustrate and evaluate observable

implications of my theory.

In this introductory chapter, I begin by defining indigeneity for the purpose of

the study and providing a brief history of the international discourse surrounding

indigenous rights.5 I then discuss in greater detail the predictions of the dominant

perspective in political science for the effects of collective recognition on state con-

solidation and democratic representation and elaborate on the theoretical linkages

between these two dependent variables. Next, I briefly summarize my argument and

5I discuss the history of this issue in the Philippines specifically in Chapter 3.

21

list the main contributions of this work. Finally, I describe the methodology I employ

in this study and outline the rest of the dissertation project.

1.1 Definitions

While the rights of indigenous peoples are established in international law, there

is no universally accepted definition of the term “indigenous peoples.” Given the

wide variation in the meaning of this term across country contexts and the fact that

international law emphasizes self-definition and self-ascription, the United Nations

has explicitly refrained from adopting such a definition (Charters and Stavenhagen

2009, UNHCR 2013).6 However, several documents provide working definitions or

frameworks that are frequently referenced.7 The Chairperson-Rapporteur of the UN’s

Working Group on Indigenous Populations, for examples, has offered the following

characteristics as central to the general understanding of the term (UNHCR 2013, 7):6According to one account by a participant in negotiations over early drafts of UNDRIP, the

decision not to include a definition in this document specifically was due to objections raised byAfrican and Asian governments: “African and Asian governments generally held the view that adefinition of the term ‘indigenous peoples’ should be included in the text in order to identify thebeneficiaries. It was clear that some of these states were more interested in obtaining a definitionwhich would exclude indigenous peoples in their own countries from becoming beneficiaries of theDeclaration. It was frequently stated by African and Asian States that they did not have anyindigenous peoples in their countries and that everyone was indigenous . . . Eventually, African andAsian governments dropped their insistence on a definition, and no such definition was included inthe Declaration as adopted by UNGA.” (Charters and Stavenhagen 2009, 80).

7For example,the Indigenous and Tribal Peoples Convention of 1989 (ILO Convention 169), theonly binding international convention on the rights of indigenous peoples, applies to the following:“a) tribal peoples in independent countries whose social, cultural and economic conditions distinguishthem from other sectors of the national community, and whose status is regulated wholly or partly bytheir own customs or traditions or by special laws or regulations; b) peoples in independent countrieswho are regarded as indigenous on account of their descent from the populations which inhabitedthe country, or a geographical region to which the country belongs, at the time of conquest orcolonisation or the establishment of present state boundaries who, irrespective of their legal status,retain some or all of their own social, economic, cultural, and political institutions” (ILO 1989).Another commonly cited “working definition” comes from the Martinez Cobo Study: “Indigenouscommunities, peoples and nations are those which, having a historical continuity with pre-invasionand pre-colonial societies that developed on their territories, consider themselves distinct from othersectors of the societies now prevailing on those territories, or parts of them. They form at presentnon-dominant sectors of society and are determined to preserve, develop and transmit to futuregenerations their ancestral territories, and their ethnic identity, as the basis of their continuedexistence as peoples, in accordance with their own cultural patterns, social institutions and legalsystem” (UNHCR 2013).

22

a) Priority in time, with respect to occupation and use of a specific territory;

b) The voluntary perpetuation of cultural distinctiveness, which may include the

aspects of language, social organization, religion and spiritual values, modes of

production, laws and institutions;

c) Self-identification, as well as recognition by other groups, or by State authorities,

as a distinct collectivity; and

d) An experience of subjugation, marginalization, dispossession, exclusion or dis-

crimination, whether or not these conditions persist.

This modern concept of indigenous peoples differs from the common usage of the word

“indigenous” meaning native to a particular place. While many of the first working

definitions emphasized aboriginality or continuity with pre-colonial societies, debates

over the definition in African and Asian contexts, where European settler colonialism

was generally less widespread than in the Americas, have led to a reduced emphasis on

this particular criterion (Charters and Stavenhagen 2009, Jefremovas and Perez 2011,

UNHCR 2013).8 Accordingly, the framework of indigenous rights has been applied to

some groups that are explicitly not descended from pre-colonial populations, such as

particular groups of African descent in Latin America.9

What has been consistently central to the understanding of indigenous peoples

is a lack of integration and assimilation into the dominant or mainstream societies

of contemporary states. The concept of indigenous peoples is therefore directly tied

to the concept of the nation-state. This connection is clear in historical accounts of

8In Asia, groups commonly referred to as indigenous peoples include Adivasi or scheduled tribes(India), “hill tribes” (Thailand), and Chittagong Hill Tracts (Bangladesh) (Anderson 1987, Chartersand Stavenhagen 2009, Inguanzo 2014). In Africa, the term indigenous has been claimed by nomadic,semi-nomadic, and pastoralist groups such as the Hadza (Tanzania), the San (Botswana), the Maasai(Kenya), and the Tuareg (Mali).

9The Brazilian government, for example, recognizes similar self-governance and collective prop-erty rights for indigenous Brazilians (American Indians) and the Afro-Brazilian descendants ofquilombos, or runaway slave communities.

23

the development of the concept in international law, and specifically its origins in the

era of global decolonization. The application of the term “indigenous” to populations

residing in independent states, as opposed to colonial possessions, began in the pe-

riod following the Second World War. It was explicitly used to describe populations

“not yet integrated into the national [communities]” of newly independent former

colonies (Rodríguez-Piñero 2006, 164, emphasis mine). Assimilating these popula-

tions to create homogenous national societies characterized by a universal concept

of citizenship was seen by policymakers at the time as a pre-condition for realizing

the self-determination rights of newly independent former colonies. Early interna-

tional legal documents on the issue of indigenous rights, such as the first Indigenous

and Tribal Populations Convention of 1957 (ILO Convention 107), emphasize inte-

gration of these populations as an explicit goal. As Rodríguez-Piñero (2006) writes,

“international law first defined indigenous peoples to see them disappear.”

Owing to efforts of indigenous rights advocates and an evolution in the concept

of human rights in the last decades of the 20th century, this discourse has shifted

radically.10 This shift is explicitly spelled out in the Indigenous and Tribal Peoples

Convention of 1989 (ILO Convention 169), with reference to the earlier convention:

“Considering that the developments which have taken place in international law since

1957, as well as developments in the situation of indigenous and tribal peoples in all

regions of the world, have made it appropriate to adopt new international standards

on the subject with a view to removing the assimilationist orientation of the earlier

standards...” (ILO 1989). Today, the idea that states should aspire to assimilate

indigenous populations has been replaced by a virtual consensus that these commu-

nities have collective rights to self-governance and self-determination. In other words,

while indigenous groups are still largely defined in terms of their distinctiveness from

“mainstream” societies in states where they reside, the stated goals of the interna-

10A full of this evolution is beyond the scope of this project and has been extensively documentedelsewhere. See Rodríguez-Piñero (2006) and Charters and Stavenhagen (2009).

24

tional legal framework (and the states that subscribe to it) are to protect and to

maintain this distinctiveness rather than attempting to eliminate it.

While this idea is now broadly accepted in international law and by governments

and policymakers around the world, tracing its evolution highlights important ways

in which it exists in tension with dominant perspectives in political science.11 Perhaps

most obviously, the idea of self-determination for groups within sovereign states poses

a direct challenge to the idea of national sovereignty, a premise of a large bodies of

work in international relations (Shadian 2010, Chowdhury 2018). However, I focus

instead on the problems the idea of collective self-determination rights poses for schol-

arship in comparative politics on state-building and democratic governance, much of

which is premised on the nation-state ideal.12 More specifically, the modern indige-

nous rights framework implies a kind of societal heterogeneity that is largely seen as

threatening to both state effectiveness and the functioning of liberal democracy. It

poses a direct challenge to the argument that institutional and cultural homogeniza-

tion are necessary for state-building and political legitimacy. In the next two sections,

I characterize and discuss this perspective primarily as it manifests within literature

on the political economy of development, first with respect to state effectiveness and

second with respect to democratic accountability.

11I am referring specifically to the political science tradition based in the (mainland) UnitedStates. As Ferguson (2016) notes, political science departments in Canada, Alaska, and Hawaii dohave dedicated scholarship focusing on First Nations experiences.

12For a more comprehensive critique of how U.S.-based political science handles Native issues instudying the United States, see Ferguson (2016). The tension motivating this project is consistentwith components of this critique; for example, the idea that the discipline is “structured around apresumption of nation-states.” In other ways, this project is squarely within the disciplinary traditionFerguson (2016) critiques. For example, it centers interactions between indigenous communities andformal state institutions rather than indigenous concepts of legitimacy, indigenous epistemology, orpolitics within indigenous political institutions.

25

1.2 Indigenous recognition, national identity, and the

state

The idea that societal heterogeneity presents challenges for state consolidation is

ubiquitous in the state-building literature. The contemporary strength of states is

often explained in terms of their success in subordinating competing sources of au-

thority (Weber 1976, Migdal 1988, Tilly 1992, Herbst 2000, Acemoglu et al. 2014)

and attachments to sub-national identity groupings such as ethnicity (Anderson 1983,

Gellner 1983, Miguel 2004). Migdal (1988), prominently, contrasts “web-like” societies

– characterized by the existence of many autonomous social organizations and “nu-

merous systems of justice [operating] simultaneously” – with “pyramid-like” societies

in which the state has managed to achieve predominance. Rulers of the former, he ar-

gues, have greater difficulty achieving their policy goals and providing public goods,

such as health services, education, and security, because they lack the legitimacy

necessarily to mobilize their populations toward these ends.

The well-documented association between various forms of societal fragmenta-

tion and state weakness implies that recognition of sub-national collectives, including

indigenous groups, should hinder states’ abilities to project authority across their

territories. Holzinger et al. (2018), for example, take this as a premise in their cross-

national study of indigenous rights provisions, describing these policies as “clipping

the wings” of government. This literature suggests two primary mechanisms through

which this may occur. First, ceding additional powers to non-state authorities may

increase the population’s reliance on these authorities and empower them to resist

state priorities (Migdal 1988, Levi 1989, Soifer 2016). Second, to the extent that

these policies calcify ethnic divisions or strengthen sub-national identities (Chan-

dra 2006), they may weaken solidarity with the national community (Mill 1861, Ekeh

1975, Lemarchand 1972, Fukuyama 2018), fuel ethnic conflict (Horowitz 1985, Lieber-

26

man and Singh 2017), or contribute to sub-optimal public goods provision (Alesina,

Baqir and Easterly 1999, Miguel 2004, Miguel and Gugerty 2005) which may in turn

compromise state legitimacy (Levi, Sacks and Tyler 2009).

However, in addition to a paucity of empirical evidence on the effects of collec-

tive recognition, there are important reasons to question predictions derived from

the conventional wisdom within the state-building literature. I highlight two here.

The first is empirical, relating to challenges in inferring the causal direction in the

relationship between societal fragmentation and state strength. If it is the case, as

Scott (1998), Migdal (1988), and others have argued, that states generally seek to

achieve predominance and to establish monopolies not only on violence but on many

forms of social control, it is presumably the least capable states that have failed to

do so. In other words, state weakness may be a cause rather than a consequence of

states’ failure to centralize authority and homogenize populations within their bor-

ders (Wimmer 2016). With respect to recognition specifically, decisions by states

to empower non-state authorities rather than trying to sideline them may reflect an

unfavorable pre-existing balance of power between national and local elites (Boone

2003, Gerring et al. 2011).13

The second critique is theoretical and substantive, building on work outside the

dominant perspective on nationalism and state-building that offers reasons recog-

nition may strengthen rather than undermine state authority. Recent work on the

relationship between traditional and state authority suggests that they can be com-

plementary and mutually reinforcing rather than competing (Logan 2013, Van der

Windt et al. 2018, Mershon 2020). Baldwin (2015) documents how traditional au-

thorities can serve as “development brokers” who facilitate connections between the

central state and peripheral populations. Strengthening pre-existing indigenous lead-

13Fragmentation may also result from circumstances of state formation that disincentive costlyaction to wrest control from powerful local elites and simultaneously produce state weakness throughother channels (Herbst 2000, Chowdhury 2018).

27

ership structures and rationalizing their relationship to state authorities may enhance

this facilitation role and, in doing so, increase the state’s legitimacy (Englebert 2002).

Relatedly —though with different normative implications —collective recognition

may strengthen the state’s coercive capacity by simultaneously increasing traditional

authorities’ loyalty to the state and their power to exact compliance with state prior-

ities (Mamdani 1996, Baldwin 2011).14 The process of recognition may also increase

state control by making indigenous communities more “legible” (Scott 1998) and re-

defining indigenous authority structures on terms controlled by the state (Kyed and

Buur 2007, Mamdani 2012).

1.3 Indigenous recognition, democratic representa-

tion, and accountability

Policies recognizing collective self-governance rights for indigenous communities raise

concerns not only about the “strength” of states, but also about the quality of democ-

racy. In particular, the proliferation of non-state governing authorities and sub-

national identities —both of which this form of recognition may encourage —are

thought to have negative consequences for democratic representation and account-

ability. From a normative perspective, granting collective rights to particular groups

exists in tension with the liberal democratic emphasis on individual autonomy (Tay-

lor et al. 1994, Barry 2002, Kuper 2003, Appiah 2007b). According to this argument,

if individuals are subject to different “rules of the game” based on their membership

in a particular community or ethnic group, they are not fully able to exercise their

individual rights as citizens of a liberal democratic polity.

Research in political science points to a specific mechanisms through which this

14However, it is important to note that much of this literature emphasizes electoral benefits forincumbents rather than loyalty to the state, and implies that immediate political pay-offs come atthe expense of longer-term state capacity (see, for example Ntsebeza 2005, Acemoglu et al. 2014.)

28

type of intermediation between citizens and the state institutions can compromise

the exercise of democratic rights and the accountability functions of democratic in-

stitutions. A large body of work, rooted largely (but not exclusively) in the African

context, argues that traditional authorities often act as clientelistic vote brokers who

deliver the votes of their communities in exchange for private rents, entrenching pow-

erful political elites at the expense of their communities’ interests (Mamdani 1996,

Ntsebeza 2005, Anderson, Francois and Kotwal 2012, Acemoglu, Reed and Robinson

2014, De Kadt and Larreguy 2015, Koter 2013, Boone 2014, Mershon 2015). If state

recognition of indigenous institutions gives leaders of these institutions more material

power over members of their community, it may increase their ability to mobilize

votes for their preferred candidates.15 State recognition may also confer symbolic

legitimation, endowing leaders with a kind of “credibility surplus” that translates into

an ability to mobilize votes, regardless of candidate quality (Platow and van Knip-

penberg 2001). In this way, individuals subject to indigenous governing structures

may be prevented from casting their votes for candidates they individually prefer or

who, if elected, would improve their material outcomes. If candidates are accountable

only to leaders, rather than to constituent as a whole, this creates opportunities for

rent-seeking behavior. This is thought to be especially problematic when traditional

institutions themselves are non-democratic.

However, this prediction relies on several assumptions that may not hold uni-

versally. The first is that the interests of non-state authorities —chiefs, indigenous

authorities, or traditional leaders —are misaligned with the interests of individual

members of the communities they govern. In other words, when individuals “vote

with the chief” they are voting against their own interests, or at least for a candidate

other than the one they would choose in the absence of coercion. However, as Baldwin

15Following a similar logic, Baldwin (2011) argues that, in the African context, decisions to devolvepower over land to traditional chiefs are motivated by the prospect of mobilizing non-coethnic votesfor the ruling party.

29

(2015) argues in the case of African chiefs, the interests of these authorities are often

more aligned with the interests of their communities than are those of elected offi-

cials due to chiefs’ comparatively longer time horizons.16 Similarly, Tsai (2007) argues

that non-democratically-elected authorities embedded in their communities are more

likely to act in the communities’ interests because they are constrained by informal

accountability mechanisms. As Ferguson (2016) notes, political science has devoted

little attention to accountability mechanisms and legitimacy within indigenous polit-

ical institutions. However, the assumption that such accountability mechanisms do

not exist in the absence of, for example, democratic elections is not self-evident.

A second assumption concerns the counterfactual upon which critics of traditional

authority and group-differentiated rights implicitly or explicitly rely: namely, that

those subject to these institutions or classifications would be better off encounter-

ing formal democratic institutions as undifferentiated individuals. This assumption

may not hold in contexts where elections in general are not contested along ideolog-

ical or programmatic dimensions, but are instead characterized by vote-buying and

clientelism, or the contingent exchange of votes for individual- or community-level

goods (Kitschelt and Wilkinson 2007, Hicken 2011). In these cases, communities

that are able to “pool their political resources” and coordinate in candidate selection

have an advantage in attracting club goods from the state in lieu of or in addition to

small payments for individual votes (Boone 2003, Smith and Mesquita 2012, Gottlieb

and Larreguy 2016). In the absence of these coordination mechanisms, individuals

exert more limited influence over their political representatives and in fact may be

locked into relationships of “perverse accountability” with the politicians who pay for

their votes (Stokes 2005).17 Voter coordination may be particularly welfare-enhancing

16Furthermore, when the provision of state public goods requires coordination between state andnon-state authorities, alignment between a candidate and traditional leaders may be relevant forindividuals evaluating likely candidate performance (Baldwin 2013).

17The idea that collective action in the context of democratic elections can advance the interests ofindividuals within a collective is ubiquitous in literature on labor movements (Rueschemeyer, Huberand Stephens 1992), corporatist interest representation (Hall 2001), civil society (Edwards, Foley

30

for individuals from historically marginalized communities who otherwise lack access

to established patronage networks through which state resources are regularly dis-

tributed. There is a broader question of whether collective recognition exacerbates

distributive politics on a systemic level and serves to prevent the formation of, for

example, a broad-based class politics. However, in cases where mainstream politics

are not structured along class lines, it is not obvious that decisions about how to

incorporate groups on the periphery will change this.

In summary, predictions from literatures on state-building and democratic ac-

countability, rooted in a paradigm that assumes the primacy of the nation-state and

the importance of individual autonomy within liberal democracies, imply negative un-

intended consequences resulting from the granting of collective self-governance rights

to indigenous communities. More recent work challenges the specific logic of this

prediction for each type of outcome. Before outlining my argument, I briefly discuss

the decision to focus on these two outcomes and the ways in which they are linked

theoretically.

1.4 State consolidation, democratic accountability,

and systemic legitimacy

The dominant perspectives in political science on state consolidation and democratic

representation are rooted in a common liberal paradigm based on a model of universal

citizenship in which ethnic or cultural based differences are not recognized as relevant

within the public sphere (Rawls 2005). The decision in this project to focus on these

two categories of outcomes stems primarily from their common intellectual roots

and from historical policy discourse and political science literature that emphasizes

and Diani 2001), and social capital (Boix and Posner 1998). However, this perspective is rarelyapplied to identity-based groups. I explore this distinction in greater detail in Chapter 6.

31

societal heterogeneity as a threat to both. In recent decades, political science as a

discipline has for the most part considered these two outcomes separately. Yet earlier

work links them more directly in ways that are particularly relevant to post-colonial

and other weakly consolidated democracies.

In particular, work on state consolidation (Huntington 1968) and institutional

design (Lijphart 1977) in post-colonial societies directly links prospects of represen-

tation within the formal political system to systemic legitimacy. Huntington (1968)

considers participation in formal state-sanctioned channels as an important indicator

of state legitimacy, alongside more direct forms of compliance such as tax payment

(Levi 1989), military conscription, cooperation with law enforcement, and compliance

with public health directives (Lieberman 2009, Blair, Morse and Tsai 2017). The

ability (perceived or actual) to make successful claims on the state through formal

political channels is likely to be a key determinant of this participation.

Importantly, the existence of liberal democratic institutions alone does not guar-

antee the ability of all individuals within a state to make successful claims. Earlier

cross-national work (Verba, Nie and Kim 1987) and more recent work in the United

States context (Bartels 2010, Schlozman, Brady and Verba 2018), highlights how

socioeconomic inequality limits substantive representation for certain groups within

nominally egalitarian political systems. Groups that are defined by their histori-

cal marginalization from the state, such as indigenous groups, are especially likely

to be at a disadvantage. Under conditions of incomplete state consolidation, where

these groups have viable “exit options,” they may instead decide to pursue their aims

through extra-systemic means (such as armed unrest or support for rebel groups).18

In this way, failures of democratic representation can weaken state legitimacy and

directly undermine state consolidation. By contrast, group-based representation, to

18As Huntington (1968) writes, “...social groups demand participation in the political system, andthe system either provides for this participation in ways that are harmonious with the continuedexistence of the system, or it alienates the groups from the system and produces overt or covert civilstrife and secession” (140).

32

the extent that it enhances the claim-making capabilities of marginalized groups, may

help to boost it (Lijphart 1977, Jung 2008).19 A systematic study of how collective

recognition affects both outcomes provides an opportunity to consider both issues

separately and revisit the theoretical connections between them.

1.5 Summary of the argument, findings, and contri-

butions

In this project, I argue that collective recognition (and specifically the granting of

collective self-governance rights) to marginalized groups can both enhance state le-

gitimacy and improve the quality of democratic representation. First, recognition

allows the state to extend its reach in a way that is compatible with the incentives

of local elites who have the ability to mobilize credible resistance to it. Second,

it enhances the ability of these groups to make successful collective claims through

the formal political system, by establishing a political identity that legitimates these

claims (Jung 2008) and by facilitating collective mobilization that increases commu-

nities’ leverage in interactions with elected officials and the state bureaucracy. By

contrast, alternative governing strategies that are more consistent with the dominant

liberal paradigm —assimilation and egalitarian governance —are likely to engender

continued resistance to or evasion from the state. Given underling conditions of state

weakness, collective recognition offers a more viable approach to incorporating periph-

eral populations such as indigenous communities, and improves the representation of

these communities within democratic politics.

I present a variety of empirical evidence in support of my argument, using data

from the Philippines, which I briefly preview here. I find that the granting of collec-

19In fact, Elkins and Sides (2007) find suggestive evidence that group-based representation canincrease affective attachment to the state among minority groups.

33

tive land titles to indigenous communities leads to increases self-identification with

indigenous ethnic groups on the census, a finding that on its face may seem to affirm

concerns that recognition of indigenous self-determination rights exacerbates ethnic

divisions. Yet I also find that titling leads to increases in birth registration and census

data quality, two important indicators of compliance with the state among popula-

tions that have often resisted being made “legible” (Scott 2010). Gains for the state

on this dimension are meaningful: for example, I estimate that the increase in birth

registration from titling accounts for up to 21% of the baseline difference in registra-

tion between indigenous and non-indigenous communities. Findings from a survey

experiment support the idea that collective recognition, rather than de-legitimizing

the state, increases identification with the nation as a whole. I also examine the effects

of collective recognition on electoral behavior and expectations of the formal political

system. I find that land titling is associated with greater electoral cohesion at the

precinct level, pointing to an increase in local-level bloc voting. However, contrary

to concerns about elite capture, it does not appear to entrench the power of exist-

ing political elites or reduce political competition. Survey results are consistent with

the idea that recognition instead increases claim-making behavior directed toward

obtaining local public goods from the state.

The project makes at least three contributions to the existing literature. First,

it offers new empirical evidence, using detailed sub-national data, on an important

policy question that has been actively debated among policymakers and political the-

orists, but until recently has received little attention in the empirical political econ-

omy literature. Second, it draws from usually disparate literatures on state-building,

democratic accountability, and ethnic and identity politics to provide a micro-level be-

havioral theory of state capacity relevant to many post-colonial democracies. Third, it

contributes to our understanding of ethnic or identity group politics by highlighting

the degree of integration into the state system as a relevant characteristic of soci-

34

etal groups and proposing conditions under which group-based representation can

contribute to, rather than undermine, the quality of governance. While the project

focuses on indigenous recognition specifically, the theory and insights may apply to

other types of marginalized groups. It speaks to broader debates about the implica-

tions of “identity politics” within liberal democracies, particularly but not exclusively

in post-colonial settings (Lilla 2017, Fukuyama 2018).

Below I discuss the methodology used in the project, before briefly outlining the

following chapters.

1.6 Methodology

I study the effects of collective recognition for indigenous communities in a single

country context. As mentioned above, the Philippines is a qualitatively important

case for this question given its comprehensive policy of indigenous recognition (dis-

cussed in greater detail in Chapter 3) (Inguanzo 2014). In addition, the existence of

sub-national variation in the implementation of this policy allows for a plausibly more

convincing analysis of the effects of recognition, holding state-level factors constant.

Within the Philippines context, I apply a mixed methods approach to study the

effects of collective recognition at the community level. First, I use in-depth inter-

views with tribal leaders, policymakers, elected officials, and advocates conducted

over 25 months of fieldwork in the Philippines. These interviews were central to

both the development and refinement of my theory, and the design of strategies to

test its observable implications more widely using quantitative data. Second, I draw

on an original face-to-face survey conducted among indigenous respondents in three

provinces, developed and designed in part based on insights gathered from the qual-

itative data. In addition to examining observable implications of the theory more

systematically and across a larger set of communities, this allows me to understand

35

whether views held by non-elite community members are consistent with those ex-

pressed by elites.20 I also use an embedded survey experiment to test the effects of

priming state policies of recognition on relevant individual attitudes. Third, I analyze

administrative data, leveraging spatial and temporal variation in the granting of com-

munal land titles to study the effects of recognition on indicators of compliance with

the state, electoral behavior, and local public goods provision on a national scale. I

briefly describe each of these data sources below.21

1.6.1 Qualitative interviews

Over several visits to the Philippines ranging in duration from two weeks to four

months between 2014 and 2016, I conducted exploratory interviews with current

and former government officials, indigenous advocates, non-governmental organiza-

tion staff, representatives of intergovernmental organizations such as the World Bank

and the Asian Development Bank, academics, and tribal leaders. Interviews with

practitioners and policymakers took place in the national capital, Manila, as well as

the regional centers of Baguio City and Davao City. Exploratory interviews with

tribal leaders were conducted in the provinces of Davao, Benguet, Ifugao, and Nueva

Vizcaya between March and December 2016. Following this exploratory phase, I

conducted semi-structured, audio-recorded interviews with tribal leaders in Oriental

Mindoro, Occidental Mindoro, and Palawan between June and October 2017, in a

total of 10 tribes.22 In total, I spoke with 107 individuals in an unstructured or semi-

structured interview setting. Figure 1-1 shows all provinces where these interviews

were conducted. This work also benefits from countless hours interacting with vari-

20In this context, conducting in-depth interviews with non-elite individuals was particularly chal-lenging given language barriers and my positionality as an outsider.

21Data sources are discussed in greater detail in subsequent chapters.22These locations were selected because of the opportunity to conduct survey research in these

provinces, in conjunction with an NGO partner, Legal Network for Truthful Elections (LENTE).This component of the research received exempt approval from MIT’s IRB (COUHES Protocol #1705974195).

36

Figure 1-1: Qualitative interview sites

ous levels of the Philippines’ bureaucracy for the purposes of collecting administrative

data and survey implementation, participation in national and regional conferences

and meetings on indigenous issues, and participant observation at community meet-

ings.23

1.6.2 Community survey

The second major source of data used in this chapter comes from an original sur-

vey of leaders and community members in indigenous communities in the same three

provinces: Oriental Mindoro, Occidental Mindoro, and Palawan. The survey was

23This included, for example, community celebrations, meetings for livelihood projects, and nego-tiations between indigenous communities and foreign hydropower and mining companies.

37

conducted between May and September 2018, in partnership with Legal Network

for Truthful Elections (LENTE), a Philippines-based NGO. The development of sur-

vey questions was informed by prior qualitative interviews and fieldwork. Survey

questions were developed in conjunction with LENTE and initially piloted in Ifugao

province in September 2017, and in out-of-sample communities in Palawan in Decem-

ber 2017.24 Communities were selected for the survey example to achieve variation

in recognition status (specifically variation in the granting of collective land titles).25

In each barangay, enumerators were assigned to interview five randomly-selected

community members in each of two sitios (neighborhoods), yielding a total target

sample of 1,000 community respondents. Due to security concerns, survey teams

were unable to visit 20 of the 100 target barangays.26 The final sample therefore in-

cludes data from 80 barangays. The survey included a survey experiment intended to

test the effects of priming state recognition policies on relevant individual attitudes.

Unless otherwise noted, all analyses of the original survey data —descriptive and ex-

perimental —are pre-registered.27 Additional information about survey methodology

and implementation is included in the Appendix.

1.6.3 Administrative data

Finally, I analyze nation-wide administrative data on collective land titling, elec-

toral returns, various measures of compliance with the state, and local public goods

provision to estimate the effects of community-level recognition. I employ a num-

ber of strategies to address potential sources of confounding. Where possible I use24The survey locations correspond to areas where LENTE, in conjunction with the Philippines’

Commission on Elections (COMELEC), implemented a pilot intervention that involved establishingseparate polling places for indigenous communities. The survey was intended in part as an evaluationof this intervention, and is discussed in greater detail in Chapter 5

25Within the target sample of barangays 46 out 100 fell within an area covered by an ancestralland title at the time of the survey.

26The survey period coincided with military operations against the New People’s Army in certainlocations, and teams were not permitted to visit these areas.

27The pre-analysis plan is filed in the Evidence in Government and Politics (EGAP) registry,under ID 20181010AA.

38

differences-in-differences or panel estimation and restrict the analysis to subsets of the

data for which parallel trends are more plausible. Where outcomes are not measured

over time, I use matching and make a selection-on-observables assumption.

1.7 Plan of the dissertation

The subsequent chapters proceed as follows. In the next chapter (Chapter 2), I

present my theory of recognition as integration. Chapter 3 provides a description

of the Philippines context, including a brief history of policies toward indigenous

communities, details of the current policy that is the focus of my empirical analysis,

and a discussion of the sub-national variation in the implementation of this policy. I

then illustrate key propositions in my theory in this context using qualitative interview

data, primarily taken from elite interviews in Oriental Mindoro, Occidental Mindoro,

and Palawan.

Chapter 4 provides quantitative empirical evidence of the first key observable im-

plication of the theory: that collective recognition leads to greater engagement and

voluntary compliance with the state. First, I leverage spatial and temporal variation

in the granting of collective land titles, known as Certificates of Ancestral Domain

Title (CADTs), to indigenous communities. I estimate the effects of collective titling

on indigenous identity and multiple indicators of state compliance. Using differences-

and-differences and panel specifications with a variety of robustness and sensitiv-

ity checks, I find evidence that titling increased both indigenous self-identification

and voluntary compliance. Second, I provide evidence from a survey experiment in

which some respondents were primed with educational materials about the Philippine

governments’ policy of recognition and asked to indicate the relative importance of

different aspects of their identity in a ranking activity. I find that the experimen-

tal treatment increased the importance of respondents’ national (Filipino) identity,

39

relative to their tribe and other characteristics.

Chapter 5 evaluates predictions about the effects of recognition on participation

in the formal political system, focusing on electoral behavior and again using a com-

bination of administrative and survey data. Analyzing precinct-level electoral returns

from mayoral elections, I show that land titling is associated with greater electoral

cohesion at the precinct level, but with greater competition at the municipal level.

Descriptive comparisons of titled and untitled communities within the survey sam-

ple support the idea that this greater cohesion is due in part to greater mobilization

capacity on the part of indigenous leaders. Additionally, I provide evidence from

a pilot intervention that established separate polling places for indigenous commu-

nities. I find that this intervention increased electoral competition at the barangay

(village) level and decreased incumbent vote share. Next, I provide evidence from

survey experiments that recognition shapes expectations about engagement in the

formal political system. Finally, I provide evidence that collective recognition in the

form of titling is associated with differential increases in some types of local public

goods.

I conclude by considering medium- to long-term implications of these findings

for indigenous recognition in the Philippines and beyond, connecting the issue of

indigenous rights to broader insights about state-building, democracy, and “identity

politics” in post-colonial societies, and outlining directions for future research.

40

Chapter 2

Recognition as Integration

This chapter offers a new theory about the effects of state recognition of collective

self-determination rights for indigenous communities on state consolidation and rep-

resentation in developing democracies. Rather than weakening states and undermin-

ing democratic quality as much of the conventional wisdom suggests, it argues that

collective recognition of indigenous communities and other groups defined by their

historical marginalization from the state can both increase the integration of these

communities into the state system and improve their representation in democratic

politics. While this theory is relevant to normative debates regarding the desirability

of multiculturalism and collective political rights (Taylor et al. 1994), it does not

speak directly to this question. Instead, it engages empirical claims in the political

science literature regarding the implications of societal diversity and applies a polit-

ical economy approach, focusing on the incentives and constraints of the state, local

indigenous elites, and indigenous communities as a whole.

The conventional wisdom in political science, grounded in canonical literatures on

state-building and nationalism, suggests that assimilating populations under a unified

governing system and national identity is preferable, and even necessary, for estab-

lishing both state authority and democratic legitimacy. Colloquially, societies that

41

are “melting pots,” where minority groups assimilate into a broader national identity

and culture, are though to be more successful than multicultural “salad bowls” where

groups within the same state maintain distinctive characteristics. This difference

is further heightened when this entails maintaining group-differentiated rights and

governing institutions (Kymlicka 1996).

However, achieving the melting pot ideal is a non-trivial task for states. This is

particularly the case for states that, due to the circumstances of their formation, lack

pre-existing de facto control over populations within much of their territory, have

limited coercive resources, and do not face external threats that have historically

incentivized costly thorough-going state-building projects (Herbst 2000, Chowdhury

2018). These weaker states are unable to unilaterally forge a cohesive national iden-

tity and undermine competing sources of authority in society, giving peripheral com-

munities who have historically existed outside the state’s reach (such as indigenous

communities) agency to shape the extent and the terms of their incorporation.

I argue that collective recognition —relative to other possible modes of state-

society interaction —can facilitate the extension of state authority into areas where

this authority is contested by altering the incentives of traditional elites and of com-

munities. For both sets of actors, collective recognition increases expected returns

to engagement with the state, creating incentives for these communities to channel

demands through the formal state system rather than remaining outside of it or ac-

tively resisting it. First, by designating local elites as intermediaries between the

state and the community, collective recognition creates conditions whereby contact

with the state reinforces, rather than undermines, elites’ status within their own

communities. Second, collective recognition legitimates and enables collective claim-

making through formal channels, allowing these communities to benefit more from

participation in formal politics than they would in its absence. In doing so, it not

only bolsters the legitimacy of formal state institutions but also offers the promise of

42

improved representation and accountability.

This chapter begins by describing the central dilemma facing incompletely consoli-

dated states attempting to extend authority across their territories. Next, it describes

potential governing strategies and discusses how each shapes the incentives of both

traditional elites and communities, concluding that only collective recognition does so

in a way that incentivizes (and does not actively disincentivize) systemic integration.

I then extend the argument to consider implications for democratic responsiveness

and representation. I conclude by presenting contrasting predictions from the con-

ventional wisdom and from my theory for both sets of outcomes, predictions which I

evaluate empirically in subsequent chapters.

2.1 The statebuilder’s dilemma

The conventional wisdom in the literature and among policy-makers that has moti-

vated concerns about indigenous and other forms of collective recognition suggests

that nation-building is a prerequisite for effective state-building (Weber 1976, Gellner

1983, Hobsbawm 1992). No state can achieve its policy goals through coercion alone;

as a result, all states rely to some degree on voluntary compliance on the part of

citizens (Levi 1989). Voluntary compliance is enhanced by state legitimacy, or the

belief that the state has the right to govern (Tyler 1997). Ethnic and institutional

heterogeneity compromise state legitimacy. If citizens do not feel a part of the na-

tional community, they are unlikely to make sacrifices for their fellow citizens (Gellner

1983, Fukuyama 2018). If they rely on authorities other than the state, they are less

likely to comply with the state’s directives (Migdal 1988, Scott 2010). As a result,

states that have succeeded in homogenizing and centralizing are stronger and more

successful than those that have not. In order to make themselves more legitimate

(and ultimately more effective), according to this logic, states should pursue assim-

43

ilative nation-building. Collective recognition, by contrast, risks reinforcing societal

fragmentation and undermining state effectiveness in the long run.

Yet taking this logic on its face reveals a “catch-22” facing weak states. Attempts

to centralize power in the hands of the state are inherently threatening to non-state

power structures. Efforts to homogenize the population under a single (exclusive)

national identity are inherently threatening to existing identity attachments. Assim-

ilative nation-building requires political will and capacity to overcome these sources of

resistance. States that are weaker a priori are precisely those that are least capable of

overcoming this resistance. While successful assimilation may produce outcomes that

states would prefer (compared to a more heterogeneous counterfactual), achieving it

may not be possible. In other words, what is often treated as a choice on the part of

states may not truly be one in practice.1

Scholars and policymakers warning about the effects of collective recognition may

therefore be drawing incorrect inferences about its likely effects. It may be the case

that societal fragmentation weakens states, but it may also be that states with frag-

mented societies are too weak to successfully implement strategies of assimilative

nation-building. When applied to weak states, the appropriate question is not “what

is the relationship between fragmentation and state strength?” Instead given back-

ground conditions of state weakness, does attempting centralization and thorough-

going nation-building increase state capacity? Or, from the state’s perspective, is it

preferable to collectively recognize existing groups authority structures, even if this

may have the effect of reinforcing or exacerbating societal fragmentation? In the next

section, I present an argument for the latter, focusing on the incentives of three key

actors: the state, local elites, and communities.

1For example, Gerring et al. (2011) et al argue that states are more likely to devolve power tosub-national units that are themselves more centralized because delegation to centralized units ismore feasible.

44

2.2 Recognition or assimilation?

2.2.1 The state

This theory applies to states that, at a particular point in time (e.g. independence

from a colonial power) have incompletely consolidated authority over the population

living within the territory over which they claim sovereignty. Within this territory,

some subset of the population is governed by traditional elites whose legitimacy does

not derive entirely or predominately from a direct association with the state. This

subset may also be differentiated ethnically, culturally, and/or linguistically from

populations over which the state exerts a greater degree of de facto authority.

All states, at some point, seek to mobilize their populations in order to fulfill core

functions, such as raising revenue through taxation and the exploitation of natural

resources. At a minimum, mobilization is necessary to prevent and contain crises that

threaten the state’s integrity such as internal armed conflict and large-scale disease

epidemics. More ambitious states may attempt mobilization in order to achieve de-

velopment and “modernity,” for ideological reasons or to increase their standing on

the international stage (Migdal 1988, Scott 2010, McGovern 2012).

To mobilize the population toward any of these goals, all states rely on a combi-

nation of coercion and voluntary compliance (Levi 1989). Weak states lack coercive

capacity and rely to a greater extent on voluntary compliance to mobilize repeat-

edly.2 At the same time, they lack the pre-existing legitimacy necessary to obtain

this compliance, particularly (by definition) within the subset of the population over

which they have weaker de facto control. How does the state go about establishing

the legitimacy necessary to obtain voluntary compliance toward its policy goals?3 I

2This may include, for example, co-production of public goods, to the extent that providing suchgoods is a priority for the central state (Baldwin 2015).

3I assume here that assimilation is not a policy goal in and of itself; instead, it is a meanstoward establishing legitimacy that allows the state to fulfill other functions deemed as necessary.This instrumental view of assimilation is consistent with its treatment in much of the state-buildingliterature (see, for example Migdal (1988)).

45

define three possible strategies or modes of state-society interaction states may em-

ploy to extend authority across their territories. These strategies are distinguished by

how the state relates to existing groupings and sources of non-state authority within

society; more specifically whether it attempts to eliminate them, ignores them, or

actively recognizes them. The three strategies are summarized in Figure 2-1, posi-

tioned along a spectrum from attempted elimination to recognition, with examples

of policies that characterize each.

The first strategy is assimilation, which involves actively attempting to elimi-

nate competing identities and sources of authority in order to bolster loyalty to the

state. Examples of this strategy include conscription into national service, forced re-

moval of individuals from their communities combined with re-education in officially-

sanctioned languages and cultural practices (e.g. residential schools established to

re-educate Native American children in Canada and the United States, so-called “re-

education” camps for the Uighur minority in China), the establishment of a national

language combined with bans on use of alternative languages in public life, the es-

tablishment of a state religion and banning of non-sanctioned religious and ritual

practices,4 and the abolishment of customary land rights and subsequent nationaliza-

tion or redistribution of land to individuals.5

The second I call egalitarian governance. In this strategy, the state does not

actively try to eliminate group-specific cultural practices, sub-national identities, or

non-state authority structures. Instead, it simply ignores them and interacts with the

population as a single group of undifferentiated individuals. State policies may apply

differentially to individuals based on their geographic location, occupational status,

or level of income —e.g. for means-tested poverty alleviation programs —but indi-

4An example of this is the Demystification Program in post-colonial Guinea. The program, whichinvolved the destruction of masks and ritual objects within polytheistic forestier communities was,as McGovern (2012) writes, “so urgent it even preceded the building of a national road system.”

5Examples of this latter policy include the ujamaa under Tanzania’s Julius Nyerere (Scott 1998,Miguel 2004) and the Dawes Act in the United States.

46

Governing strategies

Assimilation

Residential schoolsNational conscriptionRe-education campsLanguage bansReligion bansLand nationalization

No ethnic countingNational language with-out bansCustomary land tenureneither abolished norprotectedNovel structures for com-munity governance

Group-differentiatedstate servicesCustomary justiceCustomary land-tenureTraditional authoritiesconsulted on projects

Egalitarian governance Collective recognition

Active destruction Indifference Recognition

Relation to existing non-state authority structures

Figure 2-1: Governance strategies differentiated by modes of state-society interaction.

viduals are not differentiated along ethnic, tribal, or linguistic lines. State agencies

interact directly with individuals; these interactions are not mediated by any kind of

non-state authority, and traditional within-community hierarchies are not respected.

Public education may be delivered primarily in a national language, but use of other

languages is not actively banned. The government may avoid collecting data on the

population differentiated by ethnic or linguistic group (e.g. through the census). Of-

ficially, all individuals are subject to the same “rules of the game” within a single legal

system. Customary authority over land and local justice systems may not be actively

challenged, but is also not actively recognized or protected. To the extent that the

state interacts with geographically-defined communities (e.g. for the purpose of deliv-

ering community-level development projects) it ignores existing leadership structures

and may attempt to create decision-making mechanisms on an ad hoc basis. A com-

mon example of this is Community-Driven Development (CDD) programs which set

up novel governing structures for the purpose of project implementation and oversight

(Mansuri 2012, Casey 2018).

The third is collective recognition, which involves explicitly recognizing group-

47

specific sub-national identities and associated governing structures. In this model,

existing non-state authorities are authorized to set the rules of the game for groups

of citizens in key areas of local governance, including, for example, dispute resolution

and land allocation. Education and other services, even if state-funded, may be

tailored to specific communities or operated by community authorities (e.g. tribal

schools). All or most interactions between the state and the community are mediated

through these authorities, who are authorized to make or to convey decisions on the

community’s behalf. To contrast with the previous strategy, a CDD-type program

would recognize existing governing structures to manage and make decisions about

state-funded development projects rather than creating new structures for this specific

purpose. Programs targeted at individuals would be implemented in consultation

with community leadership. Modern tribal sovereignty in the United States is a

strong form of collective recognition, in which federally-recognized Native American

tribes are treated as “domestic dependent” nations within the United States, with

(limited) criminal jurisdiction through tribal courts and rights to operate services

such as health and education and exploit natural resources within their lands.

In the absence of any constraints, I assume the state prefers assimilation to the

other two strategies. Assimilation, by disrupting existing leadership structures, com-

munity ties, and strategies of survival, creates the conditions under which the state

can govern most effectively in the long term. If implemented successfully, this strategy

would increase the state’s ability to mobilize its population in the future by establish-

ing an effective monopoly on social control. This social control serves as a “currency”

(Migdal 2001) or a reservoir (Tyler 2006) upon which the state can continuously

draw. Collective recognition, by contrast, risks strengthening alternative sources of

authority and sub-national identities in ways that exacerbate societal fragmentation

and, in doing so, create barriers to state mobilization.

As discussed above, however, weak states in particular are constrained. Given this,

48

it is important to understand the incentives that societal actors face under different

strategies in their decisions to engage with the state and channel demands through

the formal political system. I discuss this next, beginning with pre-existing non-state

elites that exert localized authority over the populations in question.

2.2.2 Traditional elites

Traditional elites are actors who exercise authority over territorially-defined groups

and whose legitimacy to govern these groups either predates the state or does not

rely primarily on the direct delegation of authority from the state. I assume these

elites wish to maintain their privileged status within their communities. Due to pre-

existing legitimacy among the population, I assume they have some ability to influence

the beliefs and actions of their constituents, and that this influence can be used to

encourage or discourage cooperation and engagement with the state. However, as I

discuss in greater detail when considering the community’s perspective, this influence

is not unlimited.

Assimilationist approaches, by definition, directly threaten the status of tradi-

tional elites. As a result, they are almost certain to encounter resistance (Bandy-

opadhyay and Green 2013, Linz 1993, Scott 2010). In the face of assimilationist

policies, elites’ optimal strategy for maintaining their legitimacy within their own

communities is to encourage disengagement from or resistance to the state system on

the part of their constituents. Unless states are willing and able to bring overwhelm-

ing force to bear —which is unlikely in the absence of external threats (Herbst 2000,

Centeno 2003) —they will be unable to prevent such evasion. For most weak states,

therefore, assimilation is likely to be counterproductive.

Egalitarian governance, while it does not involve direct attempts to eliminate or

disempower traditional elites, is also threatening to their privileged position within

the community. Attempts by the state to interact with individual citizens while

49

ignoring these elites also undermine the latter’s authority and position within their

communities. If the state distributes resources to the population in a bid for “hearts

and minds” the threat that communities may shift their loyalties away from traditional

elites and toward the state is further heightened. While traditional elites enjoy pre-

existing legitimacy, they cannot compete with the state in terms of the provision of

material resources. Under this mode of state-society interaction, contact between a

community and the state directly threatens these elites’ positions. In the face of this

threat, elites are again likely to discourage the community from interacting with or

complying with the state. Community members, for their part, are likely to take cues

from local elites they know over those from a distant central state, particularly when

there is a history of exploitation by and marginalization from the center.6

Collective recognition produces a different set of incentives on the part of local

elites. Through recognition, the state designates existing elites as legitimate inter-

mediaries between the state and the community. Recognition of their status by the

state reinforces their right to govern. These elites may be able to capture some of

the benefits of state-led development projects, by claiming credit for these projects,

selectively targeting resources to their supporters within the community, and shaping

projects to better reflect community needs. When state contact with the population

is mediated through traditional authority structures, traditional elites gain a greater

advantage from encouraging cooperation and engagement with the state than from

resisting it. Collective recognition not only avoids active resistance to the state cre-

ated by the other strategies, but creates incentives for elites to encourage engagement

with the state and participation in the formal political system.

6The idea that citizens would turn against the state even when the state proves material assis-tance may be counter-intuitive. However, development is not always considered an unalloyed good,particularly for communities that define themselves in opposition to “mainstream” society associatedwith the central state. This is apparent in indigenous rights advocacy discourse around “developmentaggression” (Declaration on the Rights of Asian Indigenous Peoples 2000). Misinformation aboutstate assistance, for example the idea that vaccines are harmful, is another example.

50

2.2.3 Communities

Finally, I consider the incentives of community members separately from those of

elites. I refer specifically here to individuals who see themselves as part of an existing

collective associated with non-state authority structures. It is clear why traditional

elites prefer collective recognition to other possible modes of state-society interaction.

But if recognition has the effect of strengthening traditional elites at the expense of

other members of the community, it may compromise elites’ legitimacy and reduce

their ability to influence the community’s behavior, including vis-à-vis the state. Tra-

ditional elites’ leverage in interactions with the state stems from their ability to mo-

bilize resistance. In the absence of this ability, the state would not face prohibitive

costs in pursuing its preferred strategy. Non-elite community incentives under differ-

ent modes of state-society interaction are therefore also important to consider.

From the community’s perspective, assimilationist policies pose a threat to exist-

ing identities and ways of life. The establishment of an exclusive national identity

that outlaws and denigrates and eliminates these likely requires violence on the part

of the state and is likely to meet with community resistance, above and beyond re-

sistance directly mobilized by elites. This violence, at the very least, creates fertile

ground for resistance mobilized by elites whose privileged positions are threatened

by the same strategy. Material inducements and the promise of participation in the

modern economy may overcome resistance for some; however, in communities defined

by their historical marginalization from the state, it may not be sufficient to overcome

baseline mistrust.

Egalitarian governance does not require individuals to part from their identities

and may initially appear less threatening. For historically marginalized groups (in-

cluding, for example, indigenous peoples and formerly enslaved people), the promise

of equal citizenship and individual rights may even prove attractive. However, the

ability to realize these rights and make successful claims on the state on an individual

51

basis is by no means guaranteed. For example, while the privatization and redistri-

bution of land rights may in theory reduce power differentials between traditional

elites and non-elite community members compared to customary land tenure systems

(Goldstein and Udry 2008), individual property rights are in many cases all but impos-

sible to formalize and protect (Boone 2014). Similarly, equal rights to vote in a liberal

democracy theoretically afford individuals equal voice. In practice, however, groups

that are socioeconomically disadvantaged have systematically less political influence

(Verba, Nie and Kim 1987). The ability to make successful claims on the basis of in-

dividual citizenship alone is difficult for members of historically marginalized groups,

particularly in contexts where membership in existing patronage networks is required

to access state resources and benefit from state legal protection.

Collective recognition, by conferring collective rights, allows communities to make

collective claims on the state. This in part symbolic: state recognition of a sub-

national identity group creates a basis for the formation of political identity among

communities otherwise defined by their exclusion from the formal state system (Jung

2008). But it also empowers communities to extract greater concessions, including

from political actors associated with the formal political system. As Baldwin (2015)

and others have recognized, traditional elites can mobilize collective action toward

the provision of local public goods, including goods that are co-produced with the

state. However, they can also use this mobilizing capacity to coordinate collective

bargaining. This can take place in the context of elections, where traditional elites

help to coordinate vote choice (Conroy-Krutz 2017). By strengthening traditional

institutions, collective recognition increases elites’ effectiveness in mobilizing collective

action for this purpose. This coordination can in turn increase the community’s

bargaining power vis-a-vis candidates and political parties (Smith and Mesquita 2012,

Gottlieb and Larreguy 2016). Collective bargaining can also take place in the context

of direct interactions with agencies of the state. When state interactions with the

52

community are mediated through traditional institutions, they provide opportunities

for the community to make collective demands in exchange for compliance with state

priorities.7

Both traditional elites and communities as a whole gain more from interactions

with the state under recognition than in its absence. This raises questions about

how these resources are distributed within the community. If all additional resources

gained through collective bargaining are captured by elites, some or most individuals

may be better off in a scenario without collective recognition. However, there are

reasons to believe this would not be the case. If elites have an interest in maintaining

legitimacy within their own communities, they are unlikely to take actions that would

compromise this. Particularly when these elites encounter potential competition from

the state for community members’ loyalty, they have an added incentive to maintain

their popularity internally. If they fail to do so, they risk losing their ability to

influence community decisions regarding interactions with the state, and ultimately

being eliminated or undermined.

Furthermore, as actors embedded in their communities, these elites are likely to

be subject to informal accountability mechanisms (Tsai 2007), as well as more for-

mal accountability mechanisms that exist within traditional governance structures.

Depictions of traditional authorities as unaccountable “decentralized despots” (e.g.

Mamdani (1996)) draw heavily on cases where these authorities were created by colo-

nial powers or assigned as agents of state control during the colonial period. In cases

where these institutions have persisted in the absence of formal state recognition,

they may be less likely to lack internal accountability mechanisms. Some inequality

may persist in how resources are distributed within the community; however, pure

rent-seeking behavior by elites is limited by their need to maintain legitimacy inter-

nally.

7For example, the ability to coordinate can make the expression of demands through constructivenon-compliance with the state more effective (Tsai 2015, Grossman, Phillips and Rosenzweig 2018)

53

States, particularly weak ones, rely on voluntary compliance from their popula-

tions. Assimilation, if successfully implemented, would lead to an ideal outcome from

the state’s perspective: establishing exclusive loyalty to the state and easing compli-

ance in the future. But the community’s ability to resist and evade the state —the

existence of an “exit option” —raises the costs of this strategy beyond what most

post-colonial states have the capacity or political will to incur. Collective recognition

shapes the incentives of societal actors to make engagement and compliance with the

state more attractive than it would be under other potential modes of state-society

interaction these states may pursue. By encouraging these communities to channel

participation through formal political processes rather than through extra-systemic

means, collective recognition extends the state’s authority (Huntington 1968). In

contrast to the dominant perspective predicting negative effects of collective recog-

nition on state capacity, I argue that recognition is likely to strengthen the state by

increasing its legitimacy.

In the next section, I briefly extend this discussion to consider the effects of recog-

nition on democratic accountability and representation before moving on to present

predictions of my theory compared to those derived from the dominant state-building

perspective.

2.3 Implications for representation

I have argued that collective recognition increases voluntary compliance and engage-

ment with the state in part by increasing the ability of historically marginalized com-

munities to make collective claims on the state. Collective claim-making, to the extent

that it increases perceived efficacy within the formal political system, can contribute

to systemic legitimacy by encouraging the channeling of demands through formal

channels rather than through resistance to the system. This idea stands in contrast

54

to critics of multiculturalism, who argue that state policies emphasizing (or failing to

de-emphasize) group differences are likely to be de-legitimizing. It also has important

implications for political accountability and democratic representation that further

challenge the conventional wisdom regarding the effects of collection recognition.

Traditional elites in democratic systems are often considered antithetical to demo-

cratic representation and accountability (Mamdani 1996, Ntsebeza 2005, Koter 2013,

Acemoglu, Reed and Robinson 2014). According to this argument, elites serve as

vote-brokers who extract rents in exchange for the votes of their communities. Elites

can play the role of organizing and enforcing community-level block voting, and that

collective recognition that devolves control over local governance to them may enhance

their ability to do so. However, I argue that this can improve, rather than undermine,

the representation of these communities, compared to an egalitarian counterfactual

in which individuals from these communities participate in democratic politics as in-

dividuals. Where state resources are already distributed within patronage networks,

individuals from historically marginalized groups are at a comparative disadvantage.

On the other hand, if communities can credibly commit to delivering a block

vote to officials who implement their preferred policies (and to deny the same to

officials who do not), they may be more likely to see these policies implemented.

Traditional elites can play an information-gathering role and negotiate directly with

candidates on the community’s behalf.8 They can then encourage block voting for

their preferred candidates. In doing so, they can also limit the influence of individual-

level vote-buying on voting behavior and empower the community to enforce bargains

struck for prospective public goods. Collective recognition enhances this coordination

capacity by increasing traditional elites’ control over local governance, particularly

access to land and local justice systems.

This raises concerns about “elite capture,” specifically that elites may take advan-

8As I demonstrate in the next chapter, this type of negotiation is ubiquitous in the Philippines’context and often involves formal written agreements between communities and candidates.

55

tage of these enhanced powers to extract rents. However, the opportunity to extract

rents can increase incentives for leaders to coordinate and increase overall resources

going to the community (Mansuri and Rao 2004). More importantly, as discussed

above, rent extraction is likely to be limited by traditional elites’ need to maintain

legitimacy within their own communities. Failure to do so may result in the popu-

lation shifting its loyalties toward the state or at least toward politicians associated

with the mainstream political system. In other words, collusion between traditional

elites and politicians may quickly spill over into competition, a competition that state

actors are likely to win given their comparative resource advantage. Elites therefore

have clear incentives to back candidates that deliver concrete benefits in the com-

munity. Indeed, the electoral arena provides an opportunity for traditional elites to

demonstrate their continued value in light of competition from the state.

By enabling greater electoral coordination, collective recognition can therefore

improve the representation of historically marginalized groups in democratic politics,

in addition to enhancing state legitimacy. In a sense, the two are mutually reinforcing.

Collective recognition boosts state legitimacy by increasing perceived efficacy within

the formal political system. These gains for the state are enhanced when communities

can successfully make these claims and, in the long run, may be contingent upon their

ability to do so.

2.4 Predictions

Finally, I summarize predictions about the effects of collective recognition, contrasting

predictions rooted in the dominant state-building paradigm with predictions from

the theory of recognition as integration. Table 2.1 summarizes these predictions for

two groups of outcomes that are the focus on this project. The first group relates

to nation-building and state-consolidation and the second to voting behavior and

56

democratic representation. Note that these predictions concern the effects of collective

recognition relative to the egalitarian governance strategy I describe above. Dealing

with a country’s citizens in an undifferentiated manner is considered to be a default

against which collective recognition is measured. Given that this theory explicitly

concerns weak states, I assume that the state does not have the ability to successfully

implement an assimilation policy and, for now, do not consider assimilation directly.

Table 2.1: Predictions of Effects of Collective Recognition

Outcome State-buildingParadigm

Recognition as Inte-gration

State-buildingSub-national ID ↑ ↑National ID ↓ ↑State compliance ↓ ↑

Democracy

Block voting ↑ ↑Leader influence unconditional conditionalElectoral competition ↓ ←→State public goods ↓ ↑

Beginning with state-consolidation outcomes, the state-building paradigm pre-

dicts that collective recognition will increase attachments to sub-national identity

groupings (specifically the sub-national identities the state collectively recognizes)

and simultaneously decrease identification with the nation as a whole. Following

from this, it predicts that collective recognition will result in measurable reductions

in compliance or engagement with the state. The theory presented here also predicts

an increase in identification with the relevant sub-national groupings, given that these

groupings under collective recognition form the basis of interaction between citizens

and the state. In contrast, however, it predicts that collective recognition will increase

both identification with the nation and compliance with the state. This is because

collective recognition increases expected returns to engagement with the state and

the formal political system.

In terms of democratic representation and voting behavior, the dominant paradigm

predicts that collective recognition will increase community-level block voting, by em-

57

powering traditional elites as vote brokers. By legitimating their authority and af-

fording them control over key domains of local governance, collective recognition em-

powers these leaders to deliver a community block vote to their preferred candidates.

Another key observable implication of this is that (summarized in the fifth row of

Table 2.1) is that collective recognition will increase the influence of traditional elites’

endorsements on individual community members’ vote choice. More specifically, this

increased influence will be unconditional, in the sense that the leader’s endorsement

will override other considerations that would otherwise affect individuals’ assessments

of candidates.

Furthermore, because this paradigm views traditional elites as purely rent-seeking,

it also predicts that empowering them through collective recognition will decrease

electoral competition at the relevant level (e.g. at the level of the sub-national unit in

which recognized communities represent a relevant constituency), since they are likely

to sell their community’s votes to the most well-resourced party. In a dominant or one-

party system, this may equate to an increase in votes for the dominant party. In a less

cohesive party system or in a context where governance is relatively decentralized,

it may manifest in the form of increased vote share for incumbents, regardless of

party identification. Finally, because in this view traditional authorities trade their

community’s votes for personal gains rather than local public goods, politicians have

little incentive to provide the latter. As a result, it predicts that collective recognition

will lead to a decrease in state-provided local public goods.

The theory I present here also predicts that collective recognition will increase

community-level block voting, for similar reasons: it empowers traditional elites to

enforce adherence to a decision to deliver the community’s votes for a particular can-

didate. In contrast, however, it does not predict that the influence of traditional

elites on vote choice will be unconditional. If traditional elites are subject to intra-

community accountability mechanisms and seek to maintain legitimacy within their

58

communities, they are not able to force votes for candidates that individual commu-

nity members would otherwise strongly prefer not to vote for. In other words, the

ability to enforce adherence to block voting is conditional at least to some extent on

other candidate qualities.

In addition, the alternative theory does not necessarily predict a decrease in elec-

toral competition at levels above the relevant local community. It does not predict

that politicians will be able to buy the community’s votes by simply offering higher

up-front payment or benefit to leaders. Instead, it predicts that communities will

be able to bargain for concessions over repeated interactions, offering their votes to

the candidate that makes the most credible promises to deliver community-level club

goods. This could be the incumbent or a challenger; in fact, challengers may be

able to gain a critical vote block by credibly promising to deliver state public goods

once in office, even if they lack the ability to make large up-front payments. Finally,

because collective recognition increases communities’ abilities to bargain collectively

and because communities will bargain for public goods, I expect collective recognition

to increase state-provided public goods.

In chapters 4 and 5, I evaluate predictions of the theory in the Philippines using

analysis of observational and experimental data. These chapters focus, respectively,

on state-building and voting behavior outcomes. In the next chapter, I describe

the Philippines context in greater detail, beginning with a brief discussion of the

history of collective recognition policies for indigenous communities using evidence

from in-depth interviews with tribal elites and policymakers to illustrate the proposed

theoretical mechanisms.

59

60

Chapter 3

Indigenous recognition in the

Philippines

The Philippines is an ideal setting in which to examine the effects of collective recog-

nition of indigenous communities for at least two reasons. First, it is a qualitatively

important case, given its role as a regional leader in the recognition of indigenous

rights (Inguanzo 2014). The fact that the country’s policy is considered relatively

robust makes it a “fair” test of recognition, as compared to a case where collective

recognition exists in name only. Second, while the IPRA law and related policies apply

to indigenous communities throughout the country, there is variation in the extent of

their implementation at the community level, allowing for comparisons between units

that are arguably more similar than national states.

In this chapter, I situate my inquiry into the effects of collective recognition in

the Philippines context. In the first part of this chapter, I provide a brief history of

indigenous recognition in the Philippines, tracing the shift from assimilationist and

egalitarian policies pursued by colonial and independent governments to the current

policy of collective recognition. I also describe the relevant community-level variation

in collective land titling, the primary means through which indigenous self-governance

61

rights are recognized. I use historical accounts, primary legal documents, and inter-

views with policymakers. In the second part, I use data from in-depth qualitative

interviews with tribal elites to illustrate the mechanisms proposed in my theory of

recognition as integration in this context. Importantly, these interviews are not in-

tended to empirically test of a theory developed ex ante. Instead, they contributed to

both theory development and refinement for this project. In the context of this chap-

ter, they serve to elucidate, concretize, and contextualize the proposed theoretical

mechanisms.

3.1 From assimilation to recognition

The Philippines’ constitution defines indigenous peoples (IPs) or indigenous cultural

communities (ICCs) as “homogenous societies identified by self-ascription and ascrip-

tion of others, who have continuously lived as a community on communally bounded

and defined territory, sharing common bonds of customs, traditions, and other cul-

tural traits, through resistance to political, social and cultural inroads to colonization,

non-indigenous religions and culture.” The dichotomy between minority IP groups

and the majority Christian Filipino population emerged during the Spanish colonial

period (Jefremovas and Perez 2011). Groups designated as IPs were distinguished by

their refusal to settle in pueblos, settlements established by the colonial state, adopt

Christianity, and submit to colonial rule (Bank 2002, Prill-Brett 2007). For many

communities, this resistance to colonial rule involved retreat from coastal areas into

the mountainous interiors of the archipelago’s islands (Scott 2010). As such, indige-

nous communities in some regions refer to themselves as “highlanders” and so-called

“mainstream” Filipinos as “lowlanders” (Finin 2005).

The Spanish and American colonial governments both adopted at least nomi-

nally assimilationist approaches toward these communities. Spanish colonial-era laws

62

and decrees described the “civilization” of so-called indios or non-Christian tribes as

an explicit goal, and the colonial government enacted policies to relocate them into

hacienda-like settlements similar to those in which Christian Filipinos had been or-

ganized.1 The United States, after taking control of the territory from Spain in 1898,

crafted policy toward these communities with explicit analogy to Native Americans.2

A Bureau of Non-Christian Tribes was established within the American colonial gov-

ernment to govern “backward” communities until they were sufficiently “advanced in

civilization and material prosperity” to live in regularly organized municipalities.3

While the tribes were governed separately, they, like the Native Americans, were con-

sidered to be under a “state of pupilage,” with the stated aim of integrating them into

the mainstream society, system of governance, and means of economic production.4

Elements of this policy continued following the Philippines’ independence from

the United States in 1946. The 1973 constitution gave consideration to the “customs,

traditions, beliefs, and interests of national cultural communities,”5 but subsequent

policies still invoked the integration of “national minorities” as an explicit goal.6 With

1An 1881 decree by the Governor General of the Philippines, for example stated that “...it is theduty to conscience and to humanity for all governments to civilize those backward races that mightexist in a nation” and declared that, as such, all “indios” were to be governed by common law andmoved into new settlements governed in the same manner as existing Christian settlements. Thisdecree was to be enforced by an armed contingent of “Native Christians” who were to “[punish] thetribes that, disregarding the peace, protection, and advantages offered them, [continued] in theirrebellious attitude.” (quoted in Rubi v. The Provincial Board of Mindoro, 1919)

2In his instructions to the Philippine Commission in 1907, President William McKinley wrotethat “in dealing with the uncivilized tribes of the islands, the Commission should adopt the samecourse followed by Congress in permitting the tribes of our North American Indians to maintaintheir tribal organization and government...such tribal governments should, however, be subjected towise and firm regulation; and...constant and active effort should be exercised to prevent barbarouspractices and introduce civilized customs.” (quoted in Rubi v. The Provincial Board of Mindoro,1919)

3Administrative Code of 1917, Act No. 2544However, some scholars have argued this policy of separate governance during the American

colonial period had the opposite effect of creating or reinforcing a unified “highland” consciousnessthat did not previously exist (Finin 2005, Scott 2010).

5Article XV, Section 116See, for example Presidential Decree 1414, Section 1: “It is hereby declared to be the policy of

the State to integrate into the mainstream of Philippine society certain ethnic groups who seek fullintegration into the larger community, and at the same time protect the rights of those who wish topreserve their original lifeways beside that larger community.”

63

respect to land rights specifically, the Spanish policy of the Regalian Doctrine, which

stated that all lands and natural resources in the public domain belonged to the

state, remained virtually unchanged from the period of Spanish colonial rule (Prill-

Brett 1994, Leonen 2004).

The 1987 constitution, which remains the law of the land today, represented an

important departure from this approach, asserting that the state “shall protect the

rights of indigenous cultural communities to their ancestral lands” and to “preserve

and develop their cultures, traditions, and institutions.”7 The constitution and the

subsequent Indigenous Peoples Rights Act (IPRA) of 1997 reflected changing inter-

national norms with respect to indigenous peoples, both drawing heavily in their

language from international conventions such as the Draft UN Declaration on the

Rights of Indigenous People (UNDRIP) and the Indigenous and Tribal Peoples Con-

vention of 1989 (ILO Convention 169) (Jefremovas and Perez 2011). In contrast

to other contexts where similar policies have been implemented (see, for example

Yashar (2005) on several Latin American cases), this policy shift was not precipi-

tated by geographically broad-based domestic mobilization in support of indigenous

rights. Indigenous communities in the Cordillera region of Northern Luzon did or-

ganize for self-determination rights beginning in the 1980’s (Prill-Brett 1994, Finin

2005), but the goals of this movement were limited primarily to the establishment

of a single autonomous region, and did not extend to the recognition of indigenous

rights around the country.8

7Article II, Section 22; Article XIV, Section 178The 1987 constitution was adopted at a time of significant political change, immediately follow-

ing the People Power Revolution that overthrew the regime of Ferdinand Marcos, and the indigenousissue was relatively marginal in the public discourse at the time, with the exception of the estab-lishment of the Cordillera Administrative Region. Elites familiar with the process suggested ininterviews that the IPRA was passed through largely top-down efforts from individuals in variousgovernment agencies, namely the Department of the Environment and Natural Resources, with sup-port from international donors. Some grass-roots mobilization occurred among tribes in NorthernLuzon following a controversial dam project in the 1970s and attempted sale of ancestral lands inBaguio City in the mid-1980s, but significant mobilization did not occur in other parts of the country.One individual familiar with the process told me that the IPRA was pushed through Congress wheninfluential individuals opposed to its passage were out of the country. Author interviews: Quezon

64

The IPRA established unprecedented rights for indigenous communities, particu-

larly regarding the control and ownership of ancestral lands.9 The law established a

novel tenurial instrument, known as a Certificate of Ancestral Domain Title (CADT),

as a primary means of realizing these rights. A CADT guarantees an indigenous com-

munity’s rights to land it has owned, occupied, or used since “time immemorial,” prior

to the Spanish conquest. These include rights to private but communal ownership of

the land, rights to regulate the entry of migrants into the territory, rights to develop

land and natural resources —including to negotiate the terms for natural resource

exploration within the domain and to share profits from natural resource extraction

—and rights to resolve land disputes and allocate land in accordance with custom-

ary law.10 Within ancestral domains, customary law is considered to have primacy

over the laws of the state. Furthermore, all lands certified as ancestral domains

are exempted from real property taxes, with the exception of large scale commercial

projects.11

CADTs are granted by the National Commission of Indigenous Peoples (NCIP), a

government agency established under the IPRA law. A community seeking a CADT

must provide evidence of its claims to have occupied a particular area since time

immemorial. Such evidence may come in the form of sworn testimony of elders, writ-

ten accounts of indigenous customs and traditions, photographs showing long-term

occupation and land improvements, genealogical surveys, and anthropological data

(R.A. 8371, Section 52).12 The community must also document their customary laws

and rules for succession, and designate traditional leaders and Indigenous Peoples

Organizations (IPOs) who are authorized to enter into agreements on the commu-

City, January 20, 2017 and August 9, 2018.9While the 1987 constitution also reaffirmed the Regalian Doctrine, the IPRA law has narrowly

survived a number of legal challenges alleging that the ownership rights it grants to indigenouscommunities violate this provision of the constitution (Leonen 2004).

10R.A. 8371, Sections 7-811R.A. 8371, Section 6012In practice, according to some scholars, the documentation used often comes from colonial

anthropologists (Jefremovas and Perez 2011)

65

nity’s behalf. In many cases, the preparation of these documents marks the first time

that customary law and leadership structures have been documented in writing.13

Finally, the NCIP must conduct a census of the population within the domain and

a professional land survey, funded by the claimants or their representatives. As part

of the CADT process, the community prepares an Ancestral Domain Sustainable De-

velopment and Protection Plan (ADSDPP), which describes their aspirations for the

development of their land. However, there is no budgetary allocation for implemen-

tation of these plans, nor is there any legal obligation for local governments to fund

them.14

While all indigenous communities in the Philippines have the right to apply for a

CADT, the issuing of CADTs has been implemented unevenly among eligible com-

munities. As of March 2018, a total of 223 CADTs had been granted nationwide,

covering more than 5.3 million hectares of land, about 18% of the country’s total

land area (NCIP). In addition, approximately 260 communities had applications un-

der processing. Figure 3-1 shows the geographic distribution of approved CADTs and

additional areas designated as eligible as of October 2016, the last time the approved

CADT map was updated. I leverage this sub-national variation in titling to study the

effects of collective recognition in subsequent chapters, where I also discuss the selec-

tion process in greater detail and the assumptions needed to draw causal inferences

from these comparisons.

13Author interviews: Santa Cruz, Occidental Mindoro, July 5, 2017; Bulalacao, Oriental Mindoro,July 4, 2017

14The Implementing Rules and Regulations for the IPRA state only that “the ICCs/IPs shallsubmit to the municipal and provincial government unit having territorial jurisdiction over themtheir ADSDPP in order for the said LGU [Local Government Unit] to adopt and incorporate thesame...” (Part II, Section 2)

66

●

Approved CADTEligibleProvincial boundary

Figure 3-1: Certificates of Ancestral Domain Title (CADTs) and eligible areas as ofOctober 2016. Eligible areas are designated using the centroids of barangays listed aspart of an untitled ancestral domain. Actual boundaries of eligible areas, which mayinclude portions of individual barangays, are not published until the title is issued.Data source: National Commission on Indigenous Peoples (NCIP)

67

3.2 Recognition in practice

In this section, I use data from in-depth interviews with indigenous elites to demon-

strate how these elite (and, to some extent, indigenous communities broadly) view

their interactions with the state in the context of this history and the current policy

environment. I begin with a brief discussion of how sub-national comparisons in the

Philippines connect to the theory presented in the previous chapter. I then summarize

specific propositions I intend to illustrate using the qualitative interview data. Next,

I provide a more detailed description of the local context in which I conducted the

interviews, the selection of interview sites and subjects, and the interview methodol-

ogy. I then move on to illustrate each proposition using quotes and specific examples

from the interviews.

3.2.1 Connection to the theoretical framework

In the previous chapter, I described three potential governance strategies: assimila-

tion, egalitarian governance, and collective recognition. These describe modalities of

interaction between states and peripheral communities that are in practice governed

at the local level by non-state authority structures or traditional elites. The strategies

are distinguished by how the state relates to these non-state authority structures in its

interactions with the population; specifically whether the state attempts to eliminate

them, is indifferent to them, or collectively recognizes them, respectively.

I have argued that under both assimilation and egalitarian governance strategies,

interactions between the community and the state directly threaten the position of

traditional elites within their communities. As a result, these strategies create in-

centives for traditional elites to resist and discourage interaction with the state and

integration into the mainstream political system. In addition, I argue that traditional

elites, communities as a whole, and individuals within these communities face poor

68

prospects for making successful claims through the formal political system under both

of these strategies (assimilation and egalitarian governance), thus discouraging par-

ticipation in and claim-making through this system and undermining its legitimacy.

Collective recognition, by contrast, enhances systemic legitimacy in two ways: 1) by

creating conditions whereby engagement with the state boosts rather than undermines

the legitimacy of traditional elites internally, due to their recognition as intermedi-

aries, and 2) by enhancing the community’s collective claim-making capabilities and

increasing expected returns from participation in the formal political system.

In examining contemporary sub-national variation in the Philippines, I set aside

a direct examination of assimilation. While assimilation of indigenous communities

was a stated goal of colonial and independent state powers in the Philippines, it does

not represent current policy and cannot be compared contemporaneously with the

other strategies across communities. Instead, there is a stated policy of collective

recognition which applies broadly in theory but has been implemented unevenly.

In this context, I consider egalitarian governance to be the best approximation of

“status quo” interactions between indigenous communities and the state —that is, in-

teractions in the absence of collective land titles that put broadly applicable collective

rights into practice (and, more specifically, that require government agencies, local

government officials, and other actors, like private companies, to treat indigenous

communities as a collective). In the interviews used in this chapter and in subsequent

quantitative empirical analyses, this status quo is represented by the present in com-

munities without collective titles to their land and by the recent past in communities

with and without these titles.

This particular comparison deviates in at least two important respects from an

“ideal” test of my argument. First, it does not allow for a direct comparison of assim-

ilation with the other two strategies. Not only is the assimilation “condition” absent

in the contemporary community-level comparison, but the legacy of assimilationist

69

policies almost certainly shapes current perceptions of the state within these com-

munities. This makes the effects of assimilation difficult to disentagle from the main

(contemporary) comparison between collective recognition and the status quo. Sec-

ond, non-recognition at the community level in the context of a broad national policy

of recognition (the status quo) may differ in important ways from a true egalitarian

governance approach. In particular, communities that are promised recognition and

do not receive it may be disappointed or discouraged.

I argue however, that the comparison is still important and relevant in light of the

current and historical political context. While assimilation is important in motivat-

ing the theory, the comparison of collective recognition to egalitarian governance (or

an approximation thereof) is arguably the most important to illustrate. Egalitarian

governance most closely resembles ideal state-society relations in a liberal democratic

system. The “threat” this strategy poses to traditional elites and marginalized com-

munities, compared to coercive assimilation, is less obvious on its face. In addition,

previous work has examined the effects of assimilation policies on community orien-

tations toward the state (McGovern 2012, Bandyopadhyay and Green 2013).

Furthermore, while the legacy of assimilation and the existence of official collec-

tive recognition at the national level complicate the comparison between titled and

untitled communities as a test of the theory, these particular contextual features are

applicable far beyond the Philippines. The legacy of attempted assimilation charac-

terizes the experience of many indigenous communities around the world, given the

international consensus around the desirability of assimilation and nation-building

in the immediate post-colonial period. By the same token, the experience of non-

recognition in the presence of nominal collective recognition exists widely given the

current global consensus around collective recognition. Just as particular indigenous

communities in the Philippines who lack formal recognition may be affected by the

disconnect between their actual situation and the promise of recognition, indigenous

70

peoples in countries that do not recognize collective indigenous rights may see a dis-

connect between their countries’ policies and the recognition of collective indigenous

rights in international law. In other words, while the three governing strategies I

define are theoretically distinct, none of them exist in a vacuum in practice.

3.2.2 What I hope to illustrate

The interview data presented below is meant to illustrate how proposed theoretical

mechanisms presented in the previous chapter operate in this context. Specifically,

I provide evidence for the following five mechanisms linking collective recognition

(compared to a counterfactual of non-recognition) with increased engagement in the

formal political system and cooperation with the state.

1. In the absence of recognition, indigenous elites perceive interactions between

their communities and the “mainstream” society associated with the state as a

threat to their legitimacy within their own communities. This occurs even in

the absence of targeted assimilationist policies.

2. In the absence of recognition, indigenous communities (not only elites) are sys-

tematically disadvantaged in their interactions with the formal political system,

compared to non-indigenous populations. This results in a lack of external ef-

ficacy.

3. When communities are able to act collectively, they are better able to exert

influence through the formal political system.

4. Collective recognition in the form of land titling strengthens the position of in-

digenous elites and increases the collective claim-making capacity of indigenous

communities.

71

5. External efficacy within the formal political system and systemic legitimacy can

translate into cooperation with the state.

In addition, I provide evidence for an important scope condition of the theory in this

context: that indigenous elites are subject to internal accountability mechanisms that

limit rent-seeking behavior and incentivize efforts to obtain community-level public

goods from the state.

3.2.3 Local context and interview methodology

Most interview data used in this chapter come from a set of semi-structured interviews

conducted with indigenous elites in the provinces of Oriental Mindoro, Occidental

Mindoro, and Palawan between January and October 2017. Here I provide some

brief background information about the communities in this particular region and

describe the interview methodology.

The provinces of Oriental Mindoro and Occidental Mindoro divide the island of

Mindoro, located approximately 160 km south of the capital Manila. Palawan is

a separate island, located to the southwest of Mindoro, and is covered by a single

province. The three provinces —Oriental Mindoro, Occidental Mindoro, and Palawan

—all fall within the same administrative region, known as MIMAROPA or Region

IV-B.

Indigenous communities on Mindoro are collectively referred to as Mangyans and

are typically divided into eight main tribal groups.15 Interviewees in Mindoro use the

regionally-specific term Mangyan, “IP,” or katutubo, the Tagalog word for “indigenous,”

to refer to themselves. Palawan has three “mainland” tribes, the Pala’wan, Tagbanua,

and Batak, in addition to several other groups who are considered indigenous migrants

from other islands. In Palawan, the Cebuano term lumad is also used occasionally15According to the Mangyan Heritage Center, the eight main Mangyan groups are: the Iraya,

Alangan, Tadyawan, Tau-buhid, Buhid, Bangon, Hanunuo, and Ratagnon. Gubatnon and Batanganare also Mangyan ethnic groups recognized as indigenous by the NCIP.

72

to refer to indigenous groups. Interviewees in all three locations referred to non-

indigenous populations as Tagalogs, Christians, “lowlanders” or “the mainstream.”

Table 3.1 lists the locations of interviews in these three provinces, the names of

the tribes my interviewees represented, and the tribe’s titling status (defined based

on whether any communities within the tribe have a registered CADT; in some cases,

parts of the tribe have secured CADTs and others have not). Figure 3-2 shows the

interview locations on a map of the three provinces. In Mindoro, I interviewed repre-

sentatives of seven tribes.16 My ability to secure interviews was primarily contingent

upon existing relationships between the non-government organization with whom I

partnered, LENTE, and these communities.17 In this setting, given the logistics

involved, the difficulty of contacting potential interviewees, and the regulations sur-

rounding activities involving indigenous communities (including research), it was not

feasible to arrange or conduct interviews independently. In Palawan, I interviewed

representatives from the three mainland tribal groups.18

These interviews were conducted with a LENTE representative. The arrangement

of interviews in Palawan was facilitated by NATRIPAL, a Palawan-based indigenous

advocacy organizations with whom LENTE had partnered on a number of projects.19

The involvement of indigenous advocacy organizations (combined with the presence

of a foreigner) raises concerns about potential biases; for example, tribal leaders may

have believed that discussing the benefits of recognition would encourage greater out-

side support for their land titling claims. While this is unavoidable to some extent, I

16This includes HAGURA, which is a coalition of three tribes who applied for a single CADT.HAGURA includes the Hanunuo, Gutbanon, and Ratagnon. The HAGURA CADT covers parts ofOriental Mindoro and Occidental Mindoro.

17In this particular case, LENTE had worked with representatives of five tribes in some capacity,mostly related to an election monitoring project in partnership with the Philippines Commission onElections.

18I interviewed representatives from different groups within the Tagbanua tribe. One took placeon Palawan and was with representatives of the tribe on “mainland” Palawan. Another took placein Quezon City, Metro Manila, with representatives of the tribe on Coron Island, which is a smallerisland immediately to the north of the main island.

19NATRIPAL is an acronym for Nagkakaisang Tribu ng Palawan or “United Tribes of Palawan.”

73

Province Municipality Tribe Titling StatusOccidental Mindoro Santa Cruz Alangan TitledOccidental Mindoro Magsaysay HAGURA TitledOccidental Mindoro Mamburao Iraya UntitledOriental Mindoro Bulalacao Hanunuo UntitledOriental Mindoro Socorro Tadyawan Untitled

Palawan Aborlan Tagbanua (Mainland) TitledPalawan Roxas Batak UntitledPalawan Quezon Pala’wan Untitled

Metro Manila – Tagbanua (Coron)* Titled

Table 3.1: Interview locations and tribes by titling status. Titling status refers towhether any portion of the tribe’s territory is covered by a registered CADT. * Theinterview with representatives from the Tagbanua tribe on Coron Island, Palawan,took place in Quezon City, Metro Manila.

attempted to mitigate this concern by asking about land titling and other government

or NGO interventions toward the end of the interviews. The English version of the

interview guide is included in the Appendix.

In some cases, I traveled with LENTE and/or NATRIPAL representatives di-

rectly to the communities where these leaders were based. In others, leaders traveled

to meet with us in an area closer to the road, in some cases traveling for several

hours on foot. The latter modality provided an opportunity to interview leaders

from multiple communities (within the same tribe) simultaneously. In accordance

with Philippine government regulations (under the Indigenous People’s Rights Act),

I obtained permission from the National Commission on Indigenous Peoples before

conducting these interviews.

Some interviews were conducted one-on-one, while in others subjects were inter-

viewed as a group. I conducted these interviews with a research assistant (a LENTE

staff member), in English with simultaneous translation to and from Tagalog as

needed. Audio recordings of the interviews were transcribed directly and sections

in Tagalog were translated into English from the transcripts (in addition to the si-

multaneous translation during the interview).20 Direct quotations come from the20Translation was conducted by a Tagalog-speaking research assistant based in the United States,

74

OCCIDENTAL MINDORO

ORIENTAL MINDORO

PALAWAN

Magsaysay

Mamburao

Santa Cruz

Bulalacao

Roxas

Socorro

Aborlan

Quezon

Roxas

Figure 3-2: In-depth interview municipalities in Oriental Mindoro, Occidental Min-doro, and Palawan

75

translated transcripts, and are only included where leaders consented to the use of

direct quotations. Where the locations of interviews are identified, leaders consented

to being identified by name, position, and tribe.

As shown in Table 3.1, there is variation in the titling status of these tribes.21

However, all of these tribes had submitted complete CADT applications at the time

I conducted the interviews. In most cases, those that did not yet have their ti-

tles were waiting on administrative processes under the purview of the NCIP.22 All

tribes had also received external financial and technical support for the application

process. On Mindoro, all tribes (including those I was able to interview and those

I was not) received assistance for this process from a single Catholic church-based

non-governmental organization known as the Mangyan Mission. In Palawan, tribes

had received assistance for land claims from various NGOs, including AnthroWatch

and the Philippine Association for Intercultural Development, Inc. (PAFID), often

with funding from international donors. The involvement of international donors and

NGOs in supporting indigenous land claims is common throughout the country.23

These external organizations also played a role in influencing the nature of tribal

organizations. This was particularly apparent in Mindoro. Beginning in the late

1980s (following the adoption of the 1987 constitution and subsequent decentral-

who had no affiliation with LENTE.21Note that this refers to whether any subsection of the tribe has received a title. In some cases,

particular communities within the tribe have while others have not.22Another common situation was that the issuing of the title had been delayed due to protracted

disputes between the NCIP and other, better-resourced, government agencies charged with issuingland titles. Under an administrative order adopted in 2012, a CADT cannot be registered unless theDepartment of Agrarian Reform and the Department of the Environment and Natural Resourcesissue certificates of non-overlap, certifying that the area covered by the CADT does not cover anarea titled by those agencies.

23AnthroWatch, for example, lists recent projects in Occidental Mindoro and in four provincesin Mindanao (southern Philippines) on its website and lists the European Union as a funder:http://anthrowatch.blogspot.com/p/about-us.html. PAFID’s website lists projects in Aurora (Cen-tral Luzon) and Bukidnon and South Cotabato (Mindanao) and mentions funding from FUNDESO,a Spanish foundation: http://www.pafidph.org/. The International Fund for Agricultural Develop-ment (IFAD), a specialized UN agency, lists “include securing access to land in ancestral domainsfor indigenous peoples and documenting customary laws and traditional practices” as a key com-ponent of its Philippines country strategy and has supported projects throughout the country:https://www.ifad.org/en/web/operations/country/id/philippines.

76

ization reforms) the Mangyan Mission worked with the tribes to establish Peoples

Organizations (POs) with standardized leadership structures down to the level of

the sitio or neighborhood (a named jurisdiction below the barangay).24 These POs

appeared to represent the primary political structures within the tribes, supplanting

or taking precedence over previous political structures. However, community elders

and others described as “traditional leaders” continued to play an advisory role and

to perform specialized functions outside the purview of the PO. These functions in-

cluded adjudicating disputes between members of the community, performing rituals,

determining harvesting schedules, and patrolling the community’s land.25

In Palawan, tribes had also formed POs, but generally maintained clearer distinc-

tions between POs and the political structures that had predated their establishment.

Interviewees often made explicit reference to the difference between the PO and the

“tribal sector” or “indigenous political structure.”26 In some cases traditional leader-

ship were selected according to heredity criteria, while PO leadership positions were

elected. In interviews on both islands, the establishment of POs was discussed as a

response to incursions of the state and non-indigenous people into indigenous areas

and the need to be organized in these interactions with outsiders.27

In the sections below I provide evidence based on these interviews for the theoret-

ical mechanisms listed above. The fact that all interviews in this chapter are all with

24A Peoples Organization is a type of non-governmental organization with official legal status.25One interviewee described the distinction as follows: “Because in customary law today, besides

the elders and our guragnons, the sitio leader is separate because the guragnon is the one known forhandling problems. That’s their duty today. The sitio leader was created so that the guragnon canget an assistant.” The guragnon is a particular type of elder who is in charge of dispute resolution.The sitio leader is an elected position within the PO. In his description, both are considered part ofa “modern” version of customary law. Author interview: HAGURA.

26The term “indigenous political structure” (IPS) appears in the IPRA law and refers to “orga-nizational and cultural leadership systems, institutions, relationships, patterns, and processes fordecision-making, identified by ICCs/IPs such as, but not limited to, Council of Elders, Council ofTimuays, Bodong Holders, or any other tribal or body of similar nature.” (R.A. 8371, section 3.i).

27For example, a leader from the Batak tribe in Roxas, Palawan, explained: “I think, back then,we were abandoned by the government. When people have arrived, the people that helped here[referring to NGOs that had helped them form their PO], that was the time when we gatheredcourage to stand for what’s ours, we started approaching the government, we were paid attentionto.”

77

tribal elites, including members of POs, traditional leaders, and elders, has important

implications for how this evidence should be interpreted. Given my positionality as

an outsider, the need to respect local custom, language barriers, and the fact that

many communities were remote and difficult to access, it was not possible for me

to conduct in-depth interviews with non-elites.28 Elite incentives and perspectives

on interactions with the state are central to the theory and elite accounts are ideal

for illustrating parts of it. However, these interviews have important limitations as

a source of information about non-elite interests and the following should be read

with this in mind. I address this limitation indirectly in this chapter when describing

community accountability mechanisms that constrain elites and serve to align their

interests with those of their constituents to some extent. In subsequent chapters, I

use survey data from non-elite community members to further probe implications of

my argument and address this limitation more directly.

3.2.4 Unmediated interaction with the state threatens tradi-

tional elites

Interviewees in all tribes described an increase over time in interactions between their

communities and outsiders, including non-indigenous people, politicians, private com-

panies, and the state. Descriptions of these interactions highlighted how, in the ab-

sence of collective recognition, seemingly benign interactions between the community

and “mainstream society” were seen as threatening to the position of elites within their

communities. This theme was particularly evident in how leaders discussed exposure

of community youth to “mainstream society” and linked it to a perceived reduction

in their influence over time. “If we look at the youth now who are doing their studies

28LENTE, my NGO partners, are also considered outsiders in this context given that they arebased in Manila and are non-indigenous. Realistically, conducting interviews with non-elites wouldhave entailed hiring research assistants from these communities. This strategy was employed for thesurvey.

78

outside the community,” one leader recounted, “they get a different worldview...and

in those situations, they lose their identity as a Mangyan. They become ashamed of

speaking their own language, the Mangyan language.”29 This exposure, according to

leaders, was causing youth to lose respect for customary law and challenge or ignore

leaders’ decisions.

Leaders attributed this reported loss of influence to young peoples’ exposure to

mainstream education, temporary migration to areas outside the community for work,

the entry of the mainstream political system. “Now,” he went on to say, “[our influence]

is weak, especially when it comes to the youth. Because the youth know more than

the elders...because they went to school, they know about their surroundings. But

when it comes to issues surrounding the indigenous people, they don’t know anything.

Even the culture, it disappears. That’s why we really try to enforce, [but] the older

generation’s grip on the youth is weakening.”30 A leader in another tribe similarly

lamented: “Fifty years ago, if the chair [of the tribe] said something, people would do

it. Now, people just listen, then do what they want. Some members of the tribe are

already educated...Because of the political structure, people aren’t listening to tribal

law and don’t believe in their traditional religion.”31

The “political structure” or “politics” were frequently used as shorthand to discuss

interactions with the state and were cited as threats to traditional leaders’ influence.32

As one leader stated in response to a question regarding change in the influence of in-

digenous leadership over time: “it’s more difficult now than back then...because today,

politics is on top of everything.”33 In many cases, this language referred not to elec-

toral politics but to the implementation of government projects over which traditional

authorities had little control or influence. Leaders in one community, for example,

29Author interview: Tadyawan30Ibid.31Author interview: Hanunuo.32Author interviews: Tadyawan, Hanunuo, Iraya33Author interview: Tadyawan

79

described how the national government’s flagship conditional cash transfer (CCT)

program undermined their authority by failing to acknowledge leaders’ positions and

organizing community members outside of their purview.34 As the leader explained:

“even if you are a member [beneficiary of the CCT program], like me who is a leader,

I am not acknowledged as a member of the council. I am only acknowledged because

of me.”35 He went on to describe how when representatives of the Department of

Social Welfare and Development, the CCT implementing agency, called a meeting of

CCT beneficiaries, community members would go but when he tried to do so, they

would not.

Similar complaints arose regarding projects implemented by the local government.

The same leader described how the tribe had made a request to the municipality

for a housing project and funds granted for the project were directed through the

barangay government (one level below the municipality). While this process is in

accordance with the Philippines Local Government Code, he argued that the tribal

leadership should be the ones handling the funds. Barangay officials managing the

project was seen as problematic for two reasons: they were able to take credit for

the project (further undermining the authority of tribal leadership) and they were

seen as diverting resources to the Tagalog community and away from the indigenous

community who had initially requested the project.

The discussion of politics as a threat to leaders’ influence also referred to the entry

of electoral competition and attempts by rival political candidates (from outside the

community) to sow division by recruiting brokers from within the community. One

leader described how the establishment of paid elected positions at the barangay level

prompted candidates to begin recruiting supporters from within indigenous commu-

34This refers to the Pantawid Pamilyang Pilipino Program (4Ps). The 4Ps program has a versionspecific to indigenous communities in which tribal leadership is involved in beneficiary selection andprogram administration and certain program content is tailored to take into account different normsand customs in indigenous communities. In the Tadyawan tribe, some community members werepart of the “regular” program and others were part of the IP-specific program.

35Author interview: Tadyawan

80

nities, which ended up dividing the tribe: “In 1989, the tribe divided...That’s the

time they [candidates] saw the importance of [elected positions in] the barangay. I

need to have an [indigenous broker] in that barangay because the community of that

leader will be my supporters.36 Another described how the creation of the position

of Indigenous Peoples Mandatory Representative (IPMR), a paid barangay council

position, had created divisions in the tribe and further undermined the authority of

elders and other community authorities.37

In all of these domains, it was clear that indigenous elites considered exposure of

their communities to mainstream society and the state to threaten their authority and

influence. In some cases, this perceived threat manifested in actions to restrict this

exposure. Interviewees in two separate regions mentioned efforts by tribal leadership

to restrict the construction of cellphone towers in their areas to avoid youth being

influenced by contact outside of the tribe through mobile phone usage.38 Leaders in

one tribe described efforts to discourage women in the tribe from delivering babies

in government hospitals and how they had lobbied the local government to remove

penalties for not doing so.39

3.2.5 Egalitarian governance limits claim-making

It is perhaps not surprising that the incursion of the state and mainstream society dis-

advantages traditional elites. Individuals in these communities may benefit from de-

taching themselves from traditional authority structures, abandoning their indigenous

identity, and integrating —i.e. by seeking education and obtaining individual-level

government benefits and pay-outs from political candidates. Yet the interviews also

highlighted the ways in which indigenous communities and non-elite indigenous in-36Author interview: Iraya. He is likely referring to the adoption of the Local Government Code

in 1991. Similar points came up in interviews with the Hanunuo and the Batak.37Author interview: Hanunuo38Author interviews in Manila (with a representative of the Tagbanua tribe on Coron Island,

Palawan) and in Santa Fe, Nueva Vizcaya with a representative of the Ikalahan tribe.39Author interview: Hanunuo

81

dividuals were, in the absence of collective recognition, systematically disadvantaged

in their interactions with the formal political system and the mainstream economy.

The absence of formal property rights combined with comparatively lower levels

of wealth was also seen as a particular threat. Several interviewees mentioned how,

as companies and non-indigenous communities entered into indigenous lands, they

were able to convince or coerce individuals into selling plots of land for nominal

amounts of money. For example: “there are some lowlanders who live with us. And

those lowlanders also claim it is their land. [Interviewer: but did they pay money

for it?] Yes, but a little amount...they will just give like 1 ganta of rice, a bottle of

whiskey...The lowlander always told them that it [the land] is for everybody. You have

no title yet. You have no paper to present”40 This statement highlights how a property

rights regime that is theoretically universal (the land is “for everybody”) systematically

disadvantages those with pre-existing connections, higher levels of wealth, and the

ability to navigate the legal system.

Systematic disadvantages and discrimination were also mentioned in the context

of elections. The physical distance between many indigenous communities and polling

places and widespread illiteracy made these communities vulnerable to harassment

and coercion. Several discussed a practice known as hakot, where candidates round

up indigenous voters, drive them to polling stations, and refuse to drive them back

until votes are counted. Some described being pushed to the back of the line while

trying to vote. One interviewee described how “they don’t have time to vote because

the lowlanders or Tagalogs will go to the front. There are instances where they can’t

vote because of the line...the teachers [polling station workers] prioritize the Tagalogs

because they are well known in the community.”41 Others recounted how “assistors”

at polling stations would write-in the names of their preferred candidates, taking

advantage of the fact that many indigenous people are illiterate, or how election

40Author interview: Batak41Author interview: Pala’wan

82

officials would throw out indigenous votes on the basis that they were filled out

incorrectly. 42 One interviewee described the attitude of politicians and polling

workers toward indigenous communities as follows: “You don’t belong here. Them

first. Because you’re just an IP. Those sorts of things.”43 A leader in another tribe

recounted how the community was treated in the past, prior to changes in national

government policy toward indigenous communities: “When it comes to elections...they

[saw] us as not having rights. They see us as just pigs being fed. I experienced that

when I first voted in 1993....Back then, we walked like we were sheep of the Tagalog

leaders.”44

Leaders frequently expressed frustration at their perceived inability to make suc-

cessful claims through the elected local government, unfair distribution of government

resources to non-indigenous communities, and inability to prevent exploitation of

their land. Leaders in several different tribes described requesting projects on behalf

of their communities only to have these requests denied or told their requests might

be addressed if the local government unit had “extra funds.”45 As one interviewee

put it, “our [biggest] problem is our relationship with the government.” He went on

to describe how the tribal leadership “keep following up, but still nothing. Because

sometimes it means they were not able to do some budgeting....It’s that simple. The

governments do admit it, that there’s no clear funds for the indigenous people.”46

Leaders also described how indigenous communities were overlooked in the context

of supposedly universal government programs. One, for example, described how a na-

tional community-driven development (CDD) program had coordinated with elected

barangay officials to select members of the project oversight committee and com-

42“Sometimes things happen where they judge us indigenous people because we don’t know how towrite. Some assistors are called sometimes....when there’s an invalid vote they say that it’s from theindigenous people, that the indigenous people made those mistakes.” Author interview: Tagbanua

43Author interview: Batak44Author interview: Iraya45Author interviews: Hanunuo, Tadyawan, Tagbanua46Author interview: Tadyawan

83

plained that officials failed to include any members of the indigenous community.47

Leaders in another tribe recounted how companies seeking usage of their lands would

go directly to barangay officials for permission and ignore the indigenous communities

altogether, despite the fact that the company’s activities on the land directly affected

their livelihoods: “Back then...companies would arrive in our places. They would only

acknowledge the barangay captain to ask for permission to go to the mountains. But

us indigenous people, they would not recognize us. The reason being that we were

seen as having low status and did not have important responsibilities.”48

This lack of efficacy was particularly evident when it came to bargaining with

state agencies and other external parties for prospective community benefits, and

translated into pessimism about the utility of participation in the formal political

system. In the electoral realm, leaders in multiple communities were frustrated at

their inability to hold candidates accountable for agreements they made to help the

community. As one described, “our experience [is that] when candidates approach

us, it’s all promises...For example when they win they promise us electricity in the

barangay. but it doesn’t happen.” Another speaker added: “Every election, it’s the

same promises.” A third concluded “We don’t know how to vote...for the previous

candidate who gave false promises or for the new candidate who will also give false

promises.”49

In one tribe, leaders expressed similarly pessimistic sentiments regarding promises

by government agencies and private companies seeking usage of their land. They

described how they planned to reject a recently-proposed hydropower project because

companies implementing similar projects in the past had not honored their agreements

with the community. “There’s a lot of people who would not follow the rules....we have

evidence,” one leader explained. “Even if it’s a good agreement with the Christian

47Author interview: Tadyawan48Author interview: Tagbanua49Author interview: Tadyawan

84

organization [referring to the hydropower contractor]...it’s not being followed....We

can’t really do anything. But this taught us a lesson.”50 Past experiences in which the

community was unable to hold external actors accountable translated into a reluctance

to engage with similar actors in the future.

3.2.6 Collective claim-making increases efficacy

This lack of efficacy was not present everywhere, however. There was considerable

variation in the extent to which tribal leaders felt they could make successful claims

on the state on behalf of their communities. Across communities, however, there

was agreement that unity and the ability to act collectively in interactions with the

state increased the community’s bargaining leverage. As one interviewee explained:

“the barangay [government] becomes willful if people come together as one. Just like

in our community, doing some painting, if the community is not coming together

then it looks like there is no good will between people. The good thing is when a

barangay sees the indigenous people bonding.”51 Those who expressed pessimism in

their ability to influence the state attributed it in part to a lack of unity and those

that expressed optimism attributed it to their ability to unite the community around

a common purpose.

Again, this was particularly clear in the context of elections and interactions with

candidates for elected local government office. In almost all areas where I conducted

interviews, leaders described a procedure by which they considered candidates and

came to a consensus about which candidates to support.52 This process often in-

50Author interview: Hanunuo. Note that the hydropower company was private company with noreligious affiliation. Leaders in this community used the English word “Christian” to refer generallyto individuals and organizations associated with “mainstream” society.

51Author interview: Alangan. A leader of the Tagbanua tribe in Aborlan, Palawan expressed asimilar sentiment, arguing that unity was just as important as, if not more important than, thanrelations between the tribal leadership and barangay officials: “If he [a tribal leader], is not agreeingwith the barangay officials, the officials would not give support. If [the leader] knows how to getalong, everything is fine. But the key is being united.”

52Author interviews: Tadyawan, Hanunuo, Tagbanua, Batak

85

cluded direct meetings between candidates and tribal leadership or community-wide

meetings in which candidates discussed future plans to help the community and high-

lighted past accomplishments. In one tribe, leaders referenced written agreements

signed with candidates to deliver specific projects to the indigenous community.53

Delivering a block vote was seen across all communities as important to ensure that

these agreements about prospective community-level benefits were adhered to. One

interviewee explained the logic as follows: “if you [the candidate] don’t follow up, next

year your support disappears.”54

On the other hand, harassment at polling stations, vote-buying, the presence of

political brokers who divided the community, and the waning influence and credibil-

ity of tribal leadership were seen as posing a threat to successful execution of this

collective bargaining strategy. The same leaders who lamented their waning influence

within their communities expressed specific frustration about their ability to convince

the community to vote together. “The elders would say go ahead vote for that candi-

date,” one explained, “[but] voting for one person really can’t happen. The Mangyans

each have their own candidate. So they are cutting the voting population of that

community.”55 Leaders in another tribe complained that “people in the tribe don’t

vote together. [They] don’t listen to leaders because of the political structure.”56 One

interviewee even cited recent shifts in norms to focus on individual choice and human

rights as a barrier to indigenous communities influencing elected officials.57 “Actually

for me, it [the focus on individual rights] is good,” she explained. “But on the other

53Author interview: Alangan54Author interview: Batak55Author interview: Tadyawan56Author interview: Hanunuo57“Actually in the 90s it changed. Individual rights has [sic] become stronger. The right for

everyone to decide who is your choice, even if your father says ”no, no.” You have rights...humanrights have been stronger...[Interviewer: Is that something that was taught in school?] Yes, fromthe school. [Interviewer: And you think most people have that idea now, even among the IPs?] Yes,each individual has been given a right to select on their own....[Interviewer: So do you think that’spositive? From the perspective of the IP community?]” Author interview: Batak. This particularindividual was a long-time advocate working on indigenous issues in Palawan, but did not identifyas indigenous herself.

86

hand it is not good. Because the integrity of your tribe disappears. The cause of

the IPs is divided. Before...[tribal leaders] would say ‘okay, [the whole] village we are

supporting this one candidate.’ That’s what happened really. Before, we were not

using the money to bid [for votes]. What was important back then were the words,

[saying] we will help your community.”58

Vote-buying was seen as a particular threat to community-level electoral cohe-

sion, and tribal leaders pursued a number of strategies to limit its influence within

their communities. Leaders in one tribe described how members of the community

found to be working as brokers were sentenced to have their feet tied to a wooden

board for up to one month.59 In another tribe, leaders described mixing together

contributions given by multiple candidates so individuals in the community would

not feel an obligation to one candidate over another: “We’re all given food together

[before the election]....We collect the food altogether even when it comes from two

candidates. We get a cook from each sitio who comes.”60 A pilot project by the

Philippines Election Commission (COMELEC) to establish separate polling stations

for indigenous communities in 2016 was also seen as helpful in promoting electoral

cohesion, particularly by limiting coercion and harassment.61

Similar themes arose in discussions surrounding land use and resource exploita-

tion. Per the Indigenous Peoples Rights Act, companies and government agencies are

required to undergo a process known as Free, Prior, and Informed Consent (FPIC)

before beginning any operations on indigenous lands. Before a project moves forward,

it must receive formal approval from representatives of the affected communities.62

58Ibid.59One leader from this tribe explained “for [the candidates], vote-buying cannot be avoided. They

really find a way. But for the indigenous people, it’s not the basis of why they follow. The importantthing is we come together as one on the day of the election.” Author interview: Alangan

60Author interview: Hagura61This project was implemented in partnership with LENTE, the non-governmental organization

with whom I collaborated on this research.62The FPIC requirement legally applies regardless of whether a particular community holds an

ancestral domain title (CADT). However, as discussed below, the formal title has important impli-cations for communities’ bargaining capabilities and for the enforcement of agreements in practice.

87

Interviewees described how this process provided an opportunity for the community

to bargain for material benefits and environmental mitigation measures. These ben-

efits may include, for example, a portion of funds from mining royalties to be used

for community projects, employment guarantees, or discrete projects such as housing

construction.63

The ability of the community to act as a unit and withhold consent was seen as

crucial in this context as well. One leader described the process as follows: “The

community tries to see which [projects] are the most profitable. That’s the main

priority. Because it’s the whole community who makes a move and not just one per-

son...There’s a meeting to decide if that project is wanted at all. Then the community

usually asks what is that? Would we get anything out of it? How long will we get

something out of it? Those are the kinds of questions they ask.”64 A leader in another

tribe, discussing a mining project the community was currently considering, described

how he performs a particular ritual to decide: “The first [time] here in the community,

the ritual showed that it cannot be. There were obstacles. When they [the mining

company] were still not content, they came back last year. We still held the ritual

in those places. [Interviewer: it still isn’t approved?] Still not...[Interviewer: How is

that ritual? Can you share that?] We we do is, we kill a pig. [Interviewer: And how

do you know from that?] We look for something on the inside of the [pig’s] body. We

show it and then we announce. [Interviewer: What is shown?] It’s like a map. It is

a sign that you can approve the project.”65 It was clear that the pig’s entrails were

not actually determining the tribe’s decision; instead the ritual was used as a way to

communicate this decision while reinforcing the leadership’s authority to decide on

63Leaders in one community with a title described the breakdown of projects funded by royaltiesthey get from mining operations in their area. The funds were allocated to youth scholarships,livelihood projects, and environmental preservation. Author interview: Tagbanua. In another tribe,the leader described how the community had bargained for lower prices for water supply. Authorinterview: Iraya

64Author interview: Hagura65Author interview: Iraya

88

the tribe’s behalf.

Leaders in other communities, by contrast, expressed frustration about attempts

by companies to divide the community. This was described in a similar manner to the

behavior of candidates: companies and their political allies identified and even hired

individuals within the tribe to sow division or, in some cases, sign legal documents on

the tribe’s behalf.66 Leaders of the tribe considering the aforementioned hydropower

project, for example, were not confident in their ability to prevent it from going

forward, because companies were “[going] direct to the community.” This practice was

seen as directly undermining the tribe’s ability to bargain and shape the conditions

of the project.67

In the context of elections and negotiations over projects, it was clear that leaders

saw the ability to act collectively as a critical determinant of their ability to influence

the state (and other external actors associated with the state). When leaders were

able to resist harassment, brokerage, and other sources of division in the tribe to

deliver a block vote, they saw engagement in electoral politics as a viable channel for

making claims on the state. Absent the leverage this afforded, electoral participation

was seen as either largely futile or as a site of exploitation. Similarly, when leaders

were able to withhold consent for proposed projects on their land —something that

required uniting the tribe and preventing defection —they felt empowered to use

this process to bargain for community benefits. Leaders who were unable to do this

expressed futility and considered these projects to be threatening.

66For example, one leader in Santa Fe, Nueva Vizcaya, described how the mayor, an ally of amining company, had paid individuals in the tribe 500 pesos (approximately 10 USD) to vote toapprove a mining project. He recounted how the mayor and his allies had identified “people whocan easily be bribed” and used a “checklist” of names. A leader of the Hanunuo tribe in Bulalacao,Oriental Mindoro described how companies target “those people who are popular” and who haveinfluence in the community. These descriptions are similar to how vote-buying is described in thecontext of elections, in the Philippines and elsewhere (Finan and Schechter 2012, Cruz 2018).

67Author interview: Hanunuo

89

3.2.7 Collective recognition increases collective claim-making

capacity and creates incentives to engage with the state

The ability to encourage collective action within tribes increased leaders’ their efficacy

in interactions with the state. This, in turn, I argue, increases systemic legitimacy

by encouraging engagement with the formal political system. Here I illustrate how

recognition by the state in the form of collective land titling can boost systemic

legitimacy by enhancing a community’s collective claim-making capabilities. Vari-

ation in the ability to act collectively can undoubtedly be attributed to a number

of factors, including the pre-existing strength of traditional institutions, which may

also affect communities’ ability to obtain recognition. I do not argue that observed

community-level variation in cohesion, claim-making capability, and systemic legiti-

macy is explained entirely or primarily by variation in collective recognition; instead,

I argue that collective recognition has a positive marginal effect on these outcomes.

In subsequent chapters I use research designs that attempt to isolate the effect of

recognition while accounting for potential confounding. Here, I focus on illustrating

the causal mechanisms through which this may occur.

I highlight two mechanisms in particular. First, collective recognition strengthens

indigenous authorities and improves their ability to mobilize collective action in in-

teractions between the community and the state (and other external actors such as

private companies). This occurs due to the devolution of powers to these authorities

(especially control over land) and the elevation of their status as legitimate interme-

diaries between the community and the state. Second, collective land titling improves

the legal standing of communities when it comes to the enforcement of agreements

over project within their teritory. The ability to enforce these agreements further

strengthens the credibility of traditional elites, creating a kind of positive feedback

loop.

As mentioned above, some tribal leaders reported that their influence within their

90

communities, particularly among the youth, had decreased over time. They blamed

this for their inability to unite and organize the community and, by extension, for their

inability to influence the state. Anecdotally, however, leaders of titled communities

appeared less likely to describe this as a problem and more likely to state that their

influence had increased or remained the same, rather than decreased, in recent years.68

One leader in a titled community described “the trust they [the community] are giving

in their leaders” as the community’s greatest strength.69 More importantly, these

leaders directly linked their influence to government recognition.“In my view, there’s

more influence [of tribal leaders] today,” one explained, “because...our rules today,

which are the rules of the indigenous people, rules that came from our ancestors,

[are] being supported by our government. Our positions have become stronger today

than in the past...the leadership of the tribe is more influential today because the

leadership is supported by government.”70

In particular, communal control over land provided material incentives for indi-

viduals to maintain ties to the tribe. In doing so, it strengthened leaders’ ability to

enforce adherence to customary law. In many communities, leaders’ power stems in

part from their ability to allocate land to community members. Losing control of

the land —due, for example, to land-grabbing by individuals from outside the com-

munity —undermines this material basis of their authority. One community leader

articulated this explicitly, suggesting that when they obtain their CADT “the land

can be distributed again to the youth. If I’m a youth and if I have ancestral land,

if I have farmland there, it’s mandatory to follow the culture and the customs of

the tribe.”71 The availability of ancestral land gives individuals who might otherwise

distance themselves from the tribe (due, for example, to a history of discrimination) a

material incentive to retain these ties. One interviewee, comparing two communities

68Author interviews: Alangan, Tagbanua, Pala’wan69Author interview: HAGURA70Author interview: Tagbanua71Author interview: Tadyawan

91

within the same tribe —one with a title and one without —described how many in

the latter community no longer spoke the tribal language, due to “fear of discrimina-

tion.” By contrast, in the titled community “their identities got enhanced. Now they

have a strong affinity and reason to identify themselves as Tagbanuas. Because of the

CADT.”72

The state’s recognition of traditional elites as legitimate intermediaries between

the community and the state also enhanced these elites’ standing within their com-

munities. In contrast to leaders —primarily from untitled communities —who com-

plained about local government officials going around them to implement projects

within their territory, leaders in titled communities described how officials would ap-

proach them first, allowing them to determine who participates in the project and

influence its implementation. One leader described the process as follows: “if the

barangay has a project, the barangay officials call first for the leaders of the indige-

nous people to inform them there is a project at the barangay. For example what

needs to be implemented in the barangay and whoever wants to work, the leader of

the indigenous people is the one who will call for members who will work.”73

The CADT, by formalizing the tribe’s jurisdiction over territory, requires the local

government units to engage with community leadership in this way. Several intervie-

wees made this point explicitly. As one argued: “when they got the title, the rules of

engagement changed. Now, whether they like it or not, the LGU [local government

unit] has to engage with them [the tribal leadership].”74 Another interviewee spoke

to how these mediated interactions with the state changed the community’s views of

tribal leadership: “when one has a CADT, it’s different from not having a CADT.

72Author interview: Manila (interview with representative of the Tagbanua tribe on Coron Island,Palawan)

73Author interview: Tagbanua. A similar point came up in the context of elections, where candi-dates had to first talk to leaders rather than doing “directly to the community.” One leader describedhow the candidate and their representatives “first talk to us [leaders] before they go to the landsthat we rule. Because we are...the recognized leaders of the indigenous community.”

74Author interview: Manila. This quote comes from an NGO worker who had worked with thetribe.

92

They [community members] don’t look up to you as a leader, they don’t have much

respect...Now that there’s a CADT...there are talks about issues that can be included

in programs. Projects are discussed. Back when there was no CADT yet, the people

weren’t attending the meetings to talk about the projects.”75

Beyond enhancing the influence of leaders in the community and their ability

to coordinate collective action, the external recognition that resulted from having a

CADT was seen as strengthening the community’s hand in bargaining with private

companies and other external actors. Certain collective rights, such as the right to

Free, Prior, and Informed Consent (FPIC), apply to indigenous communities through-

out the Philippines, regardless of titling status. The implementation of this process,

however, varies considerably between communities (Co 2008). Some interviewees re-

ported that the process did not occur at all prior to the issuing of their CADT, despite

it being required by law. A leader in a titled community contrasted current negotia-

tions over a road project to similar interactions in the past as follows: “Just now what

happened for example is that [the government proposed] to clear up a road that goes

from Sablayan to Buritan. This is because there’s a CADT, we are consulted...but

when we didn’t have the CADT yet, whatever the government wants, that’s what

[happens]...There was no CADT yet then so there was still no FPIC happening.”76

Even where a formal FPIC process did occur, leaders’ bargaining capabilities

were limited in the absence of formal recognition. As mentioned above, interviewees

in untitled communities described how their ability to negotiate with and impose

conditions on companies was undermined by the latter’s practice of approaching (and

often bribing) individuals in the tribe to approve projects. Leaders of the tribe may

try to continue bargaining, but the company moves forward based on individual

approvals. Companies can then use the ambiguity surrounding who legally “speaks

for” the tribe to go against the will of the tribal leadership or to avoid punishment for

75Author interview: Alangan76Author interview: Alangan

93

failing to uphold agreements made with them. The CADT reduces this ambiguity,

giving the leadership a stronger footing to bargain on the community’s behalf and

to contest violations of the process. Leaders explicitly articulated the idea that the

CADT would improve their ability to enforce agreements with private actors. For

example, a leader in an untitled community complained, recalling a recent interaction

with a hydropower company, that “they [the operator] don’t totally listen because we

don’t have the certification yet...[since] we have no CADT, they don’t totally believe

us. They do not totally respect us, because we have no certification.”77 Recognition

of collective rights by the state not only facilitates collective claim-making but also

makes it more likely that the outcomes of collective bargaining processes will be

respected.

Finally, collective recognition may have a direct effect on state legitimacy outside

of its effects on collective claim-making capacity. Leaders I interviewed in many com-

munities directly associated the government’s recognition of indigenous rights with

dignity and inclusion in the broader national community. The promise of collective

rights signaled a departure from a history of marginalization, attempted assimilation,

and targeted discrimination that had contributed to mistrust of the state. Several

leaders invoked animal imagery to convey how the government had viewed their com-

munities in the past. As mentioned above, one leader described how the government

and politicians saw them as “pigs being fed” and as “sheep.”78 A leader in another

tribe described how in the past (during the Marcos government, prior to the pas-

sage of the IPRA law), the government did not recognize their tribe as part of the

Philippines and referred to indigenous people as “monkeys.” 79 These leaders saw the

passage of the IPRA law and the granting of collective rights in general as a credible

signal that these attitudes were changing.

77Author interview: Hanunuo78Author interview: Iraya79Specifically, he said that the g-strings the community traditional wear were described as “actual

tails.” Author interview: Hanunuo

94

Critically, however, these leaders did not equate inclusion with assimilation into

mainstream society; the collective nature of their recognition by the state was paramount.

Leaders in one tribe expressed how recognition, and specifically the CADT, allowed

them to be simultaneously separate from and included within the broader national

community. In recounting the tribe’s motivation to apply for a CADT, one explained

how “as long as we didn’t have land with boundaries we would still be in the gov-

ernment’s premises. That’s why we created a plan to have a boundary...When the

CADT comes, then we become separated from the government. We receive a right

to tell the government that these lands are ours.”80 Within the same conversation,

another leader from the tribe described how the process of documenting the tribe’s

unique culture and history (part of the CADT application process) had contributed

to a sense of inclusion and prioritization as a collective: “[Before], the culture [was]

unknown by the government and the people such as you. It [wasn’t] fully out there.

It’s like there’s a fence in between but now it’s all out there... to let them know we

are part of society. We shouldn’t be the ones who are least prioritized.”81 Collective

recognition allows the possibility of integrating into society and being prioritized by

the state while maintaining a distinct collective identity. It may therefore increase

external efficacy both through a technology mechanism (by enhancing collective claim-

making capabilities and increasing bargaining leverage) and by signaling the inclusion

of historically marginalized groups in the state’s sphere of concern.

3.2.8 Systemic legitimacy can affect cooperation with the state

I have argued that collective recognition enhances state legitimacy by increasing ex-

pected returns to engagement in the formal political system for marginalized com-

munities and traditional elites within these communities. By increasing collective

80Author interview: Alangan81In the context of the conversation, the personal pronoun “you” referred to my Tagalog-speaking

research assistant. Ibid.

95

claim-making capacity and efficacy, it can encourage cooperation with certain state

priorities and reduce the likelihood of resistance to the state (and specifically attempts

by the state to extend its authority). In the context of a weak or incompletely consol-

idated state, the ability of peripheral communities to withhold cooperation with state

priorities constrains the state and affords these communities some degree of leverage.

The examples presented so far may seem to suggest that indigenous communities in

this region of the Philippines are limited in their ability to mobilize credible resistance

to the state. Many leaders have been unable to effectively prevent harassment by

political candidates, the influence of “mainstream” culture, or the entry of private

companies and non-indigenous individuals into their territory. I have argued that the

resulting lack of external efficacy within the formal political system translates into

de-legitimating attitudes towards the state. But by the same token, the state may

not actually constrained at all by a lack of legitimacy within these communities.

The interviews also revealed, however, specific ways in which these communities

can choose to resist or cooperate with the state. The interview questions did not

directly address resistance to the state or anti-state activities (in part because such

questions were, at least initially, considered sensitive). Still, interviewees indepen-

dently mentioned a number of behaviors that could be reasonably classified as resis-

tance or non-cooperation. These included, for example, opposition to the extension

of infrastructure and communication technology into their communities (not with the

goal of securing a better bargain but because these projects might facilitate the entry

or influence of outsiders),82 active opposition to women giving birth in government

hospitals using government-sanctioned methods,83 resistance to registering with the

government (e.g. birth registration, marriage certificates),84 and simply “fleeing” or

running away when government employees, the military, or other outsiders attempted

82Author interviews: Manila (with leader from Coron, Palawan), Santa Fe, Nueva Vizcaya83Author interview: Hanunuo84Author interview: Batak

96

to enter their communities.85

Another set of behaviors, especially salient from the central state’s perspective,

related to interactions between indigenous communities and the New People’s Army

(NPA), a communist insurgency active since the late 1960s. The NPA operates

throughout the country, primarily in rural and mountainous areas, and often at-

tempts to recruit from indigenous communities who live in these same areas. While

no leaders I interviewed on Mindoro or Palawan reported active support for the NPA,

several alluded (unprompted) to NPA recruitment of community members or men-

tioned interactions with local NPA leadership (i.e. in negotiations about land use

or conducting land surveys for their CADT application). Recruitment of indigenous

individuals by the NPA and cooperation with the military against the NPA on the

part of indigenous communities has been a particularly salient issue in Mindanao in

the southern Philippines.86

The government has also sought cooperation from some indigenous communities

related to President Rodrigo Duterte’s “war on drugs.” At the time the interviews

took place, the Department of the Interior and Local Government (DILG) had di-

rected barangay governments to participate in a community-based anti-drug cam-

paign known as MASA MASID (“the masses are watching”).87 While leaders from

some communities suggested that there were no drugs in their areas and they did not

need to participate, others mentioned active cooperation with this campaign.

Anecdotally, leaders who reported greater efficacy in dealings with the state also

tended to exhibit more legitimating attitudes and report more cooperative behaviors.

85Ibid. This particular form of resistance is discussed at length in Scott (2010).86During my visit to Davao City in Mindanao, hundreds of indigenous people were occupying the

compound of the United Church of Christ in the middle of the city to protest “militarization” ontheir lands by the government. This demonstration was understood by many I spoke with as asign of support for the NPA and its political allies. “Voluntary evacuation” or bakwit, where indige-nous communities temporarily leave their communities and occupy urban areas has been repeatedlydeployed as a strategy of resistance against the government (Canuday 2006).

87MASA MASID is also an acronym for the Tagalog Mamamayang Ayaw sa Anomaliya, Mama-mayang Ayaw sa Iligal na Droga or “citizens against anomalies, citizens against illegal drugs.” TheDILG formally suspended the program in December 2017 after allegations of abuse.

97

The tribes where leaders spontaneously mentioned interactions with the NPA were

both untitled communities that had expressed particular frustration in their inability

to influence government.Leaders from one (untitled) community that had expressed

frustration with their inability to influence government also expressed particular hos-

tility toward “mainstream” education, mentioned that they had attempted to reject

nearly all infrastructure projects proposed in their area and that tribal leadership

was actively campaigning for an exemption to efforts to require women to give birth

in healthcare facilities.88 By contrast, the community that mentioned active partic-

ipation in the anti-drug campaign was a titled community where leaders had been

successful in lobbying the local government for funds and asserted that, from the

government’s perspective, “the indigenous people [are] really the priority.”89

This association alone does not provide strong direct evidence in support of the

idea that collective recognition increases compliance with the state. However, to

the extent that collective recognition increases external efficacy within the formal

political system, these examples suggest that increases in state legitimacy have the

potential to affect community behavior in domains where the state seeks compliance

and cooperation.

3.2.9 Community accountability mechanisms limit elite rent-

seeking

Finally, I address the issue of elite accountability and elite capture. As noted earlier,

the fact that interviews used in this chapter are all with elites has important implica-

tions for the interpretation of their accounts. Characterizations about what is good

or bad for the community (e.g. acting as a unit in elections or negotiations with the

state, maintaining customary law), may more accurately be described as what serves

88“We have our own way of delivering children at home, so we want the government to get rid ofthe penalty for not delivering at a hospital.” Author interview: Hanunuo

89Aborlan

98

elite interests. If elites can coerce individuals into complying with a particular course

of action and trade the unified support of the tribe for personal gain, individuals in

the tribe may prefer a scenario where they are free to act individually and sell their

vote or their land. Here, I provide evidence for the idea that elites are accountable

to members of their community and that, as such, they are constrained at least to

some extent in their ability to engage in rent-seeking behavior or act against the

community’s interest (as defined by the community).

First, it was clear from the interviews that even leaders who expressed confidence

in their ability to mobilize collective claim-making saw their own influence as limited.

In the context of elections, leaders who reported being able to deliver a block vote

made a point (typically unprompted) to stress that individuals in the tribe were free to

vote for whomever they chose.90 One interviewee claimed that tribal leaders were able

to deliver “solid command votes.”91 Most others, however, were more measured. For

example: “sometimes there are a few who will still vote for who they think deserves it

[other than the agreed-upon candidate]. But the majority, they will follow what was

agreed upon. Because that has been our tradition ever since.”92 When asked whether

he told his community to support his preferred candidate, a leader in another tribe

responded “Of course. What my role is I show the community what this person has

achieved. But I let them choose.”93

Second, interviewees did not equate the influence of tribal leadership structures

with the influence of individual leaders. Individual leaders could be replaced and often

were, even in the absence of formal elections. As one leader explained: “if you have

90In the aforementioned example about punishing community members with the wooden board,for example, it was clear that the punishment was for members of the tribe working as brokers fora non-preferred candidate, not for individuals who voted for that candidate.

91This referred to the Tagbanua community in Coron, Palawan and comes from an interviewconducted with a Tagbanua leader in Manila.

92Author interview: Tagbanua. Similarly, leaders of the Batak in Roxas, Palawan estimated that75% would vote with the leadership and leaders from the Iraya tribe in Mamburao, OccidentalMindoro estimated 80%.

93Author interview: Iraya

99

done good things, you will have followers. You will have the position for a long time. If

you are not doing any good, you will be replaced. It’s that simple.”94 The disbursed

nature of leadership within many of the tribes —and particularly the co-existence

of the “traditional” structure and the Peoples Organizations —prevented any one

individual from accumulating too much power and acted as a checking mechanism.

One leader described how elders could intervene to replace a poorly performing sitio

leader.95 A leader in another tribe described how elders who were not seen as having

integrity would simply not be called to mediate disputes or perform other duties

assigned to them.96 A leader in a third tribe recounted how another leader had

been replaced as the head of the tribe’s foundation for misusing funds. The same

person later ran for municipal councillor (a relatively rare event in a context where

the vast majority of elected politicians are non-indigenous) and failed to gain the

tribe’s votes.97

Furthermore, the evaluation of tribal leadership (particularly those associated

with POs) was often tied to their performance in attracting public goods to the

community and solving problems by liaising with politicians or government officials.98

One interviewee, a sitio leader, explained his duties as follows: “The first duty of the

leader is to take care of their sitio. Second is communicating to the government...I am

the one who fixes the problems here in our sitio. For example when a leader doesn’t

know anything about government matters, it would be hard to attract the politicians.

Now if you can’t get any politicians’ attention, what that means is that your political

94Author interview: Alangan95“Sometimes the sitio leaders cannot fulfill their duties, [they] abandon their responsibilities. Like

there’s a lot of problems today. If he doesn’t know what to do, those guragnons [elders] say theyneed to be replaced.” Author interview: HAGURA

96Author interview: Sante Fe, Nueva Vizcaya97In the same interview, the leader explained how the tribal leadership had to be transparent

about the tribe’s budget, because “otherwise, you know, there will be infighting.” This particulartribe had a CADT in part of their territory and had regular income from tourist activities that wasmanaged by a tribal foundation. Author interview: Tagbanua (Coron)

98This is similar to how Baldwin and Mvukiyehe (2015) characterizes the incentives of chiefs inthe Zambian context

100

leadership is weak.”99 Leaders who exhibited a lack of external efficacy described how

their failures to influence government undermined their positions within the tribe.

One explained, for example, how he was blamed when a project associated with the

CDD program failed to include members of his community.100

Collective recognition (and specifically collective land titling) in this context in-

volves the recognition of customary law within particular territory. However, this “top

down” empowerment by the state does not necessarily invalidate “bottom-up” forms

of accountability that exist within the community. If anything, it may strengthen

intra-community accountability by improving the ability of tribal leadership to fulfill

the duties expected of them by members of the community (for example, by improv-

ing enforcement of agreements signed with companies and other external actors). The

existence of bottom-up accountability weighs against the idea that individuals from

marginalized communities are strictly better off in a system where they are treated

as individuals rather than as members of a collective.

3.3 Conclusion

In this chapter, I aimed to make the idea of collective recognition concrete by ground-

ing it in the Philippines context. I described the sub-national variation in recognition

that provides the empirical leverage for the study of the effects of titling. I then

used evidence from in-depth interviews with indigenous elites in three provinces to

illustrate how collective recognition can increase state legitimacy within historically

marginalized communities by increasing their ability to make claims through the for-

mal political system. In the absence of recognition, I illustrate how contact between

the state and these communities leads to sub-optimal outcomes from community-

state interactions, for both elites and communities and makes cooperation with the

99Author interview: Iraya100Author interview: Tadyawan

101

state less attractive. In the next two chapters, I test observable implications of this

argument quantitatively using a combination of administrative and survey data.

102

Chapter 4

Recognition and integration

In this chapter, I use administrative and survey experimental data from the Philip-

pines to evaluate predictions of my theory related to the effects of collective recogni-

tion on state-consolidation outcomes, specifically sub-national identity, national iden-

tity, and compliance with the state. The dominant state-building paradigm predicts

that recognition of sub-national collectives will increase attachments to the relevant

sub-national identity groups (Chandra 2006, Lieberman and Singh 2017). In doing

so, it will both reduce attachments to a broader national identity and undermine the

state’s legitimacy (Fukuyama 2018). This, in turn, will manifest in reduced engage-

ment or voluntary compliance with the state. The recognition as integration theory

also predicts an increase in sub-national identification. In contrast, however, it pre-

dicts that collective recognition, by increasing expected returns to participation in the

formal political system, will increase identification with the nation and compliance

with the state among the relevant communities.

I evaluate these competing predictions in two ways. First, I leverage spatial and

temporal variation in the granting of communal land titles to indigenous communi-

ties throughout the Philippines to investigate the effects of granting collective self-

governance rights on communities’ identities and compliance with the state. More

103

specifically, I use difference-in-differences and panel designs to estimate the effects

of receiving a Certification of Ancestral Domain Title (CADT) on indigenous self-

identification and two measures of compliance: birth registration and legibility, a

measure of census data quality. Second, to examine the effects of collective recogni-

tion on sub-national vs. national identity at the individual level, I use evidence from

a priming experiment embedded in an original survey of indigenous communities in

three provinces: Oriental Mindoro, Occidental Mindoro, and Palawan.

By focusing on variation within a single country, this design holds constant state-

level characteristics that may affect both the decision to adopt a strategy of recogni-

tion and state-building outcomes. Combining nation-wide administrative data with

individual-level experimental data allows for examination of “real world” outcomes

alongside more causally well-identified individual level mechanisms.

Consistent with the recognition as integration theory and contrary to predictions

derived from the dominant state-building paradigm, I find in the administrative data

analysis that titling increases both indigenous self-identification and compliance with

the state. In the survey experiment, I find that priming respondents with informa-

tion about the government’s policy of recognition. Results from an original survey

experiment suggest that granting collective self-governance rights increases, rather

than decreases, identification with the nation.

In this chapter, I first present the observational data analysis, describing the data

and empirical strategy, presenting the results and robustness checks. Next, I present

the experimental component, describing the survey, experimental design, and results.

I conclude the chapter by considering alternative interpretations of my findings.

104

4.1 Observational Data Analysis

4.1.1 Research Design

As discussed in the previous chapter, the Indigenous Peoples Rights Act of 1997

(IPRA) established unprecedented rights for the Philippines’ indigenous communities,

particularly regarding the control and ownership of ancestral lands. In particular,

the law established a novel tenurial instrument, known as a Certificate of Ancestral

Domain Title (CADT). A CADT provides private but communal ownership of the

land, rights to regulate the entry of migrants into the territory, rights to develop

land and natural resources —including to negotiate the terms of natural resource

exploration within the domain and to share profits from natural resource extraction

—and rights to resolve land disputes and allocate land in accordance with customary

law. While the rights detailed in the IPRA apply to all indigenous communities in

the Philippines, CADTs have been issued unevenly among eligible communities and

over time. I leverage this spatial and temporal variation in titling to study the effects

of collective indigenous recognition at the community-level.

CADTs are granted by the National Commission of Indigenous Peoples (NCIP), a

government agency established under the IPRA law. A community seeking a CADT

must provide evidence of its claims to have occupied a particular area since time im-

memorial. Such evidence may come in the form of sworn testimony of elders, written

accounts of indigenous customs and traditions, photographs showing long-term occu-

pation and land improvements, genealogical surveys, and anthropological data. The

community must also document their customary laws and rules for succession, and

designate traditional leaders and Indigenous Peoples Organizations (IPOs) who are

authorized to enter into agreements on the community’s behalf.1 Finally, the NCIP

1In many cases, the preparation of these documents marks the first time that customary law andleadership structures have been documented in writing. Author interviews: Santa Cruz, OccidentalMindoro, July 5, 2017; Bulalacao, Oriental Mindoro, July 4, 2017

105

must conduct a professional land survey, funded by the claimants or their represen-

tatives.

As of March 2018, a total of 223 CADTs had been granted nationwide, covering

more than 5.3 million hectares of land, about 18% of the country’s total land area. In

addition, approximately 260 communities had applications under processing.2 Figure

4-1 shows the cumulative number of CADTs approved per year between 2002 and

2018. Figure 3-1 shows the geographic distribution of approved CADTs and additional

communities designated as eligible but not yet titled.3

The boundaries of CADTs and CADT applications do not necessarily correspond

to pre-existing political or administrative boundaries. In some cases, one CADT

covers portions of multiple barangays, the smallest administrative units in the Philip-

pines. In others, only a portion of a single barangay is covered. The 223 approved

CADT applications cover 1,882 unique barangays across 271 municipalities. Areas

included in submitted or on-process CADT applications span an additional 4,026

barangays. In addition to these areas, communities in 2,871 barangays where no ap-

plication has been submitted have been identified as part of eligible ancestral domains

by the NCIP.4 This makes a total of 8,779 barangays that may be considered part of

a “universe” of potentially title-able areas, approximately 21% of all barangays in the

Philippines.5

2This number is approximate, because the administrative data on which these numbers are baseddoes not consistently indicate which areas are covered under a single application.

3The map reflects titled and eligible communities as of October 2016, the last time the approvedCADT map had been updated by the NCIP as of this writing.

4The size of an ancestral domain is not determined until the community applies and a land surveyis performed; however, eligibility is typically designated at the level of the barangay.

5This list should not necessarily be considered exhaustive. The list of “identified” areas with noapproved CADT or on-process application was compiled manually based on submissions from eachNCIP regional office. Submissions from some offices were missing or out-of-date. To date, therehas been no comprehensive nation-wide assessment of the locations of eligible ancestral domainsthroughout the country; in other words, there may be areas that upon further investigation wouldbe considered eligible but have not come to the attention of NCIP regional offices. Indeed, identifi-cation as an eligible area by the NCIP may even constitute or reflect recognition in some form. Toaddress this, I also estimate all specifications within matched subsets, predicting titling based onpre-treatment covariates as an alternative to using the NCIP’s designation.

106

2002 2004 2006 2008 2010 2013 2015 2017

Total Approved CADTs by Year

Year

Tota

l app

rove

d C

AD

Ts

050

100

150

200

Figure 4-1: Cumulative number of Certificates of Ancestral Domain Title (CADTs)approved by year, as of March 2018. Data source: National Commission on IndigenousPeoples (NCIP)

107

4.1.2 Empirical strategy

The existence of sub-national variation in recognition through land titling provides

an opportunity to examine the effects of collective recognition of indigenous commu-

nities, holding state-level factors constant. Yet land titles are not randomly assigned,

creating challenges in inferring a causal relationship from naive cross-sectional com-

parisons between titled and untitled communities. Communities that already have

an affinity with the state may be more likely to apply for land titles and, conditional

on applying, indigenous leaders who are more effective or politically connected may

be more effective at obtaining them. As shown in Table B.1 in the Appendix, titled

and untitled communities within the eligible universe differed in important ways at

baseline. In 2000, prior to the issuing of the first title, barangays covered by titles

had, on average, a higher proportion of the total population identifying as indige-

nous and a lower proportion identifying as Catholic, compared to currently untitled

communities. By some measures, titled barangays were more integrated at baseline:

they had less rugged terrain, and were more likely to have an elementary school and

a health center. By others, they appeared less so: titled communities were on average

farther from the coast and the road network and had lower baseline rates of birth

registration.

I address this inferential challenge using a generalized differences-in-differences

approach, comparing differential change over time in titled and untitled communities.

This strategy accounts for fixed community-level characteristics that may affect both

titling status and outcomes, such as the pre-existing strength of indigenous political

institutions, historical political connections and level of integration into the state,

natural resource wealth, and level of development. It relies on the assumption there

is no time-varying confounding; in other words, that titled and untitled communities

would have followed parallel trends in the absence of the titling “treatment.” This

assumption would be violated if, for example, indigenous communities integrating at

108

faster rates were able to obtain titles earlier. While I cannot rule this out completely

(and lack data that would allow me to substantiate it using pre-treatment outcome

data), I use a number of strategies to relax or increase the plausibility of the parallel

trends assumption.

First, I implement a pre-processing step that matches communities that received

titles prior to the 2010 census to those that did not, on pre-treatment outcomes and

a number of pre-treatment predictors of titling. This method requires a less stringent

conditional parallel trends assumption to estimate the Average Treatment Effect on

the Treated (ATT) (Abadie 2005). Second, I implement specifications restricting the

sample to (a) communities designated by the NCIP as part of the eligible universe

of title-able areas and (b) communities that were eventually titled as of 2018, lever-

aging only variation in whether they received the title prior to enumeration of the

2010 census. Within these subsets, I assume that the vast majority of “untreated”

communities had applied for a title prior to 2010.6 Anecdotal evidence suggests that

the timing of when titles are issued is determined in part by factors unrelated to

community characteristics. In author interviews with relevant stakeholders, delays in

the issuing of titles were frequently attributed to a backlog of applications and lack

of capacity on the part of NCIP staff. Comparisons of communities titled before and

after 2010 provide support for this idea: as shown in Table B.2 in the Appendix,

areas titled prior to 2010 were, on average, closer to the regional offices of the NCIP

and had more difficult geography (steeper, higher elevation, and more rugged). In

addition, as described below, I use indirect tests to evaluate the plausibility of the

parallel trends assumption and conduct a number of placebo tests and robustness

checks.

6Unfortunately, systematic data on when applications were submitted is not available.

109

4.1.3 Outcomes

I estimate the effects of collective recognition through titling on two types of out-

comes theoretically linked to state consolidation: sub-national identification (in this

case indigenous self-identification) and compliance with the state. As discussed above,

strong attachments to sub-national identity groupings are associated in the literature

with state weakness and conflict. If, as suggested by the state-building literature,

state recognition of sub-national collectives strengthens identification with those col-

lectives, we might expect it to exacerbate state-building challenges attributed to eth-

nic division. I measure indigenous self-identification using data from the 2000 and

2010 waves of the Philippines Census of Population and Housing and capture the pro-

portion of individuals in a barangay who self-identify with one of the ethnic groups

recognized as indigenous by the NCIP.

If it is the case that collective recognition weakens the state —by increasing sub-

national identification and decreasing identification with the national community,

legitimizing non-state authority at the expense of state authority, or through other

mechanisms implied by the literature —we should expect to find evidence that the

state’s ability to project its authority and achieve its goals with respect to affected

populations is compromised as a result of this recognition. In contrast, the recognition

as integration theory predicts that communities that are recognized are more likely

to “opt-in” to the state system and engage with the state. I operationalize this

concept using two indicators of compliance with state priorities: birth registration

and legibility, which measure census data quality using the extent of age heaping

(Lee and Zhang 2016).

Birth registration is an important tool for states to track their populations (Hunter

and Sugiyama 2018). The Philippines, like many developing countries, has made

explicit efforts to increase rates of birth registration among its population (Abouzahr

110

2014).7 Yet birth registration rates have historically been lower among indigenous

communities compared to the rest of the population (Authority 2017). This disparity

is attributed to resistance among these communities to giving birth in government-run

health facilities, as well as difficulty in accessing these facilities. In the Philippines

and elsewhere, birth registration is often necessary to establish citizenship and the

legal identity that allows citizens to obtain all other documents required to avail of

government programs and citizenship rights (Scott 1998; Hunter and Sugiyama 2018).

This measure therefore captures not only compliance with an explicit state priority,

but also perceived benefits among the population of engagement with the state more

broadly. I define birth registration as the proportion of respondents whose births are

reported to be registered with the Civil Registration Office. This variable is measured

for three census waves: 2000, 2007, and 2010.8

Legibility is a measure of census data quality originally proposed by Lee and Zhang

(2016). This measure builds explicitly on the insight of Scott (1998), that information

gathering and the standardization of information about the population represent core

activities of states. Specifically, it captures the accuracy of age reporting data and

the extent of “heaping” around ages ending in 0 and 5. Since the actual distribution of

ages in the population is unlikely to spike at these points, deviations from a smooth

distribution can be interpreted as errors in data collection. This inaccuracy could

occur for two reasons: 1) a lack of awareness among the population about their exact

ages and 2) the inability of census enumerators to reach or gather data from the

population. Both scenarios represent a lack of interaction between the state and

society, and potentially a willingness to be measured and provide information to the

state.9 I operationalize legibility at the barangay level using a Whipple Index, which

7Efforts to increase birth registration have included registration drives targeting adults, in addi-tion to parents of newborns (Abouzahr 2014).

8In both 2000 and 2010, the Philippines government conducted a Census of Population andHousing, while in 2007 only a Census of Population was conducted.

9Lee and Zhang (2016) demonstrate that this measure of accurately captures a state’s “presenceon the ground” and is associated with other measures of state capabilities, such as tax contributions

111

involves calculating the percentage of the population with recorded ages ending in

5 or 0 and determining how much this deviates from the expected 20%. The index

ranges from 0, which represents no heaping, to 500, which means that all reported

ages end in 5 or 0. I calculate this using all three census waves: 2000, 2007, and 2010.

Table B.3 in the Appendix shows summary statistics for all barangays included in

the main analysis.

4.1.4 Estimation

The primary unit of analysis in this study is the barangay, the lowest level at which

outcomes can be measured. All administrative data analyses estimate the relationship

between a barangay’s titling status in a given year t and the value of the dependent

variable in that year. I use the following specification:

𝑦𝑖𝑡 = 𝛽𝑋𝑖𝑡 + 𝑍 ′𝑖𝑡𝛾 + 𝜆𝑡 + 𝑐𝑖 + 𝜖𝑖𝑡

where 𝑋𝑖𝑡 represents the titling status of barangay 𝑖 in year 𝑡, 𝑍𝑖𝑡 represents a vector

of time-varying covariates, 𝜆𝑡 is a year fixed effect, and 𝑐𝑖 is a barangay-level fixed

effect.

Titling status (𝑋𝑖𝑡) is operationalized in the main analysis by calculating the

proportion of a barangay’s land area that fell within a CADT in a given year, using

the map published by the NCIP in 2016. As a robustness check, I replicate the

analysis using a binary measure based on a separate list of barangays obtained from

NCIP in 2018.10 As discussed below in further detail, I also use the binary measure

to create a matched reference group using pre-treatment covariates. In the main

and public goods provision.10There are some discrepancies between the two lists as to the barangays covered and, anecdotally,

concerns about the map’s accuracy. Of the 1,793 barangays coded as having a CADT in 2016 usingthe binary measure, only 1,472 are coded as having greater than zero overlap using the continuousmeasure.

112

analysis, standard errors are clustered at the level of the barangay ; however, I also

employ alternative specifications that account for potential spatial autocorrelation. In

addition, I estimate all models excluding time-varying covariates 𝑍𝑖𝑡, due to potential

post-treatment bias.

4.1.5 Results

To test of the effects of collective recognition on the strength of indigenous self-

identification, I compare change over time in the proportion of barangay residents

self-identifying as a member of an indigenous ethnic group in barangays that received

a land title between the 2000 and 2010 census rounds and in barangays that did not.

Results from this differences-in-differences analysis appear in Table 4.1. Column 1

shows the estimate using the full sample of rural barangays. The coefficient is statisti-

cally significant and substantively large: full coverage of a barangay’s land area with

a CADT between 2000 and 2010 is associated with a more than 10 percentage-point

increase in the proportion of individuals identifying as a member of an indigenous

ethnic group.

Columns 3, 5, and 7 show estimates from the same analysis using various subsets

of the data for which the parallel trends assumption is more plausible. In column 3, I

estimate the effect of land titling within a matched sample. Using a genetic matching

algorithm (Diamond and Sekhon 2006), I match all barangays that were “treated” as

of 2010 (using the binary land titling indicator) to barangays untreated during that

period on a number of pre-treatment covariates, including indigenous population in

2000, geographic location (latitude and longitude), land area, terrain ruggedness,

distance to the 1980 road network, distance to the coast, and distance to the regional

office of the NCIP, the government agency responsible for issuing titles. Column 5

presents estimates within the set of barangays in the eligible “universe” designated by

the NCIP. This includes barangays that received a land title at any point between

113

Table 4.1: Land Titling (Continuous) and Indigenous Identification (Differences-in-Differences)

Dependent variable:

Indigenous Prop.All Rural All Rural Matched Matched Eligible Eligible Titled Titled

(1) (2) (3) (4) (5) (6) (7) (8)

Titled Prop. 0.104*** 0.103*** 0.048*** 0.050*** 0.053*** 0.052*** 0.027* 0.028**(0.009) (0.009) (0.012) (0.012) (0.010) (0.010) (0.014) (0.014)

Same Municipality −0.003 −0.058 −0.048 −0.030(0.013) (0.077) (0.050) (0.075)

Log. Population 0.012*** 0.017 0.007 0.014(0.004) (0.018) (0.011) (0.028)

Mean Age 0.003*** −0.001 0.003* −0.011**(0.0005) (0.004) (0.002) (0.005)

Mean HH Size −0.001* −0.015 0.007 −0.011(0.001) (0.010) (0.005) (0.013)

Observations 33530 33530 1674 1674 7023 7023 1430 1430𝑅2 0.01 0.011 0.011 0.013 0.004 0.005 0.003 0.008

CRSE at bgy level *p<0.1; **p<0.05; ***p<0.01

114

2000 and 2018, barangays for which CADT applications have been submitted, and

barangays not covered by a title or an application that have been identified as eligible

by an NCIP regional office. Column 7 restricts this sample even further to barangays

that, as of March 2018, were covered by a land title, leveraging variation in when

titles were received relative to the enumeration of the 2010 census. The coefficient

decreases in magnitude within these subsets, but remains positive and statistically

significant, as well as substantively significant; coefficients across the four subsets are

equivalent to between 0.20 and 0.73 standard deviations in the change in indigenous

self-identification between 2000 and 2010.

One concern with the interpretation of these results is that the estimates may

reflect changes in barangay-level population composition, as opposed to changes in

the propensity of individuals to self-identify as indigenous. It may be the case that

once an area is covered by a CADT, indigenous people decide to move to that area,

that indigenous people out-migrate at lesser rates, or that non-indigenous people out-

migrate at greater rates. However, as shown in columns 2, 4, 6, and 8 of Table 4.1, the

results remain virtually unchanged when controlling for change in total population

and the proportion of individuals in the barangay who have lived in the same munici-

pality for the past five years, both rough proxies for differential in- and out-migration.

These models also include an indicator for average age, to address the possibilities

that the results are explained by differential birth rates among indigenous commu-

nities in titled vs. non-titled areas or differential out-migration of indigenous youth

in non-titled areas. As I show in the Appendix, the results hold when controlling

for recent migration among individuals who self-identify as indigenous specifically. I

also conduct the same differences-in-differences analysis using total population and

recent migration to the municipality as the dependent variables, respectively, and do

not find consistent effects of land titling in either direction.11

11These results appear in Appendix tables B.8, B.5, B.6, respectively.

115

These results provide evidence in support of the idea that collective recognition

can increase indigenous self-identification, at least as reported on the census. This

finding is consistent with both the dominant state-building paradigm and the recog-

nition as integration theory. Next I estimate the effects of receiving a CADT on the

two measure of compliance with the state. While the former predicts a decrease in

compliance, the latter predicts an increase.

Table 4.2 shows results for the birth registration outcome using the same subsets of

data used above: the full set of rural barangays, barangays matched on pre-treatment

covariates, barangays designated as eligible by the government, and all barangays that

were eventually titled.12 The models in columns 2, 4, 6, and 8 control for a number of

potential time-varying confounders, including change in population, the presence of

a health center, and street and highway access. In all models and all subsets, titling

leads to a significant positive increase in the proportion of the barangay population

whose births are registered, with estimates ranging between 2 and 3 percentage points.

These results are both statistically and substantively significant, across all spec-

ifications. The coefficient estimates in Table 4.2 represent between 0.13 and 0.24

standard deviations in barangay-level change in birth registration between 2000 and

2010 and account for between 13 and 21% of the of baseline difference in average

birth registration rates between majority indigenous and non-majority-indigenous

rural barangays.13 As shown in Table B.12 in the Appendix, the estimated effects of

titling are significantly greater for barangays with larger indigenous populations at

baseline, suggesting that the effects on barangay-level birth registration are driven by

changes in registration among the indigenous population specifically.14

12Note that because birth registration is measured for three census waves, I use a two-way fixedeffects model here.

13In 2000, indigenous-majority barangays in the study population reported 69.8% of births asregistered on average, while non majority-indigenous barangays had an average reported registrationrate of 83.9%, a 14.1 percentage-point difference.

14One potential concern in the interpretation of the interaction effects is that they may reflect thefact that barangays with larger indigenous population had lower baseline levels of birth registrationand therefore more room to increase. In the Appendix, I show common support in birth registration

116

Table 4.2: Land Titling (Continuous) and Birth Registration (Two-Way FE)

Dependent variable:

Birth RegistrationAll Rural All Rural Matched Matched Eligible Eligible Titled Titled

(1) (2) (3) (4) (5) (6) (7) (8)

Titled Prop. 0.029*** 0.029*** 0.018** 0.016** 0.025*** 0.024*** 0.022*** 0.019**(0.006) (0.006) (0.008) (0.008) (0.006) (0.006) (0.008) (0.008)

Bgy. Health Center 0.002* 0.014** 0.008*** 0.016**(0.001) (0.005) (0.003) (0.006)

Street Pattern −0.003*** −0.005 −0.002 −0.006(0.001) (0.004) (0.002) (0.005)

Highway Access 0.001 0.008 0.008*** 0.015**(0.001) (0.005) (0.003) (0.006)

Log. Population 0.016*** −0.022 0.004 −0.020(0.005) (0.019) (0.010) (0.025)

Mean Age −0.001** −0.0002 0.0004 −0.006*(0.001) (0.003) (0.002) (0.004)

Mean HH Size −0.001 0.007 0.011*** 0.008(0.001) (0.005) (0.003) (0.006)

Observations 34000 33979 1665 1665 7146 7146 1454 1454𝑅2 0 0.001 0.002 0.007 0.001 0.005 0.003 0.013

CRSE at bgy level *p<0.1; **p<0.05; ***p<0.01

117

Table 4.3 shows results from the same two-way fixed effects model using legibility

as the dependent variable. For the main effects, the coefficients are negative in all

subsets and statistically significant in all but the matched subset, providing suggestive

evidence that titling is associated with an increase in legibility (recall that a lower

legibility score indicates greater accuracy). The coefficients in Table 4.3 represent a

range of 0.08 to 0.25 standard deviations in change between 2000 and 2010 for rural

barangays. Again, as shown in Table B.15 in the Appendix, the effect is significantly

greater in areas with larger indigenous communities at baseline, suggesting that the

change is driven by greater improvements in census accuracy within indigenous com-

munities specifically.

To address concerns about reverse causality —specifically that places more com-

pliant or legible at baseline were more likely to obtain land titles —I estimate lagged

dependent variable models, predicting birth registration and legibility with land ti-

tling in the following year. Results from these analyses appear in Tables B.16 and B.17

the Appendix. While lagged effects are observed when including all rural barangays,

neither measure exhibits significant lagged effects in any of the other three subsets.

4.1.6 Additional Robustness Checks

I conduct a range of additional tests to probe the robustness of these findings. First,

I implement indirect tests to support the plausibility of the parallel trends assump-

tion. While the differences-in-differences design accounts for unit-level time invariant

confounding, it does not account for time-varying confounding. Without outcome

data prior to 2000, I cannot substantiate the parallel trends assumption using pre-

treatment trends in the outcomes. As an alternative indirect test, I construct a

psuedo panel of birth registration by age cohort and compare age cohorts in treated

across pre-treatment indigenous population. I also replicate the main analyses and interaction modelseliminating all barangays with baseline birth registration levels greater than 90%.

118

Table 4.3: Land Titling (Continuous) and Legibility (Two-Way FE)

Dependent variable:

Whipple IndexAll Rural All Rural Matched Matched Eligible Eligible Titled Titled

(1) (2) (3) (4) (5) (6) (7) (8)

Titled Prop. −3.602*** −3.472*** −1.545 −1.344 −2.610*** −2.479*** −2.133* −1.999(0.846) (0.844) (1.052) (1.058) (0.931) (0.938) (1.202) (1.230)

Bgy. Health Center −0.030 −0.736 −0.515 −0.657(0.180) (0.791) (0.450) (0.978)

Street Pattern 0.704*** 0.078 0.648 1.360(0.176) (0.599) (0.400) (0.829)

Highway Access −0.551** −0.644 0.384 −1.569*(0.216) (0.685) (0.437) (0.817)

Log. Population −5.158*** −1.062 −12.413*** −4.745(0.844) (2.689) (1.535) (3.148)

Mean Age 0.331 0.376 0.290(0.396) (0.251) (0.499)

Mean HH Size −1.158*** −1.355** 0.525 −0.926(0.204) (0.628) (0.448) (0.629)

Observations 34000 33979 1665 1665 7146 7146 1454 1454𝑅2 0 0.005 0.001 0.006 0.001 0.012 0.001 0.01

CRSE at bgy level *p<0.1; **p<0.05; ***p<0.01

119

and untreated communities using data from the 2000 census. As shown in Figure B-2

in the Appendix, pre-treatment cohort trends do not differ substantially time trends

for these indicators between titled barangays and the various reference groups used

in the main analysis. In addition, I include specifications for all outcomes interacting

the time dummy with province fixed-effects, which require only that parallel trends

hold within provinces.

Second, I re-estimate all of main specifications using Conley standard errors that

account for both spatial and temporal autocorrelation (Conley 1999). This is impor-

tant given that Ancestral Domains do not correspond directly to barangay boundaries

and in some cases cover portions of multiple neighboring barangays. However, as

shown in in the Appendix, the results remain largely unchanged when spatial depen-

dencies are taken into account. Third, for the identity outcome only, I use distance

to the regional office of the NCIP as an instrument for receiving a land title between

2000 and 2010, among the universe of barangays titled as of 2018. This analysis,

including a discussion of the plausibility of the needed identification assumptions, is

included in the Appendix.

4.1.7 Additional Alternative Explanations

I also consider several additional alternative explanations for the pattern of results

from the observational analysis. The first is that the effects of collective titling on

indigenous identity reflects purely instrumental behavior by community members.

Individuals in these communities —regardless of whether they qualify as indigenous

by some “objective” standard —are more likely to identify in this way to the state

because they expect it will lead to material benefits, such as land ownership within the

Ancestral Domain or access to benefits from negotiated settlements between outside

actors (i.e. private companies) and the tribe. On its face, this is unlikely given the

fact that indigenous status for the purposes of allocating these benefits would not be

120

determined by an individual’s response to the census questionnaire, but by the tribal

leadership in a particular community. Furthermore, if this were the case, we might

expect differentially greater positive effects of titling in areas with more valuable

land. I test this proposition by interacting the titling variable with two proxies for

land value: the presence of mineral deposits and a soil quality index. As shown in

Table B.21 in the Appendix, we do not see consistent evidence of differentially greater

increases in indigenous self-identification in titled communities with more valuable

land.

In addition to this “demand side” explanation, I consider the possibility that the

effects of titling on indigenous identity and compliance are due not to changes in com-

munity behavior, but to greater “supply” of state capacity in titled communities. This

could operate in part as a direct effect of the titling process itself: through titling, the

state gains greater access to communities which is reflected in more accurate census

data and higher levels of birth registration. The effect on indigenous identification

could also reflect a change in state behavior if, for example, census enumerators were

more likely to classify individuals as indigenous because they knew the community

had a title.

This, too, seems unlikely to explain the results for a number of reasons. First,

as noted above, the results hold when controlling for indicators of increased state

presence, such as road infrastructure, highway access, and the presence of a health

center.15 We also do not see an increase in recorded population as a result of titling,

which we might expect to observe if the titling process granted the state greater

access directly. Second, particularly in the eligible and titled subsets, many (if not

most) of the communities that did not receive land titles between 2000 and 2010 had

gone through most of the titling process short of receiving the legal title at the time

of census enumeration. Any gains in legibility or access resulting directly from the

15Measure in 2010, these are post-treatment. However, the results remain virtually the sameregardless of whether they are included.

121

titling process should apply to these communities as well.

This fact also helps to address the possibility that census enumerators are more

likely to classify people as indigenous if they are in titled areas, explaining the effects

on the reported indigenous population. The agency charged with issuing CADTs,

the National Commission on Indigenous Peoples (NCIP), is entirely separate from

the Philippines Statistics Authority (PSA), which administers the census. Yet even

if there were direct coordination between the PSA and the NCIP that resulted in

higher levels of indigenous classification (rather than indigenous self-identification),

we would expect this to apply to all areas classified by the NCIP as indigenous

(i.e. all communities in the “eligible universe” subset), not just to communities that

had received a title at the time of census enumeration. We would also expect birth

registration drives (which, as noted above, do target indigenous communities due

to their historically lower registration rates) to affect all communities classified as

indigenous, not just titled ones.

In addition, as shown in Table B.22 in the Appendix, I also do not find consistent

evidence that the effects of titling on compliance outcomes are differentially greater in

areas with more valuable land. This also also weighs against the idea that the state is

differentially targeting communities with titles, i.e. through birth registration drives.

If this were the case, we would expect the state to focus in areas where returns to

access and legibility are higher.

Taken together, these findings present a pattern not anticipated by dominant

theories of diversity and state building. On the one hand, they are consistent with

predictions from the ethnic politics literature that collective recognition —a form of

sub-group classification by the state —can strengthen attachments to sub-national

identities. However, they are inconsistent with the prediction that, in doing so, col-

lective recognition weakens the state by hindering its ability to achieve compliance

among the populations concerned. In contrast, these findings are consistent with the

122

idea that collective recognition can strengthen the state by incentivizing cooperation

and engagement with the state.

4.2 Survey Experiment

In this section, I provide further evidence in support of my argument using survey

experimental evidence. Doing so is important, because the observed relationship

between titling and compliance is consistent with other mechanisms not directly ad-

dressed above. I argue that the observed increase in compliance from collective recog-

nition reflects increased the legitimacy of the state among affected populations, by

creating the possibility of (group-based) representation within the formal political

system. However, it is also possible that it reflects an elite bargain whereby the state

trades the devolution of control to local elites for coerced community compliance

with its directives (Mamdani 1996; Mamdani 2012). A legitimacy-based explanation

would suggest that collective recognition increases individual affinity with the state

and identification with the broader national community. The administrative data

analysis does not speak to this directly, in that the observed increase in indigenous

self-identification does not indicate how this affects national identity. There is no

option to identify simply as “Filipino” on the census.16 I adjudicate between these

indirectly using evidence of the effects of collective recognition on individual attitudes.

4.2.1 Survey Data and Priming Experiment

The survey was conducted in partnership with Legal Network for Truthful Elections

(LENTE), a Philippines-based NGO, between March and October 2018.17 Figure

16Instead, a choice to identify with an indigenous ethnic group would come at the expense ofidentifying with a “non-indigenous” ethnic group. However, recall that no major ethnic group inthe Philippines represents colonial settlers. In a literal sense of the word, all major groups are“indigenous” to the islands.

17Unless otherwise noted, all analyses of the original survey data included in this paper are pre-registered. The pre-analysis plan is filed in the Evidence in Government and Politics (EGAP)

123

A-1 in the Appendix shows the locations of targeted and surveyed barangays. The

sample includes 725 respondents respondents from 80 barangays in the provinces of

Oriental Mindoro, Occidental Mindoro, and Palawan.18 Respondents were sampled

in communities with and without CADTs.19 All interviews were conducted face-to-

face by locally-recruited enumerators and administered on tablets. Two sitios, or

neighborhoods, were randomly selected within each barangay and respondents were

selected within each sitio using a random walk procedure.20

In the embedded survey experiment, respondents were randomly assigned to one

of three experimental conditions:

1. A control condition in which no prime is administered

2. A “material” prime condition in which respondents are shown an informational

flyer about the IPRA law that emphasizes the devolution of material powers

and control over land to indigenous leaders

3. An “identity” prime condition in which respondents are shown a similar infor-

mational flyer that emphasizes the recognition of a distinct indigenous identity

and culture by the state

This priming treatment was intended to make salient to respondents the fact that

the Philippine government has recognized collective rights for the country’s indige-

nous communities, as a means of capturing the effects of this policy on individual

registry, under ID 20181010AA.18The original sample targeted 100 barangays, but survey teams were unable to reach 20 within

the original sample due to security concerns. Table A.1 in the Appendix compares included andexcluded barangays. Due to data quality concerns, interviews less than 20 minutes in length aredropped. This reduces the total sample size from 768 to 725.

19Within the target sample, 46 out of 100 communities had CADTs at the time of the survey. Inthe final sample, 30 out of 50 barangays had CADTs.

20Community respondents had to be over the age of 18 and identify as a member of an indigenouscommunity, but could neither hold any leadership position in their tribe nor share a household withsomeone in a leadership role. In cases where there is only one sitio in a sample barangay with anindigenous population, 10 respondents are interviewed in that sitio.

124

attitudes.21 The two different versions of the prime were originally intended to test

distinct channels through which these effects may occur. The prime condition is ran-

domly assigned at the level of the individual respondent. In the second and third

prime conditions, enumerators were instructed to show one of two physical flyers to

respondents and read its contents.22 The identity outcome was measured using a

ranking activity, where respondents were asked to rank four types of identity group-

ings —tribe, religion, gender, and nationality —in order of importance to them as

components of their individual identity. Respondents were presented with four show

cards representing their tribe, their gender, their nationality, and their religion, placed

in a random order assigned to the enumerator on the tablet. Respondents were then

asked to rearrange the cards in order of importance to them as parts of their identity.23

The quantity of interest for this outcome is the Average Treatment Effect (ATE) of

the priming treatments on the relative ranking of each attribute, more specifically the

probability that an attribute is ranked in one of the top two positions.24

4.2.2 Results

Table 4.4 shows estimates of the effects of the priming treatment on the ranking of

four identity attributes: tribe, nationality, gender, and religion, first pooling the two

treatment conditions and then analyzing them separately. The coefficients can be

interpreted as the effect of the treatment on the probability that a given attribute is

21The flyers draw heavily on existing educational materials produced by the Philippine govern-ment, and were extensively pre-tested and reviewed by local collaborators. English translations ofthe flyer is included in the Appendix.

22The English translation of the language used to introduce the flyer is as follows: “This is notjust a survey, but we are also trying to education the community about the basic concepts of theIPRA law and the legal rights of IPs. Here is some information about the IPRA Law. I don’t haveenough printed flyers to leave one with everyone, but we hope you will share this information withothers in the community. I am going to ask you a few questions about it later to make sure youremember the information, so please pay attention!” Enumerators were audio recorded during thisportion of the survey and required to take a photo of the flyer shown to the respondent, in order torecord compliance with the assigned treatment condition.

23Images and exact question wording are included in the Appendix24The only pre-registered analysis is the effect on the tribe attribute.

125

ranked in the top two positions.25 As shown in second column, the culture priming

treatment lead to an increase in the salience of the nationality attribute compared

to the control group. Respondents who were provided information about the govern-

ment’s policy of recognition for indigenous communities (emphasizing the protection

of distinct cultural traits) were more than 9 percentage-points more likely to select

nationality as their most important of the four attributes, compared to the control

group.26 The point estimate associated with the material prime had an effect in the

same direction, but the effect is smaller and not statistically significant (similarly,

the coefficient for the pooled treatment is positive but not statistically significant

(𝑝 = 0.17)).27

At the same time, primed respondents were, on average, less likely to rank the

tribe attribute in the top position, although none of the coefficients are statistically

significant at the 𝛼 = 0.05 level. These findings weigh against the idea that in-

creases in indigenous self-identification resulting from the recognition of indigenous

communities by the state come the at the expense of identification with the national

community. Instead, they suggest the opposite and provide support for a legitimacy-

based explanation of the observational findings.

As a more direct test of this argument, I estimate the effects of the prime on a

battery of questions measuring attitudes toward local elected officials at the barangay

and municipal levels. Unfortunately, the survey did not include any questions about

attitudes toward the national government specifically. However, in the context of

the survey areas, where the vast majority of elected officials come from majority

Tagalog-speaking (non-indigenous) communities, attitudes toward elected local gov-

25Figure B-6 in the Appendix shows the raw distribution of rankings for the tribe and nationalityattributes in both treatment conditions.

26The pre-registered hypothesis associated with this experiment stated that the prime wouldincrease the ranking of tribal identity relative to other attributes. Analyses of the effects of theprime on attributes other than tribe were not pre-registered. These were conducted as exploratoryanalyses after failing to reject the null hypotheses associated with the tribe attribute.

27A Wald test does not allow me to reject the null hypothesis of no difference between the twocoefficients at the 𝑝 = 0.05 level (𝑝 = 0.12)

126

Table 4.4: Priming Treatment and Top Ranking of Identity Attributes

Dependent variable:

nationality_top tribe_top gender_top religion_top

(1) (2) (3) (4) (5) (6) (7) (8)

Any Prime 0.053 −0.049 −0.055 0.045(0.039) (0.039) (0.037) (0.037)

Culture Prime 0.091** −0.082* −0.063 0.050(0.045) (0.045) (0.042) (0.043)

Material Prime 0.016 −0.016 −0.047 0.039(0.044) (0.044) (0.042) (0.042)

Observations 725 725 725 725 725 725 725 725R2 0.003 0.007 0.002 0.005 0.003 0.003 0.002 0.002Adjusted R2 0.001 0.004 0.001 0.003 0.002 0.001 0.001 −0.001

Note: *p<0.1; **p<0.05; ***p<0.01

ernment officials are likely to pick up, at least in part, attitudes toward the state

and “mainstream” society more broadly. Table 4.5 shows the effects of the prime on a

pre-registered index of these questions constructed using Principal Components Anal-

ysis.28 Compared to the control group, respondents presented with the culture prime

had a higher affinity toward local government. Again, the coefficient is in the same

direction for the material prime but is not statistically significant at the 𝑝 = 0.05

level.

Another possibility is that the prime treatment made some respondents believe

mistakenly that the survey was being conducted by the government, pressuring them

into expressing affinity with the national identity. To test this alternative explanation,

I estimate the effects of the priming treatment on responses to a question toward the

end of the survey asking respondents who they thought was responsible for conducting

it. As shown in Table B.25 in the Appendix, neither prime had a significant effect on

28The analysis of experimental effects on this outcome was not pre-registered. However, theindex itself was pre-registered for descriptive comparisons. Figure B-7 in the Appendix shows ATEestimates for each index component.

127

Table 4.5: Priming Effects on State Attitudes

Dependent variable:

State Attitudes Index

(1) (2)

Any Prime 0.246*(0.131)

Culture Prime 0.304**(0.151)

Material Prime 0.190(0.150)

Observations 725 725R2 0.005 0.006Adjusted R2 0.004 0.003

Note: *p<0.1; **p<0.05; ***p<0.01

the belief that a government actor conducted the survey.29

It is not immediately clear why the effects of the two priming treatments differed

from one another (although I note that the difference between them is not statis-

tically significant). One possibility suggested in post hoc interviews with research

assistants, is that the material prime was seen as inconsistent with perceptions of

tribal governance structures in particular tribes. Specifically, the material prime may

have suggested centralized control or decision-making when in fact many decisions

are made through community consensus or consultation.

The survey experimental findings provide evidence supporting the idea that in-

digenous recognition increases compliance with the state in part due to increased

state legitimacy. Rather than increasing attachments to tribal identity at the ex-

29Responses counted as “government actors” include the National Commission on Indigenous Peo-ples (NCIP), the Commission on Elections (COMELEC), the barangay government, the municipalgovernment, and any open-ended responses that referenced a government actor (e.g. “the President,”“government,” “mayor,”)

128

pense of national identity, providing information about the government’s policy of

recognition increased the relative salience of respondents’ national identity and led

to more positive attitudes toward local government officials.

4.3 Discussion

The findings presented in this chapter weigh against the idea that the devolution of

power to non-state authorities and recognition of sub-national identities threatens

the state’s ability to implement its chosen policies. Leveraging sub-national variation

in recognition in the Philippines, I find evidence that collective recognition increases

identification with indigenous ethnic groups. At first glance, this may be interpreted

in support of the idea that recognition strengthens sub-national and non-majoritarian

identities and has the potential to weaken the state. Yet I also find that collective

recognition leads to greater compliance with the state, using birth registration and

census data quality as a proxy measures. Evidence from the survey experiment sup-

ports the interpretation that recognition increases compliance by encouraging affinity

with the national community and more positive attitudes toward local government.

These findings also challenge the assumed inverse relationship between sub-national

identity and identification with a broader national identity, or at the very least raise

questions about the types of identity groups to which it applies. This assumption

may be more appropriate in contexts where national politics is seen as a zero-sum

competition between large, regionally-based ethnic groups. It may be less appropriate

for groups —such as indigenous communities —defined precisely by their historical

marginalization from the state. I explore this distinction further in the concluding

chapter. In the next chapter, I provide additional evidence for the mechanism im-

plied by the theory of recognition as integration: that recognition increases the state’s

legitimacy among marginalized groups by increasing their to make collective claims

129

through the formal political system.

Finally, the analyses presented here have important limitations that suggest ad-

ditional directions for future research. For example, these findings only speak to the

effects of recognition in the short-to-medium term. If in the long term recognition has

the effect of raising communities’ expectations beyond what the state can reasonably

deliver, these effects may not endure. In addition, this analysis focuses on the effects

of collective recognition on attitudes and behaviors among indigenous communities

eligible for this recognition. It does not speak directly to the effects on surround-

ing communities or to intra-community dynamics at a local level. In the concluding

chapter, I propose future research directions to address these and other remaining

questions regarding the interpretation of results.

130

Chapter 5

From recognition to representation

I have argued that collective recognition of historically marginalized groups increases

their voluntary engagement with the state and the formal political system. In the

previous chapter, I presented evidence that recognition in the form of collective land

titling leads to increases in indigenous self-identification, but also to multiple indi-

cators of voluntary compliance with and integration into the state. Here, I provide

evidence in support of my proposed interpretation: collective recognition increases

marginalized communities’ capacities to make collective claims on the state, including

through participation in formal democratic politics, thereby increasing the expected

benefits from this participation.

I propose two mechanisms through which collective recognition can improve claim-

making capacity. First, building on Jung (2008), collective recognition provides a

discursive basis for political claim-making, establishing “indigeneity as a political cat-

egory, and a location from which to make demands on the state” (p. 38). I provide

evidence consistent with this mechanism in the previous chapter, demonstrating that

collective recognition leads to greater identification with indigenous ethnic groups. In

addition, by legitimating and expanding the powers of indigenous political institu-

tions, I argue that recognition provides a technology for solving free-rider problems in

131

the context of democratic elections characterized by vote-buying, enabling communi-

ties to more effectively bargain, as a collective, for community-level public goods.1

The proposition that empowering traditional institutions can increase electoral

cohesion is consistent with a body of work depicting traditional authorities as elec-

toral brokers who deliver the votes of their communities in exchange for private rents,

thereby entrenching powerful political elites at the expense of their communities’ inter-

ests (Mamdani 1996, Ntsebeza 2005, Anderson, Francois and Kotwal 2012, Acemoglu,

Reed and Robinson 2014, De Kadt and Larreguy 2015, Koter 2013, Boone 2014, Mer-

shon 2015, Conroy-Krutz 2017). This raises concerns that collective recognition may

result in “elite capture” of indigenous institutions in a way that is ultimately harm-

ful to ordinary indigenous voters and detrimental to democratic accountability more

broadly. I have argued, however, that collective recognition, to the extent that it

encourages bloc voting, also has the potential to improve the democratic representa-

tion of marginalized indigenous communities. If collective recognition enhances the

ability of communities to “pool their political resources” and coordinate in candidate

selection, this may in turn improve their ability to attract public goods provided by

the state (Boone 2003, Smith and Mesquita 2012, Gottlieb and Larreguy 2016). If

communities can credibly commit to delivering a block vote to officials who imple-

ment their policy priorities (and to deny the same block vote to officials who do not),

they may be more likely to see these preferred policies implemented. The ability to

coordinate may therefore increase the benefits individuals in these communities can

expect to gain from participating in formal democratic politics, especially compared

to a status quo in which individual votes are typically bought or coerced.

In Chapter 3, I use in-depth interviews with tribal leaders to demonstrate that

indigenous elites see their ability to coordinate votes as a critical determinant of

1This stands in contrast to Jung (2008) and others (i.e. Mamdani 2012), who suggest thatrecognition based on cultural identity may have the effect of limiting demands on the state to therealm of culture (rather than, for example, state public goods).

132

their ability to influence local government. I illustrate mechanisms through which

collective recognition by the government empowers indigenous political institutions

and facilitates this process. I also provide evidence that these elites are constrained in

their ability to engage in rent-seeking behavior by accountability mechanisms within

their own communities. The latter is an important scope condition for the argument

that individual members of indigenous communities (and not only indigenous elites)

can expect higher returns from engagement in formal democratic politics when they

participate collectively. In this chapter, I analyze administrative and survey data to

provide further evidence of various observable implications of this argument.

First, I use nationwide precinct-level electoral returns data to examine the re-

lationship between collective land titling, community-level electoral cohesion, and

electoral competition. The argument presented here suggests that we should observe

more cohesive voting in communities with collective land titles, compared to similar

untitled communities. I additionally examine electoral competition in order to distin-

guish between collective bargaining and “elite capture,” assuming that, on average,

elite capture would be more likely to entrench already-powerful political elites and

reduce overall competition.2 Consistent with predictions of the theory, I show that

collective titling is associated with greater electoral cohesion at the precinct level in

municipal elections. However, titling does not disproportionately benefit incumbents.

If anything, it is associated with greater competition at the municipal level, providing

suggestive evidence against a pure elite capture interpretation of bloc voting.

Second, I analyze data from an original face-to-face survey conducted in three

provinces (Oriental Mindoro, Occidental Mindoro, and Palawan), to further probe

the argument that collective recognition facilitates collective voting behavior by em-

powering indigenous political institutions to facilitate voter coordination. Consistent

with this argument, I document higher levels of voter coordination and leader in-

2I discuss this assumption more explicitly in the context of the Philippines’ electoral systembelow.

133

fluence on voting choice in titled communities compared to untitled communities. I

use an embedded survey experiment to further examine how collective recognition

affects elite influence over vote choice. I find that collective recognition (more pre-

cisely, information about collective recognition) does not increase indigenous leaders’

influence on hypothetical vote choice unconditionally. Instead, it shifts community

members’ voting preferences to prioritize signals of future public goods provision by

the state. This weighs against the idea that collective recognition simply increases the

coercive power of elites and in support of the idea that it increases expected returns

to engagement in the formal political system for ordinary indigenous voters.

Third, I use data from a pilot intervention by the Philippines’ Commission on

Elections (COMELEC) in the same region to demonstrate how another form of col-

lective recognition that improves intra-community monitoring capacity —the estab-

lishment of separate polling places for indigenous communities —does not decrease

electoral competition or benefit incumbents. Finally, I return to the panel analysis

in the previous chapter and evaluate the effects of collective titling on access to cer-

tain local public goods. I find that collective titling is associated with differential

increases in certain types of public goods. Taken together, these individual pieces

of evidence are consistent with the theory that collective recognition can facilitate

collective participation and claim-making within formal democratic politics, thereby

increasing expected returns to engagement in the formal political system for histori-

cally marginalized groups.

5.1 Collective recognition and voting behavior: na-

tionwide

As an initial test of the idea that collective recognition facilitates greater electoral co-

ordination within indigenous communities, I compare local-level electoral cohesion in

134

titled and untitled barangays in the 2016 mayoral elections. I operationalize the titling

“treatment” at the barangay level in two different ways throughout the analysis. First,

using the map of approved CADTs provided by National Commission on Indigenous

Peoples (NCIP), I calculate the proportion of a barangay’s land area that fell within

a CADT in a given year. Second, I use a binary measure based on a separate list

of barangays obtained from NCIP.Electoral cohesion is measured using precinct-level

vote margin for mayoral races during the 2016 national and local elections, averaged

at the barangay level, the lowest level at which precincts are geo-located. Vote margin

(margin2016_prec) is calculated by taking the difference in vote totals between the

winner within a precinct and the second runner up, and dividing by the total number

of votes cast in the precinct. This measure is meant to capture the extent to which

communities at a very local level (the precinct) vote for the same candidate. I am

not concerned in this case with who communities vote for, just the extent to which

individuals in a localized geographic area vote as a block for the same candidate.

Table 5.1 shows the estimated relationship between land titling and average precinct-

level vote margin during the 2016 elections for municipal mayor, aggregated at the

barangay level among all rural barangays. All comparisons are cross-sectional.3 Unless

otherwise noted, all models include municipality fixed effects, to account for race-level

factors such as the number of candidates and the overall level of competition. Mu-

nicipalities with uncontested mayoral races are dropped from the analysis.

As shown in column 1, there is a significant positive relationship between the

proportion of a barangay’s land area covered by a CADT in 2016 and the extent

to which individuals in a precinct voted for the same mayoral candidate during the

2016 mayoral elections, on average. In column 2, I control for a number of pre-

treatment barangay-level covariates that may affect both the probability of receiving

3As of this writing, precinct-level electoral returns data are not publicly available for electionsprior to 2016. For the analysis of the COMELEC pilot intervention, I purchased these data for alimited number of provinces and manually entered them from pdf files with the help of a researchassistant.

135

Table 5.1: Land Titling (Continuous) and Average 2016 Precinct-Level Vote Margin(Barangay Level)

Dependent variable:

Average 2016 Precinct-Level Vote MarginAll Rural All Rural Eligible Eligible No OP No OP Matched Matched

(1) (2) (3) (4) (5) (6) (7) (8)

Titled Prop. 0.040*** 0.021** 0.022* 0.011 0.042*** 0.035** 0.033** 0.028*(0.008) (0.008) (0.011) (0.012) (0.014) (0.015) (0.013) (0.014)

IP Pop. 2000 0.018** 0.017 0.009 0.030(0.008) (0.011) (0.014) (0.019)

Ethfrac 2000 0.010 0.014 −0.001 0.045**(0.007) (0.012) (0.015) (0.022)

Log Pop 2000 −0.026*** −0.028*** −0.024*** −0.016*(0.002) (0.004) (0.005) (0.008)

Area 0.0002*** 0.0001 −0.00003 0.00004(0.0001) (0.0001) (0.0001) (0.0001)

Log Coast Distance 0.003** 0.0005 0.008** −0.008(0.001) (0.003) (0.004) (0.007)

Mean Elevation −0.00000 −0.00000 −0.00002 −0.0001*(0.00001) (0.00002) (0.00002) (0.00005)

Elevation SD 0.00000 0.00000 −0.00003 0.00002(0.00003) (0.0001) (0.0001) (0.0001)

Mean Slope 0.0001 0.0002 0.002* 0.002(0.0004) (0.001) (0.001) (0.002)

Soil Index −0.008* 0.014 0.021* 0.011(0.005) (0.010) (0.012) (0.022)

Log Road Dist 0.002*** 0.002*** 0.001* 0.002(0.0004) (0.001) (0.001) (0.001)

Log NCIP Dist −0.007 0.010 0.018 0.004(0.005) (0.009) (0.011) (0.016)

Mun FE Y Y Y Y Y Y Y YObservations 32,554 27,302 7,194 6,949 4,871 4,709 1,917 1,868Adjusted R2 0.761 0.781 0.765 0.769 0.766 0.769 0.782 0.788

Note: *p<0.1; **p<0.05; ***p<0.01

136

a land title and vote margin, including the proportion of the population identifying

as indigenous in 2000, ethnic fractionalization in 2000, logged population, as well as

a number of geographic variables. The coefficient on titling reduces in magnitude but

remains statistically and substantively significant, accounting for approximately 0.18

standard deviations of within-municipality variation in average precinct-level vote

margin. Columns 3 and 4 show estimates using only barangays in the eligible universe

based on the NCIP data. The positive bivariate relationship (within municipalities)

holds in this subset, and is statistically significant at the 10% level (𝑝 = 0.051).

When controls are included the coefficient remains positive is not longer statistically

significant.

Estimates within the eligible universe may be biased downward if the CADT pro-

cess, prior to the actual granting of the official title, has an independent effect on

in-group identity or the influence of indigenous leaders, and therefore on electoral

cohesion. The CADT may also become a politically salient issue before it is secured,

encouraging mobilization around a candidate who pledges to support a community’s

CADT claim (despite the fact that local elected officials have no official role in the

titling process).4 To investigate this possibility, I re-estimate the same models within

the eligible universe, excluding barangays listed as having a CADT application under

processing. The comparison in these models is between barangays that were covered

by an approved CADT in 2016 and barangays where a community has either ap-

plied for a CADT or where no application has been submitted but the barangay has

been listed as eligible by the regional NCIP office. I also create a matched sample

from among the whole set of rural barangays, matching barangays with an approved

CADT in 2016 to those without on pre-treatment covariates, excluding those with

on-process CADTs. Estimates from these analyses are shown in columns 5-8. In both

subsets excluding barangays with on-process CADTs, the coefficient is substantively

4Indeed, in some interviews I conducted in areas with on-process CADTs, the leaders mentionedthat the candidates were courting the IP vote by promising to support their CADT application.

137

and statistically significant, with and without controls.5

These findings suggest a positive association between collective land titling and

electoral cohesion at the community level. What implications does this have for demo-

cratic accountability? The Philippines has a fragmented and unstable party system.

Party switching and the formation of new parties are both common (Mendoza, Cruz

and Yap II 2014). Particularly at the municipal level, politics are instead primar-

ily organized around political family dynasties (Cruz, Labonne and Querubin 2015).

This makes it more difficult to determine whether electoral cohesion should be inter-

preted as “elite capture” or collective bargaining, especially compared to dominant

party systems (Larreguy N.d., De Kadt and Larreguy 2015). Rather than focusing

on how increased bloc voting benefits a particular party, I examine the relationship

between titling and incumbent vote share. If indigenous leaders are acting as clien-

telistic vote brokers, they may be more likely to support incumbents who can leverage

state resources to pay a higher price. Examining incumbency speaks to how collective

recognition affects electoral competition and whether it entrenches elites or empowers

challengers.

Table 5.2 shows results from the same models shown above, with incumbent vote

share replacing vote margin as the dependent variable. Incumbent vote share is

calculated at the precinct level and averaged for each barangay.6 While the bivariate

relationship between titling and incumbent vote share is positive and significant, the

coefficient shrinks considerably in size when controls are included. The relationship

is neither substantively nor statistically significant in any of the other models. This

5Results from the using the binary version of the land titling variable appear in the Appendix.While the results are similar in the full sample of rural barangays and in the matched sample,the relationship disappears in the eligible subset. This may be due to the fact that the binaryvariable does not distinguish between barangays that are fully covered by a CADT and those thatare only partially covered. If a CADT covers a small portion of a barangay, it less likely to move abarangay-level average capturing voting behavior of all voters in the barangay.

6Incumbency was determined using name-matching based on 2013 municipal-level returns data.Where there was no exact name match, the surname was used to capture family ties betweena candidate and the incumbent. Municipalities where incumbency could not be determined aredropped.

138

suggests that, if there is a positive effect of collective land titling on electoral cohesion,

it does not appear to disproportionately benefit incumbents.

Table 5.2: Land Titling (Continuous) and Average 2016 Precinct-Level IncumbentVote Share (Barangay Level)

Dependent variable:

Average 2016 Precinct-Level Incumbent VoteshareAll Rural All Rural Eligible Eligible No OP No OP Matched Matched

(1) (2) (3) (4) (5) (6) (7) (8)

Titled Prop. 0.014** −0.001 0.007 −0.001 0.006 0.002 −0.002 −0.004(0.007) (0.007) (0.009) (0.007) (0.011) (0.011) (0.010) (0.011)

IP Pop. 2000 0.008 0.008 0.004 −0.002(0.006) (0.006) (0.011) (0.014)

Ethfrac 2000 0.004 0.004 −0.007 −0.008(0.006) (0.006) (0.012) (0.017)

Log Pop 2000 −0.017*** −0.017*** −0.010*** −0.005(0.001) (0.001) (0.004) (0.006)

Area 0.0002*** 0.0002*** −0.00004 −0.0001(0.00005) (0.00005) (0.0001) (0.0001)

Log Coast Distance 0.002** 0.002** 0.002 0.006(0.001) (0.001) (0.003) (0.005)

Mean Elevation 0.00000 0.00000 −0.00001 0.00002(0.00001) (0.00001) (0.00002) (0.00003)

Elevation SD −0.00001 −0.00001 −0.0001 −0.0001(0.00002) (0.00002) (0.0001) (0.0001)

Mean Slope −0.00001 −0.00001 0.002*** −0.001(0.0003) (0.0003) (0.001) (0.002)

Soil Index −0.007* −0.007* 0.002 −0.010(0.004) (0.004) (0.009) (0.017)

Log Road Dist 0.002*** 0.002*** 0.002*** 0.001(0.0003) (0.0003) (0.001) (0.001)

Log NCIP Dist −0.003 −0.003 −0.005 −0.005(0.004) (0.004) (0.009) (0.013)

Mun FE Y Y Y Y Y Y Y YObservations 27,291 22,959 6,308 22,959 4,371 4,237 1,726 1,697Adjusted R2 0.776 0.791 0.756 0.791 0.769 0.770 0.800 0.801

Note: *p<0.1; **p<0.05; ***p<0.01

139

Next, to further probe the implications for electoral competition, I examine the

relationship between land titling and vote margin at the municipal level, the level

at which the races in question were contested. Here, the independent variable is the

percentage of a municipality’s land area covered by one or more CADTs.7 I restrict

the analysis to rural municipalities, which I define as municipalities with at least 90%

of barangays classified as rural by the Philippine Statistics Authority.8 Table 5.3

shows the results from this comparison. Across all specifications, the point estimate

is negative, meaning that elections with a higher percentage of land covered by a

CADT were more competitive, on average. In all but one model, the coefficients fail

to reach conventional levels of statistical significance; however, the coefficients are

substantively meaningful in size. The absolute value of the coefficient in column 3,

for example, exceeds the 2016 win margin of 356 of the 1,210 mayoral races in the

sample (approximately 29%).

These analyses are all cross-sectional and should not necessarily be interpreted as

causal. It is possible that communities that are better able to mobilize a bloc vote

have also been more successful lobbying for the approval of a CADT application, and

that CADTs are more likely to be granted in competitive municipalities. At mini-

mum, however, these findings provide suggestive evidence that state recognition of

indigenous communities through the granting of collective land titles is associated

with greater electoral cohesion at the community level, but greater competition over-

all. If collective recognition led to elite capture of indigenous institutions, we might

expect it to benefit already-powerful political actors and reduce competition. These

results provide suggestive evidence that this is not the case.

7Because both the binary titling and “eligible universe” classifications are designated at thebarangay level, I cannot conduct the same additional analyses at the municipal level.

8This includes 1,210 of the 1,647 barangays in the country.

140

Table 5.3: Land Titling and 2016 Mayoral Vote Margin (Municipal Level)

Dependent variable:

2016 Mayoral Vote Margin

(1) (2) (3) (4)

Titled Prop −0.024 −0.121 −0.110** −0.115(0.051) (0.078) (0.055) (0.081)

IP Pop. 2000 0.067 0.059(0.060) (0.064)

Ethfrac 2000 −0.078 −0.007(0.056) (0.074)

Log Pop 2000 −0.002 0.007(0.020) (0.021)

Area 0.00002 0.0001(0.0001) (0.0001)

Log Prov. Cap Distance −0.006 −0.009(0.008) (0.008)

Log Coast Distance 0.042*** 0.044***(0.012) (0.013)

Mean Elevation 0.00002 −0.00002(0.0001) (0.0001)

Elevation SD −0.0003 −0.0003(0.0002) (0.0002)

Mean Slope 0.007 0.008(0.005) (0.005)

Region FE N N Y YObservations 1,100 895 1,100 895R2 0.0002 0.031 0.049 0.061Adjusted R2 −0.001 0.020 0.035 0.037

Note: *p<0.1; **p<0.05; ***p<0.01

141

5.2 Collective recognition and voting behavior: sur-

vey results

Next, I further explore the relationship between recognition and voting behavior using

data from the survey of indigenous communities in Oriental Mindoro, Occidental

Mindoro, and Palawan. As described in the previous chapter, the survey included a

total of 725 community respondents from 80 barangays in these provinces. Barangays

were selected for the survey to achieve variation in titling status and participation in

the COMELEC pilot intervention, described in greater detail below.

5.2.1 Collective recognition and voter coordination

First, I descriptively compare respondents in barangays with and without CADTs in

the survey sample. With the exception of one municipality, survey data collection

began in August 2018, three months after the May 2018 barangay elections. Re-

spondents answered a battery of questions about their experiences in these elections,

with a focus on the role of indigenous leaders in directing communities’ voting be-

havior and community-level coordination in candidate selection more broadly. These

questions are summarized in two pre-registered indices, constructed using Principal

Components Analysis: a leader influence index and a voter coordination index. The

leader influence index includes a battery of survey questions about the reported role

of leaders in determining respondents’ (and the communities’) vote choice in the May

2018 elections, while the voter coordination index captures reported actions by the

community to coordinate in candidate selection, including holding community meet-

ings.9 My theory predicts that survey respondents in titled communities will report

9The leader influence index included questions about whether the respondent knew which can-didate the leader supported for barangay captain, whether the respondent supported the same can-didate as the leader, whether the respondent believed the majority of the community supportedthe same candidate as the leader, whether the respondent believed voting for a different candidatefrom the leader’s chosen candidate would damage the respondent’s relationship with the leader, and

142

higher levels of leader influence on vote choice and higher levels of voter coordination.

Table 5.4 shows results from barangay-level comparisons of index values for titled

and untitled barangays within the survey sample, excluding the 12 barangays sur-

veyed prior to the May 2018 elections. These estimates suggest higher levels of both

leader influence on vote choice and voter coordination in titled barangays. While the

coefficient on the voter coordination index drops below statistical significance when

barangay-level controls are included (𝑝 = 0.17), the coefficient remains similar in

magnitude (the lack of precision is not necessarily surprising given the small sample

size). Analyses of individual index components are included in Tables B.26 and B.27

in the Appendix.

5.2.2 Collective recognition and leader influence on vote choice

I further examine the relationship between collective recognition and leader influence

on candidate choice using a choice-based conjoint experiment. Conjoint experiments

are used frequently in political science to understand how individuals make multi-

dimensional decisions, including over candidates during elections (Hainmueller, Hop-

kins and Yamamoto 2014). In this context, the purpose of this measurement strategy

is to better understand how the endorsement of a candidate by indigenous leaders

compares to other metrics of candidate quality in determining voter preferences.

Understanding how individual indigenous voters evaluate candidates for local

elected office is important for interpreting the relationship between collective recog-

nition and community-level electoral cohesion, and ultimately for understanding the

implications of collective recognition for democratic representation. If collective recog-

nition makes individual voters more likely to prefer candidates who are endorsed by

whether the respondent believed the leader could find out who they voted for. The voter coordina-tion index included questions about whether the respondent believed others in the community couldfind out who they voted for, whether the respondent was aware of a meeting being held ahead ofthe elections to discuss candidates, whether the respondent attended the meeting, and whether thecommunity came to a consensus about which candidate to support.

143

Table 5.4: Land Titling and Voter Coordination

Dependent variable:

vote_influence_index vote_coordination_index

(1) (2) (3) (4)

CADT 2018 0.327** 0.407** 0.377*** 0.216(0.147) (0.181) (0.131) (0.157)

IP Pop 2000 −0.095 −0.437(0.320) (0.277)

Log Pop 2000 0.074 0.207*(0.134) (0.116)

Pct Catholic 2000 0.118 −0.258(0.524) (0.454)

Ethfrac 2000 0.506 0.214(0.372) (0.322)

Mean Slope 0.001 0.021(0.019) (0.017)

Log Coast Dist −0.016 0.104(0.112) (0.097)

Log Road Dist 0.033 0.016(0.022) (0.019)

Strong Roof Mat 2000 0.415 −0.285(0.539) (0.466)

Area 0.004* 0.001(0.002) (0.002)

Observations 68 68 68 68R2 0.070 0.226 0.111 0.308Adjusted R2 0.056 0.091 0.097 0.186

Note: *p<0.1; **p<0.05; ***p<0.01

144

tribal leadership but are otherwise of lower quality, this could suggest that collective

recognition empowers traditional elites in a way that is detrimental to accountabil-

ity.10 If candidates can obtain a community block vote simply by co-opting or pro-

viding private goods to tribal leadership, elected officials may not be incentivized to

provide public goods or otherwise act in the best interests of the community as a

whole.

I examine this possibility using a survey experiment. Specifically, I estimate the

effects of a randomly-assigned priming treatment on the criteria used to determine

hypothetical vote choice, as measured in the conjoint experiment. The priming treat-

ment (described in detail in the previous chapter) is intended to make salient to

respondents the fact that the national government has recognized collective rights

for the country’s indigenous communities, as a means of capturing the effects of this

policy on individual attitudes. The experiment speaks to the following question: does

being reminded of collective recognition change the criteria indigenous voters use to

make (hypothetical) voting decisions? If so, does it do so in a way that is detrimental

to democratic accountability?

Survey respondents were assigned to one of three priming treatment conditions:

1. A control condition in which no prime is administered

2. A “material” prime condition in which respondents are shown an informational

flyer about the IPRA law that emphasizes the devolution of material powers

and control over land to indigenous leaders

3. An “identity” prime condition in which respondents are shown a similar infor-

mational flyer that emphasizes the recognition of a distinct indigenous identity

10However, as Baldwin (2013) argues, it is not necessarily the case that voting with traditionalleaders is a signal of elite capture. If coordination between traditional leaders and elected officialsis required for public goods provision, the support of traditional leaders may be one of the mostimportant signals of prospective performance. In other words, evidence of unconditional voting withindigenous elites would not be sufficient to conclude that detrimental elite capture is occurring.

145

and culture by the state

I then compare the results of the conjoint experiment across treatment arms. In

the conjoint experiment, respondents are asked to choose between two hypothetical

candidates for barangay captain, or local chief executive. The conjoint experiment

was presented to respondents as a “voting game.”11 Each respondents was shown

five pairs of candidate profiles, with each profile meant to represent a hypothetical

candidate, and asked which they would choose in a hypothetical barangay election.

The candidates vary along the following five attributes, with each attribute taking

two possible values (attributes were represented using images that were explained to

the respondent prior to the start of the game):

∙ Campaigning : the respondent did (1) OR did not (0) give a gift during the

campaign period

∙ Campaign promises : the respondent made a specific promise with a plan to

improve roads in the community (1) OR made vague, non-specific promises to

help the community (0)

∙ Mayor alignment : the respondent is (1) OR is not (0) aligned with the municipal

mayor

∙ Leader endorsement : the candidate is (1) OR is not (0) supported by tribal

leadership

∙ Community work : prior to running for elected office, the candidate did (1) OR

did not (0) engage in voluntary work to help the indigenous community11The English translation of the text introducing the voting game was as follows: “This is a voting

game. We will show you two hypothetical candidates in a barangay election. Imagine two candidates,Candidate A and Candidate B. Neither of them is a real candidate. Assume that neither candidatehas been in office before (but again, they are not meant to represent real people who have previouslyrun for office). We will present you with some information about each of the candidates. Beforewe start the game, I will explain to you the different types of information you will see about thecandidates. Assume that the candidates are the same in every way, except for any differences yousee here. Please indicate which of the two candidates you would vote for. There is no right or wronganswer.” Images representing candidate attributes are shown in Figure B-8 in the Appendix.

146

The campaigning attribute is meant to signal vote-buying; however this term was

not used due to the strong negative valence of this term among indigenous communi-

ties identified during interviews and pilot testing. The campaign promises and mayor

alignment attributes are meant to signal the future delivery of public or club goods to

the community. In the Philippines context, many local public goods are funded at the

municipal level, and alignment between a barangay captain and the municipal mayor

would make it more likely that municipal resources will be directed to that particular

barangay. The community work attribute was meant to signal prioritization of the in-

digenous community in a way that would have been appropriate for a non-incumbent

candidate (both candidates were described as never having been elected previously to

minimize association with current or past barangay officials). The leader endorsement

attribute was extensively pilot-tested to ensure that the language and associated im-

age captured an endorsement by tribal leadership in a way that was applicable across

different tribal leadership structures.12

The main quantity of interest is the difference between the Average Marginal

Component Effects (AMCEs) for leader endorsement in the treatment and control

conditions: in other words, the ATE of the priming treatment on the influence the

indigenous leader’s endorsement has on hypothetical vote choice. The pre-registered

hypothesis associated with the conjoint experiment stated that the priming treatment

would increase the influence of the indigenous leaders’ endorsement on hypothetical

candidate choice.

Figure 5-1 shows the Average Marginal Component Effects (AMCE) for each can-

didate attribute across treatment conditions. Panel 5-1a shows coefficients in the

control condition compared to coefficients within a pooled primed condition. Panel

12As discussed in Chapter 3, the practice of tribal leadership endorsing candidates in local electionswas common across tribes. The image, shown in the Appendix, depicts a group of people sittingaround a fire. Another person (symbolizing the candidate) is either standing next to the group andshaking hands (to indicate endorsement) or standing apart from the group (to indicate a lack ofendorsement).

147

5-1b shows coefficients in all three treatment conditions.13 These coefficients can be

interpreted as follows: averaging across all combinations of the other attributes within

the control group, a candidate who gives a gift during the campaign is, on average,

approximately 9 percentage points less likely to be selected than one who does not.14

As is evident in Figure 5-1, I do not find evidence in support of the hypothesis

that the priming treatment increases the influence of indigenous leaders on candidate

choice. Instead, to the extent the priming treatment had any effect on candidate

preferences, it appears to have made respondents more likely to punish candidates

who engage in vote-buying,15 less likely to prefer candidates who engage in charitable

work in the community,16 and more likely to prefer candidates with attributes sig-

nalling future benefits from the state: mayor alignment17 and campaign promises.18

Table B.28 in the Appendix shows coefficients and standard errors for the pooled and

disaggregated analyses, with and without round and enumerator fixed effects. Figure

B-12 in the Appendix shows marginal means by experimental treatment condition.19

The analysis of priming effects on attributes other than leader endorsement are

not pre-registered and should therefore be considered exploratory. Taken together,

however, this pattern of findings is inconsistent with the idea that collective recogni-

tion unconditionally increases the influence of traditional elites on voting behavior.

It is instead consistent with the idea that collective recognition increases indigenous

voters’ expectations about the possibility of obtaining public goods from the state.

When primed to think about the state’s policy of collective recognition, respondents’

preferences for hypothetical candidates shifted away from those who provided short-

13Figure B-9 in the Appendix shows the AMCE in the pooled sample14This may seem surprising at face value. However, there is a strong norm (at least a stated one)

against vote buying in indigenous communities.15𝑝 = 0.09 in the pooled analysis including round and enumerator fixed effects16𝑝 = 0.015 in the pooled analysis including round and enumerator fixed effects17𝑝 = 0.065 in the pooled analysis, including round and enumerator fixed effects18𝑝 = 0.15 in the pooled analysis including round and enumerator fixed effects19Leeper, Hobolt and Tilley (2020) caution that subgroup differences in AMCEs can be mislead-

ing due to sensitivity of regression interactions to reference categories. In this case, the subgroupmarginal means are consistent with the direction implied by the comparison of AMCEs.

148

●

●

●

●

●

−0.

2−

0.1

0.0

0.1

0.2

Experimental Effects (Control vs. Either Prime)

AM

CE

gift promises mayor leader commwork

●

●

●● ●

●

●

ControlPrime

(a)

●

●

●

●

●

−0.

2−

0.1

0.0

0.1

0.2

Experimental Effects (All Prime Conditions)

AM

CE

gift promises mayor leader commwork

● ControlIdentityMaterial

(b)

Figure 5-1: Experimental effects by candidate attribute. Average Marginal Com-ponent Effects (AMCE) for each binary attribute, comparing titled and untitledbarangays, shown with 95% confidence intervals. Standard errors for each treatmentcondition are clustered at the level of the respondent.

149

term and particularistic benefits (vote-buying and charitable work) and toward those

who signaled the delivery of state public goods in the future. Contrary to fears

that recognizing self-determination rights for indigenous communities will undermine

democratic accountability by empowering traditional leaders as clientelistic interme-

diaries, these findings provide suggestive evidence that such policies can encourage

indigenous communities to make material demands on the state through democratic

elections.

Another possible interpretation of these findings (specifically the lack of a priming

effect on leader influence) is that they reflect a ceiling effect. In places where the actual

influence of traditional leaders is strongest, the priming treatment may have failed

to shift respondents’ perceptions of their influence, resulting in a null effect from the

priming experiment. As a manipulation check, I examine the effects of the priming

treatments on a pre-registered index of questions about the perceived influence of

indigenous leaders in local governance.20 I disaggregate by actual land titling status,

given that (as shown in the previous section), leader influence is higher on average in

titled communities.

Figure 5-2 shows the effects of both primes on the perceived influence of indigenous

leaders, compared to the control condition. In the full sample, the “material” prime

had a positive but statistically insignificant effect on perceptions of leader influence,

while the “identity” prime had no effect. Disaggregating by titling status, we see that

the “material” prime had a positive effect on perceived leader influence, significant

at the 𝛼 = 0.1 level (𝑝 = 0.059), but only in untitled communities. By contrast,

neither prime appears to have any effect on the perceived influence of leaders among

respondents in communities with titles. This suggests that the findings from the

priming experiment may be due to the failure of the prime to “move” respondents’

20The analysis of experimental effects on the perceived influence of indigenous leaders was notpre-registered, although the index itself was. Priming effects on individual index components areshown in Figure B-10 in the Appendix

150

−1.

0−

0.5

0.0

0.5

1.0

Experimental Effects on Leader Influence Beliefs

coef

Whole Sample Untitled Titled

IdentityMaterial

Figure 5-2: Effects of experimental primes on perceived influence of indigenous lead-ers, disaggregated by titling status, shown with 95% confidence intervals.

perceptions of leader influence in the titled sample. However, as shown in 5-3, the

primes failed to increase the influence of leader endorsement on hypothetical vote

choice even in untitled communities where one of the priming treatments (the material

prime) did increase the perceived influence of leaders.21

Finally, to probe the external validity of the experimental findings, I descriptively

compare conjoint results in communities that are actually titled in the sample to

those that are not. Figure 5-4 shows the AMCEs of each binary attribute within

the titled and untitled subsets.22 Table B.29 in the Appendix shows coefficients

and standard errors for this comparison, interacting a binary barangay-level CADT

indicator with each candidate attribute. I include specifications with round fixed

21Results for titled barangays are shown in Figure B-11 in the Appendix. Neither priming treat-ment had a significant effect on leader influence in this subset either.

22Figure B-13 in the Appendix shows marginal means for the titled and untitled subsets.

151

●

●

●

●

●

−0.

2−

0.1

0.0

0.1

0.2

Experimental Effects (Control vs. Either Prime − Untitled Only)

AM

CE

gift promises mayor leader commwork

●

●

●

●●

●

●

ControlPrime

(a)

●

●

●

●

●

−0.

2−

0.1

0.0

0.1

0.2

Experimental Effects (All Prime Conditions − Untitled Only)

AM

CE

gift promises mayor leader commwork

● ControlIdentityMaterial

(b)

Figure 5-3: Experimental effects by candidate attribute, untitled barangays only.Average Marginal Component Effects (AMCE) for each binary attribute, comparingtitled and untitled barangays, shown with 95% confidence intervals. Standard errorsfor each treatment condition are clustered at the level of the respondent.

152

effects, barangay-level covariates, and restricting the analysis to respondents in the

experimental control condition.23 The influence of leader endorsement on vote choice

is higher on average in titled barangays ; however, the difference is not statistically

significant (𝑝 = 0.23).24 Consistent with the experimental findings, the effect of

specific campaign promises on vote choice is significantly greater in titled barangays

compared to untitled barangays (𝑝 = 0.025). In other words, respondents in titled

barangays are significantly more likely to prefer hypothetical candidates who make

specific promises to provide community-level public goods in the future. However,

inconsistent with the experimental results, they are on average less likely to punish

those who engage in vote-buying25 (𝑝 = 0.003) and less likely to choose candidates

endorsed by the municipal mayor (𝑝 = 0.104).

While not conclusive, this pattern is consistent with the idea that collective recog-

nition increases the priority voters place on certain indicators of prospective benefits

from the state. As discussed in Chapter 3, elites in titled communities expressed

higher levels of efficacy and greater confidence in the enforceability of prospective

agreements with the state and other outside actors. They linked this explicitly to

collective recognition. The fact that we find similar results in a survey of “ordinary”

community members suggests that greater efficacy vis-a-vis the formal political sys-

tem in titled communities may not be limited to elites.

5.3 Collective recognition beyond land titling

In this section, I examine the effects of another form of collective recognition for

indigenous communities specific to the context of local elections. As discussed in

Chapter 3, indigenous communities are frequently subjected to targeted harassment23Standard errors are clustered at the level of the barangay in all specifications.24𝑃 -values listed here refer to the model with the full sample including barangay-level covariates.25One possible explanation for this is that in communities where leaders are stronger, vote-buying

occurs in a more communal manner. In Chapter 3, I illustrate how leaders pool contributions fromcandidates to try to mitigate the influence of vote-buying on vote choice.

153

●

●

●

●

●

−0.

2−

0.1

0.0

0.1

0.2

Marginal Effects of Candidate Attributes by Titling Status (Barangay Level)

AM

CE

gift promises mayor leader commwork

●

●

●

●

●

●

●

UntitledTitled

Figure 5-4: Average effects of candidate attributes on vote choice in titled and untitledbarangays, shown with 95% confidence intervals. Standard errors are clustered at thelevel of the respondent.

154

and discrimination during elections, limiting their ability to use electoral participa-

tion as a means of exerting influence on local government. In this context, indige-

nous elites saw strong traditional institutions as important for countering harassment

and vote-buying and successfully delivering a community-level block vote. Collective

recognition by the state, including collective land titling, increased their influence

within their own communities and —to some extent —their ability to prevent inter-

ference by candidates and brokers. However, particular forms of harassment are still

difficult to counter. Even if all members of a community intend to vote for the same

candidate, the practice of hakot, in which politicians intercept indigenous voters and

offer them transportation and food on the condition that they vote for a particular

candidate, may prevent them from doing so in practice. Communities that are lo-

cated physically farther away from polling states are particularly vulnerable to this

practice.

In May 2016, the Philippines Commission on Elections (COMELEC) partnered

with the National Commission on Indigenous Peoples (NCIP) and the Legal Net-

work for Truthful Elections (LENTE) to pilot an initiative intended to address hakot

and increase meaningful electoral participation by indigenous voters.26 The initiative

involved two separate interventions: separate polling places (SPP) and accessible vot-

ing centers (AVC) for indigenous communities. SPPs are dedicated rooms in existing

polling places for indigenous voters, meant to protect them from discrimination and

harassment by non-indigenous community members at polling centers. AVCs are sep-

arate voting centers located closer to indigenous communities, intended to reduce the

time it takes voters to travel to cast their votes and, in doing so, to reduce instances

of hakot. In both cases, ballots are transported from voting centers the SPP/AVC

on election day, then collected and returned to the voting centers after the close of

voting hours to be scanned into voting machines. For the purposes of reporting vote

26LENTE is the NGO with whom I partnered to field the survey and conduct qualitative fieldwork.

155

totals, SPPs and AVCs are considered separate precincts.

The SPP/AVC pilot intervention provides an opportunity to study the effects of a

form of collective recognition other than land titling, again utilizing sub-national vari-

ation. While not explicitly part of government policies mandated by the IPRA, the

SPP/AVC intervention is consistent with the law’s broader goals, and shares theoret-

ically important similarities with other forms of collective recognition for indigenous

communities. In particular, it represents the integration of indigenous communities

into mainstream democratic politics as a collective and makes indigenous institutions

and indigenous leaders more salient and visible to political actors.

The consequences of this type of intervention for electoral competition are not

obvious a priori. Increased visibility afforded by the separate polling station might, on

the one hand, make it easier for politicians to monitor indigenous voters and for tribal

leaders to monitor votes of their community members. This, in turn, might make it

easier for politicians to enforce clientelistic bargains with leaders.27 On the other

hand, by increasing monitoring capabilities and protecting indigenous voters from

harassment, the intervention may instead improve indigenous communities’ abilities

to credibly bargain with candidates (and, in turn, the extent to which indigenous

communities and elites view electoral politics as a fruitful avenue for claim-making).

As an initial test of these competing predictions, I use data from the 2016 pilot

intervention to study the effects of creating separate voting facilities for indigenous

communities on electoral participation and competition. Specifically, I study the ef-

fects of the intervention on three outcomes: turnout, vote margin (competitiveness),

and incumbent vote share, using both cross-sectional and temporal variation. If this

form of collective recognition does in fact facilitate “elite capture” of indigenous voters,

we might expect it to reduce electoral competition and benefit incumbents, who are

on average likely to have more resources to buy the loyalty of tribal leadership.28 Ev-

27Larreguy (N.d.) finds evidence of this phenomenon in Mexico.28Partisanship in the Philippines is weak. Unlike in the Mexican (Larreguy N.d.) or South African

156

idence that it increases or has no effect on political competition, while not conclusive,

would weigh against this.

The SPP/AVC intervention was initially piloted in 18 barangays in Oriental Min-

doro and Occidental Mindoro during the May 2016 national and local elections.29 To

study the effects of the pilot on turnout and electoral competition, I use precinct-level

data on turnout and returns from three mayoral elections (2010, 2013, and 2016), ag-

gregated at the barangay level.30 I study aggregate effects on turnout and electoral

competition at the level of the barangay (rather than the precinct) because precincts

are generally not comparable over time. In addition, because SPP/AVC votes are re-

ported under a single precinct for each barangay in 2016, they are also not comparable

cross-sectionally between pilot and non-pilot barangays.

First, I descriptively compare barangay-level turnout, vote margin, and incumbent

vote share in the 2016 mayoral elections in the pilot barangays to a set of “control”

barangays matched on pre-treatment covariates, including total population, indige-

nous population; number of registered voters, turnout, and vote margin in 2010, the

presence of a CADT, and various geographic controls.31 Table 5.5 shows results from

cross-sectional comparisons of 2016 turnout, vote margin, and incumbent vote share

for pilot and matched control barangays.32 Because the elections in question are for

(De Kadt and Larreguy 2015) contexts, both characterized by dominant parties, it is difficult topredict which party would benefit from elite capture in local elections in the Philippines. Incumbencyis used as a proxy for control of the local political machine.

29The intervention was expanded during the May 2018 barangay elections, which were originallyscheduled for October 2017 but postponed following an order from President Rodrigo Duterte. Thesurvey was implemented jointly with LENTE and was intended in part to study the effects of thepilot intervention following the expansion. However, as of this writing I do not have access toCOMELEC data on electoral returns during the barangay election. Future analysis will incorporatethese data, as well as survey data regarding experiences during elections.

30Electoral returns data are not publicly available at the level of the barangay for 2010 and 2013.LENTE purchased the 2010 and 2013 data from COMELEC for select provinces for the purposeof this analysis. Data from 2010 and 2013 were manually compiled from scanned PDFs with thehelp of a research assistant. Incumbent status in 2013 and 2016 was determined by automatedname-matching and manual inspection for each municipality.

31Matches were selected using a genetic matching algorithm. Table B.30 in the Appendix showsbalance between the pilot and matched sample.

32Note that one matched pair is dropped due to missing outcome data in a particular municipality(Bansud, Oriental Mindoro).

157

municipal mayor, I include municipality fixed effects to account for race-level factors.

In columns 2, 4, and 6, I include values of these dependent variables from past elec-

tions.33 As shown in columns 1 and 2, there is no statistically significant difference in

barangay-level turnout in 2016 between pilot and control barangays. I do, however,

find evidence that barangay-level vote margins are lower in pilot barangays. In other

words, mayoral contests are more competitive on average in areas where indigenous

communities voted in dedicated voting centers compared to similar barangays where

they did not. This difference is statistically significant in the model controlling for

2010 and 2013 vote margin (𝑝 = 0.022).34 As shown in the last two columns, I find

no statistically significant difference in incumbent vote share. Incumbent mayors did

not appear to fare better in SPP/AVC pilot barangays than in the set of comparison

barangays.

Pilot barangays were selected following lengthy consultations between COMELEC,

NCIP, LENTE and various communities and assignment to the pilot treatment can-

not be considered as-if random. The cross-sectional comparisons, while they account

for observable differences between pilot and non-pilot barangays, should therefore be

interpreted primarily as descriptive. To gain some additional leverage on the causal

effects of the pilot on election results, I additionally estimate differences-in-differences

and two-way fixed effect specifications that account for time invariant unobservable

confounding. Tables 5.6 and 5.7 show results from differences-in-differences and two-

way fixed estimation, respectively. While the coefficients are not precisely estimated,

their size and direction are broadly consistent with the cross-sectional results. In

Table 5.7, I additionally interact the treatment indicator with indigenous population

at baseline. The data are not sufficiently disaggregated to allow a direct compari-

son between indigenous and non-indigenous precincts; however, we might expect the

33Incumbency status was determined through a combination of automated and manual name-matching of winners in the previous year to candidates in the current year. As a result, I can onlycalculate incumbent voteshare for 2013 and 2016.

34In the bivariate model the 𝑝-value is 0.12.

158

Table 5.5: SPP/AVC Pilot, Barangay Turnout, Margin, and Incumbent Voteshare(Cross-Sectional)

Dependent variable:

Turnout 2016 Margin 2016 Incumbent VS 2016

(1) (2) (3) (4) (5) (6)

Pilot Bgy 0.013 0.006 −0.096 −0.125** 0.002 −0.011(0.012) (0.012) (0.059) (0.048) (0.033) (0.037)

Turnout 2013 0.338*(0.176)

Turnout 2010 −0.160(0.280)

Margin 2013 −1.257***(0.410)

Margin 2010 0.505*(0.286)

Incumbent VS 2013 −0.399(0.455)

Mun FE Y Y Y Y Y YObservations 34 34 34 34 34 34R2 0.519 0.626 0.848 0.914 0.910 0.914Adjusted R2 0.007 0.117 0.687 0.798 0.814 0.811

Note: *p<0.1; **p<0.05; ***p<0.01

159

SPP/AVC intervention to have a larger effect on voting outcomes in areas with a

larger indigenous population (since a larger proportion of votes would have been cast

at the dedicated voting center). As shown in columns 4 and 6, I do find a statisti-

cally significant interaction effect indicating differentially greater negative effects on

vote margin and incumbent vote share (i.e. larger increases in competitiveness) in

barangays with larger indigenous populations.

Table 5.6: SPP/AVC Pilot, Barangay Turnout, Margin, and Incumbent Voteshare(Difference-in-Differences)

Dependent variable:

Turnout 2016 Margin 2016 Incumbent VS 20162010 base 2013 base 2010 base 2013 base 2013 base

(1) (2) (3) (4) (5)

Pilot Bgy 0.011 0.028 −0.004 −0.024 −0.034(0.013) (0.022) (0.095) (0.101) (0.058)

Post 0.043*** 0.025 0.161 0.152 −0.014(0.014) (0.016) (0.099) (0.117) (0.082)

Pilot Bgy x Post 0.002 −0.015 −0.092 −0.072 0.025(0.017) (0.022) (0.134) (0.162) (0.110)

Observations (Clusters) 34 34 34 34 34Adjusted R2 0.255 0.031 0.008 0.007 −0.043

Cluster RSE at bgy level *p<0.1; **p<0.05; ***p<0.01

While preliminary in nature, these results are not consistent with the idea that

collective recognition of indigenous communities in the context of elections facilitates

greater capture of these voters and entrenches powerful political elites. If anything,

the SPP/AVC intervention made local elections more competitive rather than less.

There does not appear to be any effect on turnout, potentially weighing against the

idea that the intervention made participation through formal channels more attrac-

tive. However, given that the status quo in many indigenous communities involved

coerced participation and that turnout is generally high,35 turnout may not be a par-

35The mean in the set of matched controls is 79.6%.

160

Table 5.7: SPP/AVC Pilot, Barangay Turnout, Margin, and Incumbent Voteshare(Two-way Fixed Effects)

Dependent variable:

Turnout 2016 Margin 2016 Incumbent VS 2016

(1) (2) (3) (4) (5) (6)

Pilot Bgy −0.006 −0.017 −0.082 0.197 0.025 0.247*(0.017) (0.021) (0.126) (0.161) (0.107) (0.146)

Pilot Bgy x Pct. Indigenous 0.028 −0.747*** −0.593**(0.047) (0.269) (0.277)

Observations (Clusters) 34 34 34 34 34 34R2 0.002 0.008 0.007 0.067 0.002 0.103

Clustered RSE at bgy level *p<0.1; **p<0.05; ***p<0.01

ticularly meaningful indicator of participation or the legitimacy of electoral politics.

Future extensions of this analysis will incorporate data on the second wave of the pilot

intervention implemented during barangay elections in 2018 (when data availability

allows).

5.4 Collective recognition and state public goods pro-

vision

In this chapter, I have presented various forms of evidence in support of my argu-

ment that collective recognition of historically marginalized communities increases

collective mobilization directed toward receiving public goods from the state. Using

cross-sectional variation in collective land titling, I find suggestive evidence in support

of the idea that this form of recognition can facilitate greater electoral cohesion at

the community level. Descriptive comparisons between titled and untitled communi-

ties in the survey data suggest that this may be due in part to the greater capacity

of indigenous leaders in titled areas to achieve coordination in candidate selection.

But while collective land titling is associated with greater electoral cohesion at the

161

precinct level and does not appear to disproportionately benefit incumbents, weighing

against arguments that increased bloc voting represents “elite capture” of indigenous

institutions. Similarly, preliminary analysis of a pilot intervention that facilitates col-

lective electoral participation among indigenous communities does not yield evidence

of elite capture or reduced competition. If anything, facilitating collective electoral

participation increased electoral competition at the local level.

The interpretation of these results is not necessarily straightforward, however. A

finding that collective recognition of indigenous communities systematically reduces

electoral competition and benefits incumbents would constitute fairly convincing ev-

idence of elite capture.36 However, a lack of such a finding is not dispositive: it does

not constitute definitive proof that elite capture is not occurring or that individuals

within indigenous communities are better off when they vote as a block. It could be

the case, for example, that collective recognition, by empowering indigenous leaders

to coerce votes in favor of their preferred candidates, makes the “indigenous vote” less

expensive overall for politicians (i.e. paying leaders may be less expensive than pay-

ing individual indigenous voters), making challengers more viable than they would be

otherwise. Results from the survey experiment, presented above weigh against this

idea. Specifically, I find that priming survey respondents with information about state

recognition of self-determination rights increased the importance respondents placed

on signals of future community-level benefits from the state. This result supports the

interpretation that collective recognition increases indigenous voters’ expectations

about the ability to make collective claims through the formal political system, or to

use elections to lobby collectively for community public goods.

Another, potentially more straightforward, observable implication of this interpre-

tation is that collective recognition should lead to greater local public goods provision

in indigenous communities. I examine this using sub-national variation in titling,

36And has been in other contexts; see, for example De Kadt and Larreguy (2015).

162

combined with data on the presence of barangay-level public goods captured in the

2000, 2007, and 2010 census waves. In each barangay, census enumerators coded the

presence of various public services and establishments. Each public goods variable

is a binary indicator coded at the barangay level. I use data for the presence of five

public services: a barangay health center, an elementary school, a community water

system, a street pattern, and an access road reaching a national highway.37

Figure 5-5 shows coefficients from two-way fixed effects regressions estimating the

relationship between land titling and the presence of various state-provided public

goods during the 2000, 2007, and 2010 census waves. Of the five public goods tested,

only access to a highway has a coefficient that is significant and positive within all

four subsets of the data. The estimated effect on the presence of highway access roads

is substantively large: being awarded a CADT between census waves is associated

with a between 5 and 10-percentage point increase in the probability of gaining access

to a highway. With the exception of the matched sample, all coefficients are statisti-

cally significant using a 𝑝-value threshold adjusted to account for multiple hypothesis

testing (𝑝 < 0.01).38

Parallel trends may be particularly implausible in the case of highway access.

Areas that are closer to a highway (and therefore potentially more likely to build

a highway access road) may be more likely to be granted a CADT, due to ease of

access. Indeed, as shown in Table B.31, a lagged dependent variable model estimating

the relationship between land titling status and highway access during the previous

period in the full sample of rural barangays seems to suggest that the causal arrow

may run in the opposite direction. However, I find no placebo effect using the other

37A street pattern is defined as “a system of at least three streets running in parallel or right angleorientation. The streets may either be paved with cement or asphalt, or unpaved provided it is wideenough for a four-wheel vehicle to pass through.” Highway access refers to the presence of an accessroad from the barangay to the national highway wide enough for a four wheel vehicle to pass through(Philippine Statistics Authority).

38Using the Bonferroni correction, which may be too stringent given that the different publicgoods tested are mostly like not independent (Anderson 2008).

163

●

●

●

●

●

−0.

15−

0.05

0.05

0.15

Land Titling (Continuous) and Public Goods All Rural

coef

BHS Elem Street Hway Water

●

●

●

●

●

−0.

15−

0.05

0.05

0.15

Land Titling (Continuous) and Public Goods Matched Sample

coef

BHS Elem Street Hway Water

●

● ●

●

●

−0.

15−

0.05

0.05

0.15

Land Titling (Continuous) and Public Goods Eligible Universe

coef

BHS Elem Street Hway Water

●

●

●

●

●

−0.

15−

0.05

0.05

0.15

Land Titling (Continuous) and Public Goods Titled

coef

BHS Elem Street Hway Water

Figure 5-5: Effects of titling on local public goods provision. Coefficients from a panelregression of state-provided public goods on land titling status in 2000, 2007, and 2010with 95% confidence intervals. Point estimates represent the percentage-point changein the probability that each type of public good is present in the barangay followingthe granting of a CADT over a portion of the barangay.

164

three subsets of the data. The results in these subsets remain robust when when

estimating differences-in-differences specifications within provinces and municipalities

and accounting for spatial dependencies. In addition, barangays in the matched subset

are balanced in terms of distance to the 1980 highway network.

It is not entirely clear why land titling is associated with the expansion of highway

access roads, but not with other public goods measured here. One possibility is that

several of the other goods considered here (i.e. schools and clinics) require action

and budgetary commitments from multiple national government agencies (e.g. the

Department of Health and the Department of Education), while highway access roads

may be provided through action by local governments alone, and are more subject

to local-level lobbying. Another possibility is that highway access roads are more

desired than other types of public goods. I find evidence of this in the survey: when

respondents were asked about public project priorities in the barangay, roads and

road improvements were the most likely to be mentioned.

5.4.1 Discussion and conclusion

Individually, each of the analyses presented in this chapter has important limita-

tions as tests of the theory of recognition as integration. Taken together, however,

they paint a tentatively positive story about the effects of indigenous recognition

on democratic participation. I find evidence that recognition, rather than entrench-

ing powerful political actors through capture of indigenous institutions in democratic

elections, can increase political competition and encourage indigenous communities to

direct electoral mobilization toward receiving community-level public goods from the

state. To use the language of Hirschman (1970), rather than encouraging “exit” from

the formal political system, the recognition of historically marginalized communities

can enable them to effectively exercise “voice” through the democratic process.

165

166

Chapter 6

Voice, not exit

In this project, I have applied an empirical political economy approach to a question

that has been the subject of active debate among policymakers and political philoso-

phers, but that to date has received relatively little attention within the mainstream

U.S.-based political science literature. The concept of indigenous rights poses a direct

challenge to the idea of the nation-state, an idea that underlies (implicitly or explic-

itly) much of this literature. Asking how recognizing these rights affects governance

in modern democratic states is not only important in its own right, but also provides

an opportunity to name and interrogate assumptions behind certain influential ideas

in the study of comparative politics.

I begin this concluding chapter by discussing the implications of this work for the

people it concerns most directly —indigenous communities in the Philippines —as

well as implications for the implementation of indigenous rights provisions beyond

this particular context. I then discuss broader implications for relevant literature in

political science. I conclude by proposing areas for future research.

167

6.1 Implications for indigenous rights in the Philip-

pines and beyond

The theory and empirical evidence presented in this project suggest that collective

recognition of indigenous communities is mutually beneficial to the state and the

communities concerned. The state gains legitimacy and cooperation, while indige-

nous communities gain substantive representation within the formal political system.

There remain, however, important questions about broader implications, and partic-

ularly about how these dynamics may evolve over time. I address two here. First,

are the legitimating effects of recognition likely to endure in the long term? Second,

if so, what are the implications for the advancement of indigenous self-determination

rights? Answering these questions empirically is beyond the scope of the present

project; however, a discussion of them is useful in both interpreting the present find-

ings and outlining areas for future work.

As in many states that have codified indigenous rights provisions into law, policies

recognizing collective self-governance rights in the Philippines have been implemented

relatively recently. As such, the evidence presented here can only speak to the effects

of recognition in the short- to medium-term. There are reasons to believe that the

legitimating effects of recognition may not persist over time. In particular, these

effects are likely to be contingent upon the state’s ability to absorb new demands

being made upon it (Huntington 1968). Short-term increases in state legitimacy may

be driven largely by expectations within communities about their ability to influence

the formal political system. Under some conditions, increasing citizens’ expectations

of government performance can enhance democratic accountability (Gottlieb 2016).

But if recognition has the effect of raising communities’ expectations beyond what the

state can reasonably deliver, the legitimating effects of recognition may not endure

(Postero 2006).

168

If, on the other hand, increased claim-making resulting from collective recognition

does lead to greater distribution of public resources to indigenous communities, it may

trigger greater inter-group competition over finite resources. It is not clear, however,

whether such competition would strengthen or weaken the state’s control. It may be

the case, as Charnysh (2019) argues, that inter-group competition creates demand

for state intervention and ultimately strengthens state capacity. This outcome may

be particularly likely if indigenous communities are, as a result of recognition, more

likely to express demands through formal channels. Another possibility is that in-

creased claim-making by indigenous communities will produce political backlash that

eventually leads to a roll-back of indigenous rights, particularly if these policies are

seen as restricting the ability of the government or private sector actors to develop

valuable natural resources.1 Such a move could, in turn, provoke resistance that is

more threatening to the state.2 Understanding the long-term implications fully will

require repeated analysis of these outcomes over time. Below, I suggest additional

strategies for examining these various possibilities in the shorter term.

Suppose the legitimating effects of collective recognition I have documented do

endure over the long term. What are the broader implications for indigenous self-

determination rights? The scope of inquiry in this project has largely been limited

to engaging claims in the social science literature about the likely effects of collec-

tive recognition on state consolidation and democratic representation. I have not

directly considered the effects of collective recognition in reference to the stated goal

of indigenous rights provisions; namely, the self-determination of indigenous peoples.

1Anecdotally, there is some evidence of this in the Philippines, where lawmakers in 2018 threat-ened to reduce the NCIP’s budget to 1,000 PHP (approximately 20 USD). While this proposalwas put forward by a representative of the leftist parylist Bayan Muna, supposedly in objection tokillings of indigenous leaders in Mindanao, it was supported by other members of Congress, many ofwhom have commercial interests in indigenous lands. Similarly, President Jair Bolsonaro of Brazilhas taken steps to roll back indigenous rights provisions, arguing that these laws are an impedimentto development.

2Finin (2005), for example, argues that attempted assimilationist policies by the American colo-nial administration in the Philippines lead to the formation of a regional indigenous identity in theCordillera region and created demands for regional autonomy.

169

Somewhat counterintuitively, my findings suggest that collective recognition, by en-

couraging greater engagement with formal or “mainstream” politics, may have effects

contrary to this goal.

In a sense, I have argued that collective recognition enables indigenous commu-

nities to act more effectively as “interest groups” within systems set up by the state.

This shift, in legitimizing the state, may have the effect of de-legitimizing indigenous

governance as an alternative to the state (Ferguson 2016). Corntassel and Witmer

(2008) observe this dynamic in the United States, arguing that the increased partici-

pation of Native American tribes in electoral politics (in the wake of decentralization

of power from the federal to the state level in the 1970s and 1980s) has undermined

the political and cultural foundations of their claims to sovereignty. As Anderson

(1987) argues in his discussion of marginal minorities in Southeast Asia, “the costs of

going ethnic, that is, participating in ethnic majority politics and economics within

the nation-state are not to be underestimated...Almost always it means the end of

cultural autonomy and self-contained integrity.” Yet, he goes on to suggest that do-

ing so “[opens] the way to developing a necessary political and economic bargaining

power. Perhaps this is, at the end of the twentieth century, the only way out.”

6.2 Implications for political science literature

The study of indigenous politics in U.S.-based political science is relatively nascent

(Ferguson 2016). However, indigenous claims speak to fundamental questions about

the nature of the state. In particular, these claims challenge the nation-state ideal

that is central to the dominant paradigm in the field (Shadian 2010). In examining

the effects of indigenous rights policies on state consolidation and democratic repre-

sentation, this project sits squarely within the existing paradigm. Yet in doing so it

speaks to and challenges the conventional wisdom on multiple distinct areas of study

170

within political science. I discuss four of these here: ethnic and identity politics, tra-

ditional authority, connections between state-building and democratic representation,

and group-based representation within liberal democracies.

6.2.1 Ethnic and identity politics

The idea that ethnic diversity is harmful to the functioning of modern societies is

so prevalent as to be considered “one of the most powerful hypotheses in political

economy” (Habyarimana et al. 2007). Social scientists have also recognized that ethnic

categories are constructed and reinforced by the state (Chandra 2006). What often

follows from these two propositions is the argument that state policies de-emphasizing

or attempting to erase ethnic differences lead to better outcomes than those that have

the potential to reinforce them (Miguel 2004, Miguel and Gugerty 2005, Lieberman

and Singh 2017). The findings presented in this project seem to challenge this idea:

I find evidence that collective recognition of indigenous communities by the state

simultaneously reinforces sub-national in-group identity and increases the legitimacy

of the state. Why do policies emphasizing diversity appear beneficial in this case and

harmful in others? As a study of a single policy area within a single country, this

project is limited in its ability to answer this question directly. However, I propose

two types of explanations, both with broader implications for the study of societal

diversity and state-building.

First, as discussed in Chapter 1, there are reasons to question the direction of

the causal relationship between ethnic diversity and state strength (Wimmer 2016).

Given that forging a unified national identity has been an explicit goal of many

modern states, the failure to do this (observed as ethnic diversity) may be a cause

rather than a consequence of state weakness. If this is the case, it does not necessar-

ily follow that all states will succeed in consolidating authority through assimilative

nation-building policies. I have argued that weak states attempting these policies

171

are likely to see effects in the opposite direction. This idea is supported by recent

work directly examining the effects of nation-building policies (Bandyopadhyay and

Green 2013, McGovern 2012). The main empirical contribution of this project is to

examine the effects of the “opposite” type of policy —collective recognition —within

a single country context. In doing so, it reinforces the broader point that counterfac-

tual thinking is critical in understanding the relationship between diversity and state

strength.

The second set of explanations relates to scope conditions. When considering the

effects of policies that reinforce sub-national identities (i.e. collective recognition),

two dimensions seem likely to matter: the type of group identity in question and the

type of policy toward that group. I do not have variation on either of these dimensions

in the current project that would allow for a direct examination of the importance of

these conditions. Instead, I briefly discuss each as a means of generating hypotheses

to be evaluated in future research.

Type of group identity. Perhaps the findings in this project, which seem contrary

to conventional wisdom in the ethnic politics literature, are explained by the focus on

indigenous groups specifically. Indigenous identity may be, for example, inherently

less threatening to the state or more compatible with a national identity than other

types of identity groupings. Indigenous identity is often categorized as a type of eth-

nic identity, by both governments and researchers,3 yet scholars focused specifically

on indigenous politics have often rejected this idea. One common distinction is that

in many contexts indigenous groups are not defined not by common “descent-based

attributes” (Chandra 2006) but instead by a common history of marginalization or

targeted oppression by the state (Jung 2008). Writing about groups that are now con-

sidered indigenous in the Southeast Asia context, Anderson (1987) notes that these

3In the Philippines context, for example, the state designates particular ethnic groups as “in-digenous” and “non-indigenous.” Cross national studies, for example Lieberman and Singh (2017)in their study of the effects of census classification on conflict, include indigenous classification as atype of ethnic classification.

172

groups are explicitly non-ethnic, in the sense that they historically lacked a cohe-

sive identity beyond that of their local community, but in some cases have taken on a

more broadly encompassing ethnic identity in response to state policies.4 Jung (2008),

focusing on the Latin American context, similarly documents how peasant commu-

nities took on an “ethnicized” indigenous identity precisely as a means of contesting

exclusion from the state.

What implications might this distinction have for the effects of group recognition?

One possibility is that the effects of groups recognition are contingent on size and

scale of groups in question. Compared to groups that are already politically powerful

and connected to national and regional politics —Anderson (1987) calls “coalitional

minorities” or what Cederman, Wimmer and Min (2010) call “politically relevant ”

ethnic groups —the recognition of collectives (e.g. tribes) that are relatively small in

scale may not have the effect of exacerbating ethnic competition over the distribution

of resources at a national or regional level. Similarly, while collective recognition of

larger, more geographically-concentrated groups (particularly when paired with the

granting of autonomy) may fuel or enable secessionist ambitions, collective recognition

of groups that are more geographically disbursed would seem less likely to do so (Dahl

1971).

A relevant question here is how the recognition of indigenous rights may affect

the process of ethnic identity formation: where a larger-scale identity group does not

already exist, does collective recognition of smaller groups make it more or less likely

to develop? If larger-scale regional or national indigenous identities have developed

partly in response to state policies of marginalization or attempted assimilation (An-

derson 1987, Finin 2005, Jung 2008), perhaps the recognition of smaller collectives

(even if it is justified with reference to “indigenous” as a larger category) can forestall

this process. However, such a result may also be contingent upon the degree of po-

4Scott (2010) similarly describes how populations of upland Southeast Asia have adopted “pliable”ethnic identities as a means of evading state power

173

litical decentralization within a polity. Centralized systems may require that groups,

once incorporated into the formal political system, organize themselves on a larger

scale to contest power at higher levels of government (Horowitz 1985). In decentral-

ized systems, by contrast, smaller groups may be able to make effective claims on

state resources and otherwise advance their interests without doing so.

Beyond indigenous communities, a similar argument may apply to “other runaway,

fugitive, and marooned” groups (Scott 2010) including, for example, nomadic com-

munities. While these types of groups may seem insignificant from the perspective

of a discipline focused largely on contestation within formal institutions (and often

in national or regional capitals), the presence of “non-state” spaces where many of

these groups reside has long been considered problematic for states facing, for exam-

ple, counterinsurgency or terrorist threats. In addition, as mentioned in Chapter 1,

groups identified as indigenous manage land covering a quarter of the earth’s land

area (Garnett et al. 2018). They may become even more politically salient with

global population growth and climate change driving migration into these lands and

increased demand for natural resources.

Type of policy. The recognition of sub-national group membership by states can

be undertaken for a range of purposes and accompanied by a range of group-specific

policies, from targeted discrimination or assimilation to affirmative action.5 To a

greater extent than for other types of sub-national groups (e.g. racial, ethnic, or

religious groups), state recognition of indigenous groups tends to be accompanied

by a particular policy “bundle” that includes rights to collective self-determination

and/or self-governance within a particular territory. Indeed, indigenous groups are

frequently distinguished from ethnic or racial groups precisely on the basis of claims

5In their cross-national study of the effects of ethnic census enumeration, Lieberman and Singh(2017) acknowledge that ethnic categorization may be positive under some circumstances, for exam-ple when coupled with efforts to redress ethnic inequality. They argue, however, that “on balance,ethnic enumeration on the census will tend to drive a deepening of ethnic identification, and morecompetition and potential conflict . . . ”

174

to self-governance rights. Ivison (2003) writes, for example, that “indigenous peoples

should not be seen as fundamentally racial entities, but rather as political or, perhaps

more precisely, as constitutional ones” (p. 334).6 Belonging entails membership in a

distinct political community where legitimacy is derived from sources other than the

state. Ethnicity, defined in terms of descent-based attributes, is neither necessary nor

sufficient.7

What does this imply for the generalizability of the argument and evidence pre-

sented in this project to other forms of group recognition? If (as suggested above) the

size and scale of many indigenous groups makes state recognition of these groups ap-

pear relatively less threatening to the project of state consolidation, policies uniquely

associated with indigenous groups arguably makes such recognition appear more

threatening, at least on its face. In many cases, state policies toward indigenous

communities entail a direct reduction in the state’s power to regulate social relations

and economic activity in parts of the territory it claims, powers which have been seen

as crucial in the consolidation of state authority (Migdal 1988).8

I have argued, however, that collective recognition of indigenous communities can

increase state legitimacy precisely because it also (in the Philippines context and in

many others) involves the conferral of self-governance rights. I do not necessarily

6Similarly, Ferguson (2016), writing on the American context, argues that reducing indigeneityto “mere ‘ethnicity” ’ ignores the “legal and historical collective identities” that American Indianshold.

7This distinction is usefully illustrated in the controversy over Senator Elizabeth Warren’s identi-fication as Native American. The Senator took a DNA test as a means of proving Native Americanheritage. In response, representatives of the Cherokee nation, with which she had associated herself,argued that tribal membership is not conferred on the basis of racial bloodlines. Linksey, Annie andAmy Gardner “Elizabeth Warren apologizes for calling herself Native American,” The WashingtonPost, February 5, 2019. https://www.washingtonpost.com/politics/elizabeth-warren-apologizes-for-calling-herself-native-american (Accessed September 2, 2020)

8The recent ruling of the United States Supreme Court in McGirt v. Oklahoma provides anothersalient example. The court’s ruling in favor of the Muscogee (Creek) Nation effectively limits thejurisdiction of state criminal courts in much of eastern Oklahoma. In the dissent, Chief Justice JohnRoberts wrote, “the state’s ability to prosecute serious crimes will be hobbled . . . On top of that, thecourt has profoundly destabilized the governance of eastern Oklahoma.” Healy, Jack and Adam Lip-tak “Landmark Supreme Court Ruling Affirms Native American Rights in Oklahoma,” The New YorkTimes, July 9, 2020. https://www.nytimes.com/2020/07/09/us/supreme-court-oklahoma-mcgirt-creek-nation.html (Accessed July 23, 2020)

175

expect the collective recognition of historically marginalized groups to have the same

legitimating effects when it is not accompanied by policies that 1) are compatible with

elite incentives and 2) increase these groups’ ability to make collective claims through

formal politics. In fact, I argue that assimilationist policies (another potential goal

of group classification) are likely to backfire and reduce state legitimacy when state

capacity is limited. Similarly, state policies that simply recognize the existence of

groups without corresponding policies granting autonomy or self-governance rights

may also fail to produce legitimating effects, particularly if they make between-group

inequality more apparent without simultaneously providing a means of addressing it.

On the other hand, as Jung (2008) argues, the recognition of marginalized group iden-

tities alone may have the effect of giving these groups “moral standing” that can be

leveraged for claim-making. If this is the case, collective recognition may have legiti-

mating effects even if it is not accompanied by self-governance rights. Disentangling

these two mechanisms is a potentially fruitful avenue for future research.

6.2.2 Traditional authority

In addition to the literature on ethnic and identity politics, this project speaks to the

growing literature on the role of traditional authorities in modern states (Holzinger,

Kern and Kromrey 2016). With some recent exceptions (Logan 2013, Baldwin 2015,

Van der Windt et al. 2018), the existence of these authorities is typically seen as an

obstacle to both state consolidation and democratic representation. Migdal (1988), for

example, considers them problematic because they can challenge state authority and

because their existence undermines the state’s claim to monopolistic power to regulate

behavior (Migdal 1988). Work on democratic representation, somewhat contrary

to this, has often characterized them as “decentralized despots” who help entrench

the power of national political elites while undermining the accountability function

of democratic elections (Mamdani 1996, Ntsebeza 2005) (Baldwin (2015) being an

176

important recent exception).9 As in the study of ethnic identity, it follows from these

perspectives that policies that sideline or undermine traditional authorities will lead

to better outcomes than policies that empower them.

The argument and findings put forward in this project similarly seem to challenge

this view. An analogous critique to arguments about ethnic identity can be made

regarding the causal direction between state strength and the strength of non-state

authorities. As several authors have noted, state policies toward traditional author-

ities are likely to be a function of their strength (relative to the state) at the time

of state formation (Herbst 2000, Boone 2003, Gerring et al. 2011, Acemoglu et al.

2014). Inverse relationships between the existence of non-state authorities and state

strength may be explained by this fact rather than negative effects of non-state au-

thorities on state consolidation. If this is the case, policies of collective recognition

(which legitimate and empower non-state authorities) may not weaken the state as

the literature seems to suggest.10 This project has the advantage of examining the

effects of such policies using variation within a single country case and with careful

attention to causal inference where possible. However, beyond questions of internal

validity, the findings may be specific to indigenous political institutions, as opposed

to other types of traditional or non-state authority structures, or to other features of

the context.

Are indigenous political institutions like those in the Philippines qualitatively dif-

ferent from other types of traditional authority? To date, much of the political science

literature on traditional authority in modern states is rooted in the African context, of-

ten referring to “traditional leaders” or “chiefs” (Holzinger, Kern and Kromrey 2016).11

9In their review of the literature, Holzinger, Kern and Kromrey (2016) make a helpful distinctionbetween neotraditional and neoliberal paradigms. The latter, which is more critical of traditionalauthority, has been (I argue) more dominant in contemporary political science based in the UnitedStates.

10Similarly, as I argue, attempts to eliminate them may not be successful when states are alreadyweak.

11According to a review by Holzinger, Kern and Kromrey (2016), “this body of scholarship concen-trates mostly on Africa and covers only about half of that continent. Whereas large parts of Africa

177

Holzinger et al. (2018) include these types of institutions in their cross-national study

of the constitutionalization of “indigenous provisions.” Similarly, Baldwin (2015) uses

a definition of “traditional leader” that is applicable to the African context and to

authority structures associated with indigenous communities outside of Africa.12 The

comparative study of traditional authority within political science is still relatively

nascent, but early efforts have generally included both types.

However, there are theoretical distinctions between traditional authorities typi-

cally studied in Africa and indigenous authorities in Asia, the Americas, and Aus-

tralasia, which may be important for understanding the effects of state recognition

and “polity dualism” more broadly (Kyed and Buur 2007). Indeed, the concepts of

traditional authority and indigenous rights are often treated as distinct within Africa.

The language of indigenous rights is not typically used by African governments and

international organizations to refer to traditional leaders or chiefs who exercise au-

thority over broad swathes of the rural population in many African countries. Instead,

it is used with reference to specific pastoralist or hunter-gatherer minority populations

whose “cultures and ways of life differ considerably from the dominant society and

[whose] cultures are under threat, in some cases to the extent of extinction” and who

are “subject to domination and exploitation within national political and economic

structures that are commonly designed to reflect the interests and activities of the

national majority” (ACHPR 2005, 89). As in the Philippines, the term “indigenous”

in the African context refers to historically marginalized groups who have their own

governing structures as a result of their exclusion from the mainstream political sys-

tem. By implication, populations governed by chiefs or traditional leaders represent

part of the mainstream.

are insufficiently studied, a few countries (in particular South Africa, Ghana, Namibia, Somaliland,or Uganda) attract most of the attention” (p. 477)

12Baldwin (2015) defines traditional leaders as “rulers who have power by virtue of their associationwith the customary mode of governing a place-based community” (p. 21). She intentionally uses“place” rather than territory to include nomadic populations.

178

This difference arguably has its roots in distinct colonial legacies. Traditional

authorities as typically understood in the African context (i.e. “chiefs”) were often

directly integrated into the colonial state as part of a strategy of indirect rule. In some

cases, they were essentially created from scratch to govern the periphery on behalf

of the colonial power. In others, the state directly delegated power to pre-existing

authorities (or some standardized version of them) to fulfill key state functions such as

the collection of taxes (Crowder 1964, Hobsbawm and Ranger 1992, Mamdani 1996).13

While these authorities may have derived some legitimacy from their association

with pre-colonial governing structures, they were in many cases primarily considered

agents of the state. As Lord Lugard, a key intellectual architect of indirect rule in the

British empire wrote: “the chief himself must understand that he has no right to place

and power unless he renders his proper services to the state” (Lugard 1965).14 By

contrast, the term “indigenous” as originally applied to populations in independent

post-colonial states referred specifically to groups that had not been integrated into

the political and economic system established by colonial powers (Rodríguez-Piñero

2006). If African chiefs are legacies of indirect colonial rule, indigenous authorities

may be considered legacies of incomplete colonial rule.

Why might this distinction matter for the study of traditional or non-state author-

ity (and more specifically for the implications of state recognition of this authority)?

I propose and briefly discuss two ways in which the type of authority (“chiefs” vs.

indigenous authorities) may condition the effects of recognition on state-building and

democratic representation. The first relates to the integrative effects of recognition.

I have argued that collective recognition of indigenous communities in the Philip-

pines strengthens the state by creating incentives for these communities to channel

13In Sierra Leone, for example, British colonial authorities maintained a list of families from whichchiefs could be selected. During the colonial period, chiefs were assigned to collect taxes Acemoglu,Reed and Robinson (2014).

14Quoted in Acemoglu, Reed and Robinson (2014). As Crowder (1964) notes, French colonialstrategies involved an even greater level of interference with local authority structures, while oftenmaintaining the rhetoric of traditional authority.

179

demands through the formal political system, as opposed to resisting or evading the

state. This argument may not apply in cases where traditional authorities already

serve to link rural populations to the central state (Chabal and Daloz 1999). In a

sense, these populations are already integrated into the formal political system, albeit

through state-sanctioned intermediaries. In such a case, policies strengthening these

intermediaries may be more likely to exacerbate the distribution of state resources

to the periphery along ethnic lines (Ekeh 1975) or to entrench the power of national

elites (De Kadt and Larreguy 2015).

The second relates to relationships of accountability between citizens and tradi-

tional authorities, and the potential for “elite capture” of indigenous authorities. The

characterization of traditional authorities as “decentralized despots” or clientelistic

vote brokers relies on the premise that these authorities are largely unaccountable to

the communities they represent. This may be true of traditional authorities in Africa

to a greater extent, given their historical role as agents of the colonial state. Mam-

dani (1996), for example, directly attributes the anti-democratic effects of traditional

authority to the legacy of indirect rule during the colonial period. Baldwin (2015)

disputes this account even in the African context, arguing that traditional leaders’

embeddedness in their communities and relatively long time horizons create incen-

tives to provide public goods. The decentralized despots argument may be even less

likely to apply to indigenous governing structures, whose leaders are not (and have

not historically been) appointed or subject to removal by the state. If anything, given

legacies of resistance to incorporation in many cases, indigenous leaders may be more

likely to see themselves as being in competition with the state, creating incentives to

act in their constituents’ best interests.15

A natural question to consider here is whether collective recognition of indige-

15Along these lines, Clayton, Noveck and Levi (2015) find evidence that competition betweentraditional elites and elected community officials is associated with increased public goods provisionin Sierra Leone, a context where traditional leaders were directly integrated into the colonial state.

180

nous institutions by independent states may have similar effects to indirect colonial

rule in the long run. Even if indigenous institutions are not “despotic” prior to state

recognition, state recognition may distort these institutions in a way that makes them

less accountable. At the same time, there are reasons to believe the effects of grant-

ing indigenous self-governance rights may be different than the effects of delegating

powers to traditional authorities under a strategy of indirect colonial rule. Under

indirect rule, traditional authorities were directly answerable to the central, most

notably for the payment of taxes, and in some cases could be removed or appointed

by the state. In the Philippines and in many other settings, collective recognition

of indigenous communities entails the state forgoing tax revenue for economic activ-

ities on indigenous land. In most cases indigenous recognition (given its emphasis

on self-determination) does not put the state in a position to interfere directly with

existing governing structures. On the other hand, the state may indirectly alter these

structures by making determination about who constitutes a “legitimate” authority

in legally binding decisions involving the tribe (Jefremovas and Perez 2011, Mamdani

2012).

The current project is limited in its ability to address these questions directly.

This initial exploration of potentially important distinctions between traditional in-

stitutions as typically studied in Africa (which have been the focus of much of the

political science literature on traditional authority to date) and indigenous institu-

tions highlights a need for further empirical and conceptual work to better understand

the conditions under which collective recognition of non-state governing structures is

likely to be harmful or helpful. As Holzinger, Kern and Kromrey (2016) argue, “tra-

ditional governance must be recognized for what it is: a variety of political systems

governing communities, which requires analysis in the same ways political scientists

have approached state institutions (475). Studying non-state governing structures

directly may be seen as particularly challenging given the discipline’s focus on state

181

institutions and limited engagement with alternative sources of legitimacy (Fergu-

son 2016), coupled with more logistical challenges such as limited data availability.

Focusing on different types of relationships between state and non-state governing

structures (both contemporary and historical) may provide a fruitful starting point

for the application of a comparative perspective to understanding the latter’s role in

contemporary governance.

6.2.3 State-building and democracy

In discussing potential scope conditions, I have generally suggested that this work

highlights a need for “precising” of concepts in the political science literatures on

ethnic politics and traditional authority (Collier and Levitsky 1997). However, this

project potentially also has broader implications (or at least raises broader questions)

for scholarship on state-building and democracy that may be applicable beyond this

particular issue area. In particular, in focusing on marginalized populations and

framing societal heterogeneity as partly a consequence of incomplete state consolida-

tion, it highlights connections between these literatures that can inform the study of

governance in post-colonial democracies.

Influential perspectives in the political science literatures on state-building and

democratic representation are rooted —even if implicitly —in a common paradigm

based on the nation-state ideal. Both see societal heterogeneity as problematic for

their respective dependent variables. In the former, societal heterogeneity com-

promises state legitimacy and hinders the state’s ability to implement development

projects and provide public goods. In the latter, it undermines the development of a

common civic culture and encourages distributive politics along group (e.g. ethnic)

lines. This, in turn, compromises the accountability function of democratic institu-

tions. Group rights, to the extent that they exacerbate societal heterogeneity, are

seen as problematic in both cases, if for slightly different reasons.

182

However, the question of how societal heterogeneity should be addressed (or how

the homogenization both approaches agree is preferable should be achieved) reveals

tension between them, including in terms of the (implicit) normative commitments un-

derlying each approach. Approaches in the state-building literature are often premised

on the existence of an autonomous developmental state (Weber 1919, Mann 1984,

Migdal (1988)). In this view, government performance is a function of capacity and

the presence or absence of obstacles to the implementation of development interven-

tions, including incompletely consolidated control over territory, a lack of compliance

by citizens, or the capture of the state apparatus by societal forces. This approach

would seem to suggest that policies to homogenize society and undermine competing

sources of authority, including coercive measures, are justified as long as they are

“successful” and do not produce a backlash against the state.

The democratic accountability approach offers less clear prescriptions. This ap-

proach sees the state as either rent-seeking or as an aggregation of interests in soci-

ety. Government performance is largely a function of institutional constraints placed

on rent-seeking behavior (World Bank 2003, Adserà, Boix and Payne 2003). Well-

functioning liberal democratic institutions, in which voters can sanction poor per-

formance and reward good performance, promote good governance (Besley 2006).

When society is homogenous, liberal democratic institutions function better, because

citizens vote on the basis of evaluation and policy preferences. But it is not clear

how states should achieve this state of affairs, given that assimilative policies, al-

most by definition, directly compromise individual freedoms and are unlikely to be

implemented by democratic institutions that are truly responsive to the interests of

minority groups.

One approach to reconciling the two perspectives theoretically is to recognize a

potential trade-off between state-building and democracy and argue for sequencing:

liberal democracy is only possible once a certain baseline level of homogenization and

183

centralization has been established (Huntington 1968). Another is to treat representa-

tive institutions as endogenous to state-building and highlight how they have emerged

historically from a process of bargaining between would-be rulers and those they are

attempting to rule (Bates and Lien 1985, Levi 1989, Tilly 1992). From a policy per-

spective, however, these arguments seem to have limited applicability in post-colonial

contexts. Due to the circumstances of their formation, post-colonial states enjoy de

jure sovereignty without having to first establish de facto legitimacy throughout their

territories (Herbst 2000, Chowdhury 2018). Many of these states adopted democratic

institutions over a relatively short period of time, influenced just as much by interna-

tional forces as domestic ones (Huntington 1993). Neither legitimacy nor democratic

institutions needed to be established through a process of negotiation or bargaining

with the population, at least not to the same degree as in Western Europe.

On the other hand, the focus in this project on marginalized populations —specifi-

cally those defined by their lack of incorporation into the dominant society —suggests

a way in which similar insights can apply even in post-colonial contexts. In pockets

where state authority is incompletely consolidated, states can still use the promise

of representation to boost legitimacy. They can accomplish this not by establishing

formally democratic institutions (which already exist in many cases), but by taking

measures that improve the ability of marginalized populations to make successful

claims within existing democratic institutions. By the same token, a failure to do

so not only undermines democratic representation but also compromises state legit-

imacy, particularly when communities within a state’s territory have realistic “exit

options” due to incomplete state consolidation (Scott 2010).

Relatedly, this project invites further exploration of the relationship between forms

of contestation that occur within state-sanctioned channels and forms that occur

in opposition to or outside of it. The literature focused on democratic representa-

tion tends to assume that elections are the most important form of contestation and

184

claim-making within democratic polities. “Anti-state” forms of contestation, such as

protest, civil conflict, non-compliance, and “everyday” forms of resistance are often

studied separately, particularly in single-country studies. This project highlights an

important way in which the two are related, by showing how prospects for successful

claim-making within the formal political system can affect decisions to engage with

it rather than resisting it. This connects to insights from work on the causes of civil

war, for example the idea that anocracies, partially democratic societies, are at par-

ticular risk because groups have freedom to organize but are not able to pursue their

goals effectively through formal political channels (Fearon and Laitin 2003, Hegre and

Sambanis 2006).

6.2.4 Collective participation, identity politics, and liberal democ-

racy

This project also speaks to larger themes regarding collective participation in liberal

democracies and the desirability of what is frequently referred to in both popular and

scholarly discourse as “identity politics” (Appiah 2007a, Lilla 2017, Fukuyama 2018).

While liberal democratic systems are rooted in principles of individual rights and

individual autonomy, participation in collectives —whether they take the form of

political parties, civil society organizations, unions, or social movements —is widely

recognized as one of the primary avenues through which individuals pursue their

interests in liberal democratic systems. Similarly, social scientists have identified

“social capital” as an important ingredient in making democracy work. A large body

of work has been devoted to understanding the determinants of social capital and

collective action within communities (Olson 2009, Putnam, Leonardi and Nanetti

1993, Boix and Posner 1998, Portes 1998, Edwards, Foley and Diani 2001). While

much of this literature focuses on the benefits of social capital for the provision of

public goods within communities, many have also linked it (implicitly or explicitly) to

185

collective political participation directed toward claim-making on the state (Nannicini

et al. 2013).

At the same time, particularly in the study of contexts in the Global South, col-

lective political participation organized around identity categories, such as ethnicity,

race, or religion is seen as antithetical to a well-functioning democracy.16 Collective

participation along identity lines —ethnic politics or identity politics —is considered

especially pernicious because identity-based cleavages are seen as particularly rigid

and conducive to zero-sum conflict (Horowitz 1985). In addition, the recognition that

many identities, including ethnicity, are constructed by the state (Hobsbawm and

Ranger 1992, Chandra 2006) makes these types of categories seem arbitrary relative

to, for example, categories based on ideology or class.17 If identity labels are both

uniquely problematic and uniquely arbitrary, it follows that de-emphasizing them

within the public sphere will lead to better outcomes.

Another school of thought suggests, however, that the constructed nature of

politically-salient identity categories makes them more, not less appropriate as an

organizing principle for collective political action (Nagel 1973, Thomson 1973, Fran-

cis 2018). In many contexts, ethnic and racial identities have great bearing on an

individual’s life chances, position in society, and ability to make claims through lib-

eral democratic institutions precisely because these categories have historically been

used by the state as a basis for differential treatment (Marx 1997, Katznelson 2006,

Weaver and Lerman 2010). These categories can matter long after the end of officially-

sanctioned institutional discrimination (Shapiro 2005, Bonilla-Silva 2013, Sharkey

2013). Arguments along these lines have gained prominence with respect to the issue

16At the same time, the discipline has also recognized that ethnic groups as institutions are oftenparticularly conduce to solving collective action problems (Miguel and Gugerty 2005, Habyarimanaet al. 2007). This tension is evident in recent work showing positive effects of exposure to civil warviolence on cooperation. This finding is often framed in terms of “positive” effects of war; howeverthe narrative is much different when it becomes apparent that these effects on cooperation onlyextend to in-group members (Bauer et al. 2016).

17Fukuyama (2018), for example, attributes attachments to identity categories to a psychologicalneed for “self-actualization” and “public recognition of...inner worth” (107).

186

of race in the United States and may seem self-evident to many American readers.18

Similar insights may be applied to the study of ethnicity, identity, and nationalism

within the comparative political science literature. While homogenous societies may

be more successful on many dimensions compared to heterogenous ones, ignoring

sub-national identity categories in the name of national unity does not make these

categories irrelevant for the de facto distribution of political power.

By the same token, official ignorance or de-emphasis of identity-based divisions

will not necessarily reduce the potential for identity-based conflict if members of

marginalized groups see calls for unity as an effort to legitimate a status quo in which

they are disadvantaged. The African Commission’s Working Group of Experts on

Indigenous Populations/Communities articulates this point in response to arguments

that calls for indigenous rights undermine national unity: “Multiculturalism is a living

reality. Giving recognition to all groups, respecting their differences...does not lead

to conflict...What rather creates conflict is that certain dominant groups force a sort

of unity that only reflects the perspectives and interests of certain powerful groups

within a given state, and which seeks to prevent weaker marginalized groups from

voicing their particular concerns and perspectives” (ACHPR 2005, 88). Ivison (2003)

similarly writes, with respect to indigenous peoples in North America, “given their

history of being subject to coercive assimilation by the state —often through the very

language of ‘equal citizenship rights’ —it is unreasonable to expect indigenous peoples

to see citizenship rights as providing, in themselves, an unproblematic framework for

a ‘common emancipatory project’.”

The recognition that membership in particular identity groups has material im-

plications for individuals within the group makes it possible to justify policies of

collective recognition even without abandoning (implicit or explicit) commitments to

18See, for example, arguments in favor of reparations for slavery (Coates 2014) and against “color-blindness” as an approach to combatting racial discrimination (Alexander 2020).

187

political liberalism (Ivison 2003, Seymour 2017).19 In particular, if collective recog-

nition makes it possible for individual members of historically marginalized groups

to achieve political equality in a way they would not otherwise be able to, it can

contribute to rather than undermine individual rights (Ivison 2003). Where a lack of

political equality and ongoing discrimination along identity lines contributes to the

illegitimacy of the state, it can also contribute to rather than undermine national

unity. Whether or not these prove true is an empirical question. In this project, I

have proposed specific mechanisms to support both of these claims, drawing on ex-

isting theories that recognize the importance of collective participation for interest

representation in democracies. So far, I have provided empirical evidence of these

mechanisms within a single country context. As I discuss in greater detail below,

future research will shed light on the conditions under which this holds, within and

beyond the Philippines.

6.3 Directions for future research

In this final section, I outline a number of specific directions for future research. First,

as discussed above, more work is needed to understand the long-term implications of

collective recognition for the state in the Philippines context. The legitimating effects

of recognition may not endure if the state is unable to meet demands increasingly

expressed through the formal political system. In other words, the extent to which

short-term support for the state is contingent or represents a “reservoir” upon which

the state can draw is not immediately apparent (Dahl 1971, Tyler 2006). A straight-

forward way to address this would be to examine longer-term trends in the types of

compliance outcomes in the current analysis, as data availability allows. Another is to

examine in the short- to medium term the effects of collective recognition on “costlier”19This is notably distinct from justifications of collective recognition and multiculturalism that

emphasize the inherent value of diversity and cultural preservation (see, for example Kymlicka(1996)).

188

forms of compliance, including compliance with public health directives or support

for non-state armed groups. Future research on in the Philippines could include col-

lecting data on the attitudes of non-indigenous populations and incorporating data

on inter-communal conflict specifically.

Second, it will be important to test this theory beyond the context in which it was

developed. I propose Brazil as a “theory-testing” case, due to the existence of a simi-

lar policy of indigenous recognition to that of the Philippines, similarly decentralized

governance, and the availability of fine-grained administrative data. The Brazilian

context provides several sources of additional variation that allow for testing of spe-

cific components of the theory. For example, the government’s policy includes the

establishment of indigenous schools, providing an opportunity to examine a form of

collective recognition that does not involve the granting of formal property rights.

Brazil’s federal structure and party system also create opportunities to leverage vari-

ation in political conditions, particularly the extent of clientelism.20 This will allow

me to examine more directly an implication of my theory: that collective recognition

increases the democratic representation of marginalized communities to a greater ex-

tent in areas where access to government services is more contingent upon connections

to established patronage networks.

Third, I hope to test an observable implication of my theory that is not tested

empirically in the current project: the idea that attempted assimilation policies lead

to resistance to and reduced cooperation with the state. Existing research has looked

cross-nationally at the effects of nation-building policies on conflict (Bandyopadhyay

and Green 2013), but has not focused specifically on compliance or political partic-

ipation outcomes. One potential opportunity to examine this is in the context of

the United States, focusing on the legacies of Termination Policy in the 1950s and

20Several studies have distinguished between “clientelistic” and “programmatic” parties in theBrazilian context. See, for example Hagopian, Gervasoni and Moraes (2008), Muniz (2015), Johan-nessen (2020)

189

1960s. During this period, the federal government terminated its government-to-

government relationships with more than 100 Indian tribes, affecting around 13,000

individuals (Ulrich 2013). Depending upon data availability, it may be possible to

study the effects of termination on outcomes such as census response, voter turnout,

and political representation. Another possibility (contingent upon data availability)

would be to systematically examine the legacies of assimilation on compliance with

public health directives in the midst of the Ebola crisis in West Africa. McGovern

(2012) documents in depth the Demystification Program which sought to assimilate

“forest” communities in post-colonial Guinea. During the Ebola crisis, McGovern

(2016) suggested that communities in the forest region of Guinea were substantially

more mistrustful of government compared to similar communities across the border

in neighboring Sierra Leone and Liberia.

A fourth research direction would involve further unpacking the causes of col-

lective recognition, both cross-nationally and at the sub-national level. In focusing

on its consequences, I have set this question aside and attempted to isolate sources

of exogenous variation rather than centering the selection process in the theory or

analysis.21 While the theory suggests that states benefit from recognition, it is not

necessarily the case that they implement it in a strategic manner. However, to the

extent that they do, it implies that they may opt for recognition in, for example,

particularly restive areas. Holzinger et al. (2018) provide suggestive evidence of this

in their cross-national analysis of the constitutionalization of indigenous provisions,

finding that countries with a legacy of conflict are more likely to recognize indigenous

group rights. This analysis could be extended to include statutory provisions and

to incorporate over-time variation in policy adoption. In both the Philippines and

Brazil, additional historical data could be incorporate into a more careful analysis of

21In addition to shedding light on states’ motivations, theorizing and studying the causes ofrecognition more carefully will also help address remaining concerns about the identification of thecausal effects of recognition.

190

the predictors of sub-national recognition.

191

192

Appendix A

Survey and Interview Details

A.1 Survey Methodology

A.1.1 Overview

In partnership with Legal Network for Truthful Elections (LENTE), a Philippines-

based NGO, I conducted a survey among indigenous communities in the provinces

of Oriental Mindoro, Occidental Mindoro, and Palawan. The survey was funded in

large part by a grant from the MIT Governance Lab (MIT GOV/LAB).

The purpose of the survey was twofold: first, we sought to gain a better un-

derstanding of the effects of collective land titling, using survey experiments and

descriptive comparisons between titled and untitled communities in the sample. Sec-

ond, we sought to evaluate a pilot intervention jointly implemented by LENTE, the

Philippines’ Commission on Elections (COMELEC), and the National Commission

on Indigenous Peoples (NCIP) to establish separate polling places for indigenous com-

munities. It was initially implemented in 20 barangays during the May 2016 general

election. At the time the survey sample was drawn, an additional 15 barangays had

been selected for inclusion during the barangay elections (originally scheduled for

October 2017 and postponed to May 2018). The pilot intervention is described in

193

greater detail in Chapter 5.

A.1.2 Sample selection

The survey targeted 100 barangays (villages) across the three provinces. A-1 shows the

provinces and barangays included in the target sample. Barangays were selected for

inclusion using a multi-step procedure. First, the 35 pilot barangays were included.

Then, an additional 35 were selected as matched controls to the pilot barangays

using pre-treatment covariates. The additional 30 barangays were selected to achieve

variation in titling (CADT) status within the sample. In all cases, LENTE and

its local collaborators verified the presence of known indigenous communities in the

barangay.

In each barangay, enumerators were assigned to interview five (5) randomly-

selected community members in each of two sitios (neighborhoods), yielding a total

target sample of 1,000 community respondents.1 Because no household listing were

available, households were selected using a random walk procedure. Respondents

were randomly selected within households from among eligible adults who planned to

sleep in the house that night.2 The survey additionally targeted indigenous leaders

in each community3 and, where applicable, Indigenous Peoples Mandatory Represen-

tatives (IPMRs), individuals selected by indigenous communities to represent them

in elected barangay councils.1Community respondents had to be over the age of 18 and identify as a member of an indigenous

community, but could neither hold any leadership position in their tribe nor share a household withsomeone in a leadership role. In cases where there was only one sitio in a sample barangay with anindigenous population, 10 respondents were interviewed in that sitio.

2Specifically, survey teams are instructed to choose a central point in the sitio and walk in oppositedirections, skipping two houses and visiting the third. If no eligible adult is available at the selectedhousehold, enumerators are instructed to move to the next immediately adjacent household in thesame direction. If an adult is home, the enumerator asks to list the names of all adults (aged 18and over) in the household in alphabetical order. Then the household head is asked select a numbercard from a hat and the enumerator will interview the person whose name in the alphabetical listingcorresponds to the selected number.

3In Oriental and Occidental Mindoro, the survey targets the indigenous leader in each sitio,yielding two leader interviews per barangay. In Palawan, most tribal organizations only extend tothe barangay level, so only one leader interview is targeted per barangay.

194

(a) The Philippines, with target provinces indicated.

ORIENTAL MINDORO

OCCIDENTAL MINDORO

PALAWAN

surveyednot surveyed

(b) Barangays targeted for survey data collection

Figure A-1: Target survey sample195

A.1.3 Implementation

The survey was administered between May and September 2018.4 All interviews were

conducted face-to-face and administered on Android tablets using SurveyCTO.

Approximately 30 survey enumerators were recruited by LENTE and two partner

organizations: the Mangyan Mission (in Oriental Mindoro and Occidental Mindoro)

and NATRIPAL (in Palawan). All enumerators were members of indigenous com-

munities in their respective regions. Hiring enumerators from the community was

considered necessary for reasons of access and language. Wherever possible, enu-

merators avoided surveying in their home barangays. Quality control was ensured

using a combination of automated and manual analysis of incoming data (including

examination of short voice recordings of enumerators), callbacks to respondents, and

in-person shadowing by LENTE staff.

Due to security concerns, survey teams were ultimately unable to visit 20 of the

100 target barangays.5 The final sample therefore includes data from 80 barangays.

A.1 compares characteristics of the barangays that were included and excluded. Ta-

ble A.2 includes demographic summary statistics for the final sample of community

respondents.

A.1.4 Ethics approval and permissions

The survey received exempt approval from MIT’s Committee on the Use of Human

and Experimental Subjects (COUHES) under protocol # 1709098195. In addition,

in accordance with the Indigenous Peoples Rights Act, permission was obtained from

the National Commission on Indigenous Peoples. Permission was also sought from

tribal leadership in each community before surveyors began their work.

4Survey enumeration initially began in Palawan and was delayed in Mindoro due to a particularlysevere typhoon season that lead to delays in training and deployment of enumerators.

5The survey period coincided with military operations against the New People’s Army in twomunicipalities, and teams were not permitted to visit these areas.

196

Table A.1: Covariate Balance (Targeted vs. Actual Sample)

Included Mean Excluded Mean T Pvalcadt_bin2018 0.364 0.429 0.603

cadt_prop2016 0.206 0.299 0.274logpop_2010 7.735 7.755 0.908

ip_2010 0.392 0.355 0.631ip_2000 0.337 0.315 0.763

ethfrac_2000 0.517 0.536 0.741area_sqkm 52.739 65.029 0.602elev_mean 205.046 146.658 0.109

elev_sd 50.391 41.186 0.318slope_mean 7.212 5.532 0.087

soil_index 1.928 2.031 0.092log_ncip_dist 12.602 12.237 0.000

log_coast_dist 8.780 8.728 0.854log_road_dist 2.878 1.508 0.108

religion_catholic_2000 0.692 0.769 0.152strong_mat_roof_2000 0.290 0.318 0.590

strong_mat_outerwalls_2000 0.284 0.336 0.387own_lot_2000 0.672 0.721 0.459

have_street_pattern_2000 1.909 1.524 0.094have_highway_2000 1.273 1.286 0.931

have_church_2000 1.078 1.143 0.447have_market_2000 1.675 1.667 0.942

have_elem_2000 1.065 1.048 0.756have_bhs_2000 1.299 1.333 0.771

have_water_sys_2000 1.779 1.429 0.007leg_whipple_2000 125.888 130.721 0.559

birth_reg_2000 0.719 0.817 0.059margin_bgy2016 0.301 0.457 0.057

Table A.2: Sample Demographics

variable mean1 age 40.2882 female 0.5633 completed_elementary 0.3964 completed_hs 0.1645 catholic 0.4266 evangelical 0.2457 born_bgy 0.763

197

A.2 Elite interview guide

Leadership position and governance structures

∙ What is your position in your community (i.e. your IP leadership position)?

How would you describe your duties and responsibilities in this position?

∙ How long have you held this position?

∙ How did you come to hold this position? Were you, for example, chosen due

to heredity, elected by members of your community, or chosen by a group of

community elders?

∙ Is this the manner in which people in your position have been chosen historical?

If not, what was the process historically?

∙ Compared to previous generations, do you feel that IP leaders have more influ-

ence in this community than they used to or less influence?

General conditions in the community

∙ What would you say are some of the most important sources of strength in your

community?

∙ What are some of the most important challenges facing your community right

now?

∙ Would you say your community is similar in these ways to other IP communities

in this province, or different? If you think your community is different, how so?

Relations between indigenous communities and the government

∙ How would you describe your relationship with the barangay and municipal offi-

cials in your area? Is it generally friendly and constructive, generally combative,

or somewhere in between?

198

∙ When was the last time you had a conversation with a municipal government

official in your capacity as an IP leader? What was that conversation about?

∙ Has a government official at any level ever asked you to mobilize the community

for any purpose? If so, what for? Has any government official or agency tried

to mobilize your community directly for any purpose?

∙ What are your priorities for municipal government spending (for example, on

community projects)? Do you think the municipal government shares your

priorities?

∙ Do you feel that the municipal and provincial governments care about the well-

being of IPs in this area? Do you think most people in your community feel the

same way?

∙ In general, do you feel that the central government cares about the well-being

of IPs in the Philippines? Do you think most people in your community feel

the same way?

Experience with elections

∙ During the most recent elections in 2016, were you approached by any candidate

and asked to mobilize your community in support of him or her? If yes, how

did you respond?

∙ Has this happened in previous elections?

∙ In 2016, did your community experience any harassment before or during the

election? If yes, can you describe what happened? Has his happened in previous

elections?

∙ In 2016, did your community experience any discrimination before or during

the election? If yes, can you describe what happened? Has this happened in

199

previous elections?

∙ Would you say that most members of your community usually support the same

candidate during elections, or is there a lot of disagreement? Questions about

land titles

For communities that have received land titles

∙ When did your community first apply for a Certificate of Ancestral Domain

Title (CADT)?

∙ When did your community receive your CADT?

∙ What were your original goals in applying for a CADT? What did you think

would change (or would not change) as a result of holding a CADT?

∙ I understand that applying for a CADT can be a costly process. Why did your

community decide it was worth the time and expense?

∙ How was the decision to apply for a land title originally made? Prompts:

– Did the original idea come from community leadership? From others in

the community?

– Were you encouraged to apply by the NCIP, a local government official, a

company, or anyone else outside the community?

∙ Can you please walk me through the process your community went through to

obtain the CADT, step-by-step, to the best of your knowledge and recollection?

∙ Did you have an external assistance (financial, technical or otherwise) during

the application process?

∙ Was everyone in the community involved? If so, in what parts of the process?

200

∙ What were some of the major challenges your community faced during this

process? How did you overcome them?

∙ Do you think things would be different in your community today if you had not

received a CADT? If so, how?

For communities that are in the process of applying for a land title

∙ When did your community first apply for a Certificate of Ancestral Domain

Title (CADT)?

∙ What were your original goals in applying for a CADT? What do you think will

change (or will not change) if and when you obtain your CADT?

∙ I understand that applying for a CADT can be a costly process. Why did your

community decide it was worth the time and expense?

∙ How was the decision to apply for a land title originally made? Prompts:

– Did the idea come from community leadership? From others in the com-

munity?

– Were you encouraged to apply by the NCIP, a local government official, a

company, or anyone else outside the community?

∙ Can you please walk me through the process your community has gone through

to obtain the CADT, step-by-step, to the best of your knowledge and recollec-

tion?

∙ To the best of your understanding, what are the steps that still need to be

completed before your community obtains a CADT?

∙ What are some of the major challenges your community has faced so far during

this process? Have you been able to overcome any of these? If so, how?

201

∙ What challenges are you still facing as part of this process, if any?

∙ Are you optimistic that you will be successful in obtaining a CADT within a

reasonable timeframe? Why or why not?

∙ Do you think things would be different in your community today if you had not

gone through the process of apply for a CADT? If so, how?

For communities that have opted not to apply for a land title

∙ As you know, many IP communities have applied for ancestral domain titles

(CADTs) from the government. I understand that your community has not

done so. Why would you say that is?

∙ Does your community have any plans to apply for a land title? If so, when do

you think this will happen?

∙ If your community has no current plans to apply for a land title, is there any-

thing that would make you more likely to do so? If so, what could happen that

would prompt the community to apply for a land title?

Free Prior and Informed Consent (FPIC) Processes

∙ Are there or have there been any major development projects (i.e. mining,

hydropower) happening in your community?

∙ Did you participate in an FPIC process related to any development projects

(or a similar process that involved negotiations between the community and the

project proponent)?

∙ If yes, what was the outcome of this process? For example, was the proponent

required to meet any conditions in order to move forward with the project in

your area? If so, what were these conditions?

∙ Who in the community was involved in the process?

202

Appendix B

Empirical Appendix

B.1 Supplementary material for Chapter 4

B.1.1 Summary statistics for observational analysis

Table B.1 compares barangays titled as of 2018 to those that are untitled but are

part of the eligible universe designated by the National Commission on Indigenous

Peoples (NCIP). The list of “identified” areas with no approved CADT or on-process

application was compiled manually by the author based on submissions from each

NCIP regional office. Table B.3 presents summary statistics for the two outcome

variables and titling status for the universe of barangays included in the study. This

includes all rural barangays except for those in the Cordillera Administrative Region

was granted a degree of autonomy as a majority indigenous region.1

1I exclude barangays in this region throughout the analyses in this paper. However, most resultsare robust to their inclusion.

203

Table B.1: Pre-Treatment Covariate Balance Among Eligible Barangays (Titled in2018 vs. Untitled in 2018)

Mean Titled Mean Untitled T p-valIndigenous Prop. 2000 0.322 0.173 0.000

Birth Registration 2000 0.784 0.851 0.000Log. Population 2000 7.096 6.860 0.000

Ethnic Frac. 2000 0.629 0.759 0.000Area (sq. km) 26.337 11.351 0.000

Elevation Mean 124.399 132.501 0.136Elevation St. Dev 31.283 40.691 0.000

Slope Mean 3.862 4.902 0.000Soil Quality Index 2.070 2.119 0.000

Log. NCIP Dist. 10.500 10.605 0.059Log. Coast Dist. 9.492 8.882 0.000Log. Road Dist. 3.940 3.495 0.000

Catholic Prop. 2000 0.645 0.744 0.000Roof Strong Materials Prop. 2000 0.510 0.548 0.000Wall Strong Materials Prop. 2000 0.597 0.505 0.000

Own Lot Prop. 2000 0.661 0.599 0.000Street Pattern 2000 1.737 1.707 0.186

Highway Access 2000 1.398 1.303 0.000Church 2000 1.107 1.126 0.054Market 2000 1.815 1.808 0.706

Elementary 2000 1.132 1.220 0.000Bgy. Health Ctr 2000 1.309 1.374 0.000

Water System 2000 1.544 1.530 0.368Legibility (Whipple) 2000 117.666 112.266 0.000

B.1.2 Identity robustness checks

This section presents robustness checks for the indigenous identity results, presented

in Table 4.1 in the main document. Table B.4 replicates the main analyses using the

binary land titling measure. While the coefficient estimate falls just below statistical

significance at the 𝛼 = 0.05 level in the titled subset (𝑝 = 0.056), the substantive con-

clusions remain largely unchanged. Tables B.5, B.6, and B.7 show results for models

estimating the effect of land titling on three different placebo outcomes: total pop-

ulation, migration (operationalized using the percentage of individuals in a barangay

who have resided in the same municipality for the past five years), and migration

204

Table B.2: Pre-Treatment Covariate Balance Among Titled Barangays (Titled Before2010 vs. Titled After 2010)

mean.Tr mean.Co T pvallogpop_2000 7.075 7.186 0.017

ip_2000 0.332 0.392 0.018ethfrac_2000 0.631 0.640 0.544

area_sqkm 27.707 23.790 0.083elev_mean 135.759 104.545 0.001

elev_sd 34.944 28.107 0.041slope_mean 4.196 3.698 0.081

soil_index 2.086 2.101 0.412log_ncip_dist 10.444 10.812 0.006

log_coast_dist 9.540 9.648 0.180log_road_dist 4.090 4.198 0.675

religion_catholic_2000 0.640 0.626 0.459strong_mat_roof_2000 0.538 0.472 0.001

strong_mat_outerwalls_2000 0.578 0.488 0.000own_lot_2000 0.680 0.674 0.724

have_street_pattern2000 0.305 0.283 0.466have_highway2000 0.622 0.655 0.293

have_church2000 0.876 0.911 0.070have_market2000 0.219 0.207 0.648

have_elem2000 0.880 0.859 0.342have_bhs2000 0.703 0.697 0.863

have_water_sys2000 0.452 0.454 0.960leg_whipple_2000 124.137 126.080 0.180

birth_reg_2000 0.790 0.725 0.000

among indigenous communities specifically. These analyses are meant to address

the concern that the change in indigenous identity is due to changes in population

composition While titling does appear to have significant effects in some subsets of

the data, they do not point in a consistent direction, suggesting they are unlikely to

fully explain the main effects. Table B.8 replicates the main analysis controlling for

migration among indigenous communities specifically and the results remain largely

unchanged. Table B.9 estimates the differences-in-differences specification, interact-

ing the titling variable with province fixed effects. These models require only that

parallel trends hold within provinces. The coefficients fall below conventional levels

205

Table B.3: Summary Statistics (Study Barangays)

Num. Vals Num. NA Min Max Median Mean Std. DevTitled Proportion 2010 35110.00 0.00 0.00 1.00 0.00 0.03 0.14

Titled Binary 2010 35110.00 0.00 0.00 1.00 0.00 0.03 0.18Titled Binary 2018 35110.00 0.00 0.00 1.00 0.00 0.04 0.20

Eligible Universe 35110.00 0.00 0.00 1.00 0.00 0.21 0.41Indigenous Pop. 2000 35069.00 41.00 0.00 1.00 0.00 0.08 0.23Indigenous Pop. 2010 31991.00 3119.00 0.00 1.00 0.00 0.11 0.26

Birth Registration 2000 35069.00 41.00 0.00 1.00 0.95 0.83 0.26Birth Registration 2007 34940.00 170.00 0.00 1.00 0.98 0.89 0.22Birth Registration 2010 31991.00 3119.00 0.00 1.00 0.99 0.89 0.22

of statistical significance in some subsets; again, however, the substantive conclusions

remain largely unchanged. Finally, Table B.10 replicates the analysis using Conley-

type Heteroskedasticity-Autocorrelation-Robust (HAC) standard errors that account

for spatial and temporal autocorrelation. With the exception of the titled subset, all

coefficients remain statistically significant.

Table B.4: Land Titling (Binary) and Indigenous Identification (Differences-in-Differences)

Dependent variable:

Indigenous Prop.All Rural All Rural Matched Matched Eligible Eligible Titled Titled

(1) (2) (3) (4) (5) (6) (7) (8)

Titled 0.081*** 0.080*** 0.034*** 0.034*** 0.039*** 0.038*** 0.026* 0.025*(0.007) (0.007) (0.009) (0.009) (0.007) (0.007) (0.014) (0.014)

Same Municipality −0.001 −0.030 −0.040 −0.015(0.013) (0.077) (0.050) (0.076)

Log. Population 0.012*** 0.010 0.007 0.011(0.004) (0.018) (0.011) (0.028)

Mean Age 0.003*** 0.001 0.002 −0.010**(0.0005) (0.003) (0.001) (0.005)

Observations 33530 33530 1674 1674 7023 7023 1430 1430𝑅2 0.01 0.012 0.009 0.01 0.004 0.004 0.003 0.007

CRSE at bgy level *p<0.1; **p<0.05; ***p<0.01

206

Table B.5: Land Titling and Population Change 2000-2010 (Differences-in-Differences)

Dependent variable:

Log. PopulationAll Rural Matched Eligible Titled All Rural Matched Eligible Titled

(1) (2) (3) (4) (5) (6) (7) (8)

Titled Prop. 0.033*** −0.014 0.016 −0.059***(0.011) (0.014) (0.012) (0.015)

Titled 0.026*** 0.021** 0.021*** −0.041***(0.007) (0.010) (0.008) (0.016)

Observations 33530 1674 7023 1430 33530 1674 7023 1430𝑅2 0 0.001 0 0.011 0 0.002 0.001 0.005

CRSE at bgy level *p<0.1; **p<0.05; ***p<0.01

Table B.6: Land Titling and Recent Migration 2000-2010 (Differences-in-Differences)

Dependent variable:

Same MunicipalityAll Rural Matched Eligible Titled All Rural Matched Eligible Titled

(1) (2) (3) (4) (5) (6) (7) (8)

Titled Prop. 0.004 0.001 0.001 0.0001(0.003) (0.003) (0.003) (0.004)

Titled −0.002 −0.006** −0.004** −0.012***(0.002) (0.003) (0.002) (0.004)

Observations 33530 1674 7023 1430 33530 1674 7023 1430𝑅2 0 0 0 0 0 0.004 0.001 0.007

CRSE at bgy level *p<0.1; **p<0.05; ***p<0.01

207

Table B.7: Land Titling and Migration Among Indigenous Peoples 2000-2010(Differences-in-Differences)

Dependent variable:

Same Municipality (Indigenous)All Rural Matched Eligible Titled All Rural Matched Eligible Titled

(1) (2) (3) (4) (5) (6) (7) (8)

Titled Prop. −0.025*** 0.002 −0.017** −0.003(0.007) (0.009) (0.007) (0.010)

Titled −0.025*** 0.001 −0.016** −0.013(0.006) (0.009) (0.006) (0.011)

Observations 23663.5 1574 6037 1368 23663.5 1574 6037 1368𝑅2 0 0 0.001 0 0 0 0.001 0.001

CRSE at bgy level *p<0.1; **p<0.05; ***p<0.01

Table B.8: Land Titling (Continuous) and Indigenous Identification (Differences-in-Differences) - IP Migration Controls

Dependent variable:

Indigenous Prop.All Rural Matched Eligible Titled

(1) (2) (3) (4)

Titled Prop. 0.089*** 0.041*** 0.047*** 0.023(0.009) (0.012) (0.010) (0.014)

Same Municipality (Indigenous) 0.030*** 0.135*** 0.130*** 0.137***(0.004) (0.040) (0.019) (0.050)

Log. Population −0.002 0.004 −0.005 −0.007(0.007) (0.018) (0.013) (0.026)

Mean Age 0.003*** −0.004 0.001 −0.013***(0.001) (0.004) (0.002) (0.005)

Mean HH Size 0.002 −0.010 0.006 −0.011(0.002) (0.010) (0.006) (0.013)

Observations 23663.5 1574 6037 1368𝑅2 0.013 0.027 0.02 0.019

CRSE at bgy level *p<0.1; **p<0.05; ***p<0.01

208

Table B.9: Land Titling and Indigenous Identification (Differences-in-Differences)

Dependent variable:

Indigenous Prop.All Rural Matched Eligible Titled All Rural Matched Eligible Titled

(1) (2) (3) (4) (5) (6) (7) (8)

Titled Prop. 0.289*** 0.158 0.364*** −0.090*(0.026) (0.109) (0.046) (0.053)

Titled 0.351*** 0.010 0.426*** 0.084(0.018) (0.051) (0.030) (0.174)

Prov FE Y Y Y Y Y Y Y YObservations 31,955 1,674 6,535 1,335 31,955 1,674 6,535 1,335Adjusted R2 0.191 0.214 0.208 0.293 0.197 0.244 0.226 0.310

Note: *p<0.1; **p<0.05; ***p<0.01

Table B.10: Land Titling and Indigenous Identification (Conley SE)

Dependent variable:

Indigenous Prop.All Rural Matched Eligible Titled

(1) (2) (3) (4)

Titled Prop. 0.104*** 0.048** 0.053*** 0.027(0.016) (0.019) (0.009) (0.023)

Observations 33530 1674 7023 1430𝑅2 0.918 0.938 0.902 0.923

Note: *p<0.1; **p<0.05; ***p<0.01

209

B.1.3 Compliance robustness checks

This section presents robustness checks and additional analyses for the birth registra-

tion results, presented in Table 4.2 in the main document. I focus on birth registration

as opposed to legibility, given that the main results for legibility do not hold across

specifications in the main analysis; however, I present some results for this outcome as

well. Table B.11 implements the main results using the binary land titling indicator.

While the coefficient remains positive, it drops below conventional levels of statistical

significance in most models. This may be due in part to the fact that birth registra-

tion is measured as an aggregate outcome at the barangay level. If increases in birth

registration are driven by changes in behavior among indigenous residents living in

an Ancestral Domain, effects may not be as apparent in areas where the Ancestral

Domain covers a relatively small part of the barangay . Indeed, as shown in Table

B.12, the effects of titling on birth registration are significant greater in areas with

larger indigenous communities at baseline. This also holds for the legibility outcome

(Table B.15).

One potential concern in the interpretation of the interaction effects is that they

may reflect the fact that barangays with larger indigenous population had lower base-

line levels of birth registration and therefore more room to increase. However, as Fig-

ure B-1 demonstrates, there is common support in baseline birth registration across

levels of indigenous population. I also replicate the main analyses and interaction

models eliminating all barangays with baseline birth registration levels greater than

90% and using a dichotomous version of the indigenous population variable. As shown

in Tables B.13 and B.14, the results remain largely unchanged using these alternative

specifications.

Table B.16 estimates the effects of titling on birth registration in the previous

period. This tests whether titling in one period predicts birth registration rates in

the previous period, which would suggest a violation of the parallel trends assumption.

210

While I do observe a lagged effect in the full rural sample, I do not see any such effect

in the other three subsets. As shown in Table B.17, I observe similar patterns for the

legibility outcome. Table B.18 estimates the differences-in-differences specification

for birth registration, interacting the titling variable with province fixed effects. The

results remain consistent in direction in all but the matched subset. As another

indirect test of the parallel trends assumption, Figure B-2 shows birth registration

rates by age in 2000, prior to the issuance of any Ancestral Domain titles for various

subsets of the data. If the practice of registering newborn births varied over time,

this should be reflected in differential rates of registration by age. In the absence of

data measured pre-treatment this provides a rough proxy for registration over time.

For example, if an area was “integrating” more rapidly over time, we might expect

to see a steeper slope in the relationship between age and registration in that area

relative to others. As shown in the figure, “trends” in registration rates by age do not

obviously differ between those barangays with titles in 2010 and those without.

Finally, as with the identity results, I re-estimate the effects of titling on birth

registration using Conley HAC standard errors. As shown in table B.19, results for

all four subsets of the data remain statistically significant.

211

Table B.11: Land Titling (Binary) and Birth Registration (Two-Way FE)

Dependent variable:

Birth RegistrationAll Rural All Rural Matched Matched Eligible Eligible Titled Titled

(1) (2) (3) (4) (5) (6) (7) (8)

Titled Prop. 0.013*** 0.012*** 0.004 0.003 0.007* 0.007 0.004 0.004(0.004) (0.004) (0.005) (0.006) (0.004) (0.004) (0.007) (0.007)

Bgy. Health Center 0.002* 0.014*** 0.008*** 0.016**(0.001) (0.005) (0.003) (0.006)

Street Pattern −0.003*** −0.006 −0.002 −0.006(0.001) (0.004) (0.002) (0.005)

Highway Access 0.001 0.008 0.008*** 0.016***(0.001) (0.005) (0.003) (0.006)

Log. Population 0.016*** −0.024 0.004 −0.023(0.005) (0.019) (0.010) (0.025)

Mean Age −0.001** −0.0003 0.0005 −0.006*(0.001) (0.003) (0.002) (0.004)

Mean HH Size −0.001 0.008 0.012*** 0.009(0.001) (0.005) (0.003) (0.006)

Observations 34000 33979 1665 1665 7146 7146 1454 1454𝑅2 0 0.001 0 0.006 0 0.004 0 0.011

CRSE at bgy level *p<0.1; **p<0.05; ***p<0.01

212

Table B.12: Land Titling and Birth Registration (Two-Way FE), IP Pop Interaction

Dependent variable:

Birth RegistrationAll Rural All Rural Matched Matched Eligible Eligible Titled Titled

(1) (2) (3) (4) (5) (6) (7) (8)

Titled Prop. x IP 2000 0.130*** 0.115*** 0.112*** 0.106***(0.019) (0.024) (0.020) (0.020)

Titled Prop. −0.018*** −0.027*** −0.020*** −0.023***(0.006) (0.008) (0.007) (0.008)

Titled x IP 2000 0.101*** 0.094*** 0.095*** 0.093***(0.015) (0.017) (0.015) (0.015)

Titled −0.018*** −0.026*** −0.022*** −0.025***(0.004) (0.005) (0.004) (0.007)

Bgy. Health Center 0.002 0.002 0.012** 0.012** 0.008*** 0.007*** 0.015** 0.014**(0.001) (0.001) (0.005) (0.005) (0.003) (0.003) (0.006) (0.006)

Street Pattern −0.003*** −0.003*** −0.005 −0.005 −0.001 −0.001 −0.005 −0.004(0.001) (0.001) (0.004) (0.004) (0.002) (0.002) (0.005) (0.005)

Highway Access 0.001 0.001 0.008 0.008 0.007** 0.007*** 0.014** 0.014**(0.001) (0.001) (0.005) (0.005) (0.003) (0.003) (0.006) (0.006)

Log. Pop 0.017*** 0.017*** −0.019 −0.019 0.005 0.005 −0.014 −0.015(0.005) (0.005) (0.018) (0.018) (0.010) (0.010) (0.024) (0.024)

Mean Age −0.001** −0.001** 0.0003 0.0001 0.0004 0.0004 −0.006* −0.006*(0.001) (0.001) (0.003) (0.003) (0.002) (0.002) (0.004) (0.004)

Mean HH Size −0.001 −0.001 0.008 0.008 0.011*** 0.011*** 0.008 0.008(0.001) (0.001) (0.005) (0.005) (0.003) (0.003) (0.006) (0.006)

Observations 33954 33954 1665 1665 7127 7127 1451 1451𝑅2 0.003 0.002 0.02 0.022 0.009 0.009 0.026 0.03

CRSE at bgy level *p<0.1; **p<0.05; ***p<0.01

213

Table B.13: Land Titling (Continuous) and Birth Registration (Two-Way FE), IPMajority Interaction

Dependent variable:

Birth RegistrationAll Rural Matched Eligible Titled

(1) (2) (3) (4)

Titled Prop. x Indig. Majority 2000 0.095*** 0.087*** 0.082*** 0.079***(0.015) (0.017) (0.015) (0.015)

Titled Prop. −0.002 −0.013** −0.005 −0.009(0.005) (0.007) (0.006) (0.008)

Bgy. Health Center 0.002 0.012** 0.008*** 0.016**(0.001) (0.005) (0.003) (0.006)

Street Pattern −0.003*** −0.005 −0.001 −0.005(0.001) (0.004) (0.002) (0.005)

Highway Access 0.001 0.008 0.007** 0.014**(0.001) (0.005) (0.003) (0.006)

Log. Pop 0.017*** −0.019 0.005 −0.015(0.005) (0.018) (0.010) (0.024)

Mean Age −0.001** 0.0005 0.0004 −0.006*(0.001) (0.003) (0.002) (0.004)

Mean HH Size −0.001 0.008 0.011*** 0.008(0.001) (0.005) (0.003) (0.006)

Observations 33954 1665 7127 1451𝑅2 0.003 0.021 0.008 0.026

CRSE at bgy level *p<0.1; **p<0.05; ***p<0.01

214

Table B.14: Land Titling (Continuous) and Birth Registration (Two-Way FE) - Ex-cluding 90% Baseline Reg.

Dependent variable:

Birth RegistrationAll Rural All Rural Matched Matched Eligible Eligible Titled Titled

(1) (2) (3) (4) (5) (6) (7) (8)

Titled Prop. 0.018* −0.051*** 0.029** −0.040** 0.031*** −0.031** 0.036*** −0.024(0.009) (0.011) (0.013) (0.015) (0.010) (0.013) (0.013) (0.015)

Titled Prop. x IP 2000 0.149*** 0.142*** 0.132*** 0.125***(0.027) (0.034) (0.028) (0.029)

Observations 11349 11323 720 720 3210 3191 755 752𝑅2 0 0.001 0.004 0.019 0.001 0.005 0.005 0.019

CRSE at bgy level *p<0.1; **p<0.05; ***p<0.01

●

●

●

●●

●

●●●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●●

●

●

●

●●

●

●

●●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●●●●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●●

●●

●

●

●

●

●

●

●

●

●

●●

●●

●

●●●●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●●

●

●

●

●●

●

●

●

●

●

●

●●●●●

●

●●●

●

●

●

●●

●

●●●

●

●●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●●

●●●

●

●●●

●

●

●

●●●●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●●●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●●●●

●●

●

●

●●●

●●●

●

●

●

●

●

●

●●●●●●

●●

●

●

●

●

●●●●●

●

●

●●

●

●●●

●

●

●

●

●

●

●

●

●●

●

●●●●

●

●

●

●

●

●

●

●

●

●

●●

●

●●

●●

●

●●

●

●

●●●

●

●●

●

●

●

●

●

●

●

●●

●●●

●

●

●

●

●

●

●

●

●

●●

●

●

● ●●

●

●

●●

●

●

●●

●

●

●

● ●

●

●

●

●●

●

●

●

●

●

●

●

●

●●●

●

●

●

●

●

●

●

●

●

●

●●●

●

●

●●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●●

●

●

●

●●

●●

●●

●

●●

●

●

●●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●●

●

●●

●

●● ●

●● ●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●●

●

●●●

●●

●

●

●

●

●●

●●●

●

●

●

●

●●

●

●●

●

● ●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●●

●

●

●

●

●

●

●●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●●

●●

●●●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●●

●

●

●

●●●

●

●●

●

●

●●

●

●●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●●●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●●

●●

●

●●

●

●●

●

●●

●

●

● ●●

●

●

●

●

●●

●

●●

●

●

●

●

●

●

●●●

●●●

●●

●

●

●●●

●

●

●

●

●●●

●●●

●

●●●●

●

●●

●

●

●

●

●

●●

●

●

●

●

●

●●

●

●

●

●

●

●●●●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●●

●●

●

● ●

●

●●

● ● ●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●●

●

●●

●

●

●●

●

●

●

●●

●

●

●

●

●

●

●

●

● ●

●

●

●

●● ●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●●●

●

●●●●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●●

●

●●

●● ●●

●

●

●

●

●

●

●

●

●

●

●●●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●●

●

●

●

●

●

●●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●●

●

●

●

●

●●●

●

●

●

●

●

●

●●

●

●

●●

●●

●

●

●●●

●

●

●

●

●

●

●●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

● ●●

●

●

●

●

●

●

●●● ●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●●●

●

●● ●

●●●

●●●

● ●

●

●

●

●

●

●

●

● ●●

●

●

●

● ●

●

●

●●

●

● ●

●

●

●

●●

● ●●

●

●●●

●

●

●●

●●●

●

●

●

●●

●

●●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●●

● ●

●

●●

●

●

●●

●●●

●

●

●

●

●

●

●

●●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●●

●

●

●

●●

●

●

●

●

●

●

●●

●

●

●

● ●

●

●

●

●

●●

●

●

●●

●●

●

●●

●

●

●

●

●

●

● ●

●

●

●●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●●

●

●●●

●

●

●

●● ●

●

●

● ● ● ●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●●

●

●

●

●

● ●● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

● ●●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●●

●●

●●

●

●

●●

●

●

●●●

●●

●

●

●

● ●

●

●

●

● ●

●

●

●

●●●

●

●

●

●

● ●●

●

●

●

●

●

●

●

●

●●●

●

●●

●●

●

●

● ●

●●●

●

●●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●●

●●●●

●●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●●

●

●●

●

●●

●

●

●

●

●

●

●●

●

● ●

●

●

●

●

●

●

●●

●

●

●●●

●

●

●

●

●

●

●●

●

●●

●

●

●

● ●

●

●

●●

●●

● ●●●

●

●

●

●

● ●

●●●

●

●●

●

●

●●

●

●

●

●

●

●

●

●

●●

●

●●

●

●

●●

●

●

●

●

●●

●

●

●

●●●

●●

●

●●

●

●

●

●

●

●●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●●●

●

●

●

●●

●

●

●●

●

●

●

●●●

●

●● ●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●●●●

●

●●●●●●●

●

●

●

●

●

●●●

●●

●●●●●

● ● ●● ●●

● ●●●●● ●

●

●●

● ●

●

●

●

●

●

●● ●● ●

●

●

●

●

●

●●

●

●

●●

●

●●

●

●●●

●

●

●

●●

●

●●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

● ●●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●●

●

●

●

●

●

●

●

●●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●●

●

●●●●

●●

●● ●

●

●

●

●●

●

●

●

● ●●●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

● ●

●●

●

●

●

●

●

●●

●

● ●

●●

●

●● ●

●

●

●

●

●●

●

●

●

●

●●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●●●

●

●

●

●

●

●

●●

●

●●

●

●

●

● ●

●

●

●

●

●

●

●

●●

●

●

●●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●●

●

●

●

●

●●

●

●

●

●

●

●

●

●● ●●●

●

●●

●

●●

●

●

●

●

●

●●

●

●●●●●

●

●

●

●

●●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●●

●

●

●

●●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●●

●

● ●

●

●

●●

●

●

●●

●

●

●

●

●●

●●●

●

●●

●

●

●

●

●●

●

●●

●

●

●●

●●

●

●●

●

●

●

●

●

●

●●

●

●

●

●●

●

●

●

●

●●●

●●

●

●●

●

●

●

●

●

●

●●

●

●

●

●●

●●

●

●

●

●● ●

●●

● ●

●

●

●●●

●●

●●●

●●

●

●

●

●

●●●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●● ●

●

●

●

●

●

●●

●

● ●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●●●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●●

●

● ●●●●

●

●

●●

●

●●

●

●

●

●

●●

●

●● ●

●

●

●●

●

●

●

●

● ●●

●

●

●

●

●

● ●

●

●

●

●

●

●●

●

●

●

●

●●

●

●●

●

●●

●●

●

●

●

●

●

●

●

●

●

●●

●●

●

●

●

●

●

●

●●

●

●

●●

●

●

●

●

●

●● ●

●

●

●

●●

●

●

●

●

●

●●

●

●

●●●●

●●

●●

●

●●

●

●

●

●

●

●

●

●●

●●

●

●

●

●

●

●●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●●

●

●

●●●

●

●

●

●

●

●●●

●●

●●

●

●

●

●

●●

●

●

●

●

●

●

●●

●

●

●

●

●●

●

●

●●●

●

●

●●●● ●

●

●●●●●●

●

●●

●

●

●

●●●●●●●●●●

●

●●●

●

●

●

●

●

●

●●

●●

●

●●

●

●●●

●

●

●

●

●

●

●

●

●

●

●●●●

●

●●

●

●

●

●

●

●●●●

●

●

●

●●

● ●

●

●

●

● ●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●●

●●

●

●

●

●●

● ●

●

●

●

●

●

●

●

●

●●

●

●

●●

●

●●

●

●

●

●

●

● ●

●

●

● ●●

●●

●

●

●

●

●

●

●

●

●

●

●●

●●

●

●

●●

●

●●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●● ●

●

●

●

●

●

●

●●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●●

●

● ●

●●

●●

●

●

●

●

●●

●

●

●●●

●

●

●

●

●

●

●

●

●

●●

●●

●

●

●

●

●●

●●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

● ●

●

●

●

●

●

●

●

●

●

●

●

●●●

●

●

● ●

●

●

●

●

●

●●

●●

● ●

●

●●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●●●●

●

●

●

●

●

●

●

●●

●●

●

●

●

●

●

●

●

●● ●●●

●

●

●

●

●●

●

●

●

●●●

●●

●

●

●

●●●

●

●

●

●

●

●●

●

●

●●● ●

●

●

●●

● ●

●●

●

●

●

●

● ●

●

●

●

●

●●

●

●

●●

●

●

●

●

● ●

●

●

● ●

●

●

●

●

●●

●

●●

●●● ●●

●●

●

●●

●

●

●●●

●●●●●

●●

●

●

●

●

●

●

●●

●●●●

●

●

●

●

●

●●

●●●●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●●

●

●●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●●

● ●● ●

●

●●

●

●●

●●●

●

●

●● ●

●

●

●

●

●

●

●

●

●●

●

●●

●

●

●

●●

●

●

●

●

●●

●

●

●●

●●

●

●

●

●

●●

●

●

●

●

●

●

●

●●

●

●

●

●●

●

●

●

●

●

●●

●●

●

●●

●●●

●

●

●●●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●●

●

●

●

●

●●

●

●

●

●●

●●●

●●

●

●

●●

●

●

●●

●

●●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●●

●

●

●

●●●

●

●●

●●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●●

●

●

●

●

●

●●

●

●

●●

●

●

●●●

●●

●

●

●●

●

●

●●

●●●

●●

●●●

●●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●●

●

●●●

●

●●●

●●

●●●●

●

●

●●●●

●

●●●

●

●●●●●●●●●●●●●●● ●●●●●●●●

●

●

●

●●●●●●

●

●●●●●●●

●●●

●

●●

●

●●

●

●●●●●●●●●

●●●●

●●●

●●●

●●

●

●●●

●

●

●●

●

●

●

●

●

●●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●●●●●

●

●

●

●

●●

●

●

●

●

●●

●

●

●

●

●

●

●●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●●

●

●

●

●

●

●●

●

●

●●●

●

●

●

●

●●

●

●●

●

●●

●

●

●

●●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●●●

●

●

●

●

●●

●●●

●●●●●

●

●

●

●

●●

●

●●

●●

●

●

●

●

●

●

●

●

●●

●●

●●●●

●

●●

●

●

●

●●

●

●●

●●

●●●

●

●●

●

●

●

●

●●●●

●

●●

●●●

●

●

●

●●

●

●●●

●

●●

●●●●

●

●●●●●

●●●●

●●●

●●●●●● ●●

●●●

●

●

●

●

●●

●●●

●

●

●

●●

●

●

●●

●●●●●

●

●●●●

●●

●

●●●●●

●

●●

●●●●●

●

●●●●●●●●

●

●●●●●● ●●●

●

●●

●●●

●●●

●

●●●●●

●

●●●●

●

●

●

●●

●●●●

●

●●●●●

●●

● ●●●●●

●●●●●

●

●

●

●●●

●

●●

●

●

●●●

●

●

●●

●

●●

●

●

●

●

●●

●●●

●●

●●

●

●

●

●

●●●●●●

●

●●●

●

●

●●

●

●

●

●

●●●

●

●●

●●

●

●●●

●

●

●

●●●

●

●●●●

●●

●

●

●

●

●

●●

●●●

●●●

●

●●

●

●

●

●

●

●●●

●

●●●●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●●●

●

●

●●

●●

●

●

●

●

●

●

●●

●

●

●●

●●●

●

●

●

●●

●

●

●

●●●●●

●

●●

●●●●●●●

●

●●

●

●

●●●●●●

●

●●●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●●

●●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●●

●

●

●

●

●●●

●

●

●

●

●●●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●●●

●

●

●●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●●

●

●

●

●●●●

●

●●

●

●

●

●

●

●

●●

●●●●

●

●

●●●

●

●●●●

●●●●

●

●

●

●

●

●

●

●

●●

●●●●●

●●

●

●

●

●●●

●●●

●

●

●●●

●

●

●

●●●●

●●

●●

●

●

●

●

●●●●

●●●●

●

●

●●

●●

●

●

●

●

●

●

●

●

●

●

●●

●●●

●

●

●

●

●●

●

●

●

●●●

●

●

●

●●●

●●

●●

●●●

●●●

●

●●●●●

●

●

●

●●●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●●

●●

●

●

●

●

●●●●

●●●

●

●

●

●

●

●●●

●

●●

●●

●

●● ●

●

●●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●●

●●●

●

●

●

●

●

●

●

●

●

●●

●●

●

●●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●●●

●●●●

●

●●●●●

●

●

●●

●

●●●

●

●

●

●

●

●

●●

●

●●

●●

●●●●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●●●● ●

●

●●●●●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●●●

●

●

●

●

●

●

●●

●

●

●

●●●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●●●●

●

●●●

●

●

●

●

● ●●●

●●●● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●●●●

●●●●● ●

●

●

●

●●●

●

●

●

●

●●

●●●

●●

●

●

●●

●

●●

●●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●●●

●

●●●●●

●●

●

●●●●

●

●●

●●●

●●●

●

●

●●●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●●●●

●●

●●●

●●●

●

●●●

●

●●●●●●

●●

●

●●

●

●

●

●●

●●●●

●●

●

●●●

●

●

●

●

●●

●●●

●

●●●● ●●●

●

●●●●●

●

●●●●●●

●

●●

●

●●

●

●

●

●●●

●●

●●●

●

●

●

●

●

●●

●

●

●●●

●

●

●

●

● ●

●●

●●●●●●●

●

●●

● ●

●

●

●

●

●

●

●●

●●

●●

●

●

●

● ●

●●

●

●

●

●●

●

● ●

●

●●●

●●

●●●

●

●

●●●●

●

●

●

●●

●

●●

●

●●●●

●

●

●●●● ●

●

●●

●

●●

●●

●●

●

●

●

● ●

● ●

●

●

●

●

●●

●●●

●

●

●●

●

●

●

●●●

●

●

●●

●

●

●

●

●

● ●

●

●

●

●

●

●●

●

●

●

●●

● ●●

●

●

●●

●

●

●

●

●

●●●

●

●

●●●

●● ●●

●

●●●

●

●●●

●

●

●●

●

●●

●

●

●

● ●

●●●

●

●

●

●

●

●

●●●

●●●

●●●●●●

●

●

●

●

●

●

●●

●●

●

●●

●●●

●

●

●●●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●●

●●● ●●

●●

●

●

●

● ●●

●

●●

●●

●

●

●●

●

●●●

●

●

●

●

●

●

●

● ●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●● ●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●●

●

●

●

●

●●

●●

●

●

●

●

●●

●●

●

●●

●●●●

● ●

●●

●

●●●

●

●

●

●

●

●●

●●

●

●●●

●●●●●●●●●●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●●

●●

●

●

●●

●

● ●●●●●●●

●●●●

●

●●●●●●

●●●●

●●●●●

●●● ●●

●

●

●

●●●●●

●

●●●

●●●●●

●● ●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●●●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●●●

●

●

●

●●●

●●

●●

● ●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●●●

●

●

●

●●●●

●

●

●

●●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●●●

●

●

●

●

●●●

●●

● ●

●●

●

●

●●

●

●●●

●●●

●

●

●

●

●● ●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●●● ●●● ●●

● ●●

●

●

●

● ● ●

●

●

●

●●●

●

●

●●

●

●

●

●

●

●

●

● ●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●●●●

●

●

●

●●

●●●

●

●●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●●

●

●●●

●

●●●

●

● ●●● ●

●

●

●

●

●

● ●

●

●● ●

●

●●

● ●

●●●

●●●

●

●●

●●

●●●

●●●●●

●

●●●

●●●●

●

●●

●● ●

● ● ●●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●●●

●

●●

●

●

●

●

●

●

●

●●●

●

●● ●●

●

●

●

●●

●

●

●

●

●

● ● ●●

●

●

●

●●

●●

●

●

●

●

●

●● ●●●●

●

●● ●●●●

●

●●●●

●●

●

● ●●● ●

●●

●

●

●

●

●●

●●

●●●

●

●

● ● ●

●

● ●

●

● ●

●

●

● ● ●

●●

●

●● ●●

●● ●

●

●●●

●

●

●●

●

●

●

●●●

●

● ●

●

●

●

●●

●● ●● ●● ●●●

●●●

●● ●●

●

● ● ●● ● ●●●●

●●

●● ●●

●

● ●

●

●●

●●●●●

●●

●

●●●●

●

●

●●

●

●

●●

●

● ●●●

●●●●●●●

●

●● ●

●●

●

●●

●

●

●●●

●● ●●●

●●●

●

● ●●● ●● ●

●●

●

●

●●

●

●● ● ●●● ●●●

●●

●

●

● ● ●●

●●

●

●●●●●

●

●●

●

●

●

●

●●

●

●●● ●

●

●● ●

●●

●●●

●

●

●●●

●●●● ● ●

●

●●● ●

●

●●●

●

●

● ●●

●

● ●●

●●●●

●●

●●

●●●

●

●●●

●●

●●

●●●

●● ●

●●●●

●

●

●●●●●

●

●

●

●● ●

●

●●

●●

●

●●●

● ●

●

● ●

●

●

●

●

●●●

●

●●

●

●

●

●

●

●

●●

●

● ●

●

●

●

●

●● ●●

●

●

●

●

●

●●

●

●

●●

●

●

●

●●

●

●

●

●

●

●●

●●

●

●

●

●●●● ●●

●

●

●●●

●●●●

●

●

●●●●●

●●●●●

●●

●

●●

●●

●● ●●●●

●●

●●

●●●●●

●●

●

●●

●●●●

●●●●

●●●

●●●

●

●

●

●

●

●●

●

●

●

●●

●●●

●

●●●

●●●●● ●●

● ●

●●

●

●

●

●

●

●

●

●

●●

●●

●

●●

●●●●●●

●

●●

●

●●●●●●

●●

●

●

●

●●

●●●

●

●

●

●●

●

●

●●

●

●●●

●●●●

●

●●

●

●●

●●

●

●

●●

●

●●●

●

●

●

●

●

●

●

●●●

●

●

●

●

●

●●

●●

● ● ●●

●●

●

●●●

●

●●●●●

●●●●●

●●

●●

●

●

●

●

●

●

●

●●●

●

●

●

●

●

●

●●●

●

●

●

●●

●

●●

●

●

●

●

●

● ●

●

●

●●

●●

●

●●

●●●●

●

●

●

●

●●●

●

●●

●

●

●

●●

●

●

●

● ●

●●

●

●●

●

●●

●●

● ●

●●

●

●

● ● ●

●● ●

●

●●●

●

●●●

●

●

● ●

●●

●

●

●

●

●●

●

●

●

●

0.0 0.2 0.4 0.6 0.8 1.0

0.0

0.2

0.4

0.6

0.8

1.0

Indigenous Prop. 2000

Bir

th R

egis

trat

ion

2000

Figure B-1: Baseline levels of indigenous population and birth registration (eligibleuniverse)

215

Table B.15: Land Titling (Continuous) and Legibility (Two-Way FE), IP Pop Inter-action

Dependent variable:

Whipple IndexAll Rural All Rural Matched Matched Eligible Eligible Titled Titled

(1) (2) (3) (4) (5) (6) (7) (8)

Titled Prop. x IP 2000 −11.188*** −11.510*** −10.815*** −9.878***(2.622) (3.117) (2.674) (2.698)

Titled Prop. 0.603 3.027*** 1.744 1.860(1.055) (1.158) (1.196) (1.473)

Titled x IP 2000 −9.513*** −9.446*** −9.738*** −8.946***(2.090) (2.279) (2.095) (2.098)

Titled 0.798 2.093** 1.781** 1.145(0.676) (0.833) (0.718) (1.061)

Bgy. Health Center −0.026 −0.028 −0.562 −0.518 −0.481 −0.464 −0.582 −0.499(0.180) (0.180) (0.783) (0.782) (0.451) (0.451) (0.976) (0.976)

Street Pattern 0.693*** 0.693*** 0.048 0.033 0.618 0.607 1.253 1.180(0.176) (0.176) (0.599) (0.600) (0.401) (0.402) (0.835) (0.841)

Highway Access −0.532** −0.539** −0.643 −0.601 0.424 0.415 −1.403* −1.390*(0.216) (0.216) (0.681) (0.682) (0.436) (0.437) (0.810) (0.813)

Log. Pop −5.215*** −5.212*** −1.427 −1.416 −12.543*** −12.560*** −5.398* −5.333*(0.846) (0.847) (2.648) (2.621) (1.533) (1.533) (3.104) (3.057)

Mean Age 0.100 0.100 0.282 0.301 0.378 0.382 0.323 0.336(0.119) (0.119) (0.395) (0.396) (0.251) (0.251) (0.500) (0.499)

Mean HH Size −1.102*** −1.100*** −1.371** −1.358** 0.537 0.543 −0.843 −0.832(0.209) (0.209) (0.625) (0.629) (0.448) (0.449) (0.625) (0.627)

Observations 33954 33954 1665 1665 7127 7127 1451 1451𝑅2 0.005 0.005 0.013 0.014 0.014 0.014 0.016 0.019

CRSE at bgy level *p<0.1; **p<0.05; ***p<0.01

216

Table B.16: Land Titling (Continuous) and Lagged Birth Registration

Dependent variable:

Birth Registration (Lagged)

All Rural Matched Eligible Titled

Titled Prop. 0.033*** −0.001 0.007 −0.005(0.009) (0.011) (0.009) (0.012)

Observations 23336 1107 4958 1008𝑅2 0.001 0 0 0

Note: *p<0.1; **p<0.05; ***p<0.01

Table B.17: Land Titling (Continuous) and Lagged Legibility

Dependent variable:

Whipple Index (Lagged)

(1) (2) (3) (4)

Titled Prop. −4.390*** 0.214 −0.875 0.221(1.173) (1.629) (1.315) (1.658)

Observations 23336 1107 4958 1008𝑅2 0 0 0 0

Note: *p<0.1; **p<0.05; ***p<0.01

Table B.18: Land Titling and Birth Registration (Differences-in-Differences)

Dependent variable:

Birth RegistrationAll Rural Matched Eligible Titled

(1) (2) (3) (4)

Titled Prop. 0.089*** −0.461*** 0.069* 0.062(0.030) (0.090) (0.036) (0.040)

Prov FE Y Y Y YObservations 31,955 1,674 6,535 1,335Adjusted R2 0.113 0.191 0.081 0.192

Note: *p<0.1; **p<0.05; ***p<0.01

217

Table B.19: Land Titling and Birth Registration (Conley SE)

Dependent variable:

Birth RegistrationAll Rural Matched Eligible Titled

(1) (2) (3) (4)

Titled Prop. 0.029** 0.018*** 0.025*** 0.022**(0.012) (0.006) (0.008) (0.009)

Observations 34000 1665 7146 1454𝑅2 0.864 0.793 0.857 0.805

Note: *p<0.1; **p<0.05; ***p<0.01

0.5

0.6

0.7

0.8

0.9

1.0

Age in 2000

% B

irth

Reg

iste

red

(Bar

anga

y M

ean)

50 47 44 41 38 35 32 29 26 23 20 17 14 11 8 6

Titled 2010Matched Untitled 2010Universe Untitled 2010Titled Unitlted 2010

Figure B-2: Birth registration rates by age in 2000

218

B.1.4 Instrumenting for land titling

Table B.20: Land Titling and Indigenous Identity (Instrumental Variables Estimates)

Dependent variable:

ip_delta

(1) (2)

cadt_bin2010 0.622 0.340*(0.422) (0.200)

Controls N YObservations 1,073 1,035Adjusted R2 −1.761 −0.515

Note: *p<0.1; **p<0.05; ***p<0.01

As an additional robustness check for the identity result, I instrument for the

granting of a land title between 2000 and 2010 using distance from the barangay to

the regional office of the NCIP. Even within the subset that is arguably most similar

—barangays that received titles at some point between 2000 and 2018 —it may be

the case that those who received titles earlier differ in important ways from those

who received them later. Communities whose leaders are more effective, for example,

might be more likely to apply earlier or to take action to expedite processing. If

these leaders are better able to increase or maintain the salience of indigenous identity

among their members over time, the coefficients could be upwardly biased even within

the subset of eventually titled barangays.

Aside from the burden placed on the claimant community, the process of issuing

a land title requires substantial effort on the part of NCIP staff, who must travel to

communities several times to hold a series of consultations, and conduct a land survey

and a comprehensive census. Regional office staff are responsible for managing the

processing of all CADT applications in their respective regions. The agency has been

historically under-resourced, and delays in the issuance of CADTs are often attributed

to bureaucratic delays Conditional on having applied for a title, communities that

219

are more accessible from the NCIP’s regional headquarters may have an advantage

in terms of processing time and may therefore be more likely to have been granted a

title between 2000 and 2010.2

In fact, I do find a fairly strong negative relationship between logged distance to

the regional NCIP office and the probability of receiving a land title between 2000

and 2010, among communities who have received titles to date (𝐹 = 15.16). In

other words, among barangays where indigenous communities have, to date, been

granted a land title, those closer to the regional NCIP office were more likely to have

their applications approved prior to 2010. Table B.20 presents instrumental variable

estimates from a two-stage least squares model, where distance to the regional NCIP

office is used as an instrument for the binary land titling indicator.3

The instrumental variables estimate is large in size and falls just below the 𝛼 = 0.1

threshold for statistical significance. However, when I control for several covariates

that are imbalanced at baseline between barangays that received titles before and after

2010 (including logged population in 2000, elevation, slope, soil quality, distance from

the road, distance from the cost, and area), the coefficient becomes more precisely

estimated and closer in size to the differences-in-differences estimates. Under some

assumptions, this estimate can be interpreted as the effect of land titling on indigenous

self-identification among those communities who received titles between 2000 and 2010

due to their proximity to NCIP offices, which may be greater in magnitude than the

effects on communities who received titles for other reasons.

The most potentially suspect of these assumptions is ignorability: that proximity

to NCIP regional offices is uncorrelated with other factors that led to increased indige-

nous self-identification among eventually titled communities. This could be violated,

2As of this writing, I do not have consistent data on the actual submission dates of land titles thatwould allow me to instrument for processing time directly. Instead, I am inferring that communitiesthat received land titles sooner had shorter processing times, on average.

3The statistically significant negative relationship also holds when using the continuous landtitling measure; however, the 𝐹 statistic is 8.03, less than what is typically considered sufficient fora first stage (𝐹 = 10).

220

for example, if indigenous communities closer to regional offices (which are often,

but not always, located in the regional administrative center) have more exposure

to “mainstream” (non-indigenous) society, which causes their indigenous identity to

become more salient over time. 4 This seems unlikely for a number of reasons. Given

the focus of both colonial and post-colonial governments on integrating indigenous

people, it would be surprising if those closest to centers of state power were more likely

to retain their indigenous identity. As discussed above, in my interviews with tribal

leaders, many blamed contact with mainstream society for a loss of indigenous iden-

tity among the youth. None suggested that contact with outsiders had had the effect

of strengthening indigenous identity. In fact, in the full subset of rural barangays and

among the “eligible” universe designated by NCIP, the relationship between distance

to the regional NCIP office and change in indigenous self-identification is consistent

with these leaders’ assertions: in general, indigenous identification increases more in

areas that are farther away from regional centers. It is only within the subset of

communities titled as of 2018 that closer proximity is associated with an increase in

indigenous self-identification between 2000 and 2010.

B.1.5 Alternative explanations

B.1.6 Survey experiment details

Figures B-3 and B-4 show translated versions of the flyers used in the “culture” and

“material” experimental conditions, respectively. Versions used in the survey were

in Tagalog. Flyers were introduced using the following language read by survey

4Another possibility is that individuals closer to regional centers had an increasing incentive torepresent themselves as indigenous due to the institution of forms of affirmative action in governmentprograms for indigenous communities. For example, members of IP communities do not have tomeet height requirements to join the Philippine National Police. However, only some governmentprograms are administered at the regional level. Additionally, in most cases, these agencies will seekverification from the NCIP or from tribal leadership of an individual’s status as a member of an IPcommunity before they are granted special accommodation. An individual’s self-designation on thecensus would have no direct effect on their eligibility.

221

Table B.21: Land Titling (Continuous) and Indigenous Identification - Land ValueInteractions

Dependent variable:

Indigenous Prop.All Rural Matched Eligible Titled All Rural Matched Eligible Titled

(1) (2) (3) (4) (5) (6) (7) (8)

Titled Prop. x Mineral Dep 0.051 −0.003 0.052 0.054(0.088) (0.070) (0.092) (0.094)

Titled Prop. x Soil Quality 0.047 0.087** 0.033 0.030(0.039) (0.035) (0.040) (0.041)

Titled Prop. 0.103*** 0.048*** 0.052*** 0.026* 0.012 −0.130* −0.011 −0.027(0.009) (0.012) (0.010) (0.014) (0.081) (0.073) (0.083) (0.087)

Observations 33530 1674 7023 1430 32236.5 1674 6804.5 1399.5𝑅2 0.01 0.011 0.004 0.003 0.011 0.015 0.005 0.005

CRSE at bgy level *p<0.1; **p<0.05; ***p<0.01

Table B.22: Land Titling (Continuous) and Birth Registration - Land Value Interac-tions

Dependent variable:

Birth RegistrationAll Rural Matched Eligible Titled All Rural Matched Eligible Titled

(1) (2) (3) (4) (5) (6) (7) (8)

Titled Prop. x Mineral Dep 0.026 −0.022 0.030 0.016(0.039) (0.023) (0.040) (0.035)

Titled Prop. x Soil Quality −0.029* −0.080*** −0.038** −0.036**(0.016) (0.015) (0.017) (0.017)

Titled Prop. 0.029*** 0.019** 0.025*** 0.021*** 0.090*** 0.181*** 0.101*** 0.093**(0.006) (0.008) (0.006) (0.008) (0.033) (0.034) (0.036) (0.036)

Observations 51000 2498 10719 2181.5 49058 2498 10386.5 2134.5𝑅2 0 0.002 0.001 0.003 0.001 0.005 0.001 0.003

CRSE at bgy level *p<0.1; **p<0.05; ***p<0.01

222

enumerators: “This is not just a survey, but we are also trying to education the

community about the basic concepts of the IPRA law and the legal rights of IPs.

Here is some information about the IPRA Law. I don’t have enough printed flyers

to leave one with everyone, but we hope you will share this information with others

in the community. I am going to ask you a few questions about it later to make sure

you remember the information, so please pay attention.”

Figure B-5 depicts the showcards used to measure the identity outcome in the

survey experiment. These images appeared on separate laminated cards and placed

in front of respondents in a randomly-assigned order. Showcards were introduced

by enumerators as follows: “People may think of themselves as part of many differ-

ent groups. Based on your answers you have given so far, here are some groups in

which you might consider yourself to be a member.” Then the groups were listed

in the assigned order with reference to the respondent’s previous answers. For ex-

ample, if the respondent indicated earlier in the survey that they were a member of

an Evangelical church, the enumerator would say “Your religion: Evangelical” while

placing the religion card in front of the respondent. After placing all four cards, the

enumerator asked “Of these four groups, which do you consider the most important

to you? Which of these do you consider the second most important?,” etc. Figure

B-6 shows the raw distribution of this outcome in each experimental condition. Ta-

ble B.23 shows covariate-adjusted estimates of experimental treatment effects on the

probability that each identity attribute is ranked in one of the top two positions,

using pre-specified covariates. Figure B-7 shows the estimated effects of the priming

treatment on individual components of the state attitudes index. Table B.25 shows

the estimated effects of each priming treatment on the probability that a respondent

believed the government conducted the survey.

223

THE IPRA LAW: RECOGNIZING

THE RIGHTS OF INDIGENOUS

PEOPLES TO PRESERVE THEIR

DISTINCTIVE CULTURES Republic Act No. 8371 or the Indigenous Peoples Rights Act (IPRA) of 1997 is a law enacted by the government of the Philippines to recognize, protect, and promote the rights of indigenous cultural communities/indigenous peoples (IPs/ICCs).

The IPRA Law recognizes several distinctive rights for IPs/ICCs, including:

Cultural Integrity Rights to Ancestral

Domains/Ancestral Land

The IPRA allows IPs/ICCs to apply for a title to their ancestral lands, known as a Certificate of Ancestral Domain Title (CADT). Getting a CADT helps protect IP communities’ rights to maintain their distinctive customs, cultural traditions, and identity without interference.

Ifugao people celebrating the end of the rice harvest

The IPRA law defines IPs/ICCs in part as groups of people who “…[share] common bonds of language, customs, traditions and other distinctive cultural traits or who have, through resistance to political, social and cultural inroads of colonization, non-indigenous religions and cultures, become historically differentiated from the majority of Filipinos….” Sec. 3(h) R.A. 8371

Indigenous youth from the B’laan tribe performing a traditional dance in Sarangi

Figure B-3: Recognition Prime (Culture). English translation.

224

THE IPRA LAW: RECOGNIZING

CUSTOMARY LAW AND

INDIGENOUS LEADERSHIP Republic Act No. 8371 or the Indigenous Peoples Rights Act (IPRA) of 1997 is a law enacted by the government of the Philippines to recognize, protect, and promote the rights of indigenous cultural communities/indigenous peoples (ICCs/IPs).

The IPRA Law recognizes several distinctive rights for ICCs/IPs, including:

Self-Governance and Empowerment

Ancestral Domains/Ancestral Land

The IPRA allows IPs/ICCs to apply for a title to their ancestral lands, known as a Certificate of Ancestral Domain Title (CADT). Getting a CADT It empowers leaders of the ICC to decide how land and other resources with the domain are used in accordance with customary law.

Leaders perform a tribal ritual in Manila

Tribal leaders in Bukidnon, Mindanao

The IPRA recognizes the primacy of customary law within ancestral domains:

“Customary laws, traditions and practices of the ICCs/IPs…shall be applied first with respect to property rights, claims and ownerships, hereditary succession and settlement of land disputes”

Sec. 63, R.A. 8371

Figure B-4: Recognition Prime (Material). English translation.

225

Nationality Nasyonalidad

Gender Kasarian

Religion Relihiyon

Tribe Tribo

Figure B-5: Identity showcards.

226

1 2 3 4

Control

rank

0.0

0.1

0.2

0.3

0.4

0.5

1 2 3 4

Culture Prime

rank

0.0

0.1

0.2

0.3

0.4

0.5

1 2 3 4

Material Prime

rank0.

00.

10.

20.

30.

40.

5

Figure B-6: Identity rankings by experimental condition. Raw distribution of rank-ings for nationality attribute, across experimental conditions. Attributes ranked closerto one are considered more important important.

Table B.23: Priming Treatment and Top Ranking of Identity Attributes

Dependent variable:

nationality_top tribe_top gender_top religion_top

(1) (2) (3) (4) (5) (6) (7) (8)

Any Prime 0.056 −0.047 −0.065* 0.042(0.039) (0.038) (0.037) (0.037)

Culture Prime 0.090** −0.080* −0.073* 0.048(0.045) (0.044) (0.043) (0.043)

Material Prime 0.022 −0.016 −0.058 0.037(0.044) (0.044) (0.042) (0.042)

Covariate Adjustment Y Y Y Y Y Y Y YObservations 725 725 725 725 725 725 725 725R2 0.020 0.026 0.029 0.035 0.025 0.027 0.023 0.031Adjusted R2 0.011 0.011 0.020 0.020 0.016 0.012 0.013 0.016

Note: *p<0.1; **p<0.05; ***p<0.01

227

●

●

● ●

●

●

●

−0.

6−

0.4

−0.

20.

00.

20.

40.

6

Experimental Effects on State Attitudes Whole Sample

coef

indexintentions

bgyintentions

muncapable

bgycapable

muntrust bgy

trust mun

● ●

●

●

●

●

●

●

●

culturematerial

Figure B-7: Effects of priming on individual components of state attitudes index(intentions, capability, and trustworthiness of barangay and municipal government).

228

Table B.24: Prime Treatment and Ranking of Identity Attributes

Dependent variable:

tribe_rank nationality_rank gender_rank religion_rank

(1) (2) (3) (4) (5) (6) (7) (8)

ipra_prime 0.194** 0.184** −0.229** −0.223** 0.182* 0.191* −0.161* −0.157*(0.093) (0.092) (0.097) (0.098) (0.100) (0.101) (0.094) (0.094)

Covariate Adjustment N Y N Y N Y N YObservations 475 475 472 472 473 473 475 475R2 0.009 0.032 0.012 0.026 0.007 0.025 0.006 0.033Adjusted R2 0.007 0.017 0.009 0.011 0.005 0.010 0.004 0.019

Note: *p<0.1; **p<0.05; ***p<0.01

Table B.25: Recognition Primes and Belief that Government Conducted Survey

Dependent variable:

conducted_gov

(1) (2)

Any Prime −0.007(0.030)

Culture Prime 0.033(0.034)

Material Prime −0.046(0.034)

Observations 725 725R2 0.0001 0.008Adjusted R2 −0.001 0.005

Note: *p<0.1; **p<0.05; ***p<0.01

229

B.2 Supplementary material for Chapter 5

B.2.1 Additional analysis of survey outcomes

Tables B.26 and B.27 show comparisons between titled and untitled barangays in

the survey sample for individual components of the pre-registered Vote Choice and

Voter Coordination Indices, with and without barangay-level covariates. Individ-

ual outcomes for the Vote Choice Index include the proportion of respondents in

a barangay who reported knowing which candidate for barangay captain indigenous

leaders supported in the 2018 elections, the proportion who reported supporting

the same candidate as indigenous leaders, the proportion believing that over half of

individuals in the barangay supported that candidate, the proportion who believed

that their relationships with leaders would be harmed if they supported a different

candidate, and the proportion who believed that leaders could find out whom they

voted for. Outcomes for the Voter Coordination Index include the proportion of re-

spondents in a barangay who believed that others in the community could find out

whom they voted for, the proportion who reported that the community had held a

meeting to discuss candidates in the 2018 barangay elections, the proportion who re-

ported attending such a meeting, and the proportion who reported that participants

in such a meeting had come to a consensus about which candidate to support.

230

Table B.26: Land Titling and Indigenous Leader Influence on Vote Choice (Index Components)

Dependent variable:

know_leader_candidate support_leader_candidate comm_support_leader_candidate leader_harm_rel leader_find_out

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

CADT 2018 0.159*** 0.161** 0.162*** 0.167*** 0.065 0.078 0.094 0.242* 0.190 0.233*(0.059) (0.066) (0.053) (0.055) (0.056) (0.069) (0.163) (0.145) (0.126) (0.121)

IP Pop 2000 0.070 −0.027 −0.092 −0.156 0.075(0.098) (0.084) (0.105) (0.275) (0.239)

Log Pop 2000 0.111** 0.043 0.034 −0.101 0.175*(0.053) (0.048) (0.051) (0.144) (0.106)

Pct Catholic 2000 0.263 −0.015 −0.009 −0.127 0.041(0.170) (0.161) (0.165) (0.390) (0.393)

Ethfrac 2000 0.010 0.011 −0.018 0.847** 0.863**(0.150) (0.131) (0.146) (0.363) (0.365)

Mean Slope 0.005 −0.006 −0.006 −0.003 0.026**(0.007) (0.006) (0.006) (0.017) (0.013)

Log Coast Dist −0.035 −0.033 −0.029 0.124 0.038(0.046) (0.031) (0.042) (0.090) (0.089)

Log Road Dist 0.019*** 0.005 0.015** 0.015 0.027*(0.007) (0.006) (0.007) (0.017) (0.014)

Strong Roof Mat 2000 0.069 0.097 0.136 0.948** −0.209(0.188) (0.171) (0.210) (0.462) (0.402)

Area 0.002** 0.001** 0.001 0.002 0.001(0.001) (0.001) (0.001) (0.002) (0.002)

N Clusters 68 68 68 68 68 68 68 68 68 68Observations 603 603 625 625 625 625 624 624 620 620R2 0.023 0.085 0.028 0.052 0.005 0.048 0.002 0.111 0.007 0.047Adjusted R2 0.022 0.070 0.026 0.036 0.003 0.032 −0.00005 0.096 0.005 0.031

CRSE at barangay level *p<0.1; **p<0.05; ***p<0.01

231

Table B.27: Land Titling and Vote Coordination (Index Components)

Dependent variable:

community_find_out held_meeting attended_meeting candidate_consensus

(1) (2) (3) (4) (5) (6) (7) (8)

CADT 2018 0.206* 0.199 0.174*** 0.088 0.181*** 0.085 0.149*** 0.066(0.110) (0.130) (0.062) (0.067) (0.062) (0.058) (0.058) (0.066)

IP Pop 2000 0.077 −0.225** −0.329*** −0.279***(0.230) (0.112) (0.115) (0.108)

Log Pop 2000 0.081 0.130*** 0.178*** 0.105**(0.098) (0.050) (0.048) (0.045)

Pct Catholic 2000 −0.020 −0.097 −0.197 −0.202(0.327) (0.164) (0.155) (0.155)

Ethfrac 2000 0.443 0.138 0.151 0.144(0.295) (0.133) (0.132) (0.139)

Mean Slope 0.025** 0.013** 0.011** 0.010*(0.013) (0.005) (0.005) (0.005)

Log Coast Dist 0.080 0.057 0.032 0.056(0.096) (0.044) (0.038) (0.044)

Log Road Dist 0.022* 0.007 0.003 0.007(0.013) (0.008) (0.007) (0.008)

Strong Roof Mat 2000 0.097 −0.183 −0.253* −0.194(0.357) (0.200) (0.139) (0.195)

Area 0.001 −0.00001 −0.0003 0.00002(0.002) (0.001) (0.001) (0.001)

N Clusters 68 68 68 68 68 68 68 68Observations 623 623 625 625 625 625 625 625R2 0.009 0.038 0.028 0.086 0.031 0.115 0.023 0.087Adjusted R2 0.007 0.023 0.026 0.071 0.030 0.101 0.021 0.072

CRSE at barangay level *p<0.1; **p<0.05; ***p<0.01

232

B.2.2 Conjoint experiment details and robustness checks

Figure B-8 shows the images associated with each value of the five candidate at-

tributes. Each row represents an attribute. Images on the left indicate “yes” values

and images on the right indicate “no” values. For example, in the first row, the image

on the left indicates that the candidate did give gifts during the campaign period and

the image on the right indicates that she did not. Each candidate profile took on one

of these two images for each attribute. Images were explained to respondents before

the start of the voting game and enumerators were instructed to describe individual

profiles verbally for each conjoint task.

Figure B-9 shows the Average Marginal Component Effects (AMCE) for each

candidate attribute in the full sample. Table B.28 shows coefficent estimates and

standard errors for the experimental effect estimates, with interaction terms between

each binary attribute indicator and experimental treatment condition. Columns 1

and 2 show effects for the pooled experimental treatment condition (any prime com-

pared to control) while columns 3 and 4 show effects for each experimental condition.

Figure B-10 shows the effects of the priming treatments on beliefs about the influ-

ence of indigenous leaders in different areas of governance, for titled and untitled

communities.

Figure B-11 shows experimental effects on the AMCEs of each candidate attribute

in titled barangays only. Table B.29 shows coefficients and standard errors for each

AMCE interacted with actual titling status. Figures B-12 and B-13 show marginal

means for each value of each candidate attribute, across experimental conditions and

titling status subgroups, respectively.

B.2.3 COMELEC polling places intervention

Table B.30 shows balance on a number of pre-treatment barangay-level covariates

between pilot barangays for the Separate Polling Places intervention and matched

233

Campaigning

Community Work

Mayor Endorsement

Promises

Leader Endorsement

Figure B-8: Voting Game (Conjoint) Candidate Attribute Images

●

●

●

●

●

−0.

2−

0.1

0.0

0.1

0.2

Main Effects (Full Sample)

AM

CE

gift promises mayor leader commwork

Figure B-9: Average Marginal Component Effect estimates in the full sample.

234

Table B.28: Candidate Conjoint Results by Experimental Prime

Dependent variable:

choice

(1) (2) (3) (4)

gift1 −0.060** −0.128** −0.060** −0.128***(0.024) (0.064) (0.024) (0.024)

promises1 −0.008 −0.107* −0.008 −0.106***(0.020) (0.058) (0.020) (0.020)

mayor1 0.027 −0.016 0.027 −0.016(0.022) (0.065) (0.022) (0.022)

leader1 0.065*** 0.007 0.065*** 0.007(0.021) (0.053) (0.021) (0.021)

commwork1 0.101*** 0.164*** 0.101*** 0.164***(0.020) (0.053) (0.020) (0.020)

prime_any 0.002 0.012(0.031) (0.032)

gift1:prime_any −0.030 −0.048*(0.028) (0.029)

promises1:prime_any 0.034 0.037(0.025) (0.025)

mayor1:prime_any 0.037 0.048*(0.026) (0.026)

leader1:prime_any −0.007 −0.007(0.026) (0.026)

commwork1:prime_any −0.043* −0.060**(0.025) (0.025)

ipra_factorCulture 0.006 0.020(0.035) (0.035)

ipra_factorMaterial −0.001 0.006(0.036) (0.036)

gift1:ipra_factorCulture −0.030 −0.057*(0.031) (0.031)

gift1:ipra_factorMaterial −0.030 −0.041(0.032) (0.032)

promises1:ipra_factorCulture 0.025 0.024(0.030) (0.030)

promises1:ipra_factorMaterial 0.043 0.048(0.029) (0.029)

mayor1:ipra_factorCulture 0.034 0.047(0.030) (0.030)

mayor1:ipra_factorMaterial 0.040 0.048*(0.028) (0.028)

leader1:ipra_factorCulture 0.007 0.014(0.029) (0.029)

leader1:ipra_factorMaterial −0.020 −0.025(0.030) (0.030)

commwork1:ipra_factorCulture −0.050* −0.075***(0.029) (0.029)

commwork1:ipra_factorMaterial −0.037 −0.047*(0.029) (0.029)

Enum FE N Y N YRound FE N Y N Y

Note: *p<0.1; **p<0.05; ***p<0.01

235

−1.

0−

0.5

0.0

0.5

1.0

Experimental Effects on Leader Influence Beliefs Untitled Only

coef

index overall disputes resources land peace edu projects customs tenyears

IdentityMaterial

−1.

0−

0.5

0.0

0.5

1.0

Experimental Effects on Leader Influence Beliefs Titled Only

coef

index overall disputes resources land peace edu projects customs tenyears

IdentityMaterial

Figure B-10: Effects of priming treatments on perceived influence of indigenous lead-ers.

236

●

●

●

● ●

−0.

2−

0.1

0.0

0.1

0.2

Experimental Effects (Control vs. Either Prime − Titled Only)

AM

CE

gift promises mayor leader commwork

●

●

●

●

●

●

●

ControlPrime

(a)

●

●

●

● ●

−0.

2−

0.1

0.0

0.1

0.2

Experimental Effects (All Prime Conditions − Titled Only)

AM

CE

gift promises mayor leader commwork

● ControlIdentityMaterial

(b)

Figure B-11: Experimental effects by candidate attribute, titled barangays only. Av-erage Marginal Component Effects (AMCE) for each binary attribute, comparingtitled and untitled barangays, shown with 95% confidence intervals. Standard errorsfor each treatment condition are clustered at the level of the respondent.

237

Table B.29: Candidate Conjoint Results by Land Titling Status

choice

All All Control Only

(1) (2) (3)

gift1 −0.136*** 0.038 0.495(0.026) (0.280) (0.583)

promises1 −0.016 0.237 0.978**(0.029) (0.277) (0.490)

leader1 −0.00003 0.118 0.252(0.027) (0.316) (0.479)

mayor1 0.054* 0.282 0.113(0.031) (0.282) (0.575)

commwork1 0.130*** −0.115 −0.189(0.026) (0.204) (0.418)

cadt_bin2018 −0.050 −0.060** −0.087(0.034) (0.030) (0.070)

gift1:cadt_bin2018 0.067** 0.074*** 0.039(0.028) (0.025) (0.056)

promises1:cadt_bin2018 0.044 0.070** 0.077(0.030) (0.031) (0.055)

leader1:cadt_bin2018 0.039 0.037 0.065(0.034) (0.031) (0.049)

mayor1:cadt_bin2018 −0.044* −0.046 −0.036(0.026) (0.028) (0.055)

commwork1:cadt_bin2018 −0.004 −0.012 0.013(0.024) (0.024) (0.052)

Observations 7118 7118 2308Respondents 720 720 233Controls N Y YRound FE Y Y Y

Note: SE clustered at barangay level *p<0.1; **p<0.05; ***p<0.01

238

●

●

●

●

●

●

●

●

●

●

no community work

community work

(commwork)

no leader endorse

leader endorse

(leader)

no mayor support

mayor support

(mayor)

no promises

promises

(promises)

no gift

gift

(gift)

0.40 0.45 0.50 0.55 0.60Marginal Mean (Control)

Feature ● ● ● ● ●gift promises mayor leader commwork

(a)

●

●

●

●

●

●

●

●

●

●

no community work

community work

(commwork)

no leader endorse

leader endorse

(leader)

no mayor support

mayor support

(mayor)

no promises

promises

(promises)

no gift

gift

(gift)

0.45 0.50 0.55Marginal Mean (Either Prime)

Feature ● ● ● ● ●gift promises mayor leader commwork

(b)

●

●

●

●

●

●

●

●

●

●

no community work

community work

(commwork)

no leader endorse

leader endorse

(leader)

no mayor support

mayor support

(mayor)

no promises

promises

(promises)

no gift

gift

(gift)

0.45 0.50 0.55 0.60Marginal Mean (Identity Prime)

Feature ● ● ● ● ●gift promises mayor leader commwork

(c)

●

●

●

●

●

●

●

●

●

●

no community work

community work

(commwork)

no leader endorse

leader endorse

(leader)

no mayor support

mayor support

(mayor)

no promises

promises

(promises)

no gift

gift

(gift)

0.45 0.50 0.55 0.60Marginal Mean (Material Prime)

Feature ● ● ● ● ●gift promises mayor leader commwork

(d)

Figure B-12: Marginal means by candidate attribute and experimental condition.Marginal mean probability of selecting a candidate by candidate attribute, disaggre-gated by experimental treatment condition. 95% confidence intervals with standarderrors clustered at the level of the respondent.

239

●

●

●

●

●

●

●

●

●

●

no community work

community work

(commwork)

no leader endorse

leader endorse

(leader)

no mayor support

mayor support

(mayor)

no promises

promises

(promises)

no gift

gift

(gift)

0.45 0.50 0.55 0.60Marginal Mean (Untitled Barangays)

Feature ● ● ● ● ●gift promises mayor leader commwork

(a)

●

●

●

●

●

●

●

●

●

●

no community work

community work

(commwork)

no leader endorse

leader endorse

(leader)

no mayor support

mayor support

(mayor)

no promises

promises

(promises)

no gift

gift

(gift)

0.45 0.50 0.55 0.60Marginal Mean (Titled Barangays)

Feature ● ● ● ● ●gift promises mayor leader commwork

(b)

Figure B-13: Marginal mean probability of selecting a candidate by candidate at-tribute, disaggregated by actual barangay titling status, disaggregated by experimen-tal treatment condition. 95% confidence intervals with standard errors clustered atthe level of the respondent.

controls. Matches were generated using a genetic matching algorithm. Balance tables

for alternative matching strategies, including propensity score matching and Maha-

lanobis distance matching, are available upon request.

240

Table B.30: Separate Polling Places Pilot - Balance Table (Genetic Matching)

mean.Tr mean.Co sdiff sdiff.pooled var.ratio T pval KS pval qqmeandiff qqmeddiff qqmaxdiffpop2010 2984.17 2761.78 11.81 11.81 1.20 0.59 0.72 0.07 0.06 0.22

num_iccs 7.11 7.56 -8.84 -8.84 2.38 0.69 0.19 0.12 0.11 0.33num_groups 15.11 16.28 -12.78 -12.78 1.54 0.61 0.39 0.11 0.11 0.28

ip_pct_official 0.37 0.37 -3.06 -3.06 0.73 0.91 1.00 0.04 0.06 0.11ethfrac 0.45 0.51 -33.48 -33.48 0.71 0.22 0.71 0.09 0.08 0.22

reg_voters2010 1439.22 1343.50 11.33 11.33 0.99 0.55 0.43 0.08 0.06 0.28margin2010 0.28 0.30 -4.63 -4.63 1.13 0.73 0.95 0.05 0.06 0.17

bgy_turnout2010 0.76 0.75 23.37 23.37 1.07 0.47 0.72 0.07 0.06 0.22ELEV_STD 240.58 210.82 15.76 15.76 1.38 0.27 0.70 0.07 0.06 0.22

SLOPE_MEAN 13.15 13.54 -7.63 -7.63 0.59 0.76 0.69 0.08 0.06 0.22NEAR_DIST 3695.45 3879.89 -6.21 -6.21 0.59 0.83 0.44 0.09 0.06 0.28

Area 55542295.44 49927116.00 15.22 15.22 0.98 0.64 0.96 0.06 0.06 0.17Latitude 12.94 12.95 -2.86 -2.86 0.98 0.46 1.00 0.04 0.06 0.11

Longitude 121.10 121.11 -2.87 -2.87 1.04 0.52 0.94 0.05 0.06 0.17cadt_intersect_pct 0.25 0.19 19.50 19.50 1.37 0.36 0.31 0.11 0.11 0.28241

B.2.4 Public goods robustness checks

Table B.31 shows estimated effects of titling on lagged highway access. While a

significant placebo effect is estimated in the full sample, this does not hold for the

other three subsets. Table B.32 shows estimates of the effects of titling on highway

access within municipalities and provinces, respectively. The dependent variable is

a binary indicator for whether a barangay gained highway road access between 2000

and 2010, and the independent variables show change in titling status during that

period, operationalized as both continuous and binary.

Table B.31: Land Titling (Continuous) and Lagged Highway Access

Dependent variable:

Highway Access (Lagged)

(All Rural) (Matched) (Eligible) (Titled)

Titled Prop. 0.069*** 0.025 0.027 −0.048(0.024) (0.038) (0.026) (0.034)

Observations 23371 1116 5003 1017𝑅2 0 0.002 0 0.001

Note: *p<0.1; **p<0.05; ***p<0.01

242

Table B.32: Land Titling and Highway Access (Conditional Logit)

Dependent variable:

Highway Access (Change)

(1) (2) (3) (4)

Titled Prop (Change) 0.531*** 0.889***(0.164) (0.101)

Titled (Change) 0.451*** 0.667***(0.142) (0.082)

Mun FE Y Y N NProv FE N N Y YObservations 36,121 36,121 36,121 36,121R2 0.0003 0.0003 0.002 0.002LR Test (df = 1) 10.361*** 9.931*** 72.860*** 62.167***

Note: *p<0.1; **p<0.05; ***p<0.01

243

244

Bibliography

Abadie, Alberto. 2005. “Semiparametric Difference-in-Differences Estimators.” TheReview of Economic Studies 72(1):1–19. Publisher: Oxford Academic.

Abouzahr, Carla. 2014. “Strengthening civil registration and vital statistics in theAsia-Pacific region: learning from country experiences.” Asia-Pacific PopulationJournal 29(1):39–73.

Acemoglu, Daron, Isaías N. Chaves, Philip Osafo-Kwaako and James A. Robinson.2014. Indirect Rule and State Weakness in Africa: Sierra Leone in ComparativePerspective. Working Paper 20092 National Bureau of Economic Research.

Acemoglu, Daron, Tristan Reed and James A. Robinson. 2014. “Chiefs: EconomicDevelopment and Elite Control of Civil Society in Sierra Leone.” Journal of PoliticalEconomy 122(2):319–368.

ACHPR. 2005. Report of the African Commission’s Working Group of Experts on In-digenous Populations/Communities. Technical report African Commission on Hu-man and Peoples’ Rights.

Adserà, Alícia, Carles Boix and Mark Payne. 2003. “Are You Being Served? Po-litical Accountability and Quality of Government.” Journal of Law, Economics &Organization 19(2):445–490.

Alesina, Alberto, Reza Baqir and William Easterly. 1999. “Public Goods and EthnicDivisions.” The Quarterly Journal of Economics 114(4):1243–1284.

Alexander, Michelle. 2020. The New Jim Crow: Mass Incarceration in the Age ofColorblindness. The New Press.

Anderson, Benedict. 1983. Imagined Communities: Reflections on the Origin andSpread of Nationalism, Revised Edition. Revised edition ed. London ; New York:Verso.

Anderson, Benedict. 1987. Introduction. In Southeast Asian Tribal Groups and EthnicMinorities: Prospects for the Eighties and Beyond. Number 22 in “Cultural SurvivalReport” Cambridge, MA: Cultural Survival pp. 1–15.

Anderson, Michael L. 2008. “Multiple Inference and Gender Differences in the Ef-fects of Early Intervention: A Reevaluation of the Abecedarian, Perry Preschool,

245

and Early Training Projects.” Journal of the American Statistical Association103(484):1481–1495.

Anderson, Siwan, Patrick Francois and Ashok Kotwal. 2012. One Kind of Democracy.The Centre for Market and Public Organisation 12/292 Department of Economics,University of Bristol, UK.

Appiah, Kwame Anthony. 2007a. Cosmopolitanism: Ethics in a World of Strangers.Reprint edition ed. New York, NY: W. W. Norton & Company.

Appiah, Kwame Anthony. 2007b. The Ethics of Identity. F first edition edition ed.Princeton, NJ: Princeton University Press.

Authority, World Bank; Philippine Statistics. 2017. “Philippines Success in ImprovingBirth Registration.”.

Baldwin, Kate. 2011. “When Politicians Cede Control of Resources: Land, Chiefs,and Coalition-Building in Africa.” Afrobarometer Working Papers (130).

Baldwin, Kate. 2013. “Why Vote with the Chief? Political Connections and PublicGoods Provision in Zambia.” American Journal of Political Science 57(4):794–809.

Baldwin, Kate. 2015. The Paradox of Traditional Chiefs in Democratic Africa. Cam-bridge University Press.

Baldwin, Kate and Eric Mvukiyehe. 2015. “Elections and Collective Action: Evidencefrom Changes in Traditional Institutions in Liberia.” World Politics .

Bandyopadhyay, Sanghamitra and Elliott Green. 2013. “Nation-Building and Conflictin Modern Africa.” World Development 45:108–118.

Bank, Asian Development. 2002. Indigenous Peoples/Ethnic Minorities and PovertyReduction: Philippines. Asian Development Bank.

Barry, Brian. 2002. Culture and Equality: An Egalitarian Critique of Multiculturalism.Harvard University Press.

Bartels, Larry M. 2010. Unequal Democracy: The Political Economy of the NewGilded Age. 1/23/10 edition ed. New York: Princeton University Press.

Bates, Robert H. and Da-Hsiang Donald Lien. 1985. “A Note on Taxation, Develop-ment, and Representative Government.” Politics & Society 14(1):53–70.

Bauer, Michal, Christopher Blattman, Julie Chytilová, Joseph Henrich, EdwardMiguel and Tamar Mitts. 2016. “The legacy of war on social and political be-haviour.”. Library Catalog: VoxEU.URL: https://voxeu.org/article/legacy-war-social-and-political-behaviour

Behr, Daniela M. 2018. “The Developmental Consequences of the de jure State-Traditional Relationship.” Working Paper .

246

Besley, Timothy. 2006. Principled agents? : the political economy of good government/ Timothy Besley. The Lindahl lectures Oxford ; New York : Oxford UniversityPress, 2006.

Blair, Robert A., Benjamin S. Morse and Lily L. Tsai. 2017. “Public health and publictrust: Survey evidence from the Ebola Virus Disease epidemic in Liberia.” SocialScience & Medicine (1982) 172:89–97.

Boix, Carles and Daniel N. Posner. 1998. “Social Capital: Explaining Its Origins andEffects on Government Performance.” British Journal of Political Science 28(4):686.

Bonilla-Silva, Eduardo. 2013. Racism without Racists: Color-Blind Racism and thePersistence of Racial Inequality in America. Rowman & Littlefield Publishers.

Boone, Catherine. 2003. Political Topographies of the African State: Territorial Au-thority and Institutional Choice. Cambridge, UK ; New York: Cambridge UniversityPress.

Boone, Catherine. 2014. Property and Political Order in Africa: Land Rights and theStructure of Politics. New York: Cambridge University Press.

Canuday, Jose Jowel. 2006. “The Power of the Displaced.” Asian and Pacific MigrationJournal 15(1):115–136. Publisher: SAGE Publications Ltd STM.URL: https://doi.org/10.1177/011719680601500106

Casey, Katherine. 2018. “Radical Decentralization: Does Community-Driven Devel-opment Work?” Annual Review of Economics 10(1):139–163.URL: https://doi.org/10.1146/annurev-economics-080217-053339

Cederman, Lars-Erik, Andreas Wimmer and Brian Min. 2010. “Why Do EthnicGroups Rebel? New Data and Analysis.” World Politics 62(1):87–119. Publisher:Cambridge University Press.

Centeno, Miguel Angel. 2003. Blood and Debt: War and the Nation-State in LatinAmerica. 1 edition ed. University Park, Pa: Penn State University Press.

Chabal, Patrick and Jean-Pascal Daloz. 1999. Africa Works: Disorder as PoliticalInstrument. London : Bloomington: diana University Press - Indiana UniversityPress.

Chandra, Kanchan. 2006. “What Is Ethnic Identity and Does It Matter?” AnnualReview of Political Science 9(1):397–424.

Charnysh, Volha. 2019. “Diversity, Institutions, and Economic Outcomes: Post-WWIIDisplacement in Poland.” American Political Science Review 113(2):423–441.

Charters, Claire and Rodolfo Stavenhagen. 2009. Making the Declaration work : theUnited Nations Declaration on the Rights of Indigenous Peoples.

247

Chartock, Sarah. 2013. “ “Corporatism With Adjectives”? Conceptualizing Civil Soci-ety Incorporation and Indigenous Participation in Latin America.” Latin AmericanPolitics and Society 55(2):52–76.

Chowdhury, Arjun. 2018. The Myth of International Order: Why Weak States Persistand Alternatives to the State Fade Away. Oxford University Press.

Clayton, Amanda, Jennifer Lynn Noveck and Margaret Levi. 2015. When elites meet: decentralization, power-sharing, and public goods provision in post-conflict SierraLeone. Technical Report WPS7335 The World Bank.

Co, Ronilda R. 2008. Free, Prior and Informed Consent (FPIC): Does it Give Indige-nous Peoples More Control over Development of their Lands in the Philippines?Master’s thesis Massachusetts Institute of Technology.

Coates, Ta-Nehisi. 2014. “The Case for Reparations.” The Atlantic .URL: https://www.theatlantic.com/magazine/archive/2014/06/the-case-for-reparations/361631/

Collier, David and Steven Levitsky. 1997. “Democracy with Adjectives: ConceptualInnovation in Comparative Research.” World Politics 49(3):430. Publisher: JohnsHopkins University Press.

Collier, Paul. 2010. Wars, Guns, and Votes: Democracy in Dangerous Places. Reprintedition ed. New York: Harper Perennial.

Conley, T. G. 1999. “GMM estimation with cross sectional dependence.” Journal ofEconometrics 92(1):1–45.

Conroy-Krutz, Jeffrey. 2017. “Individual Autonomy and Local-Level Solidarity inAfrica.” Political Behavior pp. 1–35.

Corntassel, Jeff and Richard C. Witmer. 2008. Forced Federalism: ContemporaryChallenges to Indigenous Nationhood. University of Oklahoma Press. Google-Books-ID: B6_iQHfZXAsC.

Crowder, Michael. 1964. “Indirect Rule: French and British Style.” Africa: Jour-nal of the International African Institute 34(3):197–205. Publisher: [CambridgeUniversity Press, International African Institute].

Cruz, Cesi. 2018. “Social Networks and the Targeting of Vote Buying:.” ComparativePolitical Studies . Publisher: SAGE PublicationsSage CA: Los Angeles, CA.URL: https://journals.sagepub.com/doi/10.1177/0010414018784062

Cruz, Cesi, Julien Labonne and Pablo Querubin. 2015. “Politician Family Networksand Electoral Outcomes: Evidence from the Philippines.” Working Paper .

Cuskelly, Katrina. 2011. Customs and Constitutions: State recognition of customarylaw around the world. Bangkok, Thailand: International Union for Conservation ofNature and Natural Resources.

248

Dahl, Robert A. 1971. Polyarchy; participation and opposition, by Robert A. Dahl.New Haven, Yale University Press, 1971.

De Kadt, Daniel and Horacio Larreguy. 2015. “Agents of the Regine? TraditionalLeaders and Electoral Behavior in South Africa.” Working Paper .

Declaration on the Rights of Asian Indigenous Peoples. 2000. Asia-Pacific Journalon Human Rights & the Law 1(1):165–168. Publisher: Brill Academic Publishers.

Diamond, Alexis and Jasjeet S. Sekhon. 2006. “Genetic Matching for EstimatingCausal Effects: A General Multivariate Matching Method for Achieving Balancein Observational Studies.”.

Edwards, Bob, Michael W. Foley and Mario Diani, eds. 2001. Beyond Tocqueville:Civil Society and the Social Capital Debate in Comparative Perspective. 1 editioned. Hanover, NH: Tufts University.

Ekeh, Peter P. 1975. “Colonialism and the two publics in Africa: a theoretical state-ment.” Comparative studies in society and history 17(01):91–112.

Elkins, Zachary and John Sides. 2007. “Can Institutions Build Unity in MultiethnicStates?” The American Political Science Review 101(4):693–708.

Englebert, Pierre. 2002. State Legitimacy and Development in Africa. 2nd printingedition ed. Boulder: Lynne Rienner Pub.

Fearon, James D. and David D. Laitin. 2003. “Ethnicity, Insurgency, and Civil War.”The American Political Science Review (1):75.

Ferguson, Kennan. 2016. “Why Does Political Science Hate American Indians?” Per-spectives on Politics 14(4):1029–1038.

Finan, Frederico and Laura Schechter. 2012. “Vote-Buying and Reciprocity.” Econo-metrica 80(2):863–881.URL: https://onlinelibrary.wiley.com/doi/abs/10.3982/ECTA9035

Finin, Gerard A. 2005. The Making of the Igorot: Ramut Ti Panagkaykaysa DagitiTaga Cordillera : Contours of Cordillera Consciousness. Ateneo University Press.

Francis, Megan Ming. 2018. “The Strange Fruit of American Political Development.”Politics, Groups, and Identities 6(1). Publisher: Routledge pages = 128–137.

Fukuyama, Francis. 2018. Identity: The Demand for Dignity and the Politics ofResentment. Farrar, Straus and Giroux.

Garnett, Stephen T., Neil D. Burgess, John E. Fa, Álvaro Fernández-Llamazares,Zsolt Molnár, Cathy J. Robinson, James E. M. Watson, Kerstin K. Zander, BeauAustin, Eduardo S. Brondizio, Neil French Collier, Tom Duncan, Erle Ellis, HayleyGeyle, Micha V. Jackson, Harry Jonas, Pernilla Malmer, Ben McGowan, Amphone

249

Sivongxay and Ian Leiper. 2018. “A spatial overview of the global importance ofIndigenous lands for conservation.” Nature Sustainability 1(7):369–374.

Gellner, Ernest. 1983. Nations and Nationalism. Cornell University Press. Google-Books-ID: Aqidsw3SkDUC.

Gerring, John, Daniel Ziblatt, Johan Van Gorp and Julián Arévalo. 2011. “An Institu-tional Theory of Direct and Indirect Rule.” World Politics 63(3):377–433. Publisher:Cambridge University Press.

Goldstein, Markus and Christopher Udry. 2008. “The Profits of Power: Land Rightsand Agricultural Investment in Ghana.” Journal of Political Economy 116(6):981.

Gottlieb, Jessica. 2016. “Greater Expectations: A Field Experiment to Improve Ac-countability in Mali.” American Journal of Political Science 60(1):143–157.

Gottlieb, Jessica and Horacio Larreguy. 2016. “An Informational Theory of ElectoralTargeting: Evidence from Senegal.” Center on the Politics of Development WorkingPaper .

Grossman, Shelby, Jonathan Phillips and Leah R. Rosenzweig. 2018. “Opportunis-tic Accountability: State–Society Bargaining Over Shared Interests.” ComparativePolitical Studies 51(8):979.

Habyarimana, James, Macartan Humphreys, Daniel N. Posner and Jeremy M. We-instein. 2007. “Why Does Ethnic Diversity Undermine Public Goods Provision?”American Political Science Review 101(04):709–725.

Hagopian, Frances, Carlos Gervasoni and Juan Andres Moraes. 2008. “From Patron-age to Program: The Emergence of Party-Oriented Legislators in Brazil.” Com-parative Political Studies . Publisher: SAGE Publications Sage CA: Los Angeles,CA.

Hainmueller, Jens, Daniel J. Hopkins and Teppei Yamamoto. 2014. “Causal Infer-ence in Conjoint Analysis: Understanding Multidimensional Choices via StatedPreference Experiments.” Political Analysis 22(1):1–30.

Hall, Peter A., ed. 2001. Varieties Of Capitalism: The Institutional Foundationsof Comparative Advantage. 1 edition ed. Oxford England ; New York: OxfordUniversity Press, U.S.A.

Hegre, Håvard and Nicholas Sambanis. 2006. “Sensitivity Analysis of Empirical Re-sults on Civil War Onset.” Journal of Conflict Resolution 50(4):508–535. Publisher:SAGE Publications Inc.

Herbst, Jeffrey Ira. 2000. States and Power in Africa: Comparative Lessons in Au-thority and Control. Princeton, N.J.: Princeton University Press.

Hicken, Allen. 2011. “Clientelism.” Annual Review of Political Science 14(1):289–310.

250

Hirschman, Albert O. 1970. Exit, Voice, and Loyalty: Responses to Decline in Firms,Organizations, and States. Cambridge, Mass: Harvard University Press.

Hobsbawm, E. J. 1992. Nations and Nationalism Since 1780: Programme, Myth,Reality. Cambridge University Press.

Hobsbawm, Eric and Terence Ranger. 1992. The Invention of Tradition. CambridgeUniversity Press. Google-Books-ID: sfvnNdVY3KIC.

Holzinger, Katharina, Florian G. Kern and Daniela Kromrey. 2016. “The Dualismof Contemporary Traditional Governance and the State : Institutional Setups andPolitical Consequences.” Political Research Quarterly 69(3):469–481.

Holzinger, Katharina, Roos Haer, Axel Bayer, Daniela M. Behr and Clara Neupert-Wentz. 2018. “The Constitutionalization of Indigenous Group Rights, TraditionalPolitical Institutions, and Customary Law , The Constitutionalization of Indige-nous Group Rights, Traditional Political Institutions, and Customary Law.” Com-parative Political Studies p. 0010414018774347.

Horowitz, Donald L. 1985. Ethnic groups in conflict. Berkeley: University of Califor-nia Press. OCLC: 10948982.

Hunter, Wendy and Natasha Borges Sugiyama. 2018. “Making the Newest Citizens:Achieving Universal Birth Registration in Contemporary Brazil.” The Journal ofDevelopment Studies 54(3):397–412.

Huntington, Samuel P. 1968. Political Order in Changing Societies. Yale UniversityPress.

Huntington, Samuel P. 1993. The Third Wave: Democratization in the Late 20thCentury. unknown edition ed. Norman: University of Oklahoma Press.

ILO. 1989. “Convention C169 - Indigenous and Tribal Peoples Convention, 1989 (No.169).”.

Inguanzo, Isabel. 2014. “Indigenous Peoples’ Rights in Southeast Asia.” Asian Journalof Political Science 22(1):45–70.

Ivison, Duncan. 2003. “The Logic of Aboriginal Rights.” Ethnicities 3(3):321–344.Publisher: SAGE Publications.

Jefremovas, Villia and Padmapani L. Perez. 2011. Definiting Indigeneity: Representa-tion and the Indigenous Peoples’ Rights Act of 1997 in the Philippines. In IdentityPolitics in the Public Realm. Ethnicity and Democratic Governance Series UBCPress.

Johannessen, Peter G. 2020. “Linkage Switches in Local Elections: Evidence from theWorkers’ Party in Brazil.” Comparative Political Studies 53(1):109–143.

251

Jung, Courtney. 2008. The Moral Force of Indigenous Politics: Critical Liberalismand the Zapatistas. 1 edition ed. Cambridge ; New York: Cambridge UniversityPress.

Katznelson, Ira. 2006. When Affirmative Action Was White: An Untold History ofRacial Inequality in Twentieth-Century America. Reprint edition ed. W. W. Norton& Company.

Kitschelt, Herbert and Steven I Wilkinson. 2007. Patrons, Clients and Policies: Pat-terns of democratic accountability and political competition. Cambridge UniversityPress.

Koter, Dominika. 2013. “King Makers: Local Leaders and Ethnic Politics in Africa.”World Politics 65(02):187–232.

Kuper, Adam. 2003. “The Return of the Native.” Current Anthropology 44(3):389–402.

Kyed, Helene Maria and Lars Buur. 2007. Introduction: Traditional Authority andDemocratization in Africa. In State recognition and democratization in Sub-SaharanAfrica: a new dawn for traditional authorities? New York: Palgrave Macmillanpp. 1–28.

Kymlicka, Will. 1996. Multicultural Citizenship: A Liberal Theory of Minority Rights.Reprint edition ed. Oxford: Clarendon Press.

Larreguy, Horacio. N.d. “Monitoring Political Brokers: Evidence from ClientelisticNetworks in Mexico, urldate = 2016-05-10, journal = Working Paper, year = 2013.”. Forthcoming.

Lee, Melissa M. and Nan Zhang. 2016. “Legibility and the Informational Foundationsof State Capacity.” The Journal of Politics 79(1):118–132.

Leeper, Thomas J., Sara B. Hobolt and James Tilley. 2020. “Measuring SubgroupPreferences in Conjoint Experiments.” Political Analysis 28(2):207–221. Publisher:Cambridge University Press.

Lemarchand, René. 1972. “Political Clientelism and Ethnicity in Tropical Africa:*Competing Solidarities in Nation-Building.” American Political Science Review66(1):68–90.

Leonen, Marvic. 2004. “Weaving Worldviews: Implications of Constitutional Chal-lenges to the Indigenous Peoples Rights Act of 1997.” Journal of the Integrated Barof the Philippines 30(1).

Levi, Margaret. 1989. Of Rule and Revenue. Berkeley: University of California Press.

Levi, Margaret, Audrey Sacks and Tom Tyler. 2009. “Conceptualizing Legitimacy,Measuring Legitimating Beliefs.” American Behavioral Scientist 53(3):354–375.

252

Lieberman, Evan. 2009. Boundaries of Contagion: How Ethnic Politics Have ShapedGovernment Responses to AIDS. Princeton University Press.

Lieberman, Evan S. and Prerna Singh. 2017. “Census Enumeration and Group Con-flict: A Global Analysis of the Consequences of Counting.” World Politics 69(1):1–53.

Lijphart, Arend. 1977. Democracy in Plural Societies: A Comparative Exploration.10th printing edition ed. New Haven: Yale University Press.

Lilla, Mark. 2017. The Once and Future Liberal: After Identity Politics. Harper.

Linz, Juan J. 1993. “State building and nation building.” European Review 1(4):355–369.

Logan, Carolyn. 2013. “The roots of resilience: Exploring popular support for Africantraditional authorities.” African Affairs 112(448).

Lugard, Lord Frederick J. D. 1965. The Dual Mandate in British Tropical Africa. 1edition ed. Routledge.

Mamdani, Mahmood. 1996. Citizen and Subject: Contemporary Africa and the Legacyof Late Colonialism. 1st edition ed. Princeton, N.J: Princeton University Press.

Mamdani, Mahmood. 2012. Define and Rule: Native as Political Identity. Cambridge,Mass: Harvard University Press.

Mann, Michael. 1984. “The autonomous power of the state: its origins, mechanismsand results.” European Journal of Sociology / Archives Européennes de Sociologie25(2):185–213. Publisher: Cambridge University Press.

Mansuri, Ghazala and Vijayendra Rao. 2004. “Community-Based and -Driven De-velopment: A Critical Review.” The World Bank Research Observer 19(1):1–39.Publisher: Oxford University Press.URL: https://www.jstor.org/stable/3986491

Mansuri, Vijayendra, Ghazala Rao. 2012. Localizing Development. Policy ResearchReports The World Bank.URL: https://elibrary.worldbank.org/doi/abs/10.1596/978-0-8213-8256-1

Marx, Anthony W. 1997. Making Race and Nation: A Comparison of South Africa,the United States, and Brazil. Cambridge Studies in Comparative Politics Cam-bridge: Cambridge University Press.

McGovern, Mike. 2012. Unmasking the State.

McGovern, Mike. 2016. “Uncertain Subjectivities and Conspiracy Thinking in anEbola Zone.”.

253

Mendoza, Ronald U., Jan Fredrick Cruz and David Barua Yap II. 2014. Political PartySwitching: It’s More Fun in the Philippines. SSRN Scholarly Paper ID 2492913Social Science Research Network Rochester, NY: .URL: https://papers.ssrn.com/abstract=2492913

Mershon, Carol. 2015. “The Political Determinants of Public Goods Provision inSouth Africa.” American Political Science Association Conference Paper .

Mershon, Carol. 2020. “What effect do local political elites have on infant and childdeath? Elected and chiefly authority in South Africa.” Social Science & Medicine(1982) 251:112902.

Migdal, Joel. 2001. State in Society: Studying How States and Societies Transformand Constitute One Another. Later printing used edition ed. Cambridge ; NewYork: Cambridge University Press.

Migdal, Joel S. 1988. Strong Societies and Weak States: State-society Relations andState Capabilities in the Third World. Princeton University Press.

Miguel, Edward. 2004. “Tribe or Nation? Nation Building and Public Goods in Kenyaversus Tanzania.” World Politics 56(03):328–362.

Miguel, Edward and Mary Kay Gugerty. 2005. “Ethnic diversity, social sanctions, andpublic goods in Kenya.” Journal of Public Economics 89:2325–2368.

Mill, John Stuart. 1861. Considerations on Representative Government. Amherst,NY: Prometheus.

Muniz, Cássio. 2015. “Ideology Versus Clientelism: Modernization and ElectoralCompetition in Brazil.” Theses and Dissertations .URL: https://dc.uwm.edu/etd/1015

Nagel, Thomas. 1973. “Equal Treatment and Compensatory Discrimination.” Philos-ophy & Public Affairs 2(4):348. Publisher: Princeton University Press.

Nannicini, Tommaso, Andrea Stella, Guido Tabellini and Ugo Troiano. 2013. “So-cial Capital and Political Accountability.” American Economic Journal: EconomicPolicy 5(2):222–250.

Ntsebeza, Lungisile. 2005. Democracy Compromised: Cheifs And the Politics of theLand in South Africa. 1st edition edition ed. Leiden: Brill Academic Pub.

Olson, Mancur. 2009. The Logic of Collective Action. Harvard University Press.

Platow, Michael J. and Daan van Knippenberg. 2001. “A Social Identity Analy-sis of Leadership Endorsement: The Effects of Leader Ingroup Prototypicalityand Distributive Intergroup Fairness.” Personality and Social Psychology Bulletin27(11):1508–1519.

254

Portes, Alejandro. 1998. Social Capital: Its Origins and Applications in Modern So-ciology. In Knowledge and Social Capital: Foundations and Applications. Elsevierpp. 69–88.

Postero, Nancy Grey. 2006. Now We Are Citizens: Indigenous Politics in Postmulti-cultural Bolivia. 1 edition ed. Stanford, Calif: Stanford University Press.

Prill-Brett, June. 1994. “Indigenous Land Rights and Legal Pluralism among Philip-pine Highlanders.” Law & Society Review 28(3):687–697.

Prill-Brett, June. 2007. “Contested Domains: The Indigenous Peoples Rights Act(IPRA) and Legal Pluralism in the Northern Philippines.” Journal of Legal Plural-ism and Unofficial Law 55:11–36.

Putnam, Robert D., Robert Leonardi and Raffaella Nanetti. 1993. Making democracywork : civic traditions in modern Italy. Princeton, N.J. : Princeton UniversityPress, c1993.

Rawls, John. 2005. Political Liberalism. Expanded edition ed. New York: ColumbiaUniversity Press.

Rodríguez-Piñero, Luis. 2006. Indigenous Peoples, Postcolonialism, and InternationalLaw: The ILO Regime. 1 edition ed. Oxford ; New York: Oxford University Press.

Rueschemeyer, Dietrich, Evelyne Huber and John D. Stephens. 1992. Capitalist de-velopment and democracy / Dietrich Rueschemeyer, Evelyne Huber Stephens, andJohn D. Stephens. Chicago : University of Chicago Press, 1992.

Schlozman, Kay Lehman, Henry E. Brady and Sidney Verba. 2018. Unequal andUnrepresented: Political Inequality and the People’s Voice in the New Gilded Age.Princeton University Press.

Scott, James C. 1998. Seeing like a state : how certain schemes to improve the humancondition have failed. Yale agrarian studies New Haven [Conn.] : Yale UniversityPress, c1998.

Scott, James C. 2010. The Art of Not Being Governed: An Anarchist History ofUpland Southeast Asia. Yale agrarian studies series edition ed. New Haven, Conn.:Yale University Press.

Seymour, Michel. 2017. A Liberal Theory of Collective Rights. 2 edition ed. MQUP.

Shadian, Jessica. 2010. “From states to polities: Reconceptualizing sovereigntythrough Inuit governance.” European Journal of International Relations 16(3):485–510.

Shapiro, Thomas M. 2005. The Hidden Cost of Being African American: How WealthPerpetuates Inequality. 1 edition ed. Oxford; New York: Oxford University Press.

255

Sharkey, Patrick. 2013. Stuck in Place: Urban Neighborhoods and the End of ProgressToward Racial Equality. University of Chicago Press.

Smith, Alastair and Bruce Bueno De Mesquita. 2012. “Contingent Prize Allocationand Pivotal Voting.” British Journal of Political Science 42(2):371–392.

Soifer, Hillel David. 2016. State Building in Latin America. Reprint edition ed. S.L.:Cambridge University Press.

Stokes, Susan C. 2005. “Perverse Accountability: A Formal Model of Machine Politicswith Evidence from Argentina.” American Political Science Review null(03):315–325.

Taylor, Charles, Kwame Anthony Appiah, Jürgen Habermas, Stephen C. Rockefeller,Michael Walzer and Susan Wolf. 1994. Multiculturalism: Examining the Politics ofRecognition. Expanded paperback edition ed. Princeton, NJ: Princeton UniversityPress.

Thomson, Judith Jarvis. 1973. “Preferential Hiring.” Philosophy & Public Affairs2(4):364. Publisher: Princeton University Press.

Tilly, Charles. 1992. Coercion, Capital and European States, A.D. 990 - 1992. Revisededition ed. Cambridge, Mass.: Wiley-Blackwell.

Tsai, Lily L. 2007. Accountability without democracy : solidary groups and publicgoods provision in rural China. Cambridge studies in comparative politics NewYork, NY : Cambridge University Press, 2007.

Tsai, Lily L. 2015. “Constructive Noncompliance.” Journal of Comparative Politics47(3).

Tyler, Tom R. 1997. “The Psychology of Legitimacy: A Relational Perspectiveon Voluntary Deference to Authorities.” Personality & Social Psychology Review(Lawrence Erlbaum Associates) 1(4):323.

Tyler, Tom R. 2006. “Psychological Perspectives on Legitimacy and Legitimation.”Annual Review of Psychology 57(1):375–400.URL: https://doi.org/10.1146/annurev.psych.57.102904.190038

Ulrich, Roberta. 2013. American Indian Nations from Termination to Restoration,1953-2006. Lincoln: University of Nebraska Press.

UN Commission on Human Rights. 1996. Report of the Working Group on the draftUnited Nations declaration on the rights of indigenous peoples. Technical report.

UNHCR. 2013. The United Nations Declaration on the Rights of IndigenousPeoples: A Manual for National Human Rights Institutions. Technical ReportHR/PUB/13/2 Asia Pacific Forum of Human Rights Institutions and the Officeof the United Nations High Commissioner for Human Rights.

256

Van der Windt, Peter, Macartan Humphreys, Lily Medina, Jeffrey F. Timmons andMaarten Voors. 2018. “Citizen Attitudes Toward Traditional and State Authorities:Substitutes or Complements?” Comparative Political Studies p. 0010414018806529.

Verba, Sidney, Norman H. Nie and Jae-on Kim. 1987. Participation and PoliticalEquality: A Seven-Nation Comparison. University of Chicago Press. Google-Books-ID: 7SksV_EPkiQC.

Weaver, Vesla M. and Amy E. Lerman. 2010. “Political Consequences of the CarceralState.” American Political Science Review 104(4):817–833. Publisher: CambridgeUniversity Press.

Weber, Eugen. 1976. Peasants into Frenchmen: The Modernization of Rural France,1870-1914. 1 edition ed. Stanford, Calif: Stanford University Press.

Weber, Max. 1919. Politics As a Vocation. Oxford University Press, New York.

Wimmer, Andreas. 2016. “Is Diversity Detrimental? Ethnic Fractionalization, PublicGoods Provision, and the Historical Legacies of Stateness.” Comparative PoliticalStudies 49(11):1407–1445.

World Bank. 2003. World Development Report 2004 Making Services Work for PoorPeople. Washington, D.C.: The World Bank.

Yashar, Deborah J. 2005. Contesting Citizenship in Latin America: The Rise of In-digenous Movements and the Postliberal Challenge. Cambridge ; New York: Cam-bridge University Press.

Zolberg, Aristide R. 1967. “Patterns of National Integration.” The Journal of ModernAfrican Studies 5(4):449–467.

257