Taxation and Representation - Thomas Bramborthomas.brambor.com/uploads/5/4/7/8/5478685/brambor.2012.taxati… · duces the need for taxation and as a result arguably provides less

TAXATION, NATURAL RESOURCES, AND REPRESENTATION

A DISSERTATION

SUBMITTED TO THE DEPARTMENT OF POLITICAL SCIENCE

AND THE COMMITTEE ON GRADUATE STUDIES

OF STANFORD UNIVERSITY

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS

FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

Thomas Brambor

August 2012

Dissertation Abstract and Annotated Chapter Outline

Abstract

This dissertation re-examines the links between the sources of gov-

ernment funding, representation, and accountability and analyzes whether

and how the sources of a government’s revenues are connected to the be-

havior of its political actors. I subject our well-trodden theoretical under-

standing of the relationship between taxation and representative govern-

ment to several empirical analyses in which both sub-national and na-

tional governments experience large, exogenous changes in the origins

of their funding. Particular prominence is given to the analysis of demo-

cratic governments, in which elections are indicators of representation

and at the same time can be tools to enforce accountability.

A source of government revenue that is often cast as the quintessen-

tial non-tax revenue or ’windfall revenue’ is income from natural resources.

There is a general understanding that income from natural resources re-

duces the need for taxation and as a result arguably provides less reason

for publics to demand representation and more opportunity to use these

monies for political or personal gain. Yet, the evidence for this claim is

rather sparse. Though cross-country empirical evidence does provide

reason to believe that natural resource dependence is associated with less

democracy (Ross, 2001a), some more recent analyses dispute the rela-

tionship (Haber and Menaldo, 2011). In order to provide stronger causal

evidence of the connection between the composition of government rev-

enue and representation, one would ideally want to observe exogenous

iv

changes to the sources of government revenue without accompanying di-

rect effects on the institutional environment. The discovery of substantial

natural resource deposits plausibly qualifies.

In the first part of this dissertation, I provide an extended data set

of oil production going back to 1918 and re-examine some of the claims

made in the resource curse literature. I review the existing literature and

find that the way natural resource abundance is measured essentially pre-

determines the empirical results one obtains. In addition, most of the

empirical power of these cross-national analyses is drawn from cross-

sectional variation, rather than over time institutional changes as a re-

sult of resource wealth, which is at the heart of the argument of the re-

source curse claim. Almost all the existing analyses rely on data starting

in 1970, long after many countries discovered and started producing oil.

In an effort to improve these analyses, I collected data on oil discovery

and production starting in 1918 for 98 oil-producing countries and find

no evidence for the impact of crude oil production on the quality of insti-

tutions. Instead, pre-existing institutions (before discovery) appear to be

the main drivers for the relationship I observe.

Among the issues arising in many empirical cross-country analyses,

including my own, are the use of coarse indices, non-randomly miss-

ing data, substantial heterogeneity among countries, and close interac-

tions of politics, institutions, and the economy. Consequently, part two of

this dissertation examines whether the negative effects of resource abun-

dance can be found on the sub-national level as well.

For the second part the main empirical case is Brazil, a large democ-

racy and one of the fastest growing oil producers in the world (in 2009 it

ranked 9th in the world and produced more oil than Venezuela and Iraq).

v

Due to the country’s federalist structure, petroleum royalties accrue to the

federal government and are subsequently partially redistributed to states

and municipalities. Interestingly, this redistribution occurs largely based

on the geographic distance of the receiving political units from the area

of production independent of other factors, such as need or population.

While some of Brazil’s more than 5,500 municipalities receive over fifty

percent of the value of their GDP from natural resource royalties, other

nearby municipalities receive no such income - a promising analogy to

cross-country analysis. In addition to the geographic assignment of re-

source royalties I use a change in the distribution of resource royalties

to Brazilian municipalities in 1997 to study the effects of external fund-

ing on the allocation of local government budgets. I test whether income

obtained from royalty payments is spent differently than comparable tax

income, and whether these effects differ depending on the local political

climate.

I present evidence that while income from both oil royalties and lo-

cal tax revenue increased spending, the municipal administration chose

to apply them to different areas of the budget. Royalty income is found to

be associated with spending on administration substantially more than

comparable increases in own taxation. In addition, there is some sup-

port for the hypothesis that competitive mayoral elections may decrease

spending on administration, though similar mediating effects are not ob-

served for other spending categories.

In addition, I use data on randomized audit reports of Brazilian mu-

nicipalities to investigate whether royalty rich municipalities are more

prone to corruption. While I do find evidence that royalty recipients are

actually less likely to be found corrupt with regard to their spending from

vi

federal grants, I argue that royalty income allows municipal politicians to

shift their rent-seeking behavior to revenues from royalties which are not

checked by these federal audits and in general are scrutinized less by both

government auditors and arguably voters as well.

Though the effects on the allocation of spending are of interest, we

are ultimately concerned with whether royalty income translates into an

improvement of developmental outcomes such as income, health, edu-

cation, and housing. Compared to own tax resources, spending gener-

ated from royalty income is inefficient in raising the living standards of

the citizens in royalty rich municipalities. Comparing royalty rich mu-

nicipalities to their oil-poor peers, I find that the additional income does

increase housing quality, income, and longevity. Education outcomes on

the other hand may even be negatively affected by royalty wealth.

In order to probe deeper the connection of taxation and electoral

accountability, the final section of the analysis of Brazilian subnational

data examines the effects of resource royalties on the electoral prospects

of mayors in these royalty rich localities. I find that resource royalties are

associated with increased electoral turnout but reduced electoral compe-

tition. Royalties decrease the number of candidates running for election

and increase the chances to be reelected significantly.

In summary, I find, on the one hand, evidence that access to non-

tax resources indeed allows governments to choose inefficient spending

patterns and be less concerned about the well-being of its constituents.

On the other hand, electoral accountability remains a tool, though weak-

ened, to mediate these negative effects, particularly by informed voters.

vii

Annotated Chapter Outline

Chapter 1: Re-examining the Linkages between Taxation, Representa-

tion, and Natural Resources

The introductory chapter reviews some of the main theoretical con-

tributions to our understanding of the relationship between taxation, rep-

resentation, and accountability. In addition, it outlines an argument based

on informational asymmetry between politicians and citizens, helping to

theoretically connect non-tax resources to reduced accountability.

Chapter 2: Discovery of Resources as a Source of Exogeneity

Empirically, natural resource abundance has been found to be strongly

associated with a host of negative outcomes in cross-country time-series

analyses. I review the existing literature and find that virtually all arti-

cles employing the ratio of primary commodity exports over GDP as the

measure of resource abundance find strong evidence for the ’resource

curse’ while those using per capita production or reserves as the measure

of resource abundance find evidence against the ’resource curse’. Con-

trary to claims in the literature that institutions are the result of resource

abundance, we propose to view the quality of pre-existing intuitions as

the origin of natural resource dependence. Natural resource abundance

translates into natural resource dependence only in the presence of poor

institutions. In an empirical analysis using a panel of 98 oil- producing

countries from 1918-2004, I find no evidence for the impact of crude oil

production on the quality of institutions.

Chapter 3: Do natural resource royalties make leaders unaccountable?

The peculiar form of resource royalty distribution in Brazil provides

viii

an excellent testing ground for the relationship between taxation and ac-

countability. Resource royalties in Brazil accrue to municipalities based

on an exogenous, geographically determined rule. I test whether income

obtained from royalty payments is spent differently than comparable tax

income, and whether these effect differ depending on the local political

climate. I present evidence that while income from both oil royalties and

local tax revenue increased spending, municipal administrations chose

to apply them to different areas of the budget. Royalty income is found

be associated with spending on administration substantially more than

comparable increases in own taxation. In addition, there is some sup-

port for the hypothesis that competitive mayoral elections may decrease

spending on administration, though similar mediating effects are not ob-

served for other spending categories.

In addition, I use data on randomized audit reports of Brazilian mu-

nicipalities to investigate whether royalty rich municipalities are more

prone to corruption. While I do find evidence that royalty recipients are

actually less likely to be found corrupt with regard to their spending from

federal grants, I argue that royalty income allows municipal politicians to

shift their rent-seeking behavior to revenues from royalties which are not

checked by these federal audits and in general are scrutinized less by both

government auditors and arguably voters as well.

Chapter 4: The effect of resource royalties on public goods provision

The efficient use of limited government resources to provide public

goods is a vexing issue, in particular for many developing countries. In

ix

the previous chapter, I present evidence that while royalty rich munici-

palities exhibit above average increases in areas of administrative spend-

ing, royalty payments increase spending in all areas of the municipal bud-

get, including education and culture, health and sanitation, and hous-

ing and urbanization. In this chapter, I analyze whether these increases

translate into the desired outcomes for the developmental outcomes that

ultimately matter, such as literacy rates, child mortality, income, and the

quality of housing. I find mixed success with regard to the resources ap-

plied from natural resources royalties. Low quality housing improves,

school fees drop, and income and longevity rise in municipalities with

large royalty incomes. Educational outcomes, on the other hand, seem

not to benefit from the largesse or are even negatively affected.

Chapter 5: Resource wealth and electoral competition

Having discussed the allocation and effects of spending in the pre-

vious chapters, this section focuses on the political effects of natural re-

source income. Using electoral data on 45,000 mayoral candidates in four

municipal elections (1996 - 2008), I estimate whether incumbent may-

ors in royalty rich municipalities have a higher chance to be re-elected. I

employ a regression discontinuity design and a comparison of repeated

candidate pairs to estimate the effect of incumbency. I find that resource

royalties are associated with increased electoral turnout but reduced elec-

toral competition. Royalties decrease the number of candidates running

for election and increase the chances of incumbents to be reelected sig-

nificantly.

x

Contents

1 Re-examining the Linkages between Taxation, Representation,

and Natural Resources 1

1.1 Linking Taxation, Representation, and Accountability . . . . 12

1.1.1 Taxation and Representation . . . . . . . . . . . . . . . 12

1.1.2 Representation and Accountability . . . . . . . . . . . 23

1.1.3 Taxation and Spending . . . . . . . . . . . . . . . . . . . 25

1.2 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

I Natural Resources in Cross-National Perspective 31

2 Discovery of Resources as a Source of Exogeneity 32

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

2.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . 36

2.3 Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

2.3.1 Natural resource dependence and the security of prop-

erty rights . . . . . . . . . . . . . . . . . . . . . . . . . . 44

2.4 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

2.4.1 Production and Reserves of Crude Oil . . . . . . . . . . 48

2.4.2 Quality of Institutions . . . . . . . . . . . . . . . . . . . 50

2.5 Empirical Evidence . . . . . . . . . . . . . . . . . . . . . . . . . 54

xii

2.5.1 Level of Institutional quality . . . . . . . . . . . . . . . 54

2.5.2 Trends in Institutional quality . . . . . . . . . . . . . . 57

2.5.3 Panel Regressions . . . . . . . . . . . . . . . . . . . . . . 61

Alternative Model Specifications . . . . . . . . . . . . . 62

2.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

2.7 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

II Resource Royalties on the Subnational Level 77

3 Do natural resource royalties make leaders unaccountable? 78

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

3.2 Institutional Background . . . . . . . . . . . . . . . . . . . . . 84

3.2.1 The Role of Municipalities in the Brazilian Federal

Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

3.2.2 Royalty distribution and Petroleum Law . . . . . . . . 86

3.3 The effects of non-tax revenue . . . . . . . . . . . . . . . . . . 91

3.4 Empirical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 95

3.4.1 Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96

3.4.2 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

3.4.3 Summary Statistics and Spatial Patterns . . . . . . . . 101

Distribution of Natural Resource Revenues . . . . . . . 102

Location of Royalty Receiving Municipalities . . . . . . 107

Municipal Characteristics . . . . . . . . . . . . . . . . . 108

Budgetary Items . . . . . . . . . . . . . . . . . . . . . . . 110

Election Information . . . . . . . . . . . . . . . . . . . . 113

3.4.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116

Suitability of Arellano-Bond Dynamic Panel GMM Model116

xiii

Effects of Royalties on Municipal Spending . . . . . . . 117

Effects of Royalties and Local Elections . . . . . . . . . 119

3.4.5 Royalties and Corruption . . . . . . . . . . . . . . . . . 124

Examples of corruption in royalty receiving munici-

palities . . . . . . . . . . . . . . . . . . . . . . . 127

Data on audit reports . . . . . . . . . . . . . . . . . . . . 130

3.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136

3.6 Tables and Figures . . . . . . . . . . . . . . . . . . . . . . . . . 139

3.A Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146

4 The effect of resource royalties on public goods provision 148

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148

4.1.1 Public spending and development . . . . . . . . . . . . 149

4.1.2 Does democracy help the poor? . . . . . . . . . . . . . 153

4.2 Empirical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 157

4.2.1 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157

4.2.2 Model and Results . . . . . . . . . . . . . . . . . . . . . 157

Developmental Effects of Resource Royalties . . . . . . 157

Effects of Royalties on Education . . . . . . . . . . . . . 166

4.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176

4.4 Tables and Figures . . . . . . . . . . . . . . . . . . . . . . . . . 178

5 Resource wealth and electoral competition 183

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183

5.2 How do resource royalties affect institutions? . . . . . . . . . 185

5.3 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192

5.4 Methodology & Analysis . . . . . . . . . . . . . . . . . . . . . . 193

5.4.1 Political Competition and Outcomes in Municipalities 194

xiv

5.4.2 Resource Royalties and Incumbency Effects . . . . . . 198

Regression Discontinuity Design . . . . . . . . . . . . . 201

Difference-in-Difference Estimation . . . . . . . . . . . 207

5.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210

6 Conclusion 212

Bibliography 215

xv

List of Tables

2.1 Correlations between institutional variables . . . . . . . . . . 52

2.2 Impact of Oil Production on Polity Score . . . . . . . . . . . . 55

2.3 Impact of Oil Production on Trend of Polity Score . . . . . . . 58

2.4 Effect of Oil Production on Polity Scores . . . . . . . . . . . . 62

2.5 Effects of Natural Resources on Polity Score - Comparison

of Model Specifications: 1918 - 2000 . . . . . . . . . . . . . . . 66

2.6 Effects of Natural Resources on Polity Score - Comparison

of Model Specifications: 1970 - 2000 . . . . . . . . . . . . . . . 67

2.7 Average Polity Scores before and after Start of Petroleum Pro-

duction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

2.8 Impact of Oil Production on PolCon3 Score . . . . . . . . . . . 74

2.9 Impact of Oil Production on Trend of PolCon3 Score . . . . . 74

3.1 Historical Development of Federal Division of Tax Burden

(1960 - 2005) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

3.2 Contingency table - Types of Royalty Incomes (in 2007) . . . 103

3.3 Ranking of Municipalities by GDP per capita (1999- 2004) . . 105

3.4 Comparing Corruption Statistics of Top 100 Royalty Recipi-

ents with Oil Poor Municipalities . . . . . . . . . . . . . . . . . 132

3.5 Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . . 139


xvi

3.6 Effects of Resource Royalties on Municipal Budget Expendi-

tures in Brazil

(DV: Per Capita Spending on each Budget Category; Estima-

tion: xtabond2) . . . . . . . . . . . . . . . . . . . . . . . . . . . 142

3.7 Effects of Mayoral and City Council Elections on Municipal

Budget Expenditures in Brazil


tion: xtabond2) . . . . . . . . . . . . . . . . . . . . . . . . . . . 143

3.8 Effects of Resource Royalties on Municipal Budget Expendi-

tures in Brazil


tion: xtreg) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146

3.9 Effects of Mayoral and City Council Elections on Municipal

Budget Expenditures in Brazil


tion: xtreg) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147



4.2 Effects of Resource Royalties on Educational Outcomes in

Brazil

(DV: Passing Rates and Student-Professor Ratio in Munici-

pal Primary Schools; Estimation: xtabond2) . . . . . . . . . . 180

5.1 Descriptive Statistics of Mayoral Candidates, Elections, and

Municipalities in Oil vs. Non-Oil Municipalities (1996-2008) . 195

5.2 Effects of Resource Royalties on Electoral Competition and

Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197

xvii

5.3 Regression Discontinuity Results of the Effect of Incumbency

on Vote Margin (with varying bandwidths) . . . . . . . . . . . 207

5.4 The Mediation of Incumbency Effects by Resource Royalties

in Brazilian Mayoral Elections (2000-2008) . . . . . . . . . . . 210

xviii

List of Figures

2.1 Impact of Oil Production on Level of Polity Score . . . . . . . 56

2.2 Impact of Oil Production on Trend of Polity Score . . . . . . . 59

2.3 Impact of Oil Production on Change in Polity Score . . . . . . 60

2.4 Box and Density Plots of Polity Scores before and after Start

of Oil Production . . . . . . . . . . . . . . . . . . . . . . . . . . 70

2.5 Selected Country Examples . . . . . . . . . . . . . . . . . . . . 72

2.6 Impact of Oil Production on Level of PolCon3 Score . . . . . . 75

2.7 Impact of Oil Production on Trend of PolCon3 Score . . . . . 76

3.1 Determining royalty distribution to Brazilian states and mu-

nicipalities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

3.2 Comparing Royalty Regime before (less than 5%) and after

(5 to 10%) the Petroleum Law . . . . . . . . . . . . . . . . . . . 104

3.3 Royalties by State . . . . . . . . . . . . . . . . . . . . . . . . . . 106

3.4 Geographical Distribution of Royalties in Brazil . . . . . . . . 109

3.5 Municipal Oil and Gas Royalties in Brazil (1993-2010) . . . . . 110

3.6 Royalty Receipts vs. Municipal Spending (1996-2008) . . . . . 112

3.7 Effect of Royalties on Selected Budget Spending Categories

Comparing Oil Royalty and Non[Pleaseinsertintopreamble]Receiving Municipalities

in Brazil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

3.8 Value Audited vs. Revenue per capita . . . . . . . . . . . . . . 135

xix

3.9 Municipal budget of Quissamã in comparison to average

municipality in Brazil . . . . . . . . . . . . . . . . . . . . . . . 141

3.10 Marginal Effect of Total Royalties per Capita

conditional on competitiveness of mayoral and council elec-

tions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144

3.11 Marginal Effect of Own Revenue per Capita


tions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

4.1 Developmental Effects of Resource Royalties (1991 to 2000) . 162

4.2 Developmental Effects of Resource Royalties (2000 to 2010) . 165

4.3 Reading performance, income, and education expenditure . 167

4.4 Effects of Resource Royalties on Education . . . . . . . . . . . 169

4.5 Educational Outcomes and Sources of Revenue (1996-2006) . 171

4.6 Marginal Effect of Total Royalties per Capita on Education


tions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181

4.7 Marginal Effect of Own Revenue per Capita on Education


tions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182

5.1 Regression Discontinuity Estimate of the Incumbency Ef-

fect by Royalty Receipt . . . . . . . . . . . . . . . . . . . . . . . 204

5.2 Regression Discontinuity Estimate of the Incumbency Ef-

fect by Royalty Receipt and Election Year . . . . . . . . . . . . 205

xx

Taxation and Representation - Thomas Bramborthomas.brambor.com/uploads/5/4/7/8/5478685/brambor.2012.taxati… · duces the need for taxation and as a result arguably provides less

Documents