<<

PRIMACY AND POLITY:

THE ROLE OF URBAN POPULATION IN POLITICAL CHANGE

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

By

Robert Michael Anthony, M.A.

Sociology Graduate Program

The Ohio State University

2009

Dissertation Committee:

Professor Edward M. Crenshaw, Advisor

Professor J. Craig Jenkins

Professor Kazimierz Slomczynski

Copyright by

Robert Michael Anthony

2009

ii

ABSTRACT

The study of political change, and in particular the causes of , has a long within the social sciences among cross-national comparative scholars interested in international development. Most often political change has been explained in terms of its connection to a nation’s level of . Although the exact nature of the development/ relationship has been a point of disagreement among social scientists, the premise that there is a relationship is widely accepted, debated and tested.

Early explanations for the development/democracy relationship focused on a broad set of explanatory variables (Lipset 1959). Since then most cross-national development scholars have reduced the concept of “development” to mean economic development—at least in terms of their empirical measures. In simplifying this concept, the role of other contributing factors which were once understood to be central components of the overall development process has been largely ignored and side-lined in empirical analyses. This has been especially true of urban population.

In light of the above, this dissertation is aimed at challenging the notion that the relationship between development and democracy should be understood only as a relationship between level of economic development and democratization. Even more broadly, this dissertation challenges the notion that economic factors are the

iii most important for understanding macro political change. Indeed, while economic factors are certainly a central contributing factor for political change, urban populations and their social contexts are equally important since it is urban dwellers who are predominately engaged in modern exchanges rather than abstract

“markets.”

Thus, this dissertation explores the role urban populations and their contexts within nation-states play in eliciting political change. In particular, it focuses on the dimensions of urban population within nation-states including; the absolute size, the degree of concentration, the degree of urban primacy, and the balance of the urban hierarchy. The empirical analyses reveal that indicators for each of the above dimensions perform as well or better than economic indicators when explaining political change in developing nations. In short, the theoretical arguments and empirical evidence generated in this dissertation make it clear that cross-national comparative scholars cannot afford to ignore the role that urban populations and their contexts play in eliciting political change in an increasingly urban world.

iv

Dedicated to my wife

v

ACKNOWLEDGMENTS1

I wish to thank my wife and family for their continued support over the course of my graduate studies.

I wish to thank my advisor, Edward M. Crenshaw, for intellectual support, encouragement, and direction which made this dissertation possible, and for his patience in correcting my theoretical insights.

I thank J. Craig Jenkins for his efforts to support my tenure as a graduate student in the department in difficult financial times. I also wish to thank Kazimierz

(Maciek) Slomczynski for his insights on the comparative methodology which were used for designing this study.

I am grateful to the entire department and faculty who gave me the opportunity to grow my intellectual abilities.

I wish to thank Kristopher K. Robison who provided assistance with data collection and management.

Finally, I wish to thank the Mershon Center for International Security Studies for funding as well as Elizabeth Cooksey and Keck at the Center for Human

Resource Research who provided me with funding and a quite office space in my final year as a graduate student.

vi

VITA

February 12th, 1976………………………...... Born – Sidney, Ohio

1998………………………………………………………..B.A. Bowling Green State University, Bowling Green, OH

2002………………………………………………………. M.A. University of Toledo, Toledo, Ohio

2002-Present…………………………………………..Independent Instructor, Research Assistant, Department of Sociology, The Ohio State University

FIELDS OF STUDY

Major Field: Sociology

Topics of Interest: Sociological Theory, , International Development, , , Epistemology, Identity, Group Behavior, Sociobiology,

vii

TABLE OF CONTENTS

Page

Abstract……………………………………………………………………………………………………………...iii Dedication…………………………………………………………………………………………………………...v Acknowledgements……………………………………………………………………………………………..vi Vita……………………………………………………………………………………………………………………vii List of Tables……………………………………………………………………………………………………….xi List of Figures……………………………………………………………………………………………………xiii

Chapters:

Part I……………………………………………………………………………………………………………...... 1 Introduction…………………………………………………………………………………………...... 2

1. Classical Theories and the Role of Population in Social Change……………………7

1.1 August Comte: The Law of the Three Stages…………………………………………..7 1.2 Herbert Spencer: Population Pressure and Compounding……………………...9 1.3 Emile Durkheim: Population Pressure and the Role of Specialization…...14 1.4 Marx and Engels: Population Concentration as an Economic Outcome….19 1.5 Conclusion…………………………………………………………………………………………22

2. Contemporary Theories and the Role of Population in Social Change…………23

2.1 …………………………………………………………………….....23 2.1.1 Modernization and the …………………….24 2.1.2 and the Organization of Population…………………...... 30 2.1.3 Population Size, Density, and Production Systems………………….32 2.2 Dependency and World-System Theory………………………………………………39 2.2.1 The World-System………………………………………………………………..40 2.2.2 The Capitalist World-Economy……………………………………………...44 2.2.3 Stratification in the Capitalist World-Economy………………………46 2.2.4 Uneven Development and Population…………………...……………….49 2.3 Human Ecological Theory…………………………………………………………………..54 2.3.1 The Ecological Complex …………………………………………...…………..56 2.3.2 The Role of Population in System Change………………………………66 2.4 Conclusion…………………………………………………………………………………………77

viii

Part II………………………………………………………………………………………………………………..87 Introduction……………………………………………………………………………………………88

3. Economic Development and the Urban Transition…………………………………….89

3.1 Urban Areas and Economic Systems……………………………………………………90 3.2 Theoretical Approaches to and Economic Development.....93 3.2.1 Dominant Schools of Thought…………………………………………….....94 3.2.2 Economies of Scale and Agglomeration Effects……………………..103 3.2.3 The Urban Bias Hypothesis………………………………………………….107 3.3 and Economic Development………………………………….110 3.4 Urban Primacy, Concentration, and Economic Development……………...114 3.4.1 Empirics of Urban Primacy and Economic Development………119 3.5 Conclusion……………………………………………………………………………………….123

4. Economic Development and Democratization…………………………………………125

4.1 Economic Development as a Cause of Political Democracy…………………128 4.1.1 Development and Democracy Studies………………………………….130 4.2 Political Democracy as a Cause of Economic Growth………………………….142 4.3 A Critique of Development/Democracy Studies…………………………………149

5. Urbanization and Political Systems………………………………………………………...153

5.1 The Effects of Political Systems on Urbanization………………………………..158 5.2 The Effects of Urbanization on Political Systems………………………………..168 5.2.1 Indirect Effects of Urbanization on Political Systems…………….169 5.2.2 Linking Urbanization to Political Systems…………………………….176

PART III…………………………………………………………………………………………………………...196 Introduction………………………………………………………………………………………….197

6. Measures, Data, and Methods…………………………………………………………………206

6.1 Analytical Variables and their Measures……………………………………………206 6.1.1 Measures of Political Systems……………………………………………...206 6.1.2 Measures of Urban Population and its Structure…………………..212 6.2 Other Important Independent Variables……………………………………………226 6.2.1 Geographic Controls……………………………………………………..…….226 6.2.2 Urban Population Controls..………………………………………………...228 6.2.3 Kilometers of Paved Roads………………………………………………….228 6.2.4 Real Gross Domestic Product per Capita………………………………230 6.2.5 Education…………………………………………………………………………...230 6.2.6 Colonial History………………………………………………………………….232 6.2.7 Exports, FDI, and Primary Commodity Exporters…………………233 6.3 Statistical Methodology…………………………………………………………………….235

ix

6.3.1 OLS with Panel-Corrected Standard Errors…………………………..237 6.3.2 “To lag or not to lag?”………………………………………………………….242 6.3.3 Testing and Correcting Serial Correlation…………………………….245 6.3.4 Heterogeneity in Time-Series Cross-Section Data…………………247

7. Results…………………………………………………………………………………...... 249

7.1 Main Findings……………………………………………………………………………….…249 7.2 Alternative Models…………………………………………………………………..………264 7.3 Exploratory Models………………………………………………………………………….270

8. Conclusion………………………………………………………………………………...... 282

8.1 Summary…………………………………………………………………………………………282 8.2 Discussion……………………………………………………………………………………….285

Appendix A………………………………………………………………………………………………………291

Nations and Years of Urban Dataset………………………………………………………..291 Polity IV Dataset (Dependent Variable)…………………………………………………..295 Urban Population (Primacy) Dataset Sources………………………………………….298 Position in the World-System (Snyder and Kick [1979])………………………….300 Colonial History by Nation……………………………………………………………………..301 Communist History by Nation………………………………………………………………..301

Appendix B………………………………………………………………………………………………………302

List of Nations Tables 7.1, 7.3, 7.3.1, 7.3.2, and 7.4…………………….………….....308 List of Nations Table 7.1.1…………………………………………………………………...…309 List of Nations Tables 7.2 and 7.2.1……………………………………………………...... 310 List of Nations Table 7.2.2……………………………………………………………………...311 List of Nations Table 7.2.3……………………………………………………………………...312

Bibliography…………………………………………………………………………………………………….313 Data Sources………………………………………………………………………………………….326

x

LIST OF TABLES

Table Page

2.1 A Comparison of Modernization Theory, World-System Theory, and Human Ecological Theory on Selected Dimensions related to Population and Social Change………………………………..79

4.1 Comparisons of Development Proxies in a Sample of DevelopmentDemocracy Studies…………………………..…………………………….151

6.1 Simple Statistic Measures of All Dependent and Independent Variables…………………………………………………………………………..236

7.1 Urban Population Structure (Px, P14, Pi and Log of D) Regressed on Combined Polity Score……………………………………………………...253

7.1.1 Urban Population Structure (Px, P14, Pi and Log of D) Regressed on Combined Polity Score Controlling for Secondary Education…………………...302

7.2 Urban Population Structure (Px, P14, Pi and Log of D) Regressed on Combined Polity Score Controlling for Exports as a Percent of GDP…………266

7.2.1 Urban Population Structure (Px, P14, Pi and Log of D) Regressed on Combined Polity Score using Limited Sample from 7.2……………………………303

7.2.2 Urban Population Structure (Px, P14, Pi and Log of D) Regressed on Combined Polity Score Controlling for FDI as a Percent of GDP………………..304

7.2.3 Urban Population Structure (Px, P14, Pi and Log of D) Regressed on Combined Polity Score using Limited Sample from 7.2 and Omitting , Albania, (Former) and Iraq…………………………………..305

7.3 Urban Population Structure (Pi) Regressed on Combined Polity Score Controlling for Area, Climate/Geography, and Primary Commodity Export Status………………………………………………….277

7.3.1 Urban Population Structure (Px, P14, Pi and Log of D) Regressed on Combined Polity Score……………………………………………………...306

xi

Table Page

7.3.2 Urban Population Structure (Pi) Regressed on Combined Polity Score Controlling for Area, Climate/Geography, and Primary Commodity Export Status………………………………………………….307

7.4 Urban Population Structure (Pi) Regressed on Combined Polity Score Controlling for Area, Climate/Geography, PCE Status, and Colonial/Communist History…………………………………………279

xii

LIST OF FIGURES

Figure Page

1.1 Tested and Untested Relationships in Development Studies involving Urbanization, Economic Growth, and Political Systems…………………………………………………………………………………4

1.2 Spencer’s Model of the Effects of on Social Organization…………………………………………………………………………...... 11

1.3 Durkheim’s Model of the Effects of Population Growth on Social Change……………………………………………………………………………………..16

2.1 Modernization Theory’s Basic Model of Change………………………………………..25

2.2 The Demographic Transition……………………………………………………………………29

2.3 The Relationship between Density and Efficiency in Modern and Traditional Economic Systems………………………………………………35

2.4 Economic-Political Arrangements according to World-System Theory……………………………………………………………………………..43

2.5 The Ecological Complex…………………………………………………………………………..58

2.6 Ecological Hierarchy of Power Structure by Function’s Distance from Environment and by Degree of Function’s Specialization ………………………………………………………………………………………....67

2.7 Ecological Model of Social Change as a Function of Population Adaptation to Environment………………………………………………...71

3.1 Main Focus of Interest for Modernization/Ecology and Dependency/World-System on the Relationship between Urbanization and Economic Growth…………………….…………………………………96

3.2 Economic Growth and the Urban Transition: Modernization and Human Ecology…………………………………………………………97

xiii

Figure Page

3.3 Economic Growth and the Urban Transition for Developing of the World: Dependency/World-System………………………………….101

3.4 Predicted Relationship between Economic Returns on Urban Primacy and Level of Urban Primacy……………………………………….118

4.1 Relationship between Economic Growth and Democratization………………..126

4.2 Linear versus Curvilinear Relationship between Development and Democracy………………………………………………………………...135

5.1 Tested and Untested Relationships in Development Studies involving Urbanization and Political Systems……………………………...159

5.2 Early Causal Models of Democratic Political Development with Urbanization as Direct and Indirect Effects……………………………………..172

6.1 -Size Distributions in Guatemala 1950 Indicating a Primate Urban Hierarchy…………………………………………………….219

xiv

PART I

1

INTRODUCTION

By 2008 half of the world’s population lived in urban areas.1 Moreover, as stated in the United Nation’s World Urbanization Prospects: The 2007 Revision:

…the world urban population will likely increase by 3.1 billion between 2007 and 2050, passing from 3.3 billion to 6.4 billion. The expected rise in the urban population surpasses that for the whole over the same period (2.5 billion), implying that urban areas are expected to absorb not only all the population growth expected over the next four decades but also some of the rural population, through rural-urban migration or via the transformation of rural settlements into urban centres. (italics mine)

Even though Cohen (2003) has suggested that “most of the [urban] growth over the next 25 years will not take place in mega- … but will occur in far smaller cities and ” (p. 25), the continued world-wide urban transition will require cross- national comparative scholars to refine their understanding of the population component of the urban process. Specifically, there is a growing need to develop a greater understanding of how modern population pressures that are unique to urban contexts impact power relations and stratification systems. The first step in this endeavor is to understand the fundamental impact urban populations have on the organization of formal power structures—i.e., the nation-state. However, no such inquiries have been made within cross-nation studies of development in any discipline. This is despite the well accepted knowledge within the social sciences

1 Source: World Urbanization Prospects: The 2005 Revision “Fact Sheet 1.”

2 that large and dense populations (i.e. urban populations) require a more complex and decentralized regulatory system (e.g. Spencer, Durkheim, Smith).

Although cross-national researchers have failed to explore how urban transitions contribute to macro political change, the concern among the world’s governments as to what impact urban populations are having on their economic and political stability has increased over the years. In 2007 the reported

“85 percent of Governments expressed concern about their pattern of population distribution, a percentage comparable to that recorded in the 1970s” (World

Urbanization Prospects: The 2007 Revision). Not surprisingly the highest recorded concern came from African governments and other developing regions where rapid urbanization/population growth, disease, malnutrition, civil unrest, terrorism, genocide, civil war, and war all remain constant threats to political and economic stability. Yet these concerns are not isolated to just the undeveloped regions of the world. The UN also found that over one-third of developed nations expressed a

“major concern” with the effect urbanization was having on their population distribution (another one-third expressed a “minor concern”).

In light of the above, and with the rapid pace at which the urban transition continues to unfold around the world, it is only logical that cross-national scholars interested in macro political change begin to take the effect urban populations are having on development more seriously. Figure 1.1 offers an illustration of the various relationships between urbanization, economic development, and political

3

Urbanization

Political Systems

Economic Growth

Tested Untested Untested Relationship

Figure 1.1: Tested and Untested Relationships in Development Studies involving

Urbanization, Economic Growth, and Political Systems

systems that have been theoretically and empirically scrutinized in the last fifty to sixty years by sociologists, economists, and political scientists in cross-national comparative research. As Figure 1.1 shows, the relationship between urbanization and economic growth has received a great deal of attention (e.g. Berry 1961; El-

Shakhs 1972; Banks 1974; Moir 1976, 1977; Alonso 1980; Wheaton and Shishido

1981; Moomaw and Shatter 1996; Henderson 2000, 2003), as has the relationship between economic systems and emergent patterns of urbanization (e.g. Hoselitz

1955; Lipton 1977; Kentor 1981; Bradshaw 1985; and Smith 1988;

Crenshaw 1991, 1998; Brockerhoff and Brennan 1998). Similarly, economic growth has been linked to changes in political systems, and political systems have been partially linked to decreases and/or increases in economic growth (e.g. Lipset

1959; Cutright 1963; Neubauer 1967; Jackman 1973; Bollen 1979,1980, 1983;

Bollen and Jackmann 1985a, 1985b, 1995; Przeworski and Limongi 1997). More

4 recently there has been a building literature linking political systems (i.e. regime types) to patterns of urbanization—specifically, urban primacy and concentration have both been tied to the policies of autocratic regimes (Sawers 1989; Hansen

1990; Lyman 1992; Ades and Glaeser 1995; Davis and Henderson 2003; Henderson and Wang 2007). Yet despite the interlocking and overlapping research which has been aimed at exploring the various paralleling relationships between urban systems, political systems, and economic systems, there have been no studies in cross-national comparative research which link urban population to changes in the political systems of modern nation-states. A few partial exceptions are the work of

Owen and Witton (1973) who linked urban concentration to incidences of international war, and Crenshaw (1980) who linked urbanization to terrorism.

However, neither of the above addressed the broader more fundamental question; what affect, if any, do urban populations and their arrangements have on a nation- state’s tendency to adopt a given polity?

In order to address the above question and offer an empirical assessment of this “untested” relationship, it is important to understand and review the theory and past research which has been used to support each of the “tested” relationships.

Therefore, this dissertation is arranged into three parts. The first part provides the reader with the theoretical background needed to understand why one might expect urban population pressures and urban contexts to be related to macro level political change. The second part provides an extensive review of both the theoretical and empirical literature that has been used to support each of the “tested” relationships; it also offers insight into how past studies and theory can be applied to the central

5 problematic of this dissertation. Finally, the third part presents an empirical assessment of how urban population pressures and urban contexts contribute to political change.

In the next two chapters I present the reader with the theoretical background needed to appreciate why exploring and empirically testing the “untested” relationship between urban population and political systems is justified and necessary. Chapter 1 offers a brief but focused review of classical theoretical understandings on the relationship between population and social change. The second chapter explores how contemporary theories view the role of population in eliciting macro social change. Both chapters provide the reader with background information needed to understand the reviews contained in Part II. Therefore, the reader should be familiar with the concepts and theoretical approaches covered in both chapters since the analysis presented in Part III is based on the fundamental understanding that the redistribution of rural populations into modern (urban) settings results in social changes that have a direct impact on how societies meet their regulatory needs. In short, it is imperative for the reader to understand how population pressures work to alter social relations and create emergent contexts that are more or less compatible with specific types of modern political systems.

6

CHAPTER 1

CLASSICAL THEORIES AND THE ROLE OF POPULATION IN SOCIAL CHANGE

1.1 Auguste Comte: The Law of the Three Stages

Since the inception of sociology there has been a special focus on the relationship between population and social change. In fact, these inquiries can be traced to Enlightenment thinkers who were the first to ask questions about the human condition.2 However, while Enlightenment thinkers certainly shaped the direction of early social thought, it was Auguste Comte (1798-1857) who was the first to devise a coherent theoretical statement on the relationship between human populations and social change. Drawing from his mentor Saint Simone, as well as the influential writings of Condorcet, Turgot and Montesquieu’s The Spirit, Comte devised the Law of the Three Stages (Sica 2005). In many ways these “laws” represent the first theoretical attempt to explain modern macro social change as a function of population.

According to Comte, changes in social organization were dictated by the way a population explained the nature of reality. As a population devised more complicated philosophical methods for explaining reality, its organization and moral conduct became more complex. Moreover, Comte believed “Truth” was universal

2 Locke (1632-1704), Voltaire (1689-1755), Montesquieu (1689-1755), Rousseau (1712-1778), Smith (1723-1790), Kant (1724-1804), Hume (1711-1776), and Bentham (1748-1832) are a few examples of the early enlightenment thinkers who undoubtedly paved the way for inquiries into the impact individuals had on society and society on the individual.

7 and argued that all societies would eventually come to the same conclusions about the nature of reality regardless of cultural idiosyncrasies. Thus, every society would traverse through stages of development starting with simplistic explanations and moving on to more complicated and accurate explanations of reality. Eventually every society would go through the same stages of philosophical development and share similar forms of social organization as well as moral codes. As we will see in the next chapter modernization theory makes a similar assumption about the linear progression of nation-states.

In all Comte identified three stages including; (1) the theological stage, in which reality was explained by referring to supernatural beings, (2) the metaphysical stage, in which reality was explained in terms of abstract reasoning and the mystification of experience, and finally, (3) the positivistic stage, where careful measured observations of reality and logic were used to explain reality. The positivistic stage was seen as the height of social evolution and Comte believed he could help speed this evolution by spreading the positivistic philosophy.

Although Comte’s theory was crude, simplistic, and certainly overly optimistic he was the first social thinker to suggest that societies evolved or changed due to characteristics found in populations rather than the individual. Thus, Comte had broken away from the Enlightenment thinkers in this important respect which set the foundation for several 19th Century scholars who brought science to the study of society.

8

1.2 Herbert Spencer: Population Pressure and Compounding

Herbert Spencer (1820-1903) was one of the earliest scholars to bring science to the study of society. Specifically, he turned away from Comte’s desire to speed societal evolution and instead focused on using physics and biology as templates for how one should study social change. Ultimately, Spencer concluded that like physical objects society was governed by laws which could be understood using the scientific method. Thus, while Spencer was optimistic about human’s ability to know and discover social laws, he was pessimistic about human’s efficacy in social engineering once these laws were known.3

The core of Spencer’s sociology was based on the idea that social reality, like physical and biological matter, was guided by an underlying universal law. In fact, in First Principles (1862) he argued for a unified law of existence which suggested that all matter and life (including societies) moved from homogenous states of likeness to heterogeneous states of integrated parts. This evolutionary principle shaped how Spencer explained the relationship between population and social organization.

Because Spencer derived his model of society from physics and biology it is no surprise that the two main components of his model mimic physical and biological phenomena. Taking from the structure of biological organisms, Spencer argued that every society regardless of its size or complexity needed to have a regulatory, distributive, and operative system in order to survive. In other words, like living beings societies had to have an operative system that met the internal

3 It is important to note that while Spencer was the one who coined the phrase “survival of the fittest,” he was not a part of the social eugenics movement that used this phrase.

9 needs of the society, a regulatory system which stabilized the relationship between the system and its external environment, and a distributive system which carried needed information and materials throughout the entity. Together, Spencer referred to these three systems as the axes of differentiation since these systems were common to all societies regardless of their complexity. Although the biological underpinnings of his model were intuitive and set the stage for functional thinking, much more intuitive was how Spencer explained the effects population growth had on the organization of society. These insights into population pressure, whether acknowledged or not, continue to inform contemporary understandings of social change.

Drawing on the ideas developed in First Principles and in the natural sciences, Spencer argued that population size had profound effects on how societies met regulatory, distributive, and operative needs; therefore, population increases had predictable effects on how societies were organized. Specifically, Spencer believed population growth changed social organization through a process he called compounding. For Spencer compounding was an evolutionary process where population growth strained a system’s ability to maintain cooperation and coordination between its operative, regulatory, and distributive functions. In order to overcome the pressure brought about by population growth, societies turned exclusively to specialization as a strategy for increasing efficiency in tasks. Thus, the larger a society became the more complex, specialized, and interdependent each system had to be in order to overcome internal inefficiencies brought about by population size. Figure 1.2 offers a visualization of how Spencer believed

10

Simple Without Simple Compound Doubly Compound Head With Head

Political State Paramount Chief Chiefs Civil Bureaucracy Lieutenants Lieutenants Military Bureaucracy Local Chiefs Local Governments

Extended Kinship Extended Kinship

Extended Kinship Ecclesiastical Structures Ecclesiastical Hierarchy Religious Shaman Economic Divisions Complex Economic Roles More Complex Economic Roles Regulatory Some Artists Operative Camps, Settlements Artistic Roles Artists; Literary Specialists Distributive Informal Codes Two Ranks Permanent Communities Towns and Cities Family Sharing Informal Codes, Enforced Law

Systems of Ranks Castes

Road Systems

Travel and Trade Traders Systems: Trading Specialists Markets

Communication Specialists

*Adapted from Figure 5.2, p.72 in The Emergence of Sociological Theory by Turner, Beeghley, and Powers (2002).

Figure 1.2: Spencer’s Model of the Effects of Population Growth on Social

Organization.*

compounding unfolded—that is, how population increases affected social organization via compounding (i.e. specialization) of the regulatory, distributive, and operative systems.

11

In simple societies, or those societies with very small nomadic populations,

Spencer suggested that individuals were able to perform similar functions which fulfilled the system’s needs—this is what made them “simple.” In fact, differentiation between the axes was minimal, if it existed at all. Therefore, those members responsible for performing the regulatory functions (e.g. tribal elders or chiefs) most likely played a role in meeting operative functions (i.e. they were the hunters) and the distributive functions (i.e. they decided who received what).

However, with each additional member more resources were needed, more resources had to be distributed, and more regulation was required to maintain the distribution and gathering of those resources. Thus, as population size increased each axis had to become more defined, complex, and distinct from the others— eventually becoming completely differentiated in terms of function. However, in the face of separation and distinction greater coordination and a more complex organization was required to maintain internal system stability. All these internal changes were necessary to achieve system balance in an ever changing external environment.

Thus, Spencer argued that the way populations achieved stability (or equilibrium) in the face of population expansion was to compound each system (i.e. specialize tasks within each axis and decentralize responsibilities). Specialization and decentralization within each axis allowed populations to increase the efficiency and effectiveness of the whole system by minimizing the number of functions any one part of the system had to perform in order to meet the needs of the whole system. For instance, in simple nomadic societies with less than 50 people it was

12 possible for one leader to regulate an entire population. Yet, in a society with fifty thousand people having one leader to regulate the entire population would be inefficient and most certainly physically impossible. Thus, in the wake of population increases there is pressure to decentralize the regulatory function (e.g. the creation of lieutenants), such that power is nested in the specialized position allowing for increased efficiency in meeting the regulatory needs of a society with a larger population. In short, greater size necessitates greater specialization, increased complexity, and decentralization of functions (and power) to meet the problems of cooperation and coordination brought about by population growth.

From Spencer’s point of view this evolution from simple to complex states was in-line with other phenomena in the universe- particularly organic life. Again, the movement from simple homogenous states of independence to complex states heterogeneous states of interdependence could be observed in all physical and biological phenomena. Accepting this made it logical for Spencer to assert that human populations were no exception to these “universal” laws.

As one can see in Figure 1.2 there is a very systematic and orderly relationship between population growth, social organization, interdependency, system complexity, and the decentralization of tasks in Spencer’s model. Social change leads to greater complexity, but the process of how change occurs (i.e. population growth) and the path it takes (i.e. compounding) remain consistent at every level and within all axes. In short, for Spencer evolution in social organization unfolds as a process of population growth and as we will see is very similar to ecological theory. Population growth leads to problems of coordination and

13 cooperation within the system (i.e. integration) because its past arrangements are no longer optimal for its present environment (e.g. need for space, more resources).

Under the strains of expansion populations either find solutions to these problems

(e.g. increase density of living space, grow food more efficiently, specialize tasks etc.) which allows for future population growth, or they fail to find solutions and face disintegration and population declines. Spencer believed that societies would experience both avenues at different times and under different circumstances, but the outcome of sustaining population growth was inevitable: greater complexity and increased interdependency resulting in specialization and decentralization within the axes of differentiation.

1.3 Emile Durkheim: Population Pressure and the Role of Specialization

Drawing insights from the rich of French Enlightenment thinkers,

Emile Durkheim’s (1858-1917) approach to the relationship between population and social change is perhaps the most refined of the classical social theorists, particularly because he clearly adds the dimension of density to the relationship between population pressure and social change. Moreover, his model is grounded in Darwinian logic in which evolution is guided by competition of living space making the mechanism of change much more explicit than Spencer.

If we recall, Spencer argued that important shifts in social organization were brought about by population pressures attached to growth in absolute size. Growth placed strain on existing social systems which had achieved equilibrium with their external environment. Spencer argued that as population size increased it became

14 increasingly difficult for a society to maintain internal cooperation and coordination between its axes and the external environment with its current population size.

Thus, specialization (or compounding) within the systems (and population) emerged to overcome the strains brought on by growth. Although Spencer’s model of social change makes intuitive sense from a purely theoretical standpoint, he never really addressed exactly how specialization emerged as the result of population growth. That is, what the mechanism for compounding was. In contrast,

Durkheim’s model explicitly addresses this issue using Darwinian logic.

Durkheim begins with the notion that the volume of a population (i.e. its size and degree of concentration) as well as technological advancements in communication and transportation dictates the dynamic density of a population.

Dynamic density is simply the rate of interaction between units within a population.

Determining the degree of dynamic density in a population is important for one main reason; it dictates the degree of competition among members of a given population. Thus the more units (or individuals) in a system there are the more interaction there will be. Increased interaction results in higher levels of competition, not only for scarce resources but for functional position within the system as well. Drawing insight from Darwin’s model of evolution in which specialization allows for adaptation, Durkheim applied the same principle to human populations. In doing so, he offered a clear explanation for how specialization within the division of labor arises and leads to greater system complexity and increased interdependency.

15

Most “Fit” Remain in Present Ecological Occupation System Increased Level of Level of Level of Division of Labor Migration Material Dynamic Internal and Density Density Competition Interdependency Natural between Units Less “Fit” Exit Increase from Technological Advances Competition in Communication and and Specialize Transportation

*Adapted from Figure 4.2 p.126 in Explorations in Sociological Theory by Kenneth Allan (2005) and Figure 17.1 p. 336 in The Emergence of Sociological Theory by Turner, Beeghley, and Powers (2002)

Figure 1.3: Durkheim’s Model of the Effects of Population Growth on Social Change.*

Above, Figure 1.3 approximates Durkheim’s model. While others offer similar models (see Allan p.126 [2005], or Turner, Beeghley, and Powers p. 336 [2002]), I believe that a close reading of “Book II: The Causes and Conditions” in Durkheim’s

The Division of Labor in Society is best approximated in Figure 1.3. With this in mind let us explore this model in detail. Like Spencer, Durkheim starts by acknowledging that population growth is the of social change since it places strain on how systems meet its needs. Thus, Durkheim’s model begins with a general approximation of those components which have historically led to population growth, including; (1) the ecological system and natural geography where a population resides, (2) net natural increases, and (3) net migration. He argued that population growth increased the level of material density (or population concentration) and subsequently led to increases in the level of dynamic density

16 found within a system (again, defined as the rate of interaction). Thus, according to

Durkheim increases in material density resulted in equal increases in dynamic density; material density and dynamic density had the same effect on the division of labor although they were empirically separate concepts. This “same effect” is that increases in both types of density resulted in higher levels of competition over key positions within the labor system because the system had yet to adjust to the new demands brought about by population growth. In other words, more people were competing for the same number of functional positions already in place while at the same time the system was unable to provide sustenance (or social space) for the now larger population. Therefore, excess population and fewer occupational positions increased the level and intensity of competition in the short term when populations grew.

This brings us to the role competition plays in creating a more complex social system since it is competition which breeds specialization in Durkheim’s model. As discussed above, Durkheim argued that as dynamic density increased there were subsequent increases in the level of competition between units. Drawing insight form Darwin’s model of evolution, Durkheim suggested that in the wake of competition the “fittest” would keep their place in the division of labor (i.e. those who performed their functions the best) while the less fit would be forced to struggle for existence (i.e. find a place in the division of labor in order to acquire necessary resources provided by the system). It was this struggle for existence that led to specialization and moved society forward.

17

According to Durkheim, increased population brought about new system needs, which initially could not being met (again, the reason for increased competition). This left society with a system deficiency, but the deficiency was only temporary. Durkheim argued that those who lost out in the competition were the ones who eventually overcame the deficiencies via specialization within the division of labor. Individuals either innovated by fulfilling a new need brought about by population expansion, or they failed to integrate.

Durkheim was awed by how evolution worked in social systems, specifically within the division of labor since more often than not “opponents [were] not compelled to fight to a finish, but co-exist” (1893, 1998 p.51). It was this co- existence or interdependence which was ultimately responsible for resolving the problems of integration and system needs brought about by population growth.

However, it is important to note that Durkheim suggested the specialization of functions within the division of labor was not a random process. In fact he stated:

“A function can only become specialized if this specialization corresponds to some social need” (1893, 1998, p.52). Ironically, the solution to the problems of integration brought about by population increases (i.e. greater complexity in the division of labor via specialization) was also responsible for future increases in material density due to increases in production brought about by specialization. As

Durkheim stated it, “[Specialization is] produced by the same cause that determines the of the division of labor” (1893, 1998 p.52). Hence, the dashed line leading from “Increased Division of Labor and Interdependency” to “Level of

Material Density” approximates this statement.

18

One final note should be made about Figure 1.3 before concluding this section. It is worth commenting on the inclusion of the second dashed line from the

“technology” box to the “material density” box. This line, while not present in other reconstructions of Durkheim’s model, finds support from Durkheim’s own words when he wrote; “The creation and development of cities is an even more characteristic of [density]” (p.51). Thus, instead of adding a separate box for the effect urbanization has on material density, the dashed line indicates that modern density (i.e. urbanization) is heavily dependent upon transportation and communication networks. Assessing the effects of transportation and communication networks on the material and dynamic densities of populations will be of great importance when linking urban population structure to political change since density has been tied to both the centralization and decentralization of economic and political power (e.g. Dahl and Tufte 1973).

1.4 Marx and Engels: Population Concentration as an Economic Outcome

So far I have reviewed theories which use population growth as a central explanatory variable for social change and organization. For Karl Marx (1818-1883) and Friedrich Engels (1820-1895) population pressures were never used as a central explanatory variable. In fact, population played only a minor role in their models and was mostly discussed in response to and as a rejection of Malthus’ original take on the properties of population pressure (Hawley 1984; Petersen

1988). Despite this lack of attention, Marx and Engel’s view of population is worth recounting since contemporary cross-national Marxist scholars (i.e. those who adhere to world-system and ) have greatly expanded their

19 interest in patterns of population growth in the developing world. Specifically, they have come to relate urban developmental patterns in the periphery to the global spread of capitalism which they argue promotes uneven distribution of economic opportunities and resources.

The slightest glance at the theories outlined earlier reveals that Comte,

Spencer, and Durkheim all viewed social change in terms of an evolutionary movement form simple to complex states brought about by population pressure.

For these sociologists the concentration of population was central for explaining change in social organization because population growth forced individuals to alter their relationship with their environments including each other. While Marx and

Engels certainly agreed with the notion that population pressures were responsible for increased differentiation, they did not believe this process was central to social change (Hawley 1984). In many ways they viewed excess population as a detrimental side effect of capitalism (Petersen 1988).

According to Turner, Beeghley, and Powers (2002) population size dictated two processes in Marx’s general model; (1) the level of differentiation in society, and more importantly (2) the level of production (see Figure 9.1, p.160). Thus, population size mattered in so far as it increased the need for more production and greater differentiation within the division of labor. According to Marx and Engels the consequences of increased production and greater differentiation was greater . Marx and Engels argued that more production resulted in a greater concentration of elite power over the means of production because population growth allowed elites to expand production (i.e. there was more labor)

20 and increase profit (i.e. lower wages as competition over intensified).

Both outcomes gave elites more control over a greater number of laborers and resulted in a more unequal share of society’s resources/production.

What the above alludes to is that Marx and Engels were skeptical of the positive effects population growth could have on social outcomes under capitalist production systems, especially in terms of economic outcomes for the masses. Such skepticism of population growth under capitalism comes from the assumption that resources and opportunities are finite, and thus when one group “wins” (e.g. the elite) another “loses” (e.g. the masses). In fact, Marx and Engels adhered to this zero-sum assumption so much so that they argued the exact opposite effect of population growth on social organization. If we recall, both Spencer and Durkheim asserted that increased differentiation was a source of increased complexity as well as interdependency. For Durkheim, in particular, differentiation was the key to survival and social integration (i.e. social solidarity). For Marx and Engels differentiation brought about by population growth has the exact opposite effect—it always leads to greater inequality, and most important, to an increased concentration of power that favors elites. It is no wonder then that in the

Communist Manifesto the League of Communists included a specific clause about the concentration of population: Specifically, it calls for a “gradual abolition of all the distinction between and by a more equable distribution of the populace over the country” (Section II, Measure #9 in the Communist Manifesto

1890). As we will see in the chapter to follow, this ideal of “even” population development is central to the analyses and conclusions of world-system theory

21 which grounds its theoretical and political assumptions of Marxian thought. It argues that capitalist elites penetrate foreign markets and order urban systems and economies such that the people and resources in periphery nations are used to line the pockets of the core capitalist elite.

1.5 Conclusion

In this chapter I covered four classical theorists—Comte, Spencer, Durkheim, and Marx. While Comte set the stage for the other two functionalist theorists’ views of social change, Spencer and Durkheim added key insights for understanding the effect population pressures have on the organization of societies. More importantly, each showed that increases in population size and density bring about the decentralization of economic and political power. Conversely, Marx and Engels looked to the negative impacts of population growth and concentration. They argued that population growth allows capitalists to drive down wages while concentration of the masses makes it easier for capitalists to increase their control over production/resources. In the next chapter, contemporary theoretical perspectives of modern development are reviewed which apply both sets of logic to changes in modern nation-states and the global economy. It is from these macro- social theories that most of the research focusing on macro change has been based.

22

CHAPTER 2

CONTEMPORARY THEORIES AND THE ROLE OF POPULATION IN SOCIAL CHANGE

Now that the classical theories have been discussed, I turn to more recent approaches on the role population pressures play in eliciting macro-social change within the modern nation-state. Much like the classical theories, the schools of thought covered in this chapter can be divided into two dominant views: one, the modernization school derived from systems approaches akin to Durkheim and

Spencer who see population pressures as a key component of modern social change, and two, the world-system approach derived from Marx and Engels’ views of capitalist economies which sees population structures as the outcome of capitalist economic systems. However, a third approach (the human ecological approach) covered at the end of this chapter holds that population pressures are both a cause and outcome of social change. After discussing each approach I conclude with a comparative summary.

2.1 Modernization Theory

Modernization theory is based on the notion that macro societal development is highlighted by convergence toward similar economic, political, and socio-cultural trends in the wake of technological development and its diffusion— that is industrialization (e.g. Davis [1945]; Inkeles [1960], [1981]; Kerr et. al. [1960];

23

Rostow [1960]; Alonso [1980]; Boserup [1981]). Many of the processes associated with modernization whether political, cultural, or economic are said to conform to a universal pattern. In general, modernization theories model social change (i.e. progress) as a self-limiting process highlighted by “stages” or transitions.4 These stages are comprised of (1) an early stage of stagnation highlighted by incoherence;

(2) a middle stage of rapid “take-off” highlighted by imbalance; (3) a later stage of balance highlighted by integration or a drive to maturity, and in some cases; (4) a final stage of slow decline or recession—i.e. a “pull-back.” Below Figure 2.1 offers a basic conceptual model of these stages as they relate to predicted change in a given dependent variable over time (e.g. economic growth, population growth).

2.1.1 Modernization and the Demographic Transition

Although most versions of modernization theory do not posit population as a cause, in general modernization theory tends to argue that population trends are important indicators of modern social changes (i.e. progress). Weeks (2002) defines modernization as “the process of societal development involving urbanization, industrialization, rising standards of living, better education, and improved health that is typically associated with ‘Western’ lifestyle and world view…”, moreover it “is the basis for early explanations of the demographic transition” (p.604). As indicated in Week’s definition, the concept of modernization is linked to the demographic transition. A closer look at the demographic transition and demographic transition theory reveals how modernization theory explains the

4 While modernization theories do use specific stages of development, they are not to be treated as objective phenomena. Rather the stages are conceptual tools or ideal types which necessary for basing comparisons on.

24

Stage 1 Stage 2 Stage 3 Stage 4

Δ in y Δ in

Δ in time

Figure 2.1: Modernization Theory’s Basic Model of Change.

relationship between population pressures and modern social change.

The demographic transition was first observed empirically by Thompson

(1929), with the theory first formalized by Notestein (1945) and Davis (1945), and later expanded by Davis (1963) and Easterlin (1967,1968). As we will see, the early explanations for the demographic transition latter became the basis for modernization theory which is rooted in the ideas of Comte, Spencer, and Durkheim discussed in the previous chapter. In light of this, let us explore the macro-social argument contained in demographic transition theory which is in many ways indistinguishable from modernization theory (Crenshaw et. al. 2000).

Demographic transition theory (DTT) is based on the premise that as pre- industrial societies gain agricultural technologies they experience a predictable decline in their mortality rates that sets them on a path toward modern

25 development. Generally speaking DTT breaks the transition from pre-industrial to into four stages.

In the first stage populations are pre-industrial in the sense that they possess little or no modern technology. They have very limited control over food production, , and mortality. As a result, there is a demographic balance of high births and death rates. Populations have no control over their fertility or mortality and are subject to weather, disease, and catastrophic cycles in the . Because mortality is uncontrolled and agricultural production inconsistent, children are a highly valued asset. Not only do they provide the replacement population needed under high levels of infant and adult mortality, they also provide cheap labor and old-age security in rural agricultural environs. Thus, demographically speaking pre-industrial (or pre-transitional) societies in the first stage of development are characterized by a balanced demographic regime of high fertility and mortality with relatively small and young populations.

In the second stage, populations begin to experience a rapid decline in mortality as the diffusion and development of “modern” technologies gives populations better control over infant and adult mortality. In DTT mortality reductions are generally connected to two types of technological advancement: The first and most important are agricultural advancements. Better planting and growing techniques (e.g. crop rotation and fertilizer) as well as mechanized labor

(e.g. tractors and combines) increase crop yields and improve crop quality. An abundance of high quality food results in less overall starvation and higher levels of nutrition for the population. The second technological advancements are key

26 improvements in various public health systems; most important among them being improved sanitations, the availability of clean water, and in more recent times— cheap vaccinations. Both improvements help humans resist disease leading to increased longevity (Davis 1945). Moreover, improvement in agricultural and public health result in much lower infant mortality and more children survive into adulthood. However, even with dramatic reductions in mortality populations in

Stage II tend to retain their pre-industrial attitudes toward fertility. Thus, the defining demographic feature of Stage II is that early on in acquiring modern technologies populations maintain high fertility even in the face of rapidly declining mortality. The lagging decline in fertility rates and decreases in mortality leads to rapid population growth and very large populations. According to DTT rapid population growth places new strains on economic, social, and political systems which cannot be properly dealt with by the systems in place. It isn’t until the population begins to experience the economic disparities brought about by population increases that they begin to adjust their behavior. This marks the start of the third stage.

In Stage III, rapid population growth and the spread of mechanized agricultural production (i.e. commercial agriculture) increasingly limit economic opportunities in rural areas. The resultant economic imbalance begins to “push” individuals in large numbers to urban environments where industrial production systems tend to concentrate capital and modern . Together, the concentration of modern infrastructure and capital gives the perception or establishes the reality that emerging urban offer the best economic

27 opportunities for “excess” population (i.e. generally the young who are unemployed but mobile). However, as economic opportunities continue to shift from rural to urban industrial settings, “modern” economic systems and their realities begin to increasingly contradict the traditional fertility values of pre-industrial populations.

In other words, as economic production shifts from agriculturism to industrialism rural populations bring their reproductive attitudes and behaviors to urban landscapes.

Thus, over time maintaining high fertility in urban industrial areas becomes increasingly irrational as the cost of children for individuals rises to the point where offspring are no longer an economic asset but a burden. Toward the end of the

Stage III populations begin to respond to the economic drag of children and acquire

“modern” attitudes towards fertility that match the reality of their economic situations. Individuals begin to value the quality of a few children (e.g. well educated children) rather than a quantity of children, and therefore seek to limit family size as a means to improve economic circumstances. Because modern settings offer the contraceptive technologies and distribution systems (e.g. mass production, modern transportation and communication) people are able to actively control their fertility for the first time and have reason to want to do so. Thus, Stage

IV is reached when populations have restored a demographic balance with low fertility and mortality rates, and in doing so, populations stabilized their growth.

Figure 2.2 offers a graphical display of the demographic transition (Stage IV not shown).

28

40 10 Population

Size

Population size in billions in size Population

30

20 High Growth Transitional Incipient Potential Growth Decline

10 Vital rates Vital per 1.000persons Death Rate I II III 0 0 Time  * Adapted from Figure 3.2, p.101 in Population 8th ed. by John R. Weeks (2002). Stage IV not shown.

Figure 2.2: The Demographic Transition.*

While both modernization theory and DTT argue that every population has unique social, economic, and political circumstances which dictates the pace and trajectory of their transition, modernization theory generally argues that the pattern of the demographic transition (and its impact on economic growth) is universal: In

Stage I population exhibits high growth potential since the demographic regime is based on “traditional” demographic circumstances where populations have no control over their mortality or fertility. In Stage II, transitional growth takes place as population begin to gain control over mortality but not fertility. In Stage III modern environments favor both low fertility and low mortality rates leading to an incipient decline in population. Stage IV is reached after a population restores its demographic balance such that low fertility and mortality stabilize population

29 growth. In recent decades demographers have observed fertility levels falling below replacement levels (e.g. and Western ). Based on these findings some have argued that there is a fifth stage associated with post-industrialism. Although theoretically undeveloped, the argument is that in post-industrial societies the participation of both sexes in the labor force coupled with a rise of secular values dramatically lowers the number of offspring. Logically this speculation conforms to modernization theory and DTT but it has yet to be widely accepted.

2.1.2 Technology and the Organization of Population

While the relationship between technology and population growth is central to demographic transition theory, yet to be discussed is the assertion that technological advancements have similar effects on other key aspects of social organization. This is where modernization theory comes into play. Specifically, modernization theory asserts that as populations grow and shift from traditional to modern production systems, they go through an interrelated set of macro-social changes involving economic, political, and cultural convergences that extend beyond but are related to demographic processes. These changes in social organization are much akin to Comte, Spencer, and Durkheim’s emphasis on evolutionary movements from independent homogeneity to interdependent heterogeneity in which balance is achieved. The most important of these technological advancements which clusters populations into “pre-urban” settlements and cities is the replacement of animal power with machines—that is, industrialization.

Modernization theory argues that industrial production is so efficient for economic growth and militarization that neighboring populations are compelled to

30 adopt it or face irrelevancy (Kerr 1960). Thus, the rise and spread of industrial production is perhaps the most important technology according to modernization theory because the “logic of industrialization” brings about global convergence in a variety of social forms and living conditions. For instance, modernization theory argues that as industrialization spreads globally populations being to converge in their organizational forms (e.g. bureaucracy) and values (e.g. develop secular laws and practices). Like the shift from high to low mortality and fertility, these developments are understood to be “universal” human responses to modern living, and therefore, the effects of modernization on social organization extend beyond their Western European origins.

Global convergence stemming from modernization is said to occur in other key areas as well, particularly with respect to economic systems since industrialization entails increased complexity of operative, distributive, and regulatory systems. Unlike agricultural societies whose populations share similar skill sets and are interchangeable, under industrial production specialization of tasks and production methods is the key to economic efficiency and growth (Kerr

1960). Thus, industrial production requires both laborers and employers to develop specialized skills in order to maximize economic worth. As a result, economic power becomes more evenly diffused throughout the population as the division of labor becomes increasingly specialized and decentralized. As we will see this is contrary to the Marxian view which argues industrialization entails a leveling of workers’ skills and the concentration of economic power into the hands of the few. Thus, contrary to Marxian thought, modernization theory argues that the

31 division of labor in industrializing populations (irrespective of their geographic local or cultural heritage) will become more complex, more open, more mobile, and more educated (Kerr 1960), and thus more decentralized. That is, modernization theory argues that industrialization and technological advancements “Once under way…[does] more or less violence to the traditional pre-industrial society” (Kerr

1960, p.15). Populations go from being relative small, homogenous and evenly dispersed (i.e. low density) to much larger, heterogeneous and more highly concentrated (i.e. high density) yet interdependent. In short, modernization entails industrialization but it also entails urbanization.

2.1.3 Population Size, Density, and Production Systems

While urbanization is a key component of the modernization process, the underlying mechanism dictating changes in social organization associated with population concentration is not industrialization or urbanization per se, but rather the relative density modern populations require for efficient production and distribution. In fact modernization theory argues that industrial production systems rely on a variety of population densities, including informational, organizational, and infrastructural which spur economic growth and help maintain system stability. However, it is important to note that the relationship between and development is not linear—that is, modernization theory does not argue greater density corresponds to an equal or constant change in development. Instead, modernization theory offers a nuanced view of the effect densities have on social organization and change. With this in mind let us conclude

32 our discussion on modernization theory with a focus on the relationship between population density and social change.

The basis for the relationship between population density and social change found in modernization theory rests on the idea that optimal levels (or more precisely, optimal ranges) of densities exist for a given social system. When a population develops too much or too little density for its given means of production, structural strains are placed on the current intrasocial arrangements which forces adaptations to take place. However, while modernization theory posits evolutionary movements toward “balance” or equilibrium, the precise level of density and the amount of time it takes to achieve a specific optimal level remains relative to local conditions.

Perhaps the easiest example with which to understand how density is modeled under modernization theory is found in the relationship between population concentration and economic growth. In order to explore this relationship two economic terms must be introduced; efficiencies and negative externalities. Efficiencies are anything which reduces the costs of exchanges, while negative externalities5 are anything environmental factors external to the system which increases the costs of exchanges or lowers efficiencies. According to modernization theory, the most important negative externalities are those things which increase time and/or space between social actors, and thus increase the costs associated with production, distribution, and regulation. Similarly, the most important efficiencies are anything which reduces time and/or space between social

5 There are positive externalities (external benefits) which help lower costs.

33 actors, and thus decreases costs associated with production, distribution, and regulation. Because time and space are inseparable, population density directly impacts both efficiencies and negative externalities in all economic systems as well as for all means of economic production. This does not imply that population densities alone impact efficiencies and negative externalities; rather, it implies that density has an independent effect as well as interactive effects on social change and organization. Why is this so?

Human populations rely on structured exchanges for existence.6 However, rational exchanges in traditional societies may be irrational in modern societies and vice versa. This is so because optimal or efficient exchange is achieved much differently in traditional production systems than in modern ones. The explanation for this is similar to the explanation for why high fertility is optimal in traditional settings but irrational in modern environments.

Traditional societies tend to rely on agricultural production and possess little or no mechanical technology or scientific reasoning; as a result low density distributions are optimal for production output since the dispersion of a homogenous workforce across large areas of space (i.e. arable ) results in optimal output/production (more land is farmed or cultivated via decentralization in labor). Efficiency, then, is achieved by distributing population over maximum space resulting in low density.

In modern production systems that rely on mechanical production, technology, science, and bureaucratic organization the opposite holds true. Greater

6 Simmel is credited with suggesting that the master process of society is exchange between social actors. All other social goals and purposes flow from this human necessity for existence.

34

High

ICIENCY

EFF Low Low

Low ln DESNITY High

Industrial/Modern Production Agricultural/Traditional Production

Figure 2.3: The Relationship between Density and Efficiency in Modern and

Traditional Economic Systems

density lowers the costs associated with production since high density minimizes distances between an interdependent heterogeneous workforce—thus maximizing the frequency of exchange and minimizing the time it takes to exchange goods and services (see Chapter 3 for a related discussion on economies of scale). Therefore, for modern industrial production systems efficiency is achieved by concentrating population into a comparatively small area—high density. Figure 2.3 illustrates how modernization theory models the general relationship between density and economic efficiency in modern and traditional production systems.

35

However, even though lower levels of density are optimal for agricultural societies and higher levels for industrial societies, too low of or too high a level of density for either production system can impede efficiency by creating negative externalities that have a degrading impact on actors’ abilities to exchange, produce, or distribute goods and services. For traditional agricultural societies very low density can result in negative externalities associated with the time and energy required to overcome distance. While transportation networks can alleviate some of these costs, thus allowing very low levels of density to sustain populations, there are limitations on how dispersed a population can be and remain coherent. That is, there is a balance between space being a positive externality in production and space as a negative externality.

The opposite problem is faced by industrial populations—and this is a key point for understanding how and why modernization theory differs from world- system and dependency theories. In modern production systems too high a level of population density can lead to negative externalities (e.g. congestion, pollution) which limit production potential. Not only do space and resources become scarce and costly, very dense populations must invest in technologies which support their high density (e.g. transportation and communication systems, sanitation, government, etc.).

In both instances, low and high degrees of density can result in more negative externalities than efficiencies for economic production systems even though each system achieves optimal output under their respective optimal density ranges. Therefore, for societies in the process of modernizing, modernization

36 theory argues that in order for economic development to occur populations must not be so dispersed that they cannot engage in the efficient exchanges needed to support modern production, yet they cannot be so concentrated that even the most simplest of exchanges incur irrationally high costs (e.g. congestion in of resources and information, lack of public order due to inefficient governance, etc.).

Because optimal density in the movement toward modernization is achieved simultaneously with urbanization, it follows that the transition from low to high population density is likely to produce more externalities in early stages of the urban transition, which if overcome must take place after low fertility is achieved at latter stages of the demographic transition (Alonso 1980; Crenshaw 1992,

Crenshaw et. al. 2000). Therefore, in the process of transitioning from high to low fertility, modernization theory argues that as rapid population growth unfolds populations are more susceptible to over-concentrating in urban areas (particularly one ). This is especially true in areas where agricultural improvements quickly reduce the labor needed to maintain sustenance leaving a very large and young rural workforce without work and in greater competition with one another over scarce land. In other words, rapid rural population growth associated with mortality declines floods rural areas with unusable labor. There arises an imbalance between rural density levels and the optimal level of density needed to remain efficient (see Figure 2.3).

Central to modernization theory is the idea that this rural imbalance pushes excess labor to urban areas, areas where higher densities are necessary to achieve greater efficiency. The problem, however, is that in most cases nations in transition

37 grow urban areas much faster than what the industrial sector needs. Excessive labor leads to a variety of system imbalances and institutional strains that hinder further economic growth (e.g. overurbanization). However, from the modernization perspective such strains on production systems are said to be temporary since centralization leads to decentralization.

Over time the centralization of production and population becomes less and less necessary, and the problems associated with high density are overcome with modern technologies. Moreover, in very late stages of modernization rural lands are become increasingly integrated with urban landscapes as modern technologies become even cheaper and easier to decentralize from urban centers. Thus, just like the DTT where rapid and excessive population growth is seen as a temporary condition in the transition from high to low fertility in order to achieve a modern demographic balance, so too is excessive population concentration seen as a temporary necessity in the transition from rural to urban living in the modernization process (i.e. the urban transition). However, it is worth noting that from the modernization perspective the severity and the speed at which density imbalance occurs is historically and culturally unique. Again, what remains universal is that the development of modern production systems entails the centralization of population into one or a few urban areas which is followed by a controlled decentralization into smaller cities.

In conclusion, modernization theory connects population to social change in two ways: One is in terms of population size and the other in terms of population density. According to modernization theory the acquisition of modern technologies

38 affects population by increasing its size (i.e. demographic transition) which is followed by an increase in density around key industrial producing regions (i.e. urbanization). Therefore, technology is both the cause and solution to rapid population growth and excessive concentration (i.e. urbanization). Changes in social organization are attributed to demographic imbalances between population size/density brought about by modern production systems in which traditional values and behaviors become increasingly irrational in modern environments.

2.2 Dependency and World-System Theory

First formalized by Wallerstein (1974,1979), world-system theory (WST) is a neo-Marxist approach derived from Latin American dependency theories that explains cross-national development as a consequence of capitalist trade and production.7 It was formalized and popularized as a direct challenge to modernization theory’s key assumptions and beliefs concerning the development of nation-states which included: (1) an assumption that the nation-state represents a complete economic system, and therefore, levels of development can be explained using characteristics of individual nations; (2) an assumption that all nation-states will modernize if given enough time (i.e. pass through developmental stages); (3) the assertion that technology is the solution to problems of modern social integration; and (4) the belief that once fully modernized nation-states will experience lower levels of social inequality and more stable economic growth

(Rostow 1960).

7 For a comprehensive history of world-system theory see “World-System Theory” Chirot and Hall (1982).

39

In contrast to the above, WST believes and assumes the following: (1) that nation-states “do not represent separate systems of production…but rather…all are part of a single system of production which contains multiple political units within it” (Rubinson 1976, p.639); (2) that a single exists which is maintained and organized around capitalist trade and production; (3) that the world contains finite resources, and therefore, the world-system is a zero-sum economy; (4) that capitalism operates outside the political boundaries of nation- states using coercive tactics including military force, , aid, and foreign investment in order to secure monopoly rights to land, labor, and raw materials, and finally; (5) that the global spread of capitalism and its practices has resulted in distorted development characterized by a vertical hierarchy of nation-states.

With the preceding in mind, let us look at each assumption in greater detail using the writings of Wallerstein (1974,1976) Chase-Dunn (1975), Meyer, Boli-

Bennett and Chase-Dunn (1975), Rubinson (1976), Borschier, Chase-Dunn, and

Rubinson (1978), Chirot and Hall (1982) and Timberlake (1987). Doing so will allow for an informed understanding of how WST approaches the relationship between the urban transition and social change.

2.2.1 The World-System

World-system theory’s first assumption offers a fundamental image of the modern world’s economic and political arrangements by suggesting that a single world-system has emerged from a collection of disintegrated mini systems.

Wallerstein (1974) argued that early on in human history agricultural based societies (i.e. mini systems) existed as “self-contained” entities that possessed “a

40 complete division of labor, and a single cultural framework” (Chirot and Hall 1982, p.84). Over time some of these mini systems prospered, gaining advantages in military and transportation technologies. These advancements allowed their population to establish dominance over other mini systems from which regular tributes were collected. According to Wallerstein this was the first step toward establishing a world-system. Conceptually then, a world-system is a global economic system that integrates a collection of mini systems or sub-systems under a single division of labor. To be clear and fair to WST it is important to note that the theory does not argue that a world-system has included all areas of the globe at all times

(Timberlake 1987). Instead, WST argues that the movement toward a truly global economy under a single division of labor is the goal of dominant economic powers under a given system of production (more on this below). With this in mind

Wallerstein has argued that two types of world-systems have existed and that a third possible system has yet to be realized. Let us look at each of these world- systems in greater detail.

The first hints of a world-system emerged in ancient times and persisted to about 1500 AD. This type of world-system is called a world-empire. A world-empire is characterized by single division of labor with “a single political system [that dominates] over most of [an] area” (Wallerstein 1974, p.348). World-empires were more or less based on forced economic and political integration by a single cultured entity. However, world-empires did not survive to modern times because of the costs associated with their expansion and maintenance. Specifically, the larger empires became the more resources were required to keep bureaucrats loyal and

41 populations in order. Ironically, world-empires prospered through growth but vanished because of their need for excessive expansion to achieve growth.

Examples of some world-empires that dominated the world-system include the

Roman and Egyptian empires.

The second more modern world-system to emerge is the world-economy.

This type of world-system is characterized by a single division of labor with multiple political centers. Thus, the key difference between world-empires and world- economies is that under world-empires a single political system dominates a single division of labor, while under a world economy a single division of labor dominates over “a multiplicity of political systems” (Wallerstein 1974, p. 348). For this reason, the world-economy has the benefit of decentralized governance but the benefit of uniform production and trade. Wallerstein and WST argue that this modern and less costly arrangement is precisely why the world-economy has prospered for so long.

Wallerstein also argued that a -system arrangement was possible—a socialist world government. He described this system as “an alternative world-system that could maintain a high level of productivity and change the system of distribution…[through] reintegration of the levels of political and economic decision-making” (1974, p.348). However, Wallerstein never fully explained how such a system differed from that of a world-empire since conceptually they are similar.

Below Figure 2.4 offers a visual representation of all four arrangements.

Take note that the world-empire and world socialist government are

42

Mini Systems World-Empire/ World-Economy World Socialist Government

Note: Darker circles indicate higher concentration of resources and/or power.

Figure 2.4: Economic-Political Arrangements according to World-System

Theory.

indistinguishable since in theory a single political system dominates a single division of labor under both arrangements. For the mini systems the circles represent closed economic and political entities. For the world-empires and world socialist government the single circle represents that both the economic and political systems are one-in-the-same. As for the world-economy, the larger circle represents the single division of labor that encompasses the smaller circles which represent separate political entities. Also note that in the world-economy the circles are three different sizes and three different colors. The different sizes represent the vertical hierarchical economic division of labor in which larger (and fewer) circles represent core nations and the more numerous smaller circles periphery nations.

The different colors represent the concentration of capital and resources within separate political systems with darker colors equating to greater concentration. The details of this stratification system will be discussed in greater detail below

43

2.2.2 The Capitalist World-Economy

Given the above schematic, WST argues that it is not possible to understand or explain cross-national differences in development without acknowledging that a world-economy exists which is dominated by capitalism and has a direct and historical effect on development pattern around the world. In fact, according to

Wallerstein the world-economy has been dominated by capitalism since 1500 AD, and therefore, the present world-system is perhaps best described as a capitalist world-economy and explained with Marx’s understanding of the logic of capitalism.

Like Marxism, WST argues capitalism contains its own logic tied to its unique system of production; in this case industrialization requires mass raw materials and mass consumption. Not only does capitalism have a unique means of production

(i.e. industrialism) it also had its own unique relations of production (i.e. a highly differentiated labor force, wage labor, and private property). Both components are important for explaining economic development and inequality much in the same way they are used in early Marxian thought. However, even more important for explaining differences in development between nation-states for WST is the assertion that a finite number of resources exist in the world. Because resources are finite within an all-encompassing world-economy, capitalist production ultimately limits who has access to the world’s resources and dictates how they will be used

(mainly for profit). Thus, for WST the capitalist world-economy operates under a zero-sum system or a closed economic system in which gains in one area inevitably entail loss in another. Most important, the historic outcome of this zero-sum economy under capitalism has been uneven development along political lines (i.e.

44 nation-states) and economic exploitation which results in .

According to WST this uneven development has largely been determined by the historical rise and spread of capitalism. Why is this so?

As discussed above WST is based on the assumption that capitalism as a system of trade and production is driven by a unique logic which applies equally to both the small scale economies of nation-states and to the large scale global economy (i.e. world-economy). As Wallerstein stated it, the unique logic of the capitalist world-economy is that it “is based on the constant absorption of economic loss by political entities, while economic gain is distributed to ‘private hands’”

(1974, p.348). The absorption of economic loss is made possible because the world economy is sub-divided into disparate political interests where capitalists are able to “operate within an arena larger than that which any political entity can totally control” (1974 p.348). As a result, WST argues capitalists freely pursue their interests using undeveloped (i.e. weak) nation-states. Specifically, they use weak nation-states as resource and labor reservoirs to draw surpluses from. Thus, unlike world-empires which required costly military and bureaucratic forces to maintain centralized political and economic control, in a world-economy capitalist can

“outsource” governance to local elites and “” labor and resources from them on an as needed basis. Moreover, because weak nations are weak, capitalists can use competition between regimes to improve their economic position.

Although political elites control populations and resources within their borders, according to WST capitalist elites easily penetrate these political systems through coercion using two tactics; brute force and bribery. With brute force

45 capitalists use military intervention, colonialism, or both to secure resources and access to cheap labor from political elites in weak states (usually autocracies). With bribery capitalists flood poor nations with foreign aid and foreign investments and essentially bribe political elites to accede to their wishes by enacting favorable terms of trade or labor/production law. In each case coercion allows capitalists to secure monopoly rights to land and labor as well as establish favorable terms of trade. Both strategies allow capitalists to expand their markets, lower production costs, establish ports of exit, and secure greater control over a larger area and over a larger number of people.

2.2.3 Stratification in the Capitalist World-Economy

Ultimately, the spread of capitalism as a global economic force has resulted in distorted development characterized by a vertical hierarchy of nation-states. This distortion can be traced as a series of historical developments tied to the global spread of capitalism. While the entire history behind this argument will not be covered here (but see Wallerstein 1974) recounting the basic history leads us to

WST’s stratification model. This fundamental model informs their argument concerning the relationship between population and social change, and therefore, requires attention. Thus, let us briefly recount this history.

Around 1500 AD capitalism emerged within Western Europe through advancements in science which resulted in greater agricultural output and eventually to industrial production. Economically, these advancements disproportionately benefited the merchant classes and gave rise to their economic power which was further heightened by improvements in trade. Expanded trade

46 and more efficient production resulted in a collection of neighboring nation-states who began competing for regional dominance (mainly the Dutch, English, French and Spanish). Aided by rapid industrial production and key scientific advancements, these nations created enormous military powers. Eventually, these forces and technologies were used to conquer less technologically advanced populations (e.g. North American, Latin America and ) in order to secure more resources. However, instead of creating empires with their military conquests, these burgeoning capitalistic nations used colonization to secure resources which were funneled back to the capitalists’ homelands where their production centers and economic interests resided. Eventually, every land was worth colonizing was, and those that could not be conquered with force were bribed.

After hundreds of years of colonial control and resource extraction, these few pioneering nations became very wealthy and so technologically advanced that their production techniques started to be exported to the colonized and less developed lands. Moreover, hundreds of years of colonization had embedded friendly bureaucratic governance and the basic infrastructure needed to support modern capitalist production systems (the most important of these being transportation and communication technologies [Chirot and Hall 1982]).

The historic displacement of resources and labor under capitalist rule has resulted in the global distortion of economic and political development. Thus, while the rise and spread of capitalism around the world has taken on a variety of structural forms (including colonial and imperial rule), the logic of a capitalist world economy and its impact on the distribution of the world’s resources has remained

47 consistent. The end result has been that a few core nations have benefited economically and politically at the expense of the many peripheral nations whose economies and political systems remain in a perpetual state of underdevelopment.

Therefore, WST models the capitalist world economy as a single economic system in which a collection of nation-states are arranged in a stratification system that is organized into a vertical hierarchy of economic classes.

At the apex of this economic stratification system we find core states

(Wallerstein 1974). Core states are those political entities which are the great military and economic powers of the modern world. As Chirot and Hall (1982) write, core states are composed of “well-developed towns, flourishing manufacturing, technologically progressive agriculture, skilled and relatively well- paid labor, and high investment” (p.85). Stated plainly, these areas are composed of the world’s modernized industrial giants who possess vast amounts of wealth coupled with economic and military power—they are the founders and early adopters of capitalist production.

At the very bottom of the system we find peripheral areas. Wallerstein did not label these areas “states” because “one characteristic of a peripheral area is that the indigenous state is weak” (1974, p.349). In any event, peripheral areas are defined by their lack of political strength, their relatively undeveloped lands (i.e. very little or no infrastructure), and their inability to defend against the military strength of the core. Most importantly, what makes a peripheral area “peripheral” is that core states use them to extract raw materials and foster cheap labor.

48

The final economic “class” of the capitalist world-economy is the semi- periphery. These relatively stable political areas are “in between the core and the periphery on a series of dimensions, such as the complexity of economic activities, strength of the state machinery, cultural integrity, etc.” (Wallerstein 1974, p.349).

Core states and peripheral areas can move in and out of the semi-periphery class as the capitalist world-economy experiences economic booms and busts. However, for the most part the world’s stratification system is relatively stable thanks in large part to the semi-periphery. As Wallerstein argued, the semi-periphery was

“necessary structural elements in a world-economy” because they deflect the political anger of the peripheral areas away from the core, and furthermore, they allow capitalists to engage in unscrupulous behaviors that the political elites of the core may disavow within their own boarders (e.g. child labor) (1974, p.350).

Therefore, the semi-periphery acts as both a political and economic buffer between the core and the peripheral areas as well as an alternative arena to engage in unfair economic practices. In short, WST argues that the semi-periphery is a key component of the world-economy because it allows capitalists to maintain the uneven accumulation of the world’s wealth.

2.2.4 Uneven Development and Population

Now that the underlying assumptions, worldview, and basic structural model of WST have been outlined it is possible to discuss how the theory considers modern populations in relation to social change (i.e. development). As mentioned at the start of this section WST does not use population pressure as a causal variable.

Population is simply a proxy for observing distortions in development caused by

49 capitalism and its industrial production systems. Why is this so? The foundation for treating population as a proxy for capitalist distortion resides in two fundamental and complementary assumptions of classic Marxist ideology; one assumption is grounded in dialectic materialism and the other in Marx’s political philosophy.

The materialist assumption asserts that material realities (i.e. geography, natural resources) shape social, political, and cultural realities.8 In Marx’s dialectic materialism human populations rely on natural resources within their environments to meet their natural and species needs.9 Therefore, human populations must either locate near natural resources or near the means of production which grants them access to resources. What this implies is that both material realities (e.g. natural geography) and the means of production (e.g. industrialism) will dictate basic population distributional patterns in addition to resource distributions. This is captured in Marx’s own words when he wrote “the mode of production of material life conditions the general process of social, political, and intellectual life” (1904, p.20-21) (see Hawley [1984] p.905 for origin of quote).

Some general observations support the materialist assumption: For instance, traditional economies based in agricultural means of production generally produce population distributions that are spread evenly across arable lands resulting in a low density and spacious distribution of labor. In this case limited technology requires that humans locate near the resources in order to use them. In contrast,

8 As we will see, this materialist assumption is similar to human ecological theory. 9 Natural needs are the abilities and needs that humans share with animals (i.e. our needs for food, water, and shelter). While species needs are those needs that are uniquely human which animals do not possess. Such needs arise as humans use the environment to create their reality. (Turner, Beeghley, and Powers 2002). 50 modern economies based on industrial means of production tend to produce population distributions that are highly concentrated around the means of production, not the sources of resource materials, resulting in high density and highly concentrated distribution of labor. In such societies advanced transportation and communication systems make the centralization of production possible freeing labor from resource locales but tying them to the means of production. Thus, from the Marxian perspective every production system produces its own logic of population distribution in addition to its own logic of resource allocation.

The political assumption is based on Marx’s view of human nature. It asserts that an equitable distribution of resources and power among a given economic population10 is optimal for maximizing species being (i.e. humans meeting natural and species needs using their natural and species powers to shape limited resources within one’s environment). Given this view of human nature, the greatest human potential among any population is realized when all members of a given economic system have equal access to nature’s resources and can use them to realize their species being. Therefore, from a Marxist perspective the best population distributions are those which connect the maximum number of people to their natural surrounding and the resources contained in them, and allot them control over how resources are used. Because the means of production are tied to how populations are distributed and how labor connects to and uses resources, Marxism argues that certain production systems will produce greater material and political

10 I use the term “economic population” to mean any set of humans engaged in regular exchange and production. In many ways the world-system paradigm treats populations as such, and therefore, it ignores other characteristics of populations including size and density.

51 inequalities than others. Therefore, Marx did not see the value of constructing a concept of population which was tied to biological realities as Malthus had. Instead,

Marx rejected Malthusian views on carrying capacity and the idea of population excess and instead argued that a system’s size and density limits were related to economic conditions (Hawley 1984, p.907). When coupled with the assumption that resources are fixed, under capitalist production population growth always results in greater inequality. Thus, population imbalances with the world economy are not a temporary condition resolved by natural checks; rather, they are a persistent reality that capitalists seek to maintain.

For WST the above assumptions translate into a materialistic political ideal that asserts the true potential of the world’s economy is maximized when resources, economic opportunities, and power are evenly distributed among the multitude of political entities embedded within the world-system. In other words, for a given nation within the world-economy to meet its true potential all the world’s political entities must have an equal position within the world-system and benefit equally from the world’s limited resources. This entails that the world-system be guided by a production system that not only distributes economic opportunities evenly around the world, such that each political system has an equal share in the world’s wealth, but that also grants each political system equal control over the means of production—ideally such a system is the socialist world government (Wallerstein

1974). If such a system were realized, it follows that the world’s population would take on an “even” distribution.

52

In contrast to a socialistic world government, WST argues that the capitalist world-economy is driven by a mode of production which logically moves labor away from realizing species being because the system acts to accumulate profit and power into “private hands” thus further dividing the “haves” from the “have nots”

(Wallerstein 1974). That is, capitalism’s means of production uses mechanization and mass production to maximize profit and lower the cost of wages. As a result, the methods and goals of capitalist production violently divorce laborers from their labor by limiting their input on how resources should be used (i.e. what Marx called the alimentation of labor). Moreover, wealth and power are concentrated more and more into the hands of the few at the expense of the many as economic opportunities become unevenly distributed according to elite interests. Therefore, the capitalist world-economy inevitably produces distortions (i.e. imbalances) in population distributions which favor capitalist interest—this is particularly true of urban areas in non-core nations.

As Timberlake (1987) writes, “Among the most obvious manifestations of the global unevenness produced by world-system processes are those involved in tremendous differences in the apparent nature of urbanization in the periphery compared to the core. Although there are vast differences throughout the Third

World, observable in all are high rates of urbanization in the face of economic stagnation, imbalanced city systems, and over-employment in a labor-intensive service sector and the urban informal sector” (p.40). These imbalances are not only thought to be symptoms of underdevelopment caused by the global spread of

53 capitalism and the uneven distribution of economic opportunities, but also barriers to future development within peripheral areas.

In conclusion, WST argues that the key relationship between population and social change within the world-system is not that shifts in population lead to social change via technological innovations, rather it is that shifts in economic production systems that distort economic opportunities which changes how populations are distributed and access society’s (or the world’s) resources . Therefore, if one is to properly understand the relationship between population and social change from a

Marxian perspective, it is necessary to understand how certain production systems use resources and create economic opportunities. In the modern capitalist world- economy, capitalistic production systems rely on industrialization which requires the concentration of production and resources. As a result, economic opportunities are unevenly distributed and population distributions become distorted. The result is a global vertical hierarchy created by capitalism. These distortions can be seen in the patterns of urban development within non-core areas where overurbanization and urban primacy stagnates economic development, thus keeping non-core nations from realizing their true potential.

2.3 Human Ecological Theory

So far I have discussed how modernization theory and world-system theory deal with population in divergent ways; one treats population as an indirect cause and the other as proximate outcome of social change. Despite their differences, both macro perspectives are grounded in materialist assumptions that assert a

54 causal link between technology and social change. For modernization theory production technologies are the catalyst for ascension through developmental stages, while for world-system theory technology is the catalyst for further exploitation of labor (i.e. population) as it alters the means of production in favor of economic elites.

Despite the above agreement on material causation and the importance of production technology, the perspectives differ as to what the order of the causal relationship is. Modernization theory argues for a reciprocating relationship in which population acts as an intervening variable in the emergence of new technologies. New technology impacts demographic processes which require technological responses (i.e. technology population expansion technology).

World-system theory offers a reductionist explanation in which elites invest in more efficient production technologies that have the effect of concentrating the means of production and labor even more. That is, modern production technologies centralize where elite interests reside, and therefore, distort the relations of production by “over” concentrating labor into a few areas (i.e. elite interest technological innovationsconcentration of production and laborincreased profit).

While these two macro-social materialist perspectives offer narratives of social change which place production technologies at the center of their explanations, human ecological theory offers a broader understanding of social change by focusing on the interactions between population, environment, technology, and organization. In fact human ecological theory offers an alternative

55 materialist perspective that considers population to be both a cause and outcome of social change. And though it acknowledges the importance of technology’s role in eliciting change via population, it does not reduce macro social change to self- interest nor does it reduce it to a series of accumulated or diffused technological progressions that inevitably result in “progress.” Rather, in its most mature form human ecological theories of macro social change are based on evolutionary explanations in which “progress” can be reversed as population’s connection to environment shifts over time (e.g. Lenski 1968, 2005). With this being said, let us explore how human ecological theory views the relationship between population and social change in much broader terms than either modernization theory or world-system theory.

2.3.1 The Ecological Complex

Human ecology, while moving in and out of popularity within the field of sociology over time, has manifest in a variety of forms since its introduction as a promising theory of urban development in the 1920s (e.g. Burgess and Park 1925).

While past scholars have offered general treatments and applications of ecology to sociological issues, early on most of its scholars focused exclusively on the spatial adaptations of populations to their environment or competition between populations over resources in their environment (e.g. McKenzie 1924, 1927; Park

1936; Quinn 1940). What resulted were limited overstatements that were grounded in a narrow focus concerning the importance of competition and spatial analysis (Hawley 1944). It wasn’t until much later that Amos A. Hawley and Otis D.

Duncan solidified human ecology as a legitimate macro-social theory by formalizing

56 and restating its core problematic, principles, and theoretical concepts. In fact, no complete statement of human ecological theory has been as widely accepted or cited as Amos A. Hawley’s (1944, 1950, 1968, 1986, and 1992), and no conceptual model of human ecology’s problematic has been as useful as Duncan’s ecological complex

(1959, 1961, and 1964—see Figure 2.30).11

Duncan’s (1959) ecological complex is an analytical tool used to address the key conceptual problems of human ecology: the reciprocating and interacting relationships between population, environment, technology, and organization

(p.681) (see Figure 2.5 below). It is an organization template that has been used to build theory. Before applying this schematic to an ecological understanding of the relationship between population and social change, let us take the time to define these concepts.

At its most basic level a population is an aggregate of like things (e.g. grains of sand, stones, trees, fish, etc. [Hawley (1986), p18.]). While ecology is concerned with aggregates of living things, human ecology is primarily interested in aggregates of human beings. Like most living things, the human aggregate possesses properties that individual humans do not: For instance, the human aggregate does not possess a spatial or temporal form and its parts are interchangeable. Moreover, the human aggregate does not have a life-cycle, and further, there are no limits to its size

(Hawley 1986). According to human ecological theory these properties must be

11 Others have offered derivative and complementary statements on human ecology. For instance; Lenski (1968, 2005), Berry and Kasarda (1977), Namboodiri (1994). Their insights will be referenced when appropriate.

57

Organization

Environment Technology

Population

* Adapted from Duncan (1959) Figure 10, p.683.12

Figure 2.5: The Ecological Complex.

considered when analyzing human societies since the basic level of analysis is the human aggregate.

While the most basic level of analysis for human ecology is the aggregate,

“the focus of attention in ecology is upon the population which is either organized or in the process of becoming organized” (Hawley 1944, p.403). Therefore, population as defined by human ecology includes some form or level of organization which separates it from random aggregates. As Hawley stated it, “Population and organization enter into the definitions of one another; population is bounded by the territorial extent of organization, and organization is visible only in population

(Hawley 1969)” (Hawley 1986, p.21). In short, human ecological theory does not

12 Aside from the interactions shown, each concept (population, organization, environment, and technology) act on itself albeit at a different point in time.

58 consider unorganized human populations (or simple aggregates) because such populations do not possess unit character (i.e. a shared organized response to environment—for more on the concept see Duncan 1959; Hawley 1944, 1986). Unit character is an important feature of a population because it provides human ecologists with a means to determine the spatial and temporal boundaries of a population including its functional integration. However, the concept of unit character is much more important to human ecological theory than simply offering spatial and temporal boundaries. In fact, it is the necessity for human aggregates to achieve unit character which gives human ecology its fundamental problematic; that being, the way human populations adapt to their environment.

While it is evident from the discussion above that human populations are bounded by some form of organization (more on this below), the need to organize arises from the innate properties of the units of analysis (i.e. humans) as well as the external influences placed on those units by its environment. Although individual humans are not the unit of analysis for human ecological theory, the shared properties of individual human beings are: Specifically and mainly, the necessity for both individuals and populations to connect to environment in order to survive and reproduce (Lenski 2005). Human’s sustenance needs are supplied by the “natural”

(or ecumenic) environment, but for ecology “environment” includes much more than natural landscapes and its resources. Environment also includes the systems of organization used by populations to adapt to the ecumenic environment as well as past adaptations (i.e. symbolic and material culture).

59

Thus, for human ecology the concept of environment is an open-ended concept (Hawley 1986, p.10). In fact, what constitutes an environment is determined by what is being environed. Therefore, what constitutes an

“environment” differs between individuals and populations for obvious reasons.13

Moreover, human environments include other populations, systems of organization

(e.g. government, religions, etc.), social constructions (e.g. language and culture), and technologies (i.e. information about the environment and material artifacts which are responses to those environments). As Hawley (1984) states, “It comprises all that is external to and potentially influential on a population under observation, not excluding other social systems and the interactions they incite”

(p.906). Given the wide variations in what is included in the human environment, as well as what constitutes an “environment,” the key conceptual component as defined by human ecology is that environments both limit and permit population— they pose the problem and solution to living opportunities.

One key proposition of ecology is that living things—including humans— have an inherent tendency to “preserve and expand life to the maximum attainable under prevailing conditions (Simon 1962)” (Hawley 1986, p.5). Thus, different environments will provide different populations with different opportunities for expansion, while some will necessitate contraction for survival. Moreover, different populations will face different environments, and in some cases, the same population will face different environments at the same time. Ultimately, variance

13 For instance, a population can exist in a variety of physical environments at a single point in time. For the individual to be subject to the same variance in physical environment requires a multitude of temporal points.

60 in environment in both time and space necessitates different variations of shared organization even though the underlying purpose of organization will remain the same; to extract energy to sustain a population’s needs. In short, the most important aspect of environment for human ecological theory is that it presents human populations with the problems for survival as well as the solution to survival

(Lenski 2005). Thus, humans overcome the problems environments pose by using information about them as well as the materials contained in them. That is, they develop technology.

Duncan (1959) offers a clear definition of technology as “a set of techniques employed by a population to gain sustenance from its environment and to facilitate the organization of sustenance-producing activity” (p.682). Technologies are accumulated, replaced, diffused, invented, and acquired both within and across the populations of interest to human ecology. While these processes are important to understand (see for example Rogers 1962/2003) the most important aspect of technology is that populations use them to expand or increase their access to energy contained in the environment. Technologies permit populations to acquire sustenance at levels much higher than what natural environments provide (Boserup

1981). With technology human populations can increase their carrying capacity within a given environment—both in size and in density. However, when populations grow beyond the carrying capacity of natural environments or beyond what current technologies permit, problems of cooperation and coordination arise within the population due to energy limitations.

61

As discussed earlier, human populations are living things and need energy.

Five things can limit or expand how much energy populations can acquire from their environment. These include: (1) the existing features of the natural environment— i.e. the amount of useable and potential energy (e.g. , animals, soil, temperature, rainfall, etc.); (2) the demographic characteristics of the population in need of energy (e.g. age structure, sex ratio, fertility, mortality, etc.); (3) the technology or information used to identify, extract, and convert energy sources into useable energy (e.g. agricultural technology); (4) other human populations in competition for the same resources; and (5) the systems of organization used to acquire and distribute energy to the population (e.g. the economic and political systems). Of these five, it is organization that is often seen as the most “social” component of populations since organization gives a population its “character” or form for how a population meets its needs.

Duncan (1959) conceptualized organization as “an adaptation to the unavoidable circumstances that individuals are interdependent and that the collectivity of individuals must cope with concrete environmental condition— including, perhaps, competition and resistance afforded by other collectivities— with whatever technological means may be at its disposal” (p.683). While the concept of organization clearly relies on the other three concepts (i.e. organization can only exist within population, it arises within and as a response to environment, and it is shaped by technology or information about the environment) it does possess its own unique quality; at its most basic, organization pertains to the system of relationships which arise within a population’s response to its environment.

62

Earlier I cited Hawley who suggested that population and organization enter into the definitions of one another. Understanding why this is so is central to understanding how human ecological theory explains the emergence of organizational structure. As discussed at the start of this section, human ecology is concerned with the human aggregate. The human aggregate is composed of individuals, who by their very nature are social creatures. What this means is that while the human aggregate may be composed of individuals, humans do not act or exist in isolation. Rather, humans form interdependent relationships through cooperation and competition in the process of accessing energy in the environment.

Hawley (1986) identifies two basic forms of relationships that individuals and groups form when adapting to environment; commensal and symbiotic (p.30).

Symbiotic relationships are based on mutual support—i.e. complementary differences. Commensal relationships are based on shared similarities. Both types of relationships are present in all human populations to varying degrees with the variations coinciding with the size and density of a population. That is, the larger and denser a population becomes the more variation in environment there will be and the more technology will be required, and thus, the more variation in relationships will exist leading to more symbiotic relations than commensal. In short, for human ecological theory organization is a system of mutual interdependencies that emerge in a population’s struggle to adapt to its environment. As environments, populations, and technologies change, so too does the configuration of human relationships. Therefore, the kinds of relationships and the systems which dictate those relations will be based on; (1) characteristics of the

63 population, (2) the environment from which energy is obtained, and (3) the level of technology which is used to maximize energy extraction and distribution.

While understanding the abstract concept of interrelations within a population is important for grasping the concept of organization, the question of

“what is interrelated” is even more important to address (Hawley 1986, p.31). In order to answer this question it becomes necessary to define the systems in which interrelations take place—those being ecosystems. Hawley (1986) defines an ecosystem as the “arrangement of mutual dependencies in a population by which the whole operates as a unit and thereby maintains a viable environmental relationship” (p.26). Ecosystems, while possessing spatial boundaries, do not possess clear boundaries and are similar to the concept of environment in this respect. In other words, the boundaries of an ecosystem are largely determined by which sets of populations are included in the system being observed. Thus, conceptually ecosystems tend to overlap and infringe upon one another as do environments. However, the concept of an ecosystem is most useful for understanding how organizational structure arises within a given population. That is, structure within ecosystems is an emergent property of population seeking equilibrium with its environment. So what then is interrelated?

Using the logic that aggregates possess properties individuals do not, human ecological theory extends this to systemic relationships within populations. In other words, what is being interrelated are the emergent systems of relationships that a subset of the population performs which help it and the whole population adapt to environment. While it is true individuals perform the tasks in these subsystems,

64 from the ecological perspective individuals are interchangeable at the level of systems and thus no one individual or even one functional position is responsible for emergent social order. The idea that population creates emergent social orderings when interacting with material reality is a key difference between system theories like modernization and ecological theory and those derived from Marxian thought which argue capitalist elites and their individual interests, not emergent properties, drive organization.

In short, form the ecological perspective the tasks being performed by individuals serves a purpose much larger than the individual’s interests, and further, the necessity of the task for the whole population persists independent and beyond the individual’s limited needs. Therefore, at the level of systems individuals become meaningless to analysis and require a unit that possesses emergent properties. For human ecological theory that unit is a function. In Hawley’s words a function is “any repetitive activity that is reciprocated by another or other repetitive activities”

(1986, p.32).

While functions are subsystems of a human ecosystem that connect population and segments of population to environment, not all functions are directly connected to tasks involved in the extraction of external resources and/or conversion to internal resources. However, those functions which are directly engaged with environment are referred to as key functions. All other functions are linked to key functions either directly or indirectly through a series of commensal and symbiotic relations, and those functions most directly tied to key functions are more important to the overall survival of the system. Because functions exist in

65 hierarchic relations to one another, and further, such relationships arise from a population’s need to connect to its environment, it follows that organizational structure of a population will produce a power structure.

As Hawley (1986) points out “The model a system invariably assumes…is that of a hierarchically ordered set of functions” (p.36). In other words, because some functions are more important to the survival of a population (e.g. key functions) it follows that a function’s placement (i.e. degree of importance) within the social system will lead to differences in power between functions. Therefore, the importance of a function and its relational power is not tied to the individuals who fill its positions, but rather to the position of the function within the social system. Hawley identified two criteria which determine a function’s relative power;

(1) the degree to which a function is removed from a relationship with environment, and (2) how easily a given function can be replaced (1986, p.36). Below, Figure 2.6 provides a visual approximation of how these criteria result in a nested hierarchy in which power is attached to a function’s distance from its connection with environment and to its degree of interchangeability.

2.3.2 The Role of Population in System Change

In the previous section we left off discussing the concept of a function. The key element of a function was that it consisted of repetitive tasks undertaken by members of a population in order to connect with its environment. One key feature of a function was that it did not have to be directly engaged in activities involved in energy extraction. However, functions did have to maintain an indirect relationship with the ecumenic environment, even if mediated by other functions closer to it e.g.

66

Environment

Most Key Function Power

Function1

Function1 Function2

Function3

Function2

Function4

Niche Function Function3 3 Space

Least Function Function4 Power 4

Functional Overlap/Competition Emerging Supporting Function FunctionX Distance Removed from Key Function

Figure 2.6: Ecological Hierarchy of Power Structure by Function’s Distance from

Environment and by Degree of Function’s Specialization

a key function. Thus, it was said that population size has a direct impact on how a population is organized as well as the system of functional relationships that arise.

In fact, Hawley (1984) pointed out that a population’s “size, composition, and rate of turnover set limits on what can be realized by the way of system structure” (p.906).

Thus, the larger the population the more variation in environment there will be, and therefore, the more differentiation between function there must be to adapt. In other words, population size has a direct impact on system size and complexity.

67

When comparing small simple populations to larger more complex populations it becomes clear how population size is related to organizational structure, the types of relationships functions take on, and the power structure that emerges within the ecosystem. In simple societies, most members of a population are connected to the key function directly and power is highly centralized. For example, in hunting and gathering societies most if not all members of a tribe will hunt or gather. Because a majority of the tasks and opportunities to connect to environment are limited to the key function, and these tasks require a majority of the population to perform, a more equitable distribution of power exists in simple societies. In modern societies, however, most members of a population are not engaged in activities directly tied to sustenance production or energy extraction.

Instead, most individuals are engaged in functions that are far removed from the key function and are engaged in specialized tasks that support the key function.

This results in more variation in functions and therefore greater variation in the types of relationships that exists within the system. As a result, in large complex societies power becomes decentralized and nested within a function’s position in the hierarchy (see Figure 2.32).

As one can see, when comparing small and large societies from an ecological perspective, the observed differences between social relationships are described very similarly to Spencer, Durkheim, and modernization theory. Small societies are relatively undifferentiated and have a simple division of labor with most relationships based on commensalism (i.e. similarities), while in large societies members are highly differentiated and interdependent and engaged in a complex

68 division of labor in which most relationships are based on symbiosis (i.e. differences). Therefore, larger societies rely on much higher levels of functional interdependence and differentiation (i.e. specialization) to meet the system’s needs than do smaller societies.

While the above juxtaposition of the effect of population size on social organization is important to understand from an ecological perspective, even more important is how ecological theory explains the shift from small simple systems with limited functional relationships to larger complex systems with high levels of functional interdependence. Although population growth is a key element to how these changes occur as it was in other systems theories (e.g. Spencer, Durkheim, modernization theory), ecological models of change are grounded in “orthodox”

Darwinian evolutionary approach to population.14 That is, as mentioned at the very start of this discussion the central problematic for human ecology is how human populations connect to their environment. Like all living things human populations must adapt. Therefore, how population adapts to environment is the focus for how human ecology. Below I outline how human ecological theory understands change as a cumulative evolutionary process of population response to environment. With this in mind I start with a basic understanding of evolutionary change in human populations as discussed by Hawley.

14 This does not imply that they are exactly like Darwin’s model. Human ecology recognizes that human beings can transfer information about the environment independent of genes (e.g. culture, ideology, and artifacts). Thus, even though there is much overlap between how human ecology models social change and Darwin proposed evolution unfolded, they are not exact since humans have greater control over their response to external change than all other forms of life on earth (e.g. see “Chapter 2” in Ecological-Evolutionary Theory: Principles and Applications, Lenski 2005). However, population survival (i.e. a population’s ability to adapt to its environment) and not the survival of individuals is an important overlap between human ecology and Darwinian models of evolution.

69

Hawley (1986) points out three assumptions of evolutionary thought guide human ecological theory’s basic modeling of social change as a function of population’s response to its environment. The first pertains to the ultimate purpose of all living things—to adapt to environment in order to survive and reproduce. The second relates the first; all populations seek equilibrium with environment and will

“preserve and expand life to the maximum attainable under prevailing conditions

(Simon 1962)” (Hawley 1986, p.5). These assumptions, maximum adaptation and an indefinite pursuit of equilibrium, imply the next: That is, the origin of system change is ultimately external to the system undergoing the change.

The last proposition does not imply that systems are passive in the process of change, or that change is never the result of internal processes. Unlike world- system theory which is based on a model that assumes change always originates internally via contradictions in key functions, human ecology posits that a thing cannot cause itself, and therefore, logically change must have external origins

(Hawley 1986). In other words, while populations may go through internal changes that impact its organization and relationship to environment populations undergo these internal changes in order to adapt to changes in environment—not to adapt to changes within it. Given this, human ecological theory asserts that change is best described as a series of external and internal phases with external inputs being the first in the series. Therefore, internal phases of system change can impact external environments and effect populations at a later point in time, especially when a population depletes energy from a given environment. However, the need for a population to undergo internal changes originates in response to external

70

Consequences for Population: Loss of Technology/Information/Culture1 Less Efficient Communication and Transportation1 Connection to Lowered Complexity/Interdependency2 Environment 2 2 Decrease in Population Size and/or Density Contracted Failure to Adapt Increased Centralization of Power (More Competition)

Change in Disequilibrium Repeated in Population Failure to Extinction Equilibrium Environment* Adapt

Adaptation via Connection to Technology Environment Consequences for Population: Expanded2 Acquired Technology/Information/Culture1 (Niche Space) More Efficient Communication/Transportation1 Increased Complexity/Interdependency2 Increase in Population Size and/or Density2 Decentralization of Power

1First-order Effect 2Second-order Effect *Occurs over time Feedback

Figure 2.7: Ecological Model of Social Change as a Function of Population

Adaptation to Environment

environmental inputs. Therefore, from an ecological perspective change involves a series of syntheses and accumulations as populations adapt to their environments.

Along with the above assumptions concerning the origins of change, human ecological theory borrows from evolutionary theory the notion that not all responses to environment are successful, and in some instance populations cannot adapt to their environment (i.e. achieve equilibrium). Therefore, like Darwinian evolution, social change is not a linear or deterministic process but a probabilistic one resting on the success of populations’ ability to adapt and not the individual’s.

However, for human populations change is cumulative and past adaptations can have important and far reaching impacts on current and future successes (Lenski

71

2005). To explore this further, let us take a look at Figure 2.7 which offers a visual representation that is useful for understanding how the assumptions and propositions of ecological theory model social change as a function of population’s adaptation to its environment

In Figure 2.7, we see that system change begins with a change in environment that places a population (more specifically a population’s organizational system) into disequilibrium. Populations must adapt to environmental circumstances even if those circumstances are linked to a previous response to environment (thus, the reason for the feedback loops in Figure 2.7).

Two things about change originating in “environment” need to be noted before moving on: First, as discussed in an earlier section, “environment” is an open-ended concept that includes other human populations as well as a population’s information about other environments, social systems, beliefs, etc. Thus,

“environment” includes more than the ecumenic, although such “natural” external changes will have the most immediate impact on a population’s ability to connect to it (e.g. natural disasters). Second, environments are dynamic and change is a process that occurs over time. Therefore, not shown in the above model is time.

Ideally a second or a “Time 2” model would be placed adjacently in a three- dimensional space and linked via temporal change in environment (i.e. “Change in

Environment t1”  “Change in Environment t2”, etc.). With these two points noted I move to a full treatment of the model.

Faced with a change in environment, a population is placed into disequilibrium. Disequilibrium occurs anytime a population’s current

72 organizational system cannot supply the population with the energy or sustenance it needs to persist. Therefore, faced with a changing or new environment a population must respond via adaptations. But what type of “adaptations” must a system undergo in order to meet the needs of the population?

From an evolutionary perspective there are two possible avenues a population can take in order to adapt and approach an equilibrium state15; both depend on the properties of the environment and the properties of the “organism” adapting—in this case a human population. Because populations adapt to environments using organized systems or functions that rely on technology (i.e. information about environment), it follows that any external change16 will affect the organization and relationship of a system’s functions. Moreover, because adaptations are accumulative past information will affect the probability of successful adaptation.

A population’s response to new environmental inputs involves changes in a population that act independent of the population or environment’s idiosyncratic properties. Let us review these changes: As Hawley (1986) points out, disequilibrium results in first-order and second-order effects on a population’s ability to connect to its environment and while they can be disentangled for analytical purposes, in reality these effects are inseparable (p.62). The two first-

15 Human ecology uses the concept of “Equilibrium” as an ideal state, not an empirical reality. There are too many fluctuations in environment and within systems to reach equilibrium. Moreover, the mere extraction and use of energy changes environment. Thus, systems can approach near equilibrium states but they will never be in equilibrium. 16 It should also be noted that human ecology does not presuppose a rigid and fragile ecosystem for human populations, but rather, one whose boundaries are often blurred and flexible to minor undulations in environments. Therefore, external environmental origins which lead to system change must be large enough or persist long enough to break this flexibility.

73 order effects include: (1) increases/decreases in costs associated with communication and transportation; and (2) acquisition or loss of technology, information, or culture. The three second-order effects include: (1) enlargement or reduction in accessible resource base; (2) an increase or decrease in population size/density; and (3) increased or decreased differentiation in specialization of function. Given the above parameters of social change, let us explore how a change in environment and a population’s response leads to the above effects.

The first outcome of a population’s continued failure to respond to its environment is extinction. As with Darwinian evolution, human populations can become extinct from their failure to adapt to changes in the environment. In short, if a population can no longer access its environment it will die. Thus, while extinction is rare and can occur through one temporal sequence (e.g. natural disaster such as a tsunami, earthquake, or volcano on an island) it is most likely to occur over a long period of time (e.g. war, cultural competition, disease, droughts, etc.). Therefore, extinction is much more likely to happen through a series of de- evolutionary responses as exemplified in the top path in Figure 2.7. Let us look at this sequence in greater detail.

In some instances changes in environment may be subtle or at least subtle enough that a population does not respond as quickly as it should or can. Or, it may be that the population fails to identify the sources of changes in the environment

(e.g. water becomes poisoned) and cannot develop the technology (i.e. information) to overcome it. Under these circumstances a population may be forced to “adapt” to new conditions by reducing excess population and/or by reducing the number of

74 functions in the ecosystem. In other words, when environment changes or if a population is placed into a new environment, if the population cannot adapt or adapts inefficiently the size and complexity of the system must be reduced for equilibrium to be achieved since size and complexity require more energy (Boserup

1981). If a system is simplified enough and living opportunities become scarce, the population will be forced to contract its numbers to a size that matches the amount of energy being produced. This Malthusian path to equilibrium has important consequences for how populations are organized. In the wake of population loss, past communication and transportation systems must be simplified since there are fewer individuals to fill positions and perform tasks. In addition, populations may simplify its hierarchy by removing repetitive functions or those functions that are no longer necessary or unproductive. Taking these steps lessens the amount of energy and units required to maintain the system and thus the population’s connection to environment. Therefore, a consequence of “shedding” functions and reducing system complexity in the wake of population loss is that the system requires less interdependency and fewer specialized tasks (i.e. functions) to fulfill the population’s needs. As a result, power becomes more centralized as a greater percentage of the population engages in fewer specialized functional tasks.

In some cases, adaptation to the environment simply means maintaining system stability, size, and density (e.g. working longer to make up the difference).

Alternatively, change in environment can have long term beneficial effects for a population, especially if a population is expanding into a new environment or increases its carrying capacity or density limit. In Figure 2.7 we see that the other

75 possible response to disequilibrium is that a population acquires new information about the changes in its environment. This brings about an important point about the unique property of technology within human populations.

Human ecological theory follows a Darwinian model of evolution with the exception that human populations can transfer information about environment independent of genes (i.e. via culture) (Lenski 2005). Therefore, for human populations adaptations to environment are cumulative. What this means is that as populations are faced with new environmental conditions or expand into new environments, past adaptations and technologies can be used to overcome those new circumstances (á la Spencer, Durkheim). Thus, when populations can apply either new or old information to a new environment, opportunities for living are dramatically increased and more energy becomes available to the population.

In human ecology unrealized opportunities for living are referred to as a niche spaces (see Figure 2.6 and 2.7). A niche space is not only an opportunity to expand living space for individuals or a population’s carrying capacity; it is a chance to expand system complexity which guards against future changes in environment.

The concept of niche space is easiest to understand when a population’s response to a change in environment results in more available energy (i.e. an expanded connection to environment via emergence of a key function). When a population secures more resources it secures more living space, and it follows that populations will expand to fill those spaces. These opportunities bring about changes in population that lead to more efficient systems of communication and transportation

(i.e. increased ability to overcome variations in environment and expand networks

76 within the ecosystem), the emergence of more functions as well as more relationships between functions (i.e. increased specialization of functions and greater system complexity), increases in population size and/or density (i.e. more individuals can be supported in less area by the system since more energy is available), and a decentralization of power (i.e. more system complexity and specialization requires a more complex hierarchy and greater functional interdependence). Thus, when populations are able to adapt by acquiring more energy, they expand to the size and complexity that the new relationships allow and approach equilibrium.

In short, Figure 2.7 illustrates that social change from the ecological perspective is a series of accumulated adaptations of a population to changes in the environment or the expansion into new environments. If adaptations are successful a population’s opportunities for living are expanded as is the size of the population and the ecosystem which supports it. The larger a system becomes and the more variation in environment it covers, the more functions will arise, the greater variability in functions there will be, and thus, the more complex the relationships between functions there are. Therefore, the path from small simple to large complex societies is one of evolutionary and de-evolutionary advancements in the wake of population’s response to environmental inputs. In sum, from the perspective of human ecology population is a form of social change.

2.4 Conclusion

In this chapter three macro-social theories of social change were discussed with a focus on how each theory viewed the role of population in social change.

77

While each theory offered its own unique view on the relationship, in many cases there were points of agreement and disagreement. In light of this, I conclude the chapter with a table comparing and contrasting modernization theory, world- system theory, and human ecological theory. The similarities and differences uncovered in the table below are useful as both a conclusion to this chapter and as a point of reference for what will be empirically tested in the chapters to follow. More specifically, what will be of interest in the proceeding chapters are the underlying assumptions concerning population dynamics and their relationship to changes in modern social systems.

As discussed in the introduction, around the world modern living spaces and opportunities for living are becoming more and more dominated and dependent on urban economies and urban environments. Therefore, understanding how and why populations converge toward urban living and the impact such configurations have on societies requires a firm and fundamental understanding of how population dynamics contribute to important social changes connected to urbanization. In large part the theories covered in these two chapters have provided a theoretical base to develop the hypotheses needed to explore the role population plays in modern social change. With that being said, let us turn to Table 2.1 for a brief summary comparing the theories covered in this chapter.

Although the theories covered above share a common goal—to understand social change in the global system of societies—each theory asserts a different unit of analysis. For modernization theory and human ecological theory, populations that share key organizational components are of interest. Modernization theory

78

Human Theory World-System Theory Theory

Units of Analysis1 The Nation-State The World-Economy Human Ecosystems

Centuries, beginning with Centuries, beginning with Time-Frame Considered All of human history industrialization world-empires

Ecumenic and Social Origin of Social Change Human Innovation Economic Systems Environments

Basic Model of Social Change Progressive Stages Dialectic Evolutionary

Main Determinant of Technological Innovation Contradictions within Changes in Environment Social Change or Diffusion Economic Systems

Defining Feature of a Shared Technological Shared Position in Shared Organized Population Knowledge Economic Hierarchy Response to Environment

Main Determinant of Ideology and Ecumenic Environment Means of Production Population Size and Density Technology and Technology

Role of Population Mediates Technological Serves as a Proxy for Adapts to Changes in in Social Change Progressions Distortions in Economies Environment 1All three generally take the global system of total societies into consideration (see Lenski 2005, p.127) but the type of unit focused on differs.

Table 2.1: A Comparison of Modernization Theory, World-System Theory, and

Human Ecological Theory on Selected Dimensions related to Population and Social

Change.

relies on political organization to identify system boundaries, while human ecological theory uses a population’s connection to environment or the ecosystem.

World-system theory, on the other hand, argues that the nation-state or any other similar subset is not the proper unit of analysis. Instead, it argues that the world’s population shares the same economic system (i.e. capitalism), and therefore, there are no system boundaries only positions within the world-economy. While there are merits to using each unit of analysis, in most comparative research the nation- state remains the accepted unit since information used for statistical analyses is

79 gathered at this level. More often than not this does not present a problem since ecological approaches argue similar processes for scaled ecosystems and world- system theory uses nation-states to divide societies into “classes” within the world- economy.

In terms of the time frames considered, modernization theory and world- system theory share the view that the most important social changes have occurred since the dawn of industrialization or mechanized production. Therefore, both perspectives are generally focused on social change since the 1500’s with modernization theory particularly interested in the late 1700s and early 1800s.

Human ecological theory, on the other hand, takes the whole of human history into consideration. This is so for two reasons: First, humans developed behaviors and relationships through the course of evolution and thus human nature is product of hundreds of thousands of years of adaptation. Second, humans have the unique ability to transfer information about their environment via culture in the form of tradition and technology. Therefore, human adaptations are cumulative and span over the entirety of human history. Given this, human ecological theories argue that understanding current forms of social organization and culture requires understanding past adaptations as well.

In terms of the origins of social change modernization theory and human ecological theory share the view that change has origins that are external to the system. However, modernization theory places an emphasis on human innovation and technology so much so that the assumed external origins are somewhat overlooked (e.g. modernization theory generally assumes the human condition

80 motivates individuals to seek methods to overcome environment but it does not speculate beyond this). Human ecological theory emphasizes the external origins of change (i.e. environment) while arguing that the properties of human populations interact with environments to bring about system change. World-system theory argues that change originates from internal contradictions associated with the means of production—i.e. economic systems. While world-system theory, like

Marxism, acknowledges that humans must connect to their environments to meet their species needs, it argues that the types of technologies used to meet those needs ultimately shapes how populations are organized in relation to their environments and to one another. Therefore, any external forces are secondary to internal forces.

Given the above, each theory has its own basic model of social change and points to a different determinant of social change. Modernization theory, because it emphasizes technology, argues that social change is best understood as a progression of technological accumulations which have predictable impacts on demographic processes that are best understood as a series of growth stages. From this perspective, social change is seen mostly as progressive steps that populations take to achieve better control over fertility and mortality as well as external stimuli in the environment. In contrast to human ecological theory, modernization theory sees technology as human’s attempt to overcome environment, not to adapt to it.

This leads to the view that social change is a progression towards that goal and that technology (its invention and diffusion) drives those changes.

81

Human ecological theory, while it agrees that technology is an important component of social change, does not believe that humans can overcome their environments with it. Instead, humans use technology to adapt to environmental changes which both limits and permits their existence. Since populations are continually adapting to environmental changes that are beyond their control and do so both successfully and unsuccessfully, social change is best understood as an evolutionary process with the main determinant being the environment.

Finally, world-system theory takes the view that social change is best modeled as a series of syntheses resulting from the emergence of contradictions in human’s attempts to meet their species needs and satisfy wants using their environments. This logic is derived from Marx and Engels’ view of production systems throughout history. They argued that social change is brought about the imbalance between the interests of those who own the means of production and those who do not. Specifically, production systems tend to use resources in ways that favor the interests of those who own them. Because only a small portion of the population owns the production systems, it follows that the needs of the many will not be met. Thus, over time, a majority of the population suffers, causing strain leading the masses to revolt against the means of production and those who own them. In their place, the masses create new productions systems which synthesize components of the old with the new goals of those in power. Over time contradictions in the new system arise and the process repeats itself with new productions systems put in place each time. Therefore, social change is best

82 modeled as a dialectic process with the main determinant of social change being the rise of contradictions with economic systems.

Because each perspective has its own unit of analysis, time frame, origin of change, model of change, and main determinant of change, it follows that each theory views population differently. And while modernization theory and human ecological theory overlap in their conceptions of population as well as what population’s link to social change is, there are still differences worth noting. Before looking at these two theories let us look at world-system theory which treats population as a proxy instead of a cause for social change.

World-system theory does not formerly define a population since it takes the world-system as its unit of analysis which includes all human beings. However, it does identify common boundaries for human beings which in many ways help identify a “population.” Let us look at these two boundaries. The first is the type of economic system humans share to meet their needs. The second boundary is the shared position individuals with the economic system fill (i.e. classes). While the first boundary is similar to how human ecological theory defines a population, world-system theory does not elaborate—again, because it argues that the world shares the same economic system i.e. capitalism. Therefore, position within the economic system is the key feature or closest delineation used to identify a

“population” for world-system theory.

Given the above it is not surprising that world-system theory sees the means of production as the main determinant of population size and density. While the theory does not explain growth in population numbers, it does relate population

83 size and density to population distributions within a given economic system. In other words, how populations are distributed around and in relation to the means of production is of interest to world-system theory as we will clearly see in the next chapter. Therefore, for world-system theory population is not a cause of social change although it certainly is an intervening variable in the dialectic process.

Instead, population remains a variable of interest only as it relates to distortions in the world-economy caused by the capitalist economy (i.e. as an outcome variable).

Unlike world-system theory, modernization theory and human ecological theory define population more clearly, identify its limiting factors more openly, and offer a clearer understanding of the role it plays in social change. And although the two theories offer similar explanations, they do have key differences. Let us conclude this chapter with a look at these two perspectives.

Modernization theory argues that the defining feature of a population is its level of technological know-how. In some ways this overlaps with human ecological theory which argues that the defining feature of a population is a shared and organized response to environment. Since technology is information about a given environment, in many respects shared and organized responses are “technology” if one defines “technology” as information about a given environment. The key difference, however, is that human ecology sees the need for humans to organize and respond to environment as the key feature that gives a population its spatial and temporal boundaries. Conversely, modernization theory asserts a more technological-deterministic view of population using technology, in that technology

84 is often conceived as being derived from the human being and thus is treated independent from the environment in which it was created to overcome.

Despite this fundamental difference, when it comes to identifying what the main determinant of population size and density is modernization theory and human ecological theory share very similar views on the role of population. Both argue that technology is the factor which permits populations to grow in size and increase in density since technologies produce the energy and substance required to support population’s carrying capacities. At the same time, both see population as a catalyst for social change and the creation of new technology. One minor difference is that human ecological theory emphasizes the ecumenic environment’s role while modernization theory emphasizes the role of ideology. This is not to say that modernization theory ignores the ecumenic environment or that human ecological theory ignore ideology, it simply means that the theory emphasize different complementary factors in addition to technology.

Finally, modernization theory and human ecological theory depart in an important way when it comes to the role population plays in social change. While both perspectives share views on the role population size and density have on social organization (i.e. increased complexity, specialization, and interdependency of functions), they differ with regards to why those properties ultimately arise. For modernization theory modern social systems arise from technological progress in which population gain greater control over environment. Therefore, technology permits population growth which leads to strains of social systems that require more technological responses to achieve system stability. In such a model,

85 population is a mediating factor in the path toward development. In contrast, human ecological theory argues that system complexity arises from a populations’ struggle to adapt to changes in its environment. System complexity is not the result of technological progressions although technological information is cumulative; instead complexity is a response by a living population to adapt to its environment in order to expand living to the maximum size and density that a population’s environment and its information about that environment will allow.

86

PART II

87

INTRODUCTION

Now that a theoretical background on the role of population in social change has been discussed, in the next three chapters I turn my attention to the tested relationships found in Figure 1.1. Chapter 3 explores the theory and research surrounding the reciprocal relationship between economic systems and urban transitions. I look to how and why modern economies are intimately linked to urban areas and explore how each school of thought approaches the relationship. In

Chapter 4 I look at the research surrounding the relationship between economies and political systems with a focus on the democracy/development debate that is at the core of cross-national development studies. In this chapter I also explore how the concept of development has come to be equated with level of economic development. I speculate that this may be the central reasons why the role of urban population in eliciting political change has been left out of most analyses. In Chapter

5 I explore the empirical literature concerning the effect political systems have on patterns of urbanization. In particular I look at recent research which suggests that autocratic governments promote urban concentration through their policies to maintain their control over population. I conclude Chapter 5 with a review of several studies that come closes to and compliment the notion that urban systems and urban populations can have a direct effect on political systems over time.

88

CHAPTER 3

ECONOMIC DEVELOPMENT AND THE URBAN TRANSITION

In the previous two chapters I focused on macro-theoretical perspectives that linked population to key social changes in the modern social systems. In particular I looked at the predicted effects of population size, growth, and density on the organization and evolution of societies from three perspectives; modernization theory, dependency/world-system theory, and human ecological theory. In this chapter I turn my attention to the role the urban transition plays in the formation of modern economic systems and vice versa (see Figure 1.1). Specifically, I focus on issues of comparative urbanization with an emphasis on the redistribution of rural populations to urban areas. Such an analysis inevitably involves a closer examination of “peripheral urbanization” (Walton 1982).

In order to assess the relationship between urbanization and economic development I start with a fundamental understanding of how and why urban areas are linked to economies independent of national and international contexts/boundaries. I then explore the dominant theoretical schools of thought which have been used to ascertain this relationship in developed and developing

89 regions of the world. This includes a review of one ancillary theory (i.e. urban bias theory which in some ways is an extension of dependency/world-system theories) and a review of two key concepts that inform economic geography’s understanding of urban growth and economic development (which compliments modernization/ ecological theories). After discussing the broader theories, I explore the concept of overurbanization and examine the empirical literature which has linked it and urban primacy to economic development. Aside from providing insight into why peripheral urbanization has been the most studied of all the relationships comprising the development processes (again see Figure 1.1), these extensive reviews provide insight and background for the hypotheses tested in Chapter 7.

3.1 Urban Areas and Economic Systems

Anthropological evidence suggests that the rise of cities can be traced to a series of demographic and technological progressions which occurred around

12,000 to 14,000 years ago. At this time humans began to grow their own food instead of hunting and gathering it (i.e. the Neolithic revolution—Davis 1955;

Bairoch 1988). Whenever and wherever it was possible to successfully implement agriculture and produce a food surplus (e.g. Mesopotamia), populations were able to

“free-up” excess labor and establish permanent settlements.

The world’s first cities arose from these “pre-urban” areas but the path to took thousands of years to complete because early settlements lacked the technology to support large dense populations (Boserup 1965). As Davis (1955) noted, settlement populations did not have the transportation or communication technologies to expand agricultural production. In fact, very few of the ancient

90 world’s settlements reached the size/density needed to develop a specialized division in order to truly be called “urban.” Those that did were almost always located near natural “surplus” areas such as coasts, inlets, oases, or major rivers; all places were the soil remained fertile and where transportation is made possible with little effort.

Some of the more important technologies early populations lacked which were needed to sustain large dense populations included the ability to store large amounts of agricultural surplus for long periods of time, the sanitation and irrigation technology needed to provide fresh water and sewers, the transportation technology needed to import and export agricultural surplus across great distances, and the communication technology needed to develop a political system capable of protecting production and maintaining systems of trade within a large area (Davis

1955; Hoselitz 1955). However, as Bairoch points out, the most important technological limitations imposed on pre-urban settlements were the poor transportation and communication networks which limited trade and thus surplus value. He states:

…the existence of true urban centers presupposes not only a surplus of agriculture produce, but also the possibility of using this surplus in trade. And the possibilities of trade are directly conditioned by the size of the surplus relative to the amount of ground that has to be covered in transporting it from one place to another, for distance reduces the economic value of the surplus (1988, p.11).

Despite their small size, low density, and inability to overcome vast distances and obstacles, the growth and distribution of the most successful pre-urban populations was driven by the same basic economic forces as modern urban

91 systems. This is a point worth emphasizing. In early settlements sustained growth was achieved when technological advancements aided in increasing agricultural output (e.g. the plow), which meant that the number of workers needed to produce the same amount of food decreased. Over time non-agricultural labors emerged and made their livings engaging in tasks that supported more efficient trading of surpluses. The steady release of agricultural labor resulted in the increased centralization of non-agricultural labor. Eventually, centralizing food surplus, excess labor, and non-agricultural production allowed populations to increase their economic efficiency since centralization allowed settler populations to take advantage of economies of scale (more on this below). Thus, the success of early settlements and small cities, like modern urban systems, rested on a population’s ability to produce a food surplus, free up agricultural labor, and form efficient trade systems.

As an activity, trade requires overcoming problems of distance and coordination. The formation of settlements and cities has been the most efficient means for overcoming these obstacles. Centralization increases the flow of information, reduces the costs of infrastructure and labor, as well as increases efficiency in government and production.

What can be taken from this brief sketch on the rise and importance of cities/urban areas is this: Urban areas arise from a traditional population’s need to centralize surplus materials and labor in the wake of population growth. It is a

“universal” process in the sense that the demographic changes that take place can be generalized across cultural contexts (e.g. rural to urban migration, reduction in

92 agricultural labor, lower mortality). As Schnore (1964) stated it, “The process of urbanization…[is] a reorganization of activities, or a shift in social and economic structure that places a population in a new relationship to its total environment. It is said to involve a transition from agricultural to non-agricultural pursuits, and a concomitant increase in specialization of economic activities” (p.38, italics original).

Indeed, urbanization is a central feature of modern economic and political systems of nation-states. However, despite these “universal” processes and consequences, the role the urban transition plays in economic development appears to be anything but universal. Different societies have been shown to have different experiences with how the urban transition affects their path to economic development as well as how a nation’s level of economic development affects its path to urbanization. For this reason the thrust of development studies has been aimed at explaining the relationship between modernization and economic development.

3.2 Theoretical Approaches to Urbanization and Economic Development

In Chapters 1 and 2, I discussed theoretical perspectives that made predictions about the role population played in eliciting social change. In this section I review how others have extended these same perspectives to the relationship between the urban transition and economic development.

Following London (1987), and then Crenshaw (1992), I consolidate the theoretical perspectives into two schools; the modernization/ecology school and the dependency/world-system school. Grouping is based on shared fundamental assumptions which inform predictions concerning the relationship between

93 urbanization and economic growth. These shared assumptions and predictions lead each school to form fundamental perceptions about social, economic, and political inequality as a feature of the total development process (see Figures 3.1 and 3.2 below).17 Moreover, these fundamental assumptions lead each school to emphasize one aspect of the causal direction over the other: the modernization/ecological perspectives emphasize that UrbanizationEconomic Development while the dependency/world-system perspectives emphasize that Level of Economic

Development  Urbanization.

The source for emphasizing different causal orderings has to do with how each school of thought explains rural to urban migration as a feature of economic development.18 It is also important to note that the modernization/ecology school tends to emphasize the role rural push factors play in the urban transition, and therefore, emphasize how urbanization leads to economic development (i.e. urbanization precedes development). Conversely, the dependency/world-system school tends to emphasize the role urban pull factors play in displacing rural labor.

It argues economic development (at least for the developing world) has mostly preceded urbanization in the developing world which directly affects patterns of development visible in distorted patterns of urbanization.

3.2.1 Dominant Schools of Thought

17 It is important to note that for both of the figures below, “inequality” is purposefully abstracted since it can refer to a variety of predicted imbalances associated with the overall development process (e.g. Alonso 1980).

18 It is important to note that despite emphasizing different aspects of the reciprocal relationship both theories acknowledge the total relationship (London 1987; Crenshaw 1998).

94

As it was pointed out in Chapter 2, one major difference between modernization/human ecological theory and dependency/world-system theory was that the former two theories regarded population concentration as a catalyst for social change and development while the later two theories regarded population concentration as an outcome of elite interests/global capitalist production. This same fundamental distinction is exhibited in how each theory explains the role of urbanization in economic development. Modernization/ecology sees change as an adaptive process to environment, population pressures, and demographic realities.

Dependency/world-system sees change as a dialectic process between the means of production and the emergent relations of production. Therefore, as shown in Figure

3.1, each school of thought emphasizes a different causal relationship when explaining the reciprocating relationship between urbanization and economic development: Modernization/ecology emphasizes the role of population pressures and scaled/agglomeration economies as the primary engine of change, while dependency/world-system theory emphasizes position within the global economic hierarchy, past economic relations, and the dominant economic means of production. Let us look deeper into these fundamentally different approaches.

According to modernization/ecology the urban transition can be largely explained as the result of rural push factors brought about by technological advancements that accompany the demographic transition (i.e. high fertility and low mortality “pushes” excess agricultural labor to urban areas in the wake of improved agricultural technology) (Davis and Golden 1954; Davis 1955; Firebaugh 1979;

95

Urbanization

Political Systems

Economic Growth

Main focus for Modernization/Ecology

Main focus for Dependency/World-System

Figure 3.1: Main Focus of Interest for Modernization/Ecology and

Dependency/World-System Concerning the Relationship between Urbanization and

Economic Growth

Alonso 1980). Urbanization quickly becomes a process of massive population reorganization, changing not only where people live but how and why they interact on both a cultural and economic level (Schnore 1964; Banks 1974; Inkles 1975).

Thus, the initial “costs” associated with this reorganization are extremely burdensome (Davis 1955; Lipset 1959). However, over time integration is eventually achieved resulting in the formation of cross-cutting affiliations and mutual dependencies leading to the decentralization of wealth and power. In short, modernization/ecology sees the urban transition as a part of the total modernization process or as a drive to economic maturity that brings with it a change in political systems (more in Chapter 4 on the link between development and democracy). Economic development is highlighted by a series of technological progressions and demographic shifts in which populations and power are highly

96

60

1.0

Dollars U.S. Standardized

.75 GDP per capita

.50 Inequality High Growth Transitional Incipient Potential Growth Decline

in

.25 Thousands

Agricultural Labor % Labor % in Agriculture Gini / Index I II III 0

0 Time 

*Note that this pattern is imposed upon the parallel process of the demographic transition which appears in the background.

Figure 3.2: Economic Growth and the Urban Transition: Modernization and Human

Ecology*

centralized until institutions mature and can overcome the problems of economic, cultural, and political integration. Once in place, modern institutions promote a gradual decentralization of power and wealth among groups with competing but interdependent interests. Decentralization (greater equality) is a necessary feature of modern societies.19

Figure 3.2 illustrates the fundamental predictions made by modernization/ ecology with an emphasis on inequality. In fact, the most important feature of this model is the inverted U-shape inequality takes on. Again, the rise and fall of

19 It is important to acknowledge that modernization theory and human ecological theory do disagree as to the “path” of these developments. Modernization predicts staged growth while human ecological theory predicts evolutionary growth and therefore progress is not inevitable. 97 inequality (broadly defined) is explained as the result of the high costs associated with the initial investments in infrastructure/social institutions associated with the urban transition; a process that requires the continued concentration of capital and people through intermediate stages of growth (Crenshaw 1992). At later periods the costs of urbanization shift from infrastructure to relational interdependence which is supported by the free-flow of information, services, and consumption; all of which requires interdependency and further decentralization of wealth, capital, and power. 20

The dependency/world-system perspective argues urbanization cannot be properly understood without accounting for the larger global capitalist economic system that shapes how modern economic systems develop in core and peripheral regions (Kentor 1981; Walton 1982; Timberlake and Kentor 1983; Timberlake

1987; London 1987). In particular, this approach has emerged in opposition to modernization theory on the grounds that modernization theorists mis-modeled the relationship between urbanization and economic development by suggesting urbanization has intranational origins and universal economic outcomes

(Wallerstein 1974; Moir 1976; Timberlake 1987).

The basic hypothesis of dependency/world-system is that core nations have been able to sustain steady economic growth and balanced urbanization at the expense of peripheral development. Modernization theorists have been slow to

20 Human ecology explains this process with a very similar logic but uses different language and focuses on a population’s changing connection to environment. Thus, modernization theory and human ecological theory are highly compatible with one another in their broader predictions about economic development as a process of system growth and integration despite using different language. 98 recognize this as such, and as a result, they have mistakenly come to view the development patterns of core regions as a universal process instead of a historically driven outcome that cannot be repeated in areas exploited by the core. In fact, dependency/world-system theorists argue development in non-core nations is sure to take a different, mostly distorted, path since the core uses the periphery’s resources to modernize. As Timberlake (1987) states it “Among the most obvious manifestations of the global unevenness produced by world-system processes are those involved in the tremendous differences in the apparent nature of urbanization in the periphery compared to the core” (p.40).

According to the dependency/world-system perspective distorted development is easiest to see in the periphery’s urban developmental patterns.

Unlike the urban areas in the core which are evenly distributed and internally well integrated, urban areas in peripheral nations are concentrated and disintegration with the rural population. Moreover, unlike core regions which use urban areas to create their wealth, in non-core regions primate cities act as economic outposts for core nations (see Armstrong and McGee 1985). In short, urban areas in the periphery serve the economic interests of the core, and are thus, like the economies of peripheral nations they are underdeveloped and disarticulated.

Most dependency/world-system theorists agree that urbanization and economic development in core regions follows the trends seen in Figure 3.2

(London 1987). However, they argue that this model does not illustrate how peripheral urbanization unfolds because it fails to consider how core nations use their dominance to exploit the periphery’s wealth (Timberlake 1987). In contrast to

99 the modernization/ecological approach, the dependency/world-system perspective starts with the argument that in peripheral regions rural-to-urban migration is primarily generated by urban pull factors resulting from foreign intervention.

Specifically, this perspective argues that urban pull factors can be traced to capitalist investment in one or two urban areas which creates the illusion that economic opportunities are abundant (Kentor 1981; Timberlake and Kentor 1983).

Adding to the pull factors are rural push factors that are also created by core capitalist penetration into the rural areas of the periphery. However, these factors are secondary although they do remain important. Rural push is generated by cheap agricultural imports from core nations, the rapid diffusion of agricultural technology to peripheral hinterlands, and the commercialization of agricultural production. Each distorts traditional rural economies in the periphery and reduces the value of rural labor resulting in sustained disarticulation between rural and urban economies (Kentor 1981).

What the above conditions lead to is a massive release of rural laborers who turn to urban areas for employment. Unlike core regions, peripheral nations do not have the infrastructure or economic complexity needed to accommodate the rapid influx of rural migrants once it begins. Thus, unemployment in urban areas skyrockets, social problems spin out of control (e.g. lack of housing and healthcare, crime), and peripheral governments must bear the costs of social welfare at the expense of securing future economic growth (Kentor 1981; Walton 1982).

Moreover, because peripheral nations do not have a developed urban system they tend to produce primate cities that concentrate unemployment and

100

1.0 100

Elite Share World’ Elite of

Elite/Core Control .75 of Wealth Inequality

.50 Wealths

Percent Rural

as a Percent a as

.25 Population % Rural % Population/ Gini index I II III

0 0 Time 

Figure 3.3: Economic Growth and the Urban Transition for Developing Regions of the World: Dependency/World-System

underdevelopment into a single area (Smith 1985). Thus, distorted dependent development in peripheral areas can been seen in the distorted urban patterns that emerge from core penetration in to peripheral economies. These distortions result in long-term economic stagnation.

Figure 3.3 illustrates the basic model of peripheral urbanization according to the dependency/world-system perspective. As the rural population rapidly relocates to urban areas we see a steady rise in inequality that parallels the steady rise in the amount of wealth elites/core nations control (Wallerstein 1974). For the dependency/world-system perspective this pattern of underdevelopment and distorted development is not a temporary condition as modernization/ecological theories argue, rather it is a permanent condition.

101

On a fundamental level then, the modernization/ ecology school sees urbanization as a cause of economic development which stems from the reintegration of rural labor into modern economies. The dependency/world-system school sees peripheral urbanization (i.e. distortion in the development of urban areas) as the result of core capitalist penetration into the economies of the periphery. What these two opposing dispositions entail are two very different sets of hypotheses concerning the relationship between urbanization and economic development. And while both schools of thought acknowledge that a reciprocating relationship exists (London 1987; Crenshaw 1998), each emphasizes the importance of one causal direction over the other when it comes to explaining the effect urbanization has on the distribution of labor, wealth, and technology (see

Figure 3.1).

Specifically, each school of thought predicts very different long-term patterns of development and very different consequences for rapid urbanization in developing regions of the world (more on this below). In fact, the consequences of rapid urbanization in the developing world have become a central focus in development research as a way to test the basic predictions of each school of thought (London 1987; Crenshaw 1992). Modernization/ecology emphasizes how urbanization and population concentration are temporarily conditions of transitioning economies and thus it compliments economic development.

Dependency/world-system emphasizes how a global capitalist system promotes uneven urbanization and distorted population concentration that retards

102 development. Both have competing explanations for the causes and consequences of overurbanization and urban primacy.

3.2.2 Economies of Scale and Agglomeration Effects

The complimentary concepts of economies of scale and agglomeration effects parallel the core logic of the modernization/ecological school of thought. And while these concepts are usually applied to elements of specific production systems (e.g. factory production), they can be generalized and used to explain macro social change. This is because the concepts of economies of scale and agglomeration effects describe how interactions emerge between and within populations as a population grows in size and become denser (i.e. urbanize).

In order to understand how and why economies of scale and agglomeration effects tie into modernization/ecological explanations of urbanization and economic growth it is necessary to review the idea of returns to scale. This principle was first written about by Adam Smith (1776), who in the process of visiting a pin making factory, observed that implementing a division of labor in production increased output by holding inputs constant. That is, he observed that one worker producing pins from start to finish without the aid of machines could produce only 20 pins a day, but that a band of 10 workers who divided the 18 tasks needed to complete each pin could produce 48,000 pins a day—or a 250:1 pin to worker ratio as opposed to 20:1 (World Development Report 2009, Chapter 4 “Scale Economies and

Agglomeration”). Thus, returns to scale describes the increases or decreases that

“scaling” the production process produces in terms of both the size and division of

103 labor. That is, scaled returns are dependent upon the size of the population as well as its organization.

Smith recognized that not all production processes could benefit from scaling production, and thus returns to scale were limited by two factors: The first was the thing being produced. Certain products do not benefit from increases in labor size and/or from the division of tasks. Second, market size limits increasing returns to scale because mass production requires mass consumers and mass transit/distribution. Therefore, products that are not subject to mass consumption or that incur high distributional costs relative to their market price are unlikely to benefit from returns to scale because of the limitations placed on them by market forces (i.e. externalities).

While these limits do not translate into limitations placed on whole economies, it does bring us to a second important concept of used by economic geography that does; economies of scale. Economies of scale refer to “the cost advantages that a business obtains due to expansion. They are factors that cause a producer’s average cost per unit to fall as output rises” (Wikipedia: http://en.wikipedia.org/wiki/Economies_of_scale accessed 2-23-2009). The opposite of economies of scale are diseconomies of scale, or factors that result in an increase in average cost per unit as output increases. Generally speaking economies of scale can only be realized when transportation and labor costs are low and/or held constant (thus, holding the cost of inputs constant) and when market size is sufficient or expanding (i.e. increases of production require minimal levels of

104 consumption or increases in consumption to achieve positive returns to scale). Like returns to scale economies of scale rely on the size and organization of populations.

Finally, this brings us to the last concept which compliments modernization/ecological understanding of the relationship between economic growth and urbanization; agglomeration effects. In order to understand agglomeration effects it is necessary to add one last nuance contained in economic theory—i.e., the difference between internal economies of scale and external economies of scale (i.e. agglomeration effects). Internal economies of scale refer to reductions in the cost of production that are associated with adaptations that emerge from within production systems. Examples of internal economies of scale include new technologies, reduction in management size, acquiring better interest rates due to a firm’s size, buying raw materials in bulk, and generating a positive corporate image. Internal economies of scale can be thought of as adaptations which bring balance between the system and its environment—in this case the production system and its market.

External economies of scale, or more frequently referred to as agglomeration effects, are reductions in the cost of production that are associated with external economic conditions/environments that produce a positive impact on production efficiency. In there are generally three kinds of agglomeration effects contained in urban markets which include; localization, urbanization, and pure agglomeration (examples used below are adapted from: World Development

Report 2009, p.128, Table 4.1).

105

Localization refers to the benefits producers and sellers gain from decreasing distance between each other. Although natural geography certainly plays a role in this, other examples of localization include the ability of producers to outsource production of certain items (thus benefiting from specialization in the localized division of labor), the reduced costs for consumers who travel less distance between suppliers/retailers (e.g. shopping malls, markets), the attraction of specialized labor into a central location due to an established presence (e.g. Silicon Valley), and finally, the benefits from lower costs associated with repetition in environment (e.g. planned ).

Urbanization refers to the benefits producers and sellers gain from reductions in communication and information exchanges. Again, natural geography does play a role but other examples include: (1) the knowledge gained from the diversity of local producers and sellers (i.e. knowledge of services available allows producers and sellers to take advantage of them); (2) information “spill-overs” in which producers and sellers willingly and unwillingly exchange ideas and innovations in production and selling; (3) increasing opportunities that result from high levels of specialization (i.e. niche space opportunity created via competition); and finally, (3) the interaction of all these adds to the inertia urban markets have for attracting more producers, sellers, and consumers (environmental “pull” effects).

Finally, and perhaps most important to the macro theoretical perspectives of modernization and human ecological theory, is the concept of pure agglomeration.

Pure agglomeration refers to the notion that producers, sellers, and consumers benefit from concentration because they share in the fixed costs associated with the

106 building, maintaining, and the expansion of modern infrastructure. Thus, no one entity has the burden of investing in costly projects but rather all who enter share in the costs. As we will see in the empirical literature reviewed later in this chapter, pure agglomeration effects are perhaps the most important for explaining the positive and negative effects population concentration has on economic development (Henderson 2004).

3.2.3 The Urban Bias Hypothesis

While the concepts of economies of scale and agglomeration effects parallel the modernization/ecological perspective, the urban bias hypothesis is similar in that it parallels dependency/world-system explanations for uneven development in peripheral nations.

The theory of urban bias is derived from Hoselitz (1955) who inquired as to whether or not cities had a positive (generative) or negative (parasitic) effect on economic development among developing regions of the world. That is, was urbanization good or bad for developing nations? Although Hoselitz (1955) did not coin the term “urban bias” (Lipton [1977] did), Hoselitz alluded to it when he stated the following:

The parasitic impact of colonial capitals and other administrative centers may be regarded as having been a short run impact in the sense in which this term may be employed here. Although for a time the city tended to exert a clearly unfavorable influence on the potentialities of economic development of its hinterland, we saw that factors of change developed in and around the city which had the effects of turning the parasitic character of the city into a generative one. (1955, p.282 [italics mine])

107

Thus, while Hoselitz recognized investment in urban areas could hinder development in rural areas, he did not believe that the ill effects were permanent.

So what is urban bias? Lipton (1977), who coined the term, defines the urban bias hypothesis as a symptom of uneven and “unfair” development where

“most resources in most poor countries are systematically allocated [in favor of urban areas]: that ‘developing’ polities are so structured as to provide rural people with inefficiently and unfairly few resources” (p.486). Put simply, urban bias

“creates a disparity between country and city with respect to consumption, wage, and productivity levels” (Bradshaw 1987, p.225).

An important caveat of urban bias is the notion that its origins are intranational as opposed to international as dependency and world-system theories suggest. Despite being at odds with the dependency/world-system perspective on this point, the urban bias thesis has been shown to compliment dependency/world- system findings more than challenge them (e.g. Bradshaw 1985). Specifically, it has been found that both international dependence and intranational urban bias act together and contribute to the underdevelopment of peripheral nations by retarding economic growth (London and Smith 1988). More importantly, both have been linked to overurbanization and urban primacy which are tied to underdevelopment in the developing world (Kentor 1981; Lipton 1977; Bradshaw 1985; Timberlake

1987)

According to Lipton (1977), the mechanism for long-term stagnation associated with urban bias stems from the unfair policies developing governments

108 implement which heavily favor urban areas. Like dependency/world-system theory, the urban bias thesis asserts that a nation’s true economic potential is realized when there is balanced development between agricultural and industrial/manufacturing sectors. Unlike dependency/world-system theory the urban bias thesis argues that overurbanization and primacy stem from internal interests as opposed to external systems (London and Smith 1987). For instance,

Lipton argued that government policies unfairly taxed local agricultural production, neglected infrastructural improvements in rural areas, limited welfare and public services to the rural poor, and forced potentially successful innovators living in rural areas to relocate to the city. All of these factors are outcomes of local government policies and have contributed to urban pulls and rural pushes that promote the conditions for rapid rural to urban migration in developing nations

(Lipton 1977, Chapter 2). Therefore, in neglecting rural development peripheral governments ensure long-term economic stagnation since their policies promote the underutilization of rural labor and resources which could be used to spur long- term economic growth.

As discussed in the previous section, the dependency/world-system perspective shares complimentary assumptions about the causes of rapid urbanization, the source of distorted urban development (i.e. elite interests), and the negative impact urban/rural disparities have on economic development. As a result, the urban bias hypothesis has become a very useful intranational explanation of peripheral urbanization as it relates to the overurbanization hypothesis and urban primacy. Moreover, measures of urban bias compliment the goals of the

109 dependency/world-system perspectives since the concept gives them another way to observe how global capitalism distorts economic systems in peripheral areas (e.g.

Bradshaw 1987; London and Smith 1988).

No matter what measure is used, the concept of urban bias can be used to compliment dependency/world-system theory since it seeks to capture the distortions of urban development that are associated with the means of production in peripheral regions. Therefore, despite being at odds with the basic premise of the dependency/ world-system perspective on whether or not peripheral urbanization has internal as opposed to external origins (a feature urban bias shares with modernization theory—Kasarda and Crenshaw [1991]), the urban bias thesis appears to enhance dependency findings by providing a much needed internal mechanism for underdevelopment (London and Smith 1988).

3.3 Overurbanization and Economic Development

Even though there is little disagreement about the inseparable relationship between urban areas and economic systems (Hoselitz 1955), there is a wide range of disagreement on how and why urbanization affects economic growth differently in various regions of the world. One persistent point of disagreement between the competing schools has centered on the concept and explanation of “over- urbanization.”

One of the first to address this phenomenon was Davis and Golden (1954) whose seminal essay set out to compare levels of urbanization and economic development between all the nations of the world. Although their sample of nations

110 was not exhaustive, what they observed were two important correlations that set the tone for future inquiries: The first was a positive correlation between urbanization and industrialization. The more urbanized a nation was the higher the proportion of the non-agricultural labor force was engaged in industrial labor. The second was the negative correlation between the percent urban and rural density.

They found that the higher the percent urban in a nation there was the lower the rural density. Because these correlations held for their sample, they concluded that as agricultural technologies improved rural populations were pushed from the countryside to the city.21

Davis and Golden argued that there were three reasons not to overstate the impact overurbanization had on economic development. The first was that over- urbanization had to have upper limits since urban areas were tied to modern economic systems. Thus, Davis and Golden (1954) believed that overurbanization, when and where it occurred, was a self-correcting process (i.e. rural-urban migration was not indefinite and thus overurbanization, like urbanization, had

“real” limits). Second, they argued that overurbanization could actually stimulate economic development in the long-run by concentrating and holding surplus labor.

Finally, Davis and Golden believed that overurbanization was favorable to positive political change (1954, p.19-20).

Since its conception overurbanization has been used to assess the urban transition’s effect on economic systems in nations around the world. However, this

21 Sovani (1964) shows that these correlations only hold for developing regions of the world, and therefore, suggests that the rural “push” explanation offered by Davis and Golden (1954) is questionable.

111 approach is not without criticism. One reoccurring critique has been that overurbanization is based on Western experiences with urbanization and the association between urbanization and industrial labor is higher among developing nations than developed (Sovani 1964; Kamerschen 1969 ; Linn 1982). Therefore, it is just as likely that Western nations are under-urbanized for their level of development. A second critique concerns the statistics used to establish the concept. Critics point out that overurbanization is a process involving change over time, yet the evidence used to support it was based on static observations (Sovani

1964; Kamerschen 1969).

Despite the above critiques, overurbanization has found favor among dependency and world-system theorists who point to overurbanization as proof that core-periphery relations create dependent economic systems that distort the labor sector composition in developing regions of the world as well as urban development. This leads to long-term stagnated economic growth and a disarticulated rural/urban economy (urban bias) which promotes underdevelopment in peripheral nations (Kentor 1981; Timberlake and Kentor

1983; Timberlake 1987). In many ways overurbanization has become the primary means for dependency/world-system theorists to determine how far core nations have penetrated the economies of peripheral nations (Timberlake 1985).

Timberlake (1985) discusses three ways dependency and world-system proponents use overurbanization to proxy for imbalances in development brought about by the global spread of capitalism. The argument is shaped by dependency theory which suggests the core-periphery division of labor is setup to keep

112 peripheral nations from developing a balanced economy. Lipton’s urban bias thesis supports this notion by adding intranational causation (London 1987; London and

Smith 1988).

Perhaps the most important form of distortion is the economic disarticulation caused by foreign investment. Dependency theorists argue that foreign investment distorts peripheral economies in two ways: First, foreign investments in peripheral urban areas are generally used to build manufacturing plants aimed at producing luxury and export goods. Dependency theorists have argued this type of outward looking production does not stimulate growth in the local industrial sector, but instead provides short-term growth. Second, foreign investment in rural areas destroys agricultural labor and rural land holdings among locals. Commercial farming displaces rural inhabitants by turning farmers into wage labors who no longer control or own their land (e.g. they pick fruit for Dole instead of growing the fruit) (Timberlake 1985, p.42-43).

Aside from the economic factors dependency theorists have also argued political elites play a role as well which has provided some overlap between dependency and urban bias theories (Gugler 1982). Specifically, in trying to attract foreign investors, political elites residing in urban areas use what little capital and resources they have to build urban infrastructure (Evans 1979). Such lopsided investments in urban areas create the illusion that there are economic opportunities in the city, and in many cases there is only one large city. The result is that both push and pull factors tied to foreign investment create the conditions which promote rapid rural to urban migration resulting in overurbanization and urban

113 primacy. Moreover, as both foreign investors and core elites continue to invest in one or two urban areas, urban growth becomes unbalanced (urban primacy) and the disparity between rural and urban areas widens even more over time (Smith

1985).

3.4 Urban Primacy, Population Concentration, and Economic Development

Jefferson (1939) was the first to discuss urban primacy as a feature of urbanization and economic development when he devised “the Law of the Capitals”

(p.227). His initial interpretation was that primate cities were a source of national pride because they exerted economic, political, and cultural dominance over entire regions; they were a positive feature of national development, a sign of economic and political maturity. Of course, Jefferson’s focus was on the world cities such as

London, New , and , all major cities in the developed world.

Since then the role of urban primacy in economic development has been questioned, especially with respect to the economic well-being of developing nations. That is, urban primacy is not seen as a positive development in undeveloped nation’s who lack the urban system needed to support primate cities.

General concern about urban primacy can be traced to two sources: First, Zipf

(1941, 1949) observed that the rank-size distribution of cities within nations tended to follow a log-linear pattern that was determined by the size of the largest city in the system (see Chapter 6 for a discussion of this measure). Since then several scholars have come to believe that log-normality is a sign of economic maturity and efficiency. Second, Hoselitz’s (1955) inquired as to whether or not primate cities

114 were “generative” or “parasitic” for economic and cultural development in developing regions. Together, these two inquiries have sparked a debate as to whether or not urban primacy was, as Jefferson (1939) first portrayed it, a source of political and economic national pride.

At its conception, Jefferson (1939) defined urban primacy as a condition in which a nation’s largest city was at least twice as large as the next city (p.227). True to this definition, early studies measured urban primacy as the ratio of the largest city to that of the second largest city. This simple version of the index became known as primacy 1-2 (Metha 1964; Linsky 1965). Later studies broadened the measure to include a greater sample of the urban hierarchy as rank-size became more and more linked to how primacy was understood (e.g. Metha [1964] used the ratio of the first city to that of the sum of the next four largest cities, or primacy 1-4).

Others, such as El-Shakhs (1972), developed a measure of primacy that uses information about rank-size order to determine the level of primacy in a system (for details of his measure see Chapter 6). Yet others have used a simpler urban primacy index, one that uses the ratio of the largest urban area to that of the entire urban population (Wheaton and Shishido 1981, p. 19). While problems with measures of urban primacy exist (e.g. definitions of “urban,” determining spatial boundaries etc., see Chapter 6 for a discussion) the index has been useful for linking urban system hierarchies and population concentration to economic growth in important ways.

As mentioned above, Zipf’s (1941) initial finding that urban systems tended toward a log-normality has resulted in the view that all urban systems should follow this rank-size pattern. In fact, many have argued that the relationship between

115 urbanization and economic development can be studied as a parallel process in which rank-size distributions shift from a primate urban population structure to a log-normal (or multicentric) distribution as economic systems mature (e.g. Berry

1961; Friedmann 1973; El-Shakhs 1972). Given this view the debate on the link between urban population structure and economic development has focused on two explanations for why developing nations do not have “balanced” urban systems:

The first view is argued by the modernization/ecological perspective, or what El-

Shakhs called “classic equilibrium theory” (1972, p.14). This perspective suggests that size imbalance in urban population structure is a sign of transitional economic development from traditional production systems to modern economies. Therefore, the imbalance observed in developing nations is temporary and not permanent as the transition is achieved—imbalance can be overcome. The second view is argued by the dependency/world-systems perspective. This perspective argues that urban primacy is a sign of permanent disarticulation caused by the penetration and control of peripheral economies by core nations.

From a modernization/ecological perspective the concentration of urban population into one urban area at the start of modernization is seen as a necessary and temporary part of economic evolution (El-Shakhs 1972; Kasarda and Crenshaw

1991). This perspective argues that primate cities allow poor nations to develop the basic modern economic, social, and political institutions needed to sustain long- term economic growth and build agglomeration economies. Therefore, urban primacy is something nations grow out of as decentralization or people and wealth is achieved at higher levels of development. In short, urban primacy/concentration

116 has an inverted curvilinear effect on economic growth due to the returns to scale primacy offers at different levels of development. At very low and very high levels of concentration, urban primacy retards economic growth. At low levels of development, the lack of a single economic center keeps communication and transportation costs high since economic activities remain dispersed and disintegrated. On the flip side, at higher levels of development too much concentration results in diseconomies of scale (e.g. problems of congestion, pollution). Thus, nations at intermediated stages of development and that have a primate system will benefit the most from urban concentration. This is because at intermediate levels of development urban population concentration allows developing nations to benefit from the centralization of labor, resources, infrastructure, and capital that optimize the returns to scale for burgeoning agglomeration economies (El-Shakhs 1972; DeCola 1983; Davis and Henderson

2003—see Figure 3.4 below).

Dependency/world-system theorists have interpreted urban primacy quite differently. They argue that urban primacy is a sign of capitalist penetration into peripheral economies which promotes permanent underdevelopment in the absence of a revolution (El-Shakhs 1972; Kasarda and Crenshaw 1991). This is so because primacy keeps peripheral nations from developing balanced urban population structures (i.e. follow the rank-size rule) which is necessary to sustain a balanced or articulated economic system both within the nation and within the global economy. Thus, urban primacy imposes a long-term drag on economic development in peripheral nations via imbalance in both economic production and

117

HIGH

Economic Efficiency Efficiency Economic

LOW

LOW Level of Primacy/Concentration HIGH

Dependency/World-System

Modernization/Ecology

Figure 3.4: Predicted Relationship between Economic Returns on Urban Primacy and Level of Urban Primacy

population distribution. More importantly, primate cities doom economic growth in developing nations, in that, they not only stop peripheral nations from developing a balanced hierarchy of supporting cities but they also act as economic outposts where core nations extract resources and exploit labor (Smith 1985; Walters 1985).

Similarly, urban bias theory argues primate cities create local rural/urban imbalances which becomes a structural constraint on the development of internal economic systems. In short, urban bias theory has argued that primate cities promote continued disarticulation between rural and urban economies through the

118 centralization of political power and the hoarding of infrastructural investments.

Thus primacy promotes the underutilization of rural labor and resources (Lipton

1977). From these perspectives then, primacy is modeled as having parasitic effects on economic growth. It is not that primacy stops growth entirely, but that the overall costs associated with disarticulation are such that it “eats up” any economic gains scale economies produce. Therefore, the relationship between urban primacy and economic development is a curvilinear linear function with the costs of primacy modeled exponentially (see Figure 3.4 above).

3.4.1 Empirics of Urban Primacy and Economic Development

Over the years there have been several empirical studies looking at the relationship between urban primacy/urban population concentration and economic development. Most mirror the perspectives outlined above. There are those that show primacy and concentration are bad for economic development and those that view concentration as necessary step in the development process that has different effects on economic performance at different stages of development.

Although early studies on the effect of primacy/urban concentration were limited in their empirical support (e.g. Metha 1964) those that followed quickly offered empirical evidence. Perhaps the first was El-Shakhs (1972) who looked closely at both the effects of urban population structure and concentration as dual components of urban primacy. He concluded, with what empirical evidence he had, that there was curvilinear relationship between level of development and the distribution/concentration of urban populations. When looking at the association between urban primacy and level of development, poor nation and rich nations

119 were the least likely to have a . Those at moderate levels of development had the most primate urban systems. Thus, he concluded that primacy was related to the development process and reached its peak when nations were in the transitional stage lending support to the modernization perspective.

Mera (1973), in a related study, examined the effects population concentration (i.e. primacy 1-3) had on economic development. Specifically, he tested the hypothesis that overly large urban areas resulted in diseconomies of scale

(i.e. were inefficient due to their size). This prevailing view was based on the assumptions that in very large urban areas social overhead costs (public services) were more expensive to administer and productivity levels of workers were lower than in medium-sized urban areas. In short, urban agglomerations, if too large, were thought to be an economic burden to gross national product rather than an asset. What Mera found was that growth in the degree of primacy was associated with increases in GDP per capita (1973, p.321-323). He also found evidence that growth in secondary cities (i.e. those cities smaller than the largest) were associated with growth in GDP per capita as well. Therefore, he argued that there was reason to doubt the conventional wisdom that overly large cities are necessarily “parasitic,” and further, that medium-sized cities are the optimal size for economic efficiency— large cities appear to be the most efficient for growing GDP per capita.

Wheaton and Shishido (1981) offer a very similar, but much more refined and statistically detailed, test of the diseconomies of scale hypothesis examined by

Mera (1973). Using economic theory of market areas (see Section 3.2.2 for a discussion and definition of concepts) and a simple measure of population

120 concentration (the ‘H’ index, discussed in Chapter 6) Wheaton and Shishido found strong evidence to support El-Shakhs (1972) and Mera (1973). In other words, population concentration appears to be associated with the level of development, and nations at intermediate levels of development appear to benefit the most from population concentration and agglomeration economies. It is at this point that labor intensive and capital intensive methods interact and achieve the highest returns to scale.

Junius (1999) further explored the notion that primacy was associated with levels of economic development (e.g. El-Shakhs 1972; Mera 1973). While he confirmed that there was a curvilinear relationship between primacy and economic development, he also found that the relationship was conditioned by historic, geographic, and political variables.

Henderson (2002), echoing the logic of Lipton’s (1977) urban bias thesis, looked at the impact primate cities have on the economic well-being of developing nations. He offers a specific focus on the economic impact primate cities have on other non-primate cities. As he states it, “I try to disentangle the internal costs of large city sizes from the costs that large city sizes impose on the rest of the urban system” (p.96). He finds evidence that primate cities are parasitic to other urban areas in the nation (i.e. non-primate cities) because governments funnel resources to primate cities in an effort to prop them up. In other words, there is a primacy bias when it comes to economic development.

In a later study, Henderson (2003) expands upon one of his earlier findings

(2002) that there is a best-degree of primacy which is contingent on a nation’s level

121 of development. In this article he attempts to expand on this by asking the “so what” question, meaning how great are the economic losses from being either over concentrated or under concentrated (2003, p.47). While he does not directly answer this question, he nonetheless confirms that the question is not simply whether or not urbanization affects productivity but rather if urban concentration has a direct impact on productivity. To this end he finds evidence that country size

(both land area and population) and level of development influence what the best degree of primacy is in terms of increasing or decreasing output per worker. The positive effects of primacy on productivity were found to decline with the size of a nation in terms of its area. They rose with the size of the urban population, although this effect was less significant than the land area effect. Finally, Henderson (20030 found that there are costs associated with a lack of concentration. Too little concentration was associated with lower output per worker.

Finally, and most recently, Bertinelli and Stobl (2007) offer the most compelling and complete statistical evidence that urban primacy (i.e. population concentration) has a curvilinear effect on economic development. This relationship was found to be the most pronounced among developing nations. In fact, their study suggests that urban concentration contributes to economic growth but only up to a certain level of development. After a certain point a threshold is reached and the effects of concentration become a non-factor.

122

3.5 Conclusion

It is clear from both theory and empirical evidence that there is a strong and undeniable relationship between urbanization, urban population structure, urban population concentration, and economic development. This relationship is grounded in the shift from agricultural (traditional) to non-agricultural (modern) production systems. Technology and geography appear to be two of the central determinants for how fast this transition unfolds, which directly affect how and when urbanization arises in various regions around the world as well as what forms urban systems take. In terms of research, the focus has been on how certain patterns of urbanization affect economic efficiencies.

Research on overurbanization tends to supports the fundamental propositions of dependency/world-system theory, suggesting that foreign intervention from core regions distorts development in the periphery. Still, empirical evidence supporting the negative effects of overurbanization on economic development is largely mixed. Some tend to confirm overurbanization is associated with peripheral urbanization (Kentor 1981), while critics point out that the relationship between industrialization and urbanization is highest among developing nations (Sovani 1964). Nonetheless, a key variable in the dependency argument (foreign investment) does appear to impact sectoral compositions, the location of wealth, infrastructure, and opportunities, as well as levels of population concentration (Timberlake and Kentor 1983; London 1987; London and Smith

1988). Moreover, historical events such as colonialism appear to impact rural adversity and play a role in how urbanization unfold in developing regions of the

123 world (Timberlake 1987). Yet, despite all these linkages the detrimental impact that rapid urbanization and overurbanization have on the long-term potential of peripheral economies appears to be overstated (Davis and Golden 1954; Crenshaw and Oakey 1998).

Conversely, evidence surrounding the relationship between urban primacy/concentration has meet less resistance; although there are some who still tie population imbalances to overurbanization and/or argue that urban primacy/concentration has a negative long-term impact on economic growth (Smith

1985; Walters 1985; Kasarda and Crenshaw 1991). Yet the overwhelming evidence suggests that the economies of scale and agglomeration effects derived from high levels of urban concentration in the urban transition have net positive effects on economic development (Metha 1964; El-Shakhs 1972; Metha 1973; Wheaton and

Shishido 1981; Junius 1999; Henderson 2003, 2003; Bertinelli and Stobl 2007).

Moreover, these same studies indicate that primacy is tied to more than just economic development and growth. Urban primacy is tied to geography, history, and most importantly politics. But before exploring how urban primacy/concentration is linked to political systems, it necessary to examine the other important correlation which has received a great deal of attention in development studies: The relationship between level of economic development and democratization.

124

CHAPTER 4

ECONOMIC DEVELOPMENT AND DEMOCRATIZATION

In the last chapter I explored the reciprocal relationship between urbanization and economic development. This was the first of two relationships that I argued were central to how cross-national comparative researchers explored the total development process. Looking to Figure 4.1 below, in this chapter I focus on the democratization debate—i.e., the debate as to whether or not economic development leads to democratic systems of governance and whether or not democratic systems hinder or promote positive economic growth.22 Because the theoretical approaches and concepts that were used in the last chapter are very similar to this literature, I place less of an emphasis on theory. Instead, I focus on

22 As Rao states, “In addition, economic development should be distinguished from economic growth. The latter deals merely with changes that take place in economic sectors and is expressed in terms of changes in the value of economic variables. The concept of economic growth per se does not refer to changes in sociological and political factors, for it assumes that they are stable. The concept further assumes that there is an in-built mechanism which promotes changes in economic factors via the structural relationships already in place. Economic development is a deliberate process which aims at creating the structural relationships that will facilitate the type of growth desired. It presupposes that the structural relationships currently existing do not meet the needs of the majority of the people, that an array of policies dealing with all aspects of human activity must be adopted to achieve the desired goal. A revolutionary approach is needed to foster economic development. This ‘may mean no more than the removal of obstructions opposed to the will of the people by outworn institutions controlled by groups interested in their preservation’ (Schumpeter,1962:236). The policies for such a course could be adopted by a representative group that has social (as opposed to personal) aims and the authority to implement the policies so adopted. Indeed, such a group could be the legitimate government of the country. This definition of economic development underscores the point that a normal cannot promote economic development. Government has the responsibility not only to intervene in the market, but also to give it a direction—if it is desired that the course of economic development be traversed” (1985, p.68). 125

Urbanization

Democratization

Economic Growth

Stronger Empirical Support Weaker Empirical Support

Figure 4.1: Relationship between Economic Growth and Democratization

the empirical findings supporting each direction of the democracy/development relationship.

Exploring the relationship between level of development (and also economic growth) and democracy has a long history in sociology, economics, and political science. Because the relationship has been studied form such diverse perspectives conclusions about its strength, direction, and meaning have been mixed. Some conclude that development, in terms of greater economic affluence and a modern division of labor, is positively related to democratization arguing either a linear or non-linear relationship (more on this below). Others suggest that a relationship does not exist and point to unique historic and/or cultural circumstances as the primary determinants of political democracy and economic development (e.g.

Weber). Moreover, early on most scholars were unclear as to whether the relationship was reciprocal—i.e., if democracy caused economic development or if

126 development lead to democracy. Over the years most of the research has concluded that polity has very little influence on economic growth (e.g. Burkhart and Lewis-

Beck 1994; Przeworski 2000) and that economic development supports stable (e.g. Lipset 1959; Bollen 1993; Burkhart and Lewis-Beck 1994;

Przeworski 2000). However, there is still theoretical and empirical support linking democracy to economic growth (mostly through indirect relationships) despite the weak support this portion of the relationship has received (see Sirowy and Inkles

1990; Olsen 1993; Helliwell 1994; Leblang 1996; and Feng 1997; Heo and Tan

2001). Whatever the case may be there is no lack of debate within this line of research as to what the specifics relationship between democracy, development, and economic growth is.

In the paragraphs to follow, I carefully review the empirical research relating economic development to democracy as well as research relating democracy to economic growth. I look at the empirical literature for both relationships even though there is more support for the idea that economic development results in democratization (perhaps because it has been the most studied). Because the causal direction of the association was left unclear, several empirical studies that cannot be grouped into either side of the debate. Therefore, I include these early unspecified studies with those who argue democracydevelopment since it is from this set of empirical findings that have been used to support an association between level of development and political democracy. With this in mind let us turn to the first part of the reciprocal relationship (i.e., level of developmentpolitical democracy). This view is best articulated by modernization theory.

127

4.1 Economic Development as a Cause of Political Democracy

Most of the early studies on democratization were grounded in modernization theory or what has come to be known as political modernization theory. This perspective views the relationship between democracies and dictatorships as part of the total development process in which the modernization of the economic system parallels changes in social systems that both promote and retain political democracy (e.g. de Schweinitz 1959, Lipset 1959; Cutright 1963; see

Przeworski and Limongi 1997 for a good review of this perspective).

Political modernization theory argues that like other parts of the transition process, political systems transition alongside other modern developments. That is, as societies progress from traditional agrarian economies to non-agrarian economies the political structures that regulate a population’s activities and resource distribution go through developmental stages. Specifically, political systems that are the most efficient and supportive of economic development at a given stage usually flourish until that stage is complete. In short, the progressive stages associate with economic development are: Traditional (Tribe/Hereditary

Rule)DictatorPolitical Democracy.

As Spencer discussed, in early stages of development power resides in the tribe and or clan, and at later stages when populations begin to grow in size hereditary rulers (e.g. kings) replace the tribe. This begins the rise of inequality. At transitional stages of development autocrats emerge as the strongest and most efficient leaders for economic growth. Becasuse dictators can use their central authority to mobilize resources, labor, and infrastructure by any means necessary.

128

This often includes limiting/ignoring worker’s rights (especially women’s rights) and streamlining important economic decisions without regard for fairness or justice. Although amoral, the ability for a single powerful leader to direct growth allows poor nations to experience high levels of economic growth. Therefore, during transitional growth inequality is the highest since both authority and wealth are centralized.

However, while dictatorial rule increases economic efficiency during the transitional growth stage, political modernization theorists argue that dictators sow the seeds of their own undoing. As discussed at length in the previous chapter, modernization theory argues economic growth eventually leads to and requires the decentralization of wealth and power ( i.e., growth leads to complexity which requires increased specialization and interdependence and thus the decentralization of information, resources, and power). As decentralization unfolds, key institutions and economic environments that are more conducive to political democracy begin to emerge (e.g., better schools, more/expanded public services, increases in personal wealth, greater control over property/production, social welfare, and the decentralization in decision making/authority). Therefore, at later stages of development economic maturity leads to a situation where populations out-grow the need for a dictator to centralize economic systems in order to promote economic growth. What results is the adoption of political democracy which is better suited to the economic and cultural pluralism of modern social and economic systems.

129

From the perspective of political modernization theorists dictatorships have been understood to be the “dirty” path to democracy, especially for poor developing nations who are experiencing transitional growth. The argument is that after WWII and decolonialization there has been increased international pressure for late developing nations to “join” the developed world. Therefore, many developing nations, and especially those in transition or at intermediate stages of development, have adopted autocratic systems of governance in order to achieve the high levels of economic growth needed to become developed (Przeworski and Limongi 1997;

Przeworski 2000). However, because late developers have not had the luxury of gradual and sustained growth, many of these nations face barriers that keep them from developing the economic complexity/power needed to sustain political democracy.

4.1.1 Development and Democracy Studies

Lipset (1959) was one of the first to gather data supporting the relationship between economic development and democracy. Although he did not attempt to establish causality when relating development to democracy, he accomplished three important tasks that would direct the inquiries of other scholars who followed in his footsteps: First, he broke from traditional models of Weberian thought by suggesting democracy was not just attained or sustained by ideational forces or historic accidents alone (e.g. Protestantism).23 Second, he was the first to empirically relate level of development to democratic regimes by suggesting that

23 It should be mentioned that Lipset did not deny these determinants and explores ideational causes in relation to material conditions (see pp. 86-98 in his 1959 APSR article).

130 economic development promoted institutional changes conducive to development and help sustained development once it was adopted.24 Third, he identified and collected data on sets of cross-national indicators that have become important proxies for measuring economic development ever since—a measure that has come to dominate definitions of level of development in democratization research. This last accomplishment will be discussed further in a later section since many inquiries following Lipset have used similar indicators. It will be argued that the choice of indicators for development has impacted how the relationship between development and democracy is viewed and measured to this day.

As stated above, Lipset’s (1959) theoretical argument was one of the first expressions of political modernization theory. Although he warned that no single condition or combination of conditions of the development process resulted in democracy, he did point to two basic and interrelated social structural changes that occurred in the course of the transition from traditional to modern societies which promoted political democracy. The first was a change in class structure, and the second was a change in educational attainment (literacy) among the population.

Both were tied to the modernization of economic systems which Lipset viewed as an

“economic development complex” because it was comprised of interrelated social structural systems and cultural changes that contributed to democratization (i.e. industrialization, wealth, urbanization, and education) (1959, p.71).

24 It is important to note that Lipset’s methods and data have been called into question by many (e.g. Cutright 1963, Neubauer 1967, and Jackman 1973), but his efforts to test the relationship empirically and suggest and endogenous and exogenous explanation for it is what is of importance. 131

Lipset argued that economic development was positively related to increases in the proportion of the population that was educated because the modern means of production (i.e. industrial production) required a more educated and literate labor force. In turn, education and literacy were positively related to increase in tolerance for opposing political/social norms, a key cultural aspect of pro-democratic societies. Such tolerance bolstered support for pro-democratic movements and ideas. Economic development also expanded the middle-class (i.e. produced a diamond shaped income distribution) which reduced decentralized wealth and reduced class conflict. The decentralization of wealth also increased support for moderate political ideologies as well as foster support for tolerance of oppositional political groups. Both increased education and the expanding middle-class reduced a population’s preference for anti-democratic movements (i.e. authoritarian rule).

Meanwhile greater cultural and economic interdependence resulted in a multitude of cross-cutting affiliations associated with modernization. In short, Lipset argued that the economic development complex was conducive to democratization because of its effect on other interdependent social institutions that were central to the total modernization process. Moreover, he identified urbanization as a key mechanism for this forward progression since key advancements in the economic development complex were not possible without cities. Empirically, Lipset observed a positive linear relationship between level of economic development and democracy using simple multiple correlations.

In a follow-up study, Cutright (1963) critiqued Lipset’s findings for his operational definition of democracy. He argued that Lipset’s treatment of political

132 systems as a binomial measure (i.e. either democratic or not democratic) failed to account for degree of democracy that existed in the real world. In an effort to ratify this problem and re-examine the development/democracy link, Cutright constructed an index of political development that would allow political systems to be observed as a matter of degree. His aim was to construct an index with which the

“[d]egree of political development [could] be measured”, so that “each nation

[could] be placed on a continuum of development, which [would] allow it to be compared with any other nation in the world” (1963, p.255). Cutright compiled a regression model using development indicators (the list was very similar to

Lipset’s—see Table 4.1 below) and a revised political development index. His main finding was that developments in communication technology showed the strongest correlation to his political index. Theoretically, he concluded that the diffusion of communication technology played a central role in the democratization process by allowing for a more complex political system to emerge (i.e. Spencerian social evolution) leading to more efficient coordination between institutions; what he believed to be a pre-requisite for democratization. In short, Cutright argued that political complexity was a necessary prerequisite to sustain democracy since political power in democracies was more evenly dispersed. Although his model suffered from many statistical problems (see Jackman 1973), like Lipset (1959),

Cutright’s conclusion was that a positive linear relationship between development and democracy existed.

The first to challenge the argument that a one-to-one relationship between development and democracy existed was Neubauer (1967). While Neubauer

133 acknowledged the importance of developments in communication (i.e. a mass media) for democracy, he faulted Cutright (1963) for modeling the relationship different from what he proposed was needed to properly test the relationship. That is, instead of testing a linear relationship between economic development and political development, Neubauer argued that Cutright’s index measured democratic development and not political complexity. Therefore, he asserted that Cutright’s index was biased towards democracies (see Neubauer footnote 5 p. 1003 for an explanation). In short, Cutright used democracy as a proxy for political complexity which excluded communist nations with complex political systems (p. 1004). These real-world exceptions gave little reason to believe that political complexity was solely associated with democratization since a society could have a complex political system without being democratic. Further, Neubauer was critical of both Lipset

(1959) and Cutright (1963) for modeling the relationship as being linear. He suggested that if a relationship existed it was more likely to be curvilinear with economic development having little or no effect on democratization once a certain level of economic development was obtained (Neubauer refers to this as a

“threshold” for the relationship—see Figure 4.2 for a comparison).

To challenge Cutright (1963) and Lipset (1959), Neubauer (1967) devised a refined political index. Based on his refined index and regression model Neubauer concluded that “there [was] simply no relationship between level of democratic performance and measures of socio-economic development” (1967, p. 1007).

However, after some analytical consideration Neubauer did not completely deny that a relationship between democracy and development existed on some level. He

134

High

THRESHOLD

Degreeof Political Democracy

Low Level of Economic Development High

Figure 4.2: Linear versus Curvilinear Relationship between Development and

Democracy

simply believed that the relationship was more complex and more context-driven than either Lipset (1959) or Cutright (1967) had acknowledged. He stated:

… variations in the performance of democratic countries appears to be a function of factors such as pluralism and cleavage which, though obviously related to the gross level of socio-economic development, have an effect on performance which extends far beyond this development (1967, p.1008).

Following Neubauer (1967), Jackman (1973) revisited the issue of democracy and development by suggesting that Neubauer was mostly correct in his critique of Lipset (1959) and Cutright’s (1963) linear models.

However, he asserted that Neubauer’s conclusions were premature and thus

135 favored a refined test of Neubauer’s threshold model.25 Specifically, Jackman argued Neubauer’s sample was too homogenous which limited his ability to test a curvilinear relationship across varying levels of development.

Therefore, Jackman constructed a model using improved statistical techniques and a more heterogeneous sample to capture greater variability contained in the relationship. He found modest support for a curvilinear relation; however, he was hesitant to suggest anything conclusively about development and democracy.

Following Jackman’s assessment, Bollen (1979) re-stated the debate in wake of the mounting contradictory evidence and inconclusive findings. All previous studies had found some relationship between development and democracy, but most had come across nations or sets of nations that were exceptions to the rule (i.e. they had high level of development but not democracy or low levels of development with democracy). In an effort to add greater theoretical and statistical specification to the inquiry, Bollen mimicking Weber, examined these contingencies within the confines of timing and historical/cultural contexts. The question was no longer whether or not affluence led to democracy, but rather if early adaptors of democracy faced different historical and global circumstances affecting economic development, and thus, democratization. What Bollen observed was that many of the nations which were democratic had experienced development much earlier than

25 The threshold model simply suggests a curvilinear relationship between level of development and democratic performance. Simply stated, economic development has a greater impact on developing nation’s democratic performance, but in later stages of development economic gains have little or no effect (see Jakman 1973 Figure 1 p. 614).

136 those that were not. In an effort to determine whether timing played a role in democratization Bollen tested four related hypotheses.

The first hypothesis dealt directly with timing. It stated that the earlier a nation began development the more democratic it would be. Bollen argued that many of the developing nations of the world faced greater strains in their path to development than the first developers had. He called the strains “demonstration effects” (1979, p. 573). The basic idea of demonstration effects is that developing nations feel pressure to play “catch-up” with developed nations in order to compete in the global political economy. Using this idea, Bollen argued that the global context was more important and much different for early adopters than late comers.

Thus, in modern times developing nations were placed under greater pressure to raise their consumption levels and increase their social welfare to levels similar to developed nations. These mounting pressures were too much for the political and economic systems of developing nations to bear, and therefore, forced imbalances acted to counter democratic movements in developing countries.26 Citing world- system and dependency theory, Bollen took the argument further and suggested that the problem of rapid population increases and economic dependency common among peripheral nations in the world-system may be addressed better by authoritarian governments than democratic ones (1979, p. 574). This argument is similar to the logic of de Schweinitz (1959) who saw democracy as a barrier to economic development for developing nations. Thus, late developers should be

26 Bollen’s logic follows Strain Theory which is derived from the modernization perspective for this hypothesis.

137 more prone to adopting autocracies to deal with the problems caused by disparities between the developed and developing world.

The second hypothesis dealt with the diffusion of the idea of democracy and the logic offered was straight forward. Bollen argued that as democracy spread around the world and became known, there would be greater pressure for developing nations to adopt a democratic government due to pressure from populations. Thus, the later the time of development, the greater the pressure there would be on a nation to adopt democracy or more democratic governing systems.

The third hypothesis dealt with the unique of early adopters compared to late adopters. As Weber and others had pointed out, similar cultural experiences and belief systems were shared by developed nation (i.e. Protestantism, free-markets, capitalism). This had historically set them apart from developing nations, especially in terms of economic systems which was dominated by instrumental instead of value rationality. Given this, Bollen hypothesized that the extent of Protestantism in a nation would be positively related to degree of democracy in a nation. In other words, Protestant nations had developed cultural characteristics that were favorable to democratic rule.

Finally, Bollen’s last hypothesis tested difference in state control over economic systems. Early developers had very few restrictions placed on them from the state, while late developers faced greater economic restrictions imposed by a state. For early developers low levels of state intervention allowed for greater economic growth, and it also allowed for the accumulation of personal wealth which help reduce inequality and promoted a decentralized economy and government.

138

Greater personal wealth translated into more disperses political power with which political elites could be directly and openly challenged. In contrast, developing nations had been forced to centralize both political and economic power into the hands of a few as a way to deal with the problems of late development (e.g. rapid population increases, the building of infrastructure, social welfare, etc.). In response to these pressures, developing nations had to centralize capital through the state in order to spur economic development. Moreover, developing nations had to concentrate political power in order to repress labor from organizing which would impede economic development. This synthesis of modernization/dependency theory led Bollen to hypothesize that the more control the state had over the economic system, the less democratic it would be. Taking all four hypotheses together, Bollen argued that the reason developing nations had not yet become democracies was the result of their comparative development in terms of timing within a new global context. Developing at latter periods placed nations at a demographic, political, and economic disadvantage that acted to restrain conditions which favored democratization.

Bollen did not find support for his first two hypotheses relating to the timing of development. New-comers were neither more democratic due to the diffusion of democratic ideology, nor were they less democratic due to global social/economic strains. Although development timing was not significant for predicting democracy in the first model, the level of development was. Finally, he did find support for the third and fourth hypotheses. State control over the economy and Protestantism were both positive and significant for predicting democratization. The greater the

139 control the state had over the economy the less democratic it would be. And further, the more Protestant a nation’s population was the more democratic it would be.

Lastly, Bollen found that when controlling for state economic control and

Protestantism, the level of development for a country remained a significant predictor of democratization.

One of the most recent, statistically advanced, and conclusive studies on the relationship between democracy and development was completed by Burkhart and

Lewis-Beck (1994). Burkhart and Lewis-Beck set out to challenge the findings of

Arat (1988) and also Gonick and Rosh (1988). Although both studies used time- series design, which they applauded, both studies concluded that a relationship between democracy and development did not exist. Burkhart and Lewis-Beck seriously doubted these findings. They also critiqued prior studies for not specifying the order of the relationship (i.e. did democracy lead to development or did development lead to democracy, or both?). Citing Bollen (1993), Burkhart and

Lewis-Beck suggested that the Banks (1979) data used by both Arat (1988) and

Gonick and Rosh (1988) suffered from significant problems with “validity component” measures used to estimate democracy. In light of these data issues they suggested that there was reason to question Bollen’s (1993) time-series results

(1994, p.904). After discussing these and other statistical problems with past empirical inquiries (mainly the use of cross-sectional data as opposed to time-series data), Burkhart and Lewis-Beck used a new data set (Gastil) with careful consideration for time/causal order and statistical/methodological techniques.

After great scrutiny of their own data and a full disclosure of their methodological

140 concerns, Burkhart and Lewis-Beck concluded with confidence that “…around the world, economic development works to foster democracy” (1994, p. 907). In their final statements they warn other scholars not to confuse this finding with the idea that democracy leads to economic development, which found no support in their models.

Finally, Przeworski and Limongi (1997) and Przerworski (2000) re- examined the development/democracy debate. They used hazard models to determine the probabilities of political regime change among developing nations between 1950 and 1990. Both studies point out and analyze two theoretical possibilities of Lipset’s initial assertion on the relationship between level of development and democracy. The first possibility is an endogenous explanation:

Lipset (1959) argued that modernizing pressures, independent of actors, “forced” the adoption of democracy (i.e. democracies emerge under authoritarian governments). The second possibility is an exogenous explanation; affluence sustains democracy. In this version Lipset (1959) argued that economic development created the institutions and attitudes needed to sustain a democracy.

Przeworski and Limongi (1997) found support for both explanations to varying degrees. The endogenous version held true only up to a certain level of economic development. After a nation obtained affluence beyond a given point ($5001-$6000 range using PPP standardized to 1985 US dollars), the probability that an authoritarian nation would transition to democracy began to lower (although the probability remained significantly higher than in low income nations). Thus, the relationship for the endogenous explanation was curvilinear, but there was stronger

141 support for the exogenous explanation (i.e. affluence sustains democracy). In other words, economic affluence allows democratic governments to maintain power.

Given these findings, Przeworski and Limongi (1997) and a more refined version in Przeworski (2000), offer a new evidence for the observed relationship: It is not development which caused democratization directly, rather, democratic movements randomly occur at all levels of development, and democracies that emerge under affluence are more likely to survive. Thus, over time the number of wealthy democracies accumulates because the probability of a democracy transitioning to a dictatorship under affluent conditions approaches zero.

Therefore, as more nations develop, the number of democracies accumulates giving the impression that democracies are caused by affluence. In short, both findings argue a kind of “natural selection” for democracy. Democracy is selected for in affluent conditions, or as Przeworski and Limongi state; “The emergence of democracy is not a by-product of economic development…Only once it is established do economic constraints play a role: the chances for the survival of democracy are greater when the country is richer” (p. 1997).

4.2 Political Democracy as a Cause of Economic Growth

Even with the multitude of studies suggesting that development precedes democratization and not vice versa, the issue as to whether or not political democracy “causes” economic growth is still debated. In studies looking at the link between regime type and economic growth the issue is not so much as to whether or not democracies promote economic growth but rather if dictatorships do it better. In other words, the basic question in this area of the democratization debate

142 is to what degree do political systems help or retard economic growth in the development process. From the political modernization perspective the belief is that dictatorships (i.e. autocracies) are better for economic growth than democracies. In this section I cover some of the more recent inquiries on this portion of the debate.

Sirowy and Inkeles (1990) offer a very useful summary of all the literature previous to their own article concerning the debate on whether or not democracies or autocracies are better for economic growth. As they state it, the question has been “do politics matter with respect to the pace and form of economic growth..?”(1990, p.126). In reviewing the literature they uncover three competing arguments which have been used to answer this question.

The first argument is derived from the political modernization perspective discussed earlier. They label it the “conflict” perspective.27 This perspective suggests that democracy is incompatible with the economic goals of late developing nations (i.e. after WWII and decolonialization) who are at intermediate levels of development. In other words, late developing nations cannot afford to have both political democracy (greater equality) and economic development (more dispersed wealth). The reason? High costs of building modern infrastructure and the strain of institutional changes associated with the transition from traditional to modern society requires too much material and cultural resources for weak democracies to bear. What is needed are high levels of economic and political inequality (albeit temporary) in order to centralize resources and generate the rapid economic

27 Not to be confused with Conflict Theory.

143 growth needed to overcome the burdens associated with transition to maturity.

Thus, political modernization theory argues that developing nations cannot afford to extend civil and political rights to labor or spread infrastructure evenly amongst its population. Instead, nations must choose one over the other: They either instill political democracy with the goal of having greater socioeconomic equality (in this case everyone stays equally poor), or they allow autocrats to centralize the nation’s wealth and power and forcefully direct rapid economic growth which temporarily creates high levels of political and economic inequality. Those arguing along these lines have suggested that adopting democracy too early will results in poor economic performance and unstable regimes, and thus the reason why democracy is associated with level of development—poor democracies fail. In short, the “conflict” perspective argues late developing nations are better off allowing autocratic rule until they build the economic wealth and social institutions needed to sustain democracy.

The second position is labeled the “compatibility” perspective. This perspective argues that the goals of democracy are compatible with the goals of economic development and socioeconomic equality. Therefore, “democracies are as capable as authoritarian regimes of combining redistribution and growth in such a way as to broaden markets and achieve economic expansion…” (Helliwell 1994.

P.235). Aside from critiquing the “conflict” perspective for overstating the ability of autocrats to foster high levels of economic growth with low levels of internal conflict, the compatibility perspective suggests that democracies promote more even development of economic sectors and promote a more even development

144 within nations as a whole. Thus, while dictators may aid in short-term economic goals of late developing nation, expansion democracies offer equally viable development strategies that are beneficial to long-term goals; equality and economic well-being.

The compatibility perspective’s logic for why dictators are bad for long-term growth comes from the argument that dictators’ political self-interests are not compatible with economic development. In order to maintain their power dictators must control who has information (i.e. limit opposition), control what is types of investments are made (e.g. welfare vs. military), where investments are made (e.g. vs. hinterlands), and determine who benefits from investments (e.g. urban elite or rural poor). Ultimately the steps dictators take in protecting their self-interest stifles innovation and restricts the productive potential of the nation’s workforce and has a negative impact on economic growth and development.

Conversely, democracies promote political and economic pluralism which encourages market competition/expansion which rewards innovation as well as established predictability in economic markets (i.e. laws are made by the people and not changed on a whim by a single leader). Therefore, the economic conditions created by democracies spur long-term economic growth and are a better development strategy since they promote equality and economic growth simultaneously. In short, the compatibility perspective argues that democracy and development work together and are mutually supportive and reinforcing.

Finally, a third perspective labeled the “skeptical” perspective argues that neither democracy nor dictatorships have an inherent economic advantage that is

145 tied to the way they govern. Political regime types matter very little. Instead, economic performance has more to do with how stable a political system is as well as with the specific characteristics of the political structure including the degree to which governments intervene in the economy, what types of economic activities governments invest in, and cultural characteristics (1990, p. 134). Political stability provides economies with the predictability needed for long-term investments to yield results. Therefore, the focus should be on the institutional structures that are in place, structures that both democracies and autocracies share.

In light of their extensive literature review and in-depth comparisons, Sirowy and Inkeles (1990) conclude that the only clear evidence concerning the relationship between democracy and development is that democracies do not facilitate rapid economic growth (p.150).

Since Sirowy and Inkeles’ (1990) extensive review several studies have attempted to come to a more concise conclusion on the relationship between political regimes and economic growth. Helliwell (1994), for instance, took a different approach for measuring the effects of democracy on economic growth.

Building on the compatibility perspective he tested whether or not democracies indirectly affected economic growth through its social policies which favor investment in human capital (i.e. education, health care, welfare, etc.). While he found evidence that nations at higher levels of development were more likely to be democratic, he did not find any evidence of an indirect effect via investment in human capital on economic growth. Put simply, democracies did not prove to have

146 a positive significant effect on economic growth as the compatibility perspective suggests.

Leblang (1996) takes a different approach for testing the compatibility perspective by suggesting that democracies can directly affect economic growth through the enforcement of property rights and indirectly through the continued long-term support of those rights. Specifically, he argues that democracies are more likely to protect individual property rights which give citizens an incentive to work harder since they can accumulate wealth and pass it on. That is, property rights give people confidence that their property will not be confiscated by the government in the future, but rather protected. His analysis shows that nations where property rights are protected tend to grow their economies more rapidly than nations that do not extend or respect property rights. This finding holds regardless of the nation’s current level of economic development. Moreover, he finds that political regimes can indirectly affect economic growth through its ongoing commitment to property rights.

In another study, Feng (1997) tests the hypothesis that democracies have a positive indirect effect on economic growth that may outweigh any negative economic effects democracies incur. He starts with the understanding that the direct effect of democracy on economic growth is ambiguous at best and negative at worst. However, he notes that regime stability is positively associated with economic growth. He also argues that democracies provide nations with longer periods of stability when compared to autocracies because transitions to and from democracy tend to be more “regular” as opposed to “irregular” (1997, p.397).

147

Irregular transitions are those met with force, and thus they instill great uncertainty in markets. Regular transitions are those achieved using peaceful transfers of power; i.e., they occur within framework of a constitution. Peaceful, regular transitions disrupt markets less because they are less disruptive but also because they are more legitimate (i.e. a controlled constitutional change as opposed to a chaotic violent one).

Feng (1997) finds a strong positive association between political stability and economic growth that holds for both democracies and autocracies. Feng then shows that because democracies provide nations with longer periods of stability due to regular transitions, democracy indirectly fosters positive economic growth through its effect on regime stability. Therefore, democracies are associated with long-term economic growth. As Feng summarizes;

Hence, democracy is likely to have a positive effect on growth (a) where it substantially reduces the probability of irregular government transfers and (b) where it increases the probability of major regular government change. Rapid economic growth requires both a stable regime and a political system which is capable of ‘adjusting’ to circumstance by changing the party in power or the ruling coalition of parties in a constitutional context. Democracy is likely to provide both of these conditions favourable to growth. (1990, p.409)

Most recently, and in an effort to resolve this conundrum empirically, Heo and Tan (2001) used the Granger causal approach to hash-out the strength of causality for each proposition using empirical evidence. Not surprisingly their results are inconclusive concerning the direct effects of democracy on economic growth. Not only does theory suggest that both causal directions are plausible, the statistical evidence also indicates that both outcomes are equally likely. With these

148 findings it is sure this aspect of the democratization debate will certainly go one in the foreseeable future.

4.3 A Critique of Development/Democracy Studies

Although the research concerning the relationship between economic development and democracy remains debated many scholars have come to accept that a general relationship exists which cannot be explained away by accident. As

Muller (1995) states it, “in multivariate models that take into account numerous economic and noneconomic factors, level of economic development (as measured by gross national product or energy consumption per capita) typically is the single most important explanatory variable” for democratization (p. 966). However, like the multitude of Pandora’s Boxes in the social sciences one recurring problem faced in these studies (aside from causation) has been the operational definitions used to define vague concepts such as development and democracy. Of the two concepts democracy has been the most discussed, scrutinized, and refined. For instance, and to name just a few, Lipset (1959), Cutright (1963), Neubauer (1967), Jackman

(1973), Bollen (1979, 1980), Arat (1988) Przeworski (2000), and Munck and

Verkuilen (2002) have all gone to great lengths to better define and measure democracy. Yet the operational definitions of “level of economic development” have been simplified since Lipset (1959) conceptualized it as a “complex” that included industrialization, wealth, urbanization, and education. The validity of this simplification deserves further inspection.

149

Table 4.1 compares various definitions of “level of economic development” among the studies cited above.28 Lipset’s original article went into great detail on the contributing factors between development and democracy, and even included ideational determinates (e.g. political legitimacy as a cause, attitude toward anti- democratic movements). However, as the statistical techniques become more and more refined, there has been a tendency to simplify the operational definition of

“level of economic development” to a single indicator—economic production/ wealth. Moreover, economic production is usually measured with a single indicator as well, most often as standardized energy units used (e.g. tons of coal units), or monetarily as standardized levels of GDP per capita, GNP, or PPP. Thus, as Arat

(1988) pointed out the “development/democracy debate” has become the

“economic development democracy debate” in which economic development is seen as the only determinant of development.

Although there is a strong and justifiable statistical logic for simplifying the proxy used for the “level of economic development” (see Bollen [1979] p.578 for more detail), doing so has lead to an over-simplification of the theoretical propositions being tested in the development/democracy relationship since “level of economic development” has been restricted to monetary gains and/or energy consumption.29 The result has been that social scientists interested in

28 One trend not shown in Table 1 is the reverse found among these sets of articles. Conceptualizing democracy has become a more refined effort (e.g. Bollen 1980, Bollen and Jackman 1985, and Muller 1988). Although it is statistically justifiable to pay closer attention to the operational definition of the dependent variable since it is the thing being predicted, theoretically independent variables and their specification are just as important. 29 Although it is true theory seeks the most parsimonious explanation (see Walter L. Wallace. 1971. The Logic of Science in Sociology), removing causality as a matter of statistical convince can lead to under-specification of the relationship being observed. 150

Proxies for Level of Related/Other Author(s) Indicator’s Operational Definition Development Determinants Lipset (1959) 1. Wealth 1. Per capita income, number of persons per Class Structure 2. Industrialization motor vehicle and per physician, number of Income Inequality 3. Urbanization radios, telephones, and newspapers per Protestant History 4. Education thousand persons. 2. Percent employed males in agriculture, per capita commercially produced energy consumed (tons of coal per person per year). 3. Percent of the population in places of 20,000 and over, percent in 100,000 and over, and percent residing in metropolitan areas. 4. Percent enrollment per thousand total population in primary, post-primary, and higher education, percent of total population that is literate.

Cutright (1963) 1. Educational Development 1. Literacy rates, student enrolment per 100,000 None 2. Urbanization in higher education. 3. Communication 2. Proportion of population living in cities over Development 100,000. 4. Economic Growth 3. News paper consumption, newsprint 5. Labor Force Composition consumption, telephones per capita, number of pieces of mail per capita. 4. Per capita measures of: energy consumption, steel consumption, income in U.S. dollars, and number of motor vehicles. 5. Proportion of economically active labor force employed in agriculture. Neubauer Same as Cutright (see None (1967) Neubauer [1967] p. 1006, footnote 8). Jackman (1973) 1. Economic Development 1. Energy consumption (expressed in million None metric tons of coal equivalents) per capita. Bollen (1979) 1. Economic Development 1. ln of energy consumption per capita. Timing Protestantism State Control of Economy Burkhart and 1. Economic Development 1. ln of energy consumption per capita. None Lewis- 2. World-System Position 2. Dummies based on Wallerstein (1974) Beck(1994) Przeworski and 1. Economic Development 1. 1985 PPP USD None Limongi (1997) 2. Polity (t-1) 2. Democratic and Authoritarian (Dictatorships)

Table 4.1: Comparisons of Development Proxies in a Sample of

DevelopmentDemocracy Studies

151 democratization have been unable to see how other key aspects of the development complex (Lipset 1959) affect democratization, or more generally political change, one of the more important components of that complex being urban populations.

As Przeworski and Limongi (1997) argued, certain social conditions may select for democracy. Therefore, how the urban transition unfolds and the patterns urban systems take on within a nation’s boarders may directly influence how the selection process determines the success and/or failure of any political system and not just democracy. In other words “selection” should not be understood as just an economically determined process; consideration should be given to cultural, demographic, and historical circumstances as well. Therefore, in the next chapter I examine the final piece of the puzzle contained in Figure 3.1; the reciprocal relationship between urbanization and political systems. As will be show there are several studies which link political regimes to urbanization but very few have linked urbanization to political change. In exploring this relationship I develop and argument and supporting hypotheses that allow for a test to determine what effect urban systems have on the selection of political regime types.

152

CHAPTER 5

URBANIZATION AND POLITICAL SYSTEMS

Many studies since Lipset (1959) have found evidence that level of economic development is associated with sustaining and stabilizing democratic regimes (see

Chapter 4 for a detailed review of these findings). Most explanations point to the emergence of modern production systems which alters economic relationships and cultural attitudes. Specifically, modern economic systems are said to promote an interdependent specialized division of labor centered on non-agricultural economies. As societies increase their agricultural output (while holding labor constant), both the means of production and population migrate from dispersed rural areas to dense urban settlements. Rural to urban migration changes how and why people interact (see Chapter 3 for a detailed discussion). In short, modernization is said to promote economic pluralism via a highly specialized division of labor and cultural pluralism via rural to urban migration. From the modernization perspective the end result is the emergence or adoption of democratic governance; a political system that is compatible with the material and cultural realities of modern society.

One consistent finding in this area has been that the relationship between development and democracy is curvilinear with a “threshold” limiting the effect

153 economic wealth has on democratic development (e.g. Neubauer 1967, see Figure

4.2 in Chapter 4). Despite these consistent findings the question as to whether or not economic development causes democracy has been heavily debated in recent years (e.g. Burkhart and Lewis-Beck 1994; Przeworski and Limongi 1997;

Przeworski 2000). Most suggest that the relationship is best explained as a selection bias. That is, democracies arise in and diffuse to nations at various levels of economic development but they are more likely to take hold in wealthy nations who can afford the price of democracy. Therefore, the number of wealthy democracies accumulates which gives the impression that economic development is a cause of democracy when in fact economic wealth helps retain democracies

(Przeworski and Limongi 1997; Przeworski 2000).

On a seemingly unrelated note, those looking at the causes of urban concentration/primacy have come to the conclusion that autocratic governments

“cause” primacy through policies which heavily favor the urban area from which dictators govern (Lipton 1977; Ades and Glaeser 1995; Davis and Henderson 2003).

However, those interested in democratization have yet to consider how these two findings are related. This chapter and dissertation is dedicated to such an inquiry.

As discussed at length in Chapter 3, the global urban transition is inseparable from the rise and spread of modern economic systems. As discussed in Chapter 4, economic development (a process that relies heavily on urbanization) is linked to democratization, both directly (e.g. Lipset 1959; Bollen 1979; Crenshaw 1995;

Burkhart and Lewis Beck 1994) and indirectly (Przeworski and Limongi 1997;

Przeworski 2000). However, looking back to Figure 3.1, a review of the literature

154 reveals that no one has serious considered how urban population structure or the population pressures associated with it are directly linked to political systems within nations. In fact, only a few have even considered that urbanization is linked to political regime type and those that have argue the relationship is indirect. The tendency to treat modern population pressures associated with urban systems as unimportant or tangential to political change has persisted despite the fact that other parallel processes have been directly linked to political change including; massive and rapid rural-urban migration, increased literacy and education, the centralization/ decentralization of wealth (i.e. level of inequality), and industrial production (e.g. Lipset 1959; Lipton 1977; Kasarda and Crenshaw 1991; Crenshaw

1995).

Given the above, what needs to be explored further and understood better is how urban population structure and modern population pressures contribute to the success and/or failure of certain political regimes. That is, given the classical and contemporary theoretical understandings of the role population pressure plays in eliciting social change as well as the extensive empirical findings discussed in the previous two chapters, it is likely that more than wealth or economic development alone determine political outcomes. For in fact, it may be that wealth will sustain any type of political regime regardless of its tactics or openness to descent (i.e. if people are prospering under a given regime, democratic or dictatorial, there would be little reason for those prospering to change it, see Olsen [1993] for a discussion).

Therefore, the persistence of political regimes may be better explained as a function of the structure of a nation’s urban system along with its level of socioeconomic

155 development—not just as a function of the economic system’s ability to generate or accumulate wealth. In other words, the argument that wealth sustains democracy doesn’t tell us how it does so. Looking to the urban system (a key component of modern economies) and how modern populations are dispersed throughout it offers some tangible reason as to why development (i.e. wealth) sustains democracy. In short, there is reason to believe that a nation’s urban population structure sets the tone for how political power is distributed within it and understanding this may be the key to understanding how urbanization directly affects political change.

Theory supports the above tentative hypothesis (i.e., that a nation’s urban system affects how that nation is/can be governed). Specifically, theory supports the notion that the level of urban concentration and the shape of the urban hierarchy should impact how power is distributed within modern populations which influences how much control political and economic elites can maintain over that system. From the perspective of modernization/ecological theory there are reasons to believe democracies fare better (take hold and remain more stable over time) when modern populations are located in an urban system composed of several cities (as opposed to one or two large cities), and where modern populations are distributed more or less evenly throughout the urban system. Under such an arrangement the decentralization of population and with it wealth, information, key functions, organizations, and transportation favors a national government structure that is more flexible in negotiating between competing cultural and economic interests. Therefore, nations with a history of a decentralized or a well developed urban hierarchy should be more likely to sustain democracy once acquired when

156 compared to nations at equal levels of development who have a centralized or primate urban system. In fact dependency/world-system theory and the urban bias thesis support a similar hypothesis and conclusion concerning the longevity of dictatorships who reside in primate urban systems.

From a dependency/world-system perspective nations with a centralized urban hierarchy and limited history of modern development may produce conditions that stabilize autocratic systems of government (e.g. nations that are overurbanized, primate, and/or have a high degree of urban bias). Such systems, because wealth, labor, and information are centralized and concentrated, would be governed more effectively with a centralized state that restricts civil rights, uses blunt force, and controls economic wealth to maintain the elites’ interests.

Furthermore, primate urban systems would allow elites to more easily monitor and control the movement of information and people, thus limiting the resources and opportunities needed to mobilize.

Echoing and merging the logic of Dahl and Tufte (1973), Ades and Glaeser

(1995), Crenshaw (1995), Przeworski (2000), and Davis and Henderson (2003), nations with a history of a primate urban system may be more prone to autocratic rule since these urban contexts allow dictators to maintain control over institutions, people, and wealth more efficiently and effectively than nations with multiple urban centers. In short, primate urban systems offer optimal environments for dictators to rise to power and maintain their power over time.

Whatever, the case may turn out to be the interplay between urbanization and political systems has yet to be explored in the development literature (see

157

Figure 5.1 below).30 Given this gap, in the sections to follow I explore what little information has been used to relate urbanization to political systems. First, I draw attention to the handful of studies that have linked political regimes to urban concentration/primacy. Several studies from political science and economics have shown that autocratic governments influence urban growth and urban patterns though their policies. Following this I review the sparse literature which has linked urbanization to political change. In doing so I look at two groups of findings/theory:

The first group focuses on urbanization as an indirect cause or as having indirect effects on democratic development. This has been the predominant and most accepted view of the role of urbanization in democratic development thus far. The second group does not address the relationship directly; however, they offer insights into parallel processes that can be used to support the argument that urbanization has a direct effect on polity. After examining this literature I develop testable hypotheses on the relationship between urbanization and political systems.

While I do not claim to offer a complete and/or exhaustive analysis, I do hope to offer enough insightful into this overlooked area to start what should be a new and exciting direction for cross-national comparative research.

5.1 The Effects of Political Systems on Urbanization

Figure 5.1 below illustrates the last piece of the puzzle yet to be explored. In this first section I look at the “tested” portion of this relationship (i.e., how do political systems/regimes/policies affect patterns of urbanization?). In the section

30 There are a few notable exceptions discussed below. 158

Urbanization

Political Systems

Economic Growth

Tested Relationship

Untested Relationship

Figure 5.1: Tested and Untested Relationships in Development Studies involving

Urbanization and Political Systems

to follow I begin to explore the “untested” portion of this relationship (i.e., how does urbanization impact political systems over time?).

In his essay, Jefferson (1939) emphasized that more often than not a nation’s political headquarters was located in the largest city. For him this was not a coincidence. In fact, this was why he named the law of urban primacy the “Law of the Capitols” (1939, p.227). Jefferson did not offer any detailed insight into what the relationship between capitals and primacy was. However, he did make it clear that the link between urban centers and political systems was important for explaining the emergence of urban hierarchies and primate cities. Since Jefferson there have been several follow-up studies that have found empirical evidence linking political

159 systems to patterns in urbanization—either in terms of urban concentration or primacy.31

Mutlu (1989) tested several hypotheses related to the impact political systems have on . His first hypothesis echoed Jefferson’s (1939)

“Law of the Capitals” and Lipton’s (1977) urban bias thesis. Mutlu suggested that capital cities tend toward primacy because political elites, who control a nation’s wealth, resided in capital cities and heavily invest in them. Mutlu argued biased investments make capital cities more appealing to both citizens and foreign investors. Thus, wealth concentration contributes to unbalanced urban growth.

However, Mutlu conditioned the “capital” effect on the size of the nation. He argued that larger nations are more likely to have sub-regional administrative capitals. As a result, larger nations are more likely to develop local primate cities within these regions. Therefore, a nation’s size acts to balance out urban hierarchies in terms of both primacy and concentration.

Mutlu (1989) also pointed out that historical events affecting the capitol city can influence the balance of the urban system as well. For instance, if in the past the capitol was moved (for whatever reason) it is likely that the degree of primacy and urban concentration will be less severe since growth and investment in both urban areas (the current and former capital) will lead to greater balance.

More generally, Mutlu (1989) argued that political regimes can indirectly affect urban systems through their support of certain economic or social systems;

31 As Mutlu (1989) notes, there is a difference between urban concentration and urban primacy. Measures of urban concentration capture to what degree the urban population is located in a single city while urban primacy measures how much bigger the largest city is in relation to other cities in the system. 160 e.g. free-markets versus command economies. He suggests that command economies are less likely to have primate urban systems or high levels of urban concentration because such economies are based on policies that restrict the migration of people, employment opportunities, the spatial location of production sites, and points of exit and entry in trade. Therefore, communist regimes should be less likely to produce primate urban systems due to their efforts to control economic activities.

After controlling for a number of independent variables, Mutlu (1989) found support for both the direct and indirect effects of political regimes on urban systems. In particular he observed that if a nation’s largest city is also its capitol the urban system is more likely to be more concentrated and more primate. He also found evidence that communist nations who favor a centralized state and greater control over economic relations tend to have lower levels of urban concentration and primacy. Thus, political systems exert both direct and indirect effects on urban structure.

Others have found similar results concerning the direct and indirect effects of political systems on urban structure; especially among developing regions of the world. For instance, Hansen (1990) applied the logic of Lipton’s (1977) urban bias thesis while using a compatible Marxist framework to explain the emergence of primate cites in developing regions. He argued that in order for political elites to maintain their power “governments in many developing countries have historically adopted pricing policies that have hurt agricultural productivity and thereby [favor] urban production; food subsidies that benefit the urban middle class, and urban

161 workers, have [also] discouraged farmers from producing” (1990, p.63). In short,

Hansen’s argument was that a similar “urban bias” affected those residing in small and intermediate sized cities. Thus, it is not just an urban bias against rural people/areas that leads to imbalance in urban systems of the developing world, but a large city bias that keeps developing nations from achieving “distributive justice” in its urban hierarchy (1990, p.65).

In a seminal study, Lyman (1992) examined how the political of former colonizers impacts whether or not nations developed primate or non- primate urban systems. He argued that differences between English and French colonial rule (i.e. political administration) have resulted in very different types of urban hierarchies. The English promoted localized governance where ethnic identity and language were encouraged. As a result, many of the British developed articulated indigenous communities which led to the expansion and growth of local markets resulting in a more balanced urban hierarchy.

Conversely, Lyman (1992) pointed out that the French governed their colonies as much stricter autocrats. Political administration was rationally centralized in Paris and headed by a central agency that dealt exclusively with colonial or overseas lands. Instead of promoting local governance and ethnic identity in the colonies they controlled the French promoted their own language and culture, and further, they promoted an ideology of assimilation among local ethnic groups as part of their colonial educational system. Moreover, Lyman (1992) argued that French autocratic rule was so bureaucratized and rationalized that it even trumped geography in dictating the location of capital cities and trade routes

162

(p.27.). Thus, the centralization of economic systems, political administration, and the homogenizing of culture resulted in the development of primate urban systems in former French colonies. He argued that the same should hold true for Belgian,

Portuguese, and Spanish colonies because all used similar centralized bureaucratic systems to manage their colonies. Lyman (1992) found modest support for his colonial hypothesis. Thus, what we can gather from his study is that the type of political system in place, whether imposed from near or from afar or even by past political regimes, can have a long term impact on how an urban system develops.

However, not all attempts by political regimes produce the intended effects when it comes to influencing urban structure. Sawers’ (1989) article offers an in- depth historical look at the development of Tanzania’s urban system from its time as a German and British to when it achieved independence. During that time

Dar es Salaam emerged as a primate city (becoming almost five times as large as

Dodoma) due to its favorable coastal location. After independence, however, the

Tanzanian government implemented policies in an attempt to even urban growth between urban locales and with it the distribution of services and resources.

Following the advice of Marxist urban scholars the Tanzanian government implemented strict policies in an attempt to promote even urban growth. For instance, they established price controls for agriculture (made crop prices uniform across the nation), they decentralized the location of political headquarters away from the capital and into local cities, decentralized welfare distribution centers, restricted industrial expansion within Dar es Salaam, moved the capital to in 1966, and even forcibly displaced citizens who were engaged in informal labor

163 within Dar es Salaam’s (1986, p. 841). Despite taking all these extreme steps urban primacy and an unbalanced urban hierarchy persisted and still persists to this day. Sawers (1989) argued that given the evidence of the failure of these deliberate policies, it is clear that within certain contexts political intervention will ultimately fail in the wake of the demographic pressures that accompany urbanization.

Clayton and Richardson (1989) examined the of the Soviet

Union and its effect on the nation’s urban structure. Unlike the Tanzanian example, the Soviets used population controls for both urban and rural areas beginning in the early 1900s. By limiting living areas and employment opportunities in cities the former Soviet Union was able to effectively restrict growth of their largest cities.

However, as Clayton and Richardson pointed out these policies, while they helped limit city size, they also limited agglomeration economies which stifled economic development.

In response, the Soviets began to plan satellite cities around major urban centers in an attempt to balance the benefits of agglomeration with the benefits of an evenly distributed population. However, Clayton and Richardson found that the satellite cities were growing much more rapidly than any other urban area. Yet, they did not have enough data on satellite cities to properly evaluate how it impacted urban structure. Thus, unlike the Tanzanian effort, the Soviet efforts to plan urban growth had been more effective up to a certain point. As Clayton and

Richardson pointed out in response to the rapid growth of satellite cities, there are

164 still some forces associated with urban development than not even Soviet urban planners could control.

In a more direct test of the relationship between urbanization and political systems, Ades and Glaeser (1995) examined how political regimes affect urban concentration within capital cities. Using an economic rationale in which distance affects costs, Ades and Glaeser argued that the same principle can be applied to the concentration of power within political systems. As they states it, “Politics affects urban concentration because spatial proximity to power increases political influence” (p.198). In short, they reasoned that the closer political leaders are to those they govern the more aware and the more control they have over the public’s actions.

With the above in mind Ades and Glaeser (1995) argued that dictatorial regimes gain political benefits by concentrating their power and resources into one large city. First, in primate cities dictators can control/ensure the longevity of their political power by concentrating their resources and personnel into a central location. Not only does placing all of one’s wealth and personnel into a single urban area make it easier to defend, it also makes it easier to control and react to opposition either through bribery or with the presence of a strong military force.

Second, dictators can use primate cities to disperse larger amounts of wealth and welfare within a smaller controlled area making expenditures more effective in appeasing a greater share of local urban inhabitants. And third, by supporting policies that contribute to urban primacy (i.e. low ) dictators can essentially ignore the concerns of rural inhabitants who do not have the resources needed to

165 mobilize and take action against him. Thus, urban primacy is a cost effective strategy for dictators who are likely to implement policies and distribute resources which grow a single central city. Ades and Glaeser (1995) found strong support for their hypotheses. Dictators “cause” urban concentration through their policies.

They believe their findings to be so clear and conclusive that they state “the predominant causality is from political factors to urban concentration, not from concentration to political change” (p.195).

In a follow-up study, Davis and Henderson (2003) explored a similar set of hypotheses. In constructing their models they accepted Ades and Glaeser’s (1995) finding that dictatorial regimes have a positive effect on urban concentration.

However, in an effort to expand the theoretical explanation they also looked at how democratization influences urban structure. In short, Davis and Henderson argued that certain government policies tend to be associated with specific regime types.

These policies influence whether or not a government will favor a single city over all others in the system, however, the tendency to favor a single urban area will also be conditioned by a nation’s level of development. The more developed a nation is the less favor it should show to a single urban area.

Davis and Henderson (2003) identified several scenarios in which governments enacted policies that favored the capital city. For instance, citing

Hansen (1990), they argued governments may choose not to invest in other smaller cities making them less competitive in the long run and less prone to growth. They may implement price controls, place taxes on markets outside the capital, restrict production/property rights within the city limits of the capital, and limit the

166 availability of public services to the capital area (see Davis and Henderson 2003, pp.102-103 for a complete list). Conversely, Davis and Henderson suggested that there are reasons to believe that democracies, especially those governed by federalism, should have more balanced urban systems. Aside from fewer restrictions on migration and the use of free-markets to guide their economic systems, federalism promotes the decentralization of resources across districts and cities since administrators and government officials reside in the areas they represent. This is a more nuanced version of the argument offered by Mutlu (1989).

Davis and Henderson (2003) found that a variety of political and policy variables affect urban concentration but that the same variables have very little affect on overall levels of urbanization. They found significant effects for democracy, federalism, and planned economies, all of which reduce urban concentration but have no impact on the level of urbanization.

It is clear from the various studies cited above that political systems and policies can have profound direct and indirect effects on urban structure. However, as Sawers (1989) as well as Clayton and Richardson (1989) report, not all political interventions will shape urban systems and not all intended interventions will produce the desired outcomes. And while there is growing evidence that regime type has both direct effects (Ades and Glaeser 1995) and indirect effects (Davis and

Henderson 2003) on urban structure, much more research needs to be conducted.

In fact, it makes little sense for scholars to state that causality is “predominantly” in one direction (e.g. Ades and Glaeser 1995). This is especially true when such conclusions are based on one set of empirical tests that are light on theoretical

167 insight to begin with. Instead, scholars interested in understanding the role of urban structure in the development process should be looking to original and creative tests of those relationships that have little or no empirical evidence against them.

5.2 The Effects of Urbanization on Political Systems

As illustrated above, most of the inquiries into the relationship between political systems and urbanization have been limited to exploring the effects political regimes have on urban structure. However, very few if any studies have explored how urbanization or the urban transition impacts political systems. In this section I look at what limited research/statements there are concerning this issue.

And while there have been very few empirical tests, there have been several theoretical statements which offer relevant insights. Therefore, in the sections to follow I make an attempt to uncover theoretical statements and empirical evidence relevant to the untested portion of the relationship between political systems and urbanization (see Figure 5.1). I break this inquiry up into two sub-sections; the first includes theory and evidence which supports the view that urbanization has an indirect effect on political systems.32 This has been the predominate view of the role of urbanization in the development process. The second sub-section includes a few studies that, while not direct statements, offer theoretical arguments that are highly compatible with the view that urban structure affects political outcomes directly.

32 I have found only two studies that consider the “direct” effect of urbanization on political change/democratization (discussed below). Most theory and literature has viewed the relationship as being more or less “indirect.” 168

5.2.1 Indirect Effects of Urbanization on Political Systems

Most studies and evidence on the relationship between level of development and democratization support the notion that urbanization is indirectly related to political change (e.g. Lerner 1958; Cutright 1963; and to a much lesser extent Lipset

1959). 33 As discussed in the previous chapter, the reason for this appears to be the early emphasis placed on economic development as the primary cause of democratization (see Table 4.1 in Chapter 4). In the paragraphs to follow I review the handful of studies and theoretical arguments which suggest urbanization plays either no role in democratization, a minimal role, or an indirect role through its influence on other processes associated with modernization. In other words these are the studies that argue a weak or insignificant relationship but they remain the most direct test of the untested portion of the reciprocal relationship between urbanization and political change.

Viewing urbanization as an indirect cause of democratic development and political change has undoubtedly come from the arguments of modernization theorists; but also to lesser extent dependency theorists who have accepted some of the basic premises of the modernization perspective (Crenshaw 1995). In fact,

33 It should be noted that conceptualizations of “urbanization” as the movement from traditional to modern ways/means of living have two distinct uses in the democratization literature: One meaning is the material/economic/demographic connotation that I have used thus far, the other is a cultural/ideational/normative meaning for those wishing to explore the relationship between attitudes and political behavior. Here I am referring to those who have looked at the material/economic/demographic realities of the urban transition as they related to direct/indirect causes of political change and/or democratization. I am arguing that most engaged in macro materialist/historical explanations have not viewed “urbanization” as a direct cause of political change but rather as an indirect cause. Many of those who have looked at the ideational, cultural, and normative effects stemming from the urban transition’s effect on traditional culture and norms have argued that “urbanization” is directly linked to change in political attitudes and norms (see Wirth 1938; Lerner 1958; Cornelius 1969; Inkles 1969; Schoultz 1972 for statements on this). And while there is some overlap in the modernization approach in this area, I ignore the literature looking at purely ideational forces since my argument does not depend on individuals but on populations.

169 across the board most cross-national scholars have either viewed urbanization as an outcome of agricultural technological advancements or as a pre-condition to the emergence of modern industrial/capitalist economies. In this regard urbanization has been viewed either as an outcome or as a catalyst for economic development, and therefore, it has never really been seen as a “cause” of political change. Let me expand upon this point.

Lerner (1958) offers a very good example of how urbanization has been viewed by modernization theorists in the manner described above. In his exploration of the modernization of the Middle-East, he discussed the role of urbanization in terms of its indirect effect on democratic development through its direct effect on literacy. That is, he argued urbanization affects other processes which are more directly responsible for changes in attitudes/behaviors that lead to the emergence of democratic political systems.

Specifically, Lerner (1958) argued that political democracy is achieved through a series of modernizing advancements in literacy and mass communication systems which begins with urbanization. Urbanization leads to increased literacy among the population. This is followed by the development of a mass media. It is through the use of mass media supported that literate populations can collectively demand increased political participation (i.e. democracy). Thus, Lerner’s model exemplifies the modernization approach to understanding urbanization as an indirect cause since it plays an indirect role in democratic development (see Figure

5.2 below for a very similar causal model by McCrone and Cnudde 1967).

170

While many have accepted a similar causal structure grounded in a modernization perspective (e.g. Lipset 1959, Cutright 1963) McCrone and Cnudde

(1967) were perhaps the first to empirically test whether or not urbanization had a direct or indirect effect on democratization via education and mass communication.

Using a small sample of developed and developing nations to run a series of Simon-

Blalock models, McCrone and Cnudde tested several plausible causal models in which urbanization was linked to democracy both directly and indirectly through literacy (i.e. education) and mass communication (or mass media). Based on the results of their empirical findings, they concluded that urbanization had a very weak direct (if not insignificant) effect on democratization; but a strong indirect effect through its direct effect on education (i.e. literacy). As they stated it:

[Our statistical model] indicates that the overwhelmingly important causal links in the process of democratic political development are contained in the developmental sequence from U to E to C to D. The direct effect of urbanization on democratic political development (as indicated by the use of a broken line) is negligible. (1967, p. 78) [See Figure 5.11 below]

Not long after McCrone and Cnudde’s (1967) findings were questioned by

Smith (1969). Smith offered three very important critiques which are worth reviewing: First, McCrone and Cnudde’s model lacked an operational definition for democracy. Second, they used the Simon-Blalock technique to generate their results. Due to the inherent limitations of this statistical technique, their conclusions could not be generalized to nations beyond the limited sample they had used. Moreover, Smith (1969) found that when the same technique was applied to a more inclusive sample urbanization played a more direct role in democratization

171

McCrone and Cnudde (1967) Smith (1969) E E

U D U D

C C

U=Urbanization

C= Communication Technology

E= Education *Reproduces from Figure 6, McCrone and Cnudde (1967, p.77). D= Democratic Political Development

Figure 5.2: Early Causal Models of Democratic Political Development with

Urbanization as Direct and Indirect Effects

than McCrone and Cnudde’s limited sample suggested. Moreover, Smith’s models revealed urbanization did not have a direct effect on education which was a key finding of their study and center piece of their theory (see Figure 5.2 for a comparison between the models). Finally, Smith argued that the lack of consideration for reciprocal causation, especially at more advanced levels of development, limited McCrone and Cnudde’s theory and findings. Smith suggested that at high levels of development education and communication systems should impact the levels and patterns of urbanization (i.e., there was reason to believe that causation was reciprocal).

172

The above debate was perhaps the most open and direct discussion concerning the effect urbanization had on democratic development/political change. However, the implicit treatment of urbanization as an insignificant cause of democratization can also be found elsewhere in the work of other modernization theorists. For instance, Rustow (1970), in his discussion on the fundamental prerequisites for the emergence of democratic transitions, argued that democratization was possible without “urbanization or [high levels] of per capita income” (p.352). And while his point was not to disregard these variables as unimportant, he did suggest that they were unnecessary.

Another example comes from more recent times. Midlarsky (1992) took a similar approach to understanding the role of urbanization in democratic development. His discussion centered on the role of agricultural density in eliciting political rights. In making his argument, he offers several modernization inspired models that either ignore the role of urbanization outright or treat is as an indirect cause. All three of the models center on the political modernization argument in which increased agricultural density impacts land inequality and leads to a demand/system strain that favors the expansion of political rights. However, the way inequality arises and why it leads to expanded political rights is a point of contention between the models as is the role of urbanization. In fact, two of the models he discussed ignore urbanization (insinuating that it is unnecessary altogether), while the third model treats urbanization as an indirect cause similar to

Lerner’s (1958) view.

173

The first model discussed by Midlarsky (1992) uses the history of ancient

Athens to suggest increased agricultural density can lead to a situation where inequality quickly rises as land becomes a scare commodity. The transfer of wealth and power from tribe and aristocracy to individual land holders eventually leads to a rise in political rights connected to land holders. This allows individuals to accumulate wealth and power by controlling land instead of being part of a specific group or hereditary lineage (see the next section for a competing view from

Rueschemeyer et. al. 1992 on the role of the landed classes in democratization).

Because these events are what led to develop a land-based democracy he labeled this model the Athenian path to democracy:

Agricultural DensityLand InequalityPolitical Rights

As he stated it; “Increased [agricultural] density yields increased inequality as the result of the increased land scarcity attendant on population increase…Land inequality implies graduation in wealth that would make some persons more likely to influence the political process than others…Thus land inequality (and its use, of course) [become] the sole basis for political participation “ (1992, p.485).

Therefore, as control of land shifts from hereditary and tribal hands to individuals in the wake of increased agricultural density, inequality becomes a key mechanism for granting individual land owners political rights and wealth (see

Muller 1988, 1995 for arguments linking inequality to democracy). This shift in where power originates alters how much “say” the individual has in political and economic decisions. Although Midlarsky does discuss the importance of Athens as an urban area/city, he fails to discuss how this ancient urban economy might have

174 contributed to the expansion of political rights. Moreover, it is worth mentioning that this particular model parallels the Marxist view of how the middle class bourgeoisie was able to capture control of the state away from the traditional elite and set up democratic systems that served the interests of capitalists (Arat 1988;

Rueschemeyer et. al. 1992).

The second model, or strain theory model, while agreeing that agricultural density is associated with rising land inequality, goes further to suggest that inequality triggers political violence which leads to expanded political rights. This model looks to the American and French Revolutions as evidence that violence is a necessary/central component of the democratization process.

Agricultural DensityLand InequalityPolitical ViolencePolitical Rights

Finally, the third model is closely linked to the demographic and economic modernization explanations of development discussed above, and therefore,

Midlarsky labeled it the “Developmental” model. This model follows modernization theory’s logic very closely and makes urbanization an important indirect cause of democracy (e.g. Lipset 1959). Specifically, it argues that agricultural density results in fewer rural opportunities which “pushes” excess rural population to urban areas.

Urban growth results in complex interdependent systems of trade increasing economic output and decentralizing wealth/power resulting in the expansion of the middle class and political rights. In short, economic development brings about the economic and political pluralism needed to support pro-democratic movements.

175

Agricultural DensityLand InequalityUrbanizationTrade

Economic DevelopmentPolitical Rights

While Midlarsky (1992) found support for the first and third models, the point here is that these models offer a useful summary for how urbanization has been viewed by most sociologists and political scientists as either an irrelevant/ unnecessary cause for the expansion of politic rights, or as having an indirect effect on the democratization process.

In short, when looking at how urbanization has been treated by scholars of cross-national comparative development it is clear that most attempting to explain democratization have tended to overlook, ignore, underspecify, or minimize its importance no matter what school of thought they adhere to. Modernization theorists have consistently viewed democratization as a product of industrialization while dependency/world-system theorists have viewed it as an outcome of capitalism (Crenshaw 1995). As a result, very few have ever considered how urbanization and the processes/outcomes of it may be linked to change in political systems or associated with certain political systems. And while no studies directly linking urbanization to democratization exist, there are several which can be used to support such an assertion.

5.2.2 Linking Urbanization to Political Systems

In this section, I look to a handful of scholars who have developed insights on democratization that are compatible with the notion that the demographic outcomes linked to the urban transition directly impact political systems. One such statement comes from Dahl and Tufte (1973), a second from Rueschemeyer,

176

Stephens, and Stephens (1992), and a third from Crenshaw (1995). I discuss each study in the order of the degree to which the arguments they contain can be used to support the notion that urbanization is directly linked to political systems.

Therefore, I start with Rueschemeyer et. al.’s (1992) exploration of class structure and class interest and move to Crenshaw’s (1995) analysis of the effect pre-modern history has on a nation’s ability to adopt and sustain democracy. I conclude with a detailed discussion of Dahl and Tufte’s book since it contains the most theoretical support for the hypotheses constructed in the section to follow.

In Capitalist Development and Democracy, Rueschemeyer, Stephens, and

Stephens (1992) develop a theoretical explanation for why capitalism and democracy are linked. Further, they test their class interest hypothesis using a methodological synthesis of historical comparative analysis and cross-national research. As will be shown, their argument relies on the implicit assumption that urbanization plays a key role in shifting class interests and therefore changing power relations.

In building their theoretical explanations and hypotheses, Rueschemeyer et. al. (1992) use a combination of Marxian/Weberian logic. They argue that capitalism is tied to democracy because capitalist economies alter class structure such that democracy becomes compatible with interests. In making this argument, Rueschemeyer et. al. (1992) implicitly rely on the urban transition as an assumed component of capitalist development. Therefore, urbanization, even though left underspecified is a central mechanism of their model. In order to

177 understand how and why this is so let us look more closely at their theory and reasoning.

As stated above, Rueschemeyer et. al.’s (1992) theory is an attempt to explain the widely held belief that democracy is “the characteristic political form of capitalism” (p.1). In order to build an explanation they first point out that

“democracy…is extremely rare in agrarian societies” where the “typical” regime type is autocracy or oligarchy (p.2). Given this fact, Rueschemeyer et. al. (1992) construct their theory on the premise that “power relations…determine whether democracy can emerge, stabilize, and then maintain itself even in the face of adverse conditions” (p.5). In other words, democracy requires a balance of power between class coalitions in order to persist. Too much power concentration results in a breakdown of compromise which acts against political democracy. Therefore, the real question to be answered is not why capitalism and democracy are related, but why agrarian class structures tend to be anti-democratic (i.e. promote the concentration of power) and why capitalist class structures tend to be pro- democratic (i.e. promote a more even distribution of power).

Departing from Marx somewhat, Rueschemeyer et. al. (1992) argue that

“capitalist development is associated with democracy because it transforms the class structure, strengthening the working and middle classes and weakening the landed upper class” (p.7). However, Rueschemeyer et. al. (1992) acknowledge political democracy does not favor the economic interests of all the classes within the capitalist system. Instead, under capitalism economic position within the production hierarchy leads to class formation and impacts that class’ support or

178 opposition for political democracy—this is the major component and contribution of their theory.

Rueschemeyer et. al.’s argument rests on the assertion that under capitalist production the class who favors democracy the most is the working class (i.e. the skilled and unskilled labor). Their support is based on the fact that they have neither the wealth nor an abundance of land to influence the state or counter transnational capitalist interests. Therefore, to maximize their power and secure their interests, the working class supports a political system that expands political rights to landless citizens. In contrast to Lipset (1959), Rueschemeyer et. al. (1992) argues that the middle-classes play a fickle role in this process and therefore are less important for democratization. Middle-class support wavers and depends on their specific relationship to the working class; they are not as poor as the working class yet they do not have the power of the landed aristocracy. Therefore, the more the middle-class’ interests are tied to the working class’ the more support for democracy they will show. However, the more their interests are tied to the landed aristocracy the more they will support autocracy and oligarchy. The least favorable to democracy are the landed aristocracy who rely on cheap (i.e. powerless) labor for their wealth and monopolization of property. Extending rights to the landless threatens the landed class’ economic interests. This is why agrarian societies are rarely governed by democracy.

One caveat not fully explored by Rueschemeyer et. al.’s (1992) theory, although it appears to be an important component of their logic, is the role the urban transition plays in eliciting political change through its impact on class

179 structure, class interests, and the redistribution of power from the landed to the landless. In other words, they acknowledge it is not just the working classes who are responsible for pro-democratic movements; but rather it is the urban working classes that are. As they stated:

We expected the urban working class to be the most frequent proponent of the full extension of democratic rights because this promised to include the class in the polity where it could further pursue its interests and because the working class, unlike other lower classes, had the capacity to organize itself (1992, p.6, italics mine).

This argument implies that is it not just the penetration of capitalism into economic markets that alters the class structure and erodes power away from the landed aristocracy. Instead it is the shift from rural agricultural production systems comprised of large land holdings and labor intensive production to urban industrial production systems based on a landless specialized division of labor that alters economic class structure and interests. Furthermore, the shift from low-density to high-density living increases the working (i.e. labor) class’ ability to organize, mobilize, and act in their interests (an important point developed later on).

Therefore, even though Rueschemeyer et. al. (1992) did not explicitly link the urban transition or urbanization to political democracy it is clearly a central mechanism that affects class structure and power relations. Indeed, it appears that it is urbanization which links capitalism to democratization and not industrialization

(i.e. just as you cannot develop a capitalist class structure without industrialization so too is the case with urbanization).

Crenshaw’s (1995) article on democratization offers logic that can be extrapolated to support the view that demographic pressures associated with the

180 urban transition are linked to a nation’s ability to successfully adopt political democracy. Although he does not control for urbanization in his models, his argument is nonetheless driven by a keen understanding concerning the effects demographic processes have on social/political change. Moreover, he builds an ecological-evolutionary argument centered on the idea that a population’s inclination for democracy predates either the spread of industrialism (the predominant modernization argument) or the spread of capitalism/colonialism (the predominant dependency/world-system argument). In short, he argues that denser populations necessitate cross-cutting affiliations between economic and political sectors (what he calls “proto-”) that require an investment in human capital. It is the emergence of these types of investments and affiliations (as Lipset argued) that create the underlying social structures needed to sustain modern political democracy.

Crenshaw’s argument begins with Lenski and Nolan’s (1984) finding that societies with a rich technoeconomic heritage and favorable biophysical environment have produced densely settled populations with complex institutional structures. In particular, populations with a rich technoeconomic heritage have experiences with increases in their carrying capacity as well as experiences with responding to those increases. Therefore, unlike other regions of the world which lack a rich technoeconomic heritage, “pre-modern” populations developed more complex economic and political systems because they had to, the effects of which have remained an important part of these nations’ present social and economic arrangements.

181

Crenshaw hypothesizes that because these past responses resulted in “pre- modern” institutions and social structures conducive to democratic governance, nations with a rich technoeconomic heritage should have an easier time adopting or transitioning to democracy regardless of (1) their current levels of development, (2) their current positions within the world system, or (3) their colonial pasts.34 With this in mind let us look at some of the demographic developments Crenshaw argues are linked to the emergence of pre-modern structures tied to high-density populations.

Perhaps the most important pre-modern demographic development to come about as the result of a rich technoeconomic heritage was the effect agricultural

(plow) technologies had on population growth and density. Like Spencer and

Durkheim, Crenshaw (1995) emphasizes the role isomorphic pressure plays in generating social complexity via specialization in the division of labor as well as its effect on innovation brought about by increased competition (see Chapter 1 for a review). In short, he suggests that populations with a rich technoeconomic heritage also have a history of dealing with large and dense populations both economically and politically. It is from their historic responses to demographic pressures that these populations developed the pre-modern social institutions and behaviors that are compatible with political democracy (Crenshaw 1995, p.704-705).

Perhaps the most important response has to do with how population increases affected who and how many people worked on the land. Crenshaw points out that in agrarian societies “access to land has traditionally been the single most

34 Crenshaw (1995) points out that many of the successful colonies were the same areas that had or were able to adopt agricultural technologies needed to increase carrying capacity (p. 705-706). 182 important resource determining stratification systems” (1995, p.706). He argues that because advanced agrarian economies rely on the hoarding of labor across large land holdings, their stratification system requires a more lenient tenant based farming system (i.e. a much dispersed labor base who acted mostly on their own).

Such a practical arrangement unintentionally leads to the decentralization of control over subsistence farming and use of the land. Moreover, because ownership of land is often driven by cultural based hereditary inheritance, the continued expansion of subsequent inevitably results in a greater division of land holdings among the land aristocracies’ offspring. Therefore, over time land holdings became increasingly subdivided and smaller in agrarian societies resulting in smaller land holdings and even less concentration of power among the landed classes.

Another related outcome of population growth and increased density for advanced agrarian societies that Crenshaw points out was the development of towns, , and cities. The development of these economic and population centers brought with them a more complex, interdependent, and specialized division of non-agricultural labor. This development helped further decentralize political power away from the landed classes. Decentralization resulted in a more rationalized economy, a more pluralistic economy, and the emergence of civil society (1995, p. 705). Moreover, increased population density and the rise of the nation-state required the development of complex transportation and communication systems that linked rural and urban economies. Thus, the more densely settled a population became the more the central markets supporting them

183 relied on the production of sustenance form rural areas and the more rural areas relied on the innovations and wealth generated in towns and cities.

Taken together, Crenshaw’s argument is simply that density necessitates complexity within social, political, and economic systems that requires the decentralization of wealth and power to achieve further integration in the wake of expansion. Nations with a history of dense populations are more likely to have the political institutions and economic structures needed to support a decentralized state. Democracies thrive better within societies that have already “paid” for the political institutions and economic infrastructure needed to support democracies.

Therefore, those populations that do not have a rich technoeconomic history and who attempt to adopt democracy are more likely to fail because the economic and political costs associated with democratization will be too large a burden for the population to bear.

What can be taken from Crenshaw (1995) to support the notion that there is indeed a link between urbanization and democratization is this: population density, which is a necessary feature of urban environments, is largely responsible for the emergence of complexity/interdependence within modern political institutions and economic infrastructure. One key aspect of the urban transition is that it re- distributes both the landed and landless classes into a single economic/political area, and this dramatically increases population density as well as class interaction resulting in a more rationalized/market driven social system. Therefore, if

Crenshaw’s hypotheses concerning density and proto-modernity are correct, for which he offers empirical evidence, then it should also be that nations with a rich

184 technoeconomic heritage will have an easier time urbanizing, evidenced by an urban system that is more even distributed. Moreover, and by extension, it also means that nations with a decentralized urban system should also have a more complex/interdependent political and economic foundation that is conducive to political democracy. This means that net of a nation’s technoeconomic heritage, position within the world system, or level of development those nations with a more even urban hierarchy should be more prone to democratic governance due to the effects a decentralized high density population will have on the need to develop cross-cutting affiliations between modern political and economic systems.

One key source for Crenshaw’s argument, aside from Lenski and Nolan

(1984), comes from Dahl and Tufte’s (1973) book Size and Democracy. Although he used their insights to link population density to interest group formation, a detailed follow-up review of their theoretical insights reveals that the relationship between size and democracy is more closely linked to urban structure than either they or

Crenshaw acknowledge. Therefore, it becomes useful to reassess their insights with the notion that urban systems proxy for the density of modern economic and political systems.

Dahl and Tufte (1973) address the age-old question as to whether or not excessive “size” hurts democratic governments. This now classical debate is based on the works of Plato and Aristotle and it was restated by Rousseau and

Montesquieu. Each of these scholars argued smaller sized polity was ideal for democratic government. That is, if democracies were to remain “optimal” in meeting citizens’ needs, polity had to remain small enough, in terms of both

185 citizenry and , such that participants could: (1) share experiences, values, and goals; (2) be heard as individuals; and (3) directly participate in the direction of self-governance. Dahl and Tufte suggest that these ideals supporting democracy and small polity were connected to early philosophers’ experiences with city-states as the major political units. Unlike modern nation-states with millions of citizen, direct participation in the governance of the city-state was often a reality since citizenship was restricted to the few (1973, p.4-8).

As city-states evolved into nation-states so too did the ideal of what the optimal size for an effective democracy should be. Dahl and Tufte point to French and early American philosophers who began to question the notion that democracies were optimal only when they were small. In particular, Thomas

Jefferson’s mentor Destutt de Tracy argued that a representative democracy could be just as effective for citizens of large sized polities as direct democracy was for small sized polities. And while most of the American founding fathers would adopt this same view of democracy, it was James Madison who made the most direct statement against the classical argument. He argued that a large “size” was actually preferable for the effectiveness of democratic regimes since it helped safeguard against the rise of autocracy and helped maintain individual freedom. Small sized democracies did not provide these safeguards. As Dahl and Tufte (1973) quote his words:

The greater the size, the greater the “variety of parties and interests,” and the less the probability “that a majority of the whole will have a common motive to invade the rights of the other citizens; or if such a common motive exists, it will be more difficult for all who feel it to discover their own strength, and to act in unison with each other… (p. 10-11)

186

In order to answer whether or not there was indeed an “optimal” size for democracies Dahl and Tufte (1973) looked to more fundamental processes underlying the relationship between size and democracy. It is this component of their analysis which is highly relevant to the discussion as to whether or not urbanization affects political change and the type of regime that is in place.

The first and perhaps most important point Dahl and Tufte make is better defining what is meant by “size” as a characteristic of a polity. In outlining the dimensions of size and democracy they suggest that a majority of the classical philosophers had certainly meant population when referring to size (e.g. Plato, who emphasized a citizen’s effectiveness in participation). However, they acknowledge others referred to area as well (e.g. Aristotle, who emphasized the polity’s ability to meet publicly within a common area). Moreover, size could also refer to population density which had an impact on both the effectiveness of the individual and the responsiveness of the polity. And finally, although not a measure of “size” per se,

Dahl and Tufte make it clear that “another relevant dimension of size [is] the distribution of the population” (1973, p. 18) since the location of citizenry impacted participation, the location of representative, and the degree of complexity and diversity among the nation (discussed below). Given these dimensions of polity

Dahl and Tufte suggest that most references used by past scholars referred to absolute size or the combination of both the size of the population, its density, and the area containing it.

187

In terms of the dimensions of democracy Dahl and Tufte argue for two relevant dimensions that interact with absolute size; citizen effectiveness, or how responsibly and competently citizens control the decision of the polity, and system capacity, or how efficient the polity is when responding to the collective preferences of citizens (1973, p.20). While I will not go in to great detail on these two dimensions, it is worth summarizing Dahl and Tufte’s analysis of them. They suggest that small sized democracies maximize citizen effectiveness but minimize system capacity while large sized democracies maximize system capacity but minimize citizen effectiveness. Given this relationship and the context of modernizing forces which paralleled the emergence of nation-states, Dahl and Tufte make the argument that the evolution of democracy as a form of governance has been one where direct democracy has been replaced with representative democracy due to the necessary trade-offs between effectiveness and capacity that accompanies large modern populations (1973, p. 23). Therefore, Dahl and Tufte suggest that:

No single unit size will be optimal for every purpose. An emergent complex polity of interrelated units will need units that change in size and scope as technology, communications, values, identifications, and other factors alter the balance of gains and costs. (1973, p.28)

With the relationship between size, effectiveness, and system capacity established Dahl and Tufte move to a very useful and insightful discussion on the relationship between size, complexity, and diversity. It is from this very insightful discussion that the foundation for an argument that urban systems affect the type of political regime in place can be established.

188

In looking at the relationship between size and political complexity Dahl and

Tufte look to categoric diversity as their primary proxy. Categoric diversity is composed of two kinds of diversity; cultural diversity, or the degree of variation in language, race, religion, and , and socioeconomic diversity, the degree of variation in occupation, education, income, and wealth (1973, p.31). The basic assumption is that increases in either or both types of diversity necessitate political complexity because more variables enter into the relations of each other thus impacting power relations due to the emergence of cross-cutting affiliations.

Therefore, homogenous populations with be less diverse and less complex than heterogeneous populations and will require less complex economic and political systems. Such an assumption closely echoes Spencer’s theory of compounding discussed in Chapter 1.

In an interesting set of comparisons, Dahl and Tufte argue that size (in terms of area) and level of socioeconomic development are weakly related to the degree of cultural diversity within and between nations. This is because national boundaries and the peoples contained in them arose from unique historical outcomes in which past conflicts and alliances appear to have been the primary source of national unification. In terms of socioeconomic diversity, level of development has a positive impact both within and between countries. Nations at higher levels of development are more diverse in their occupational and organizational structure.

When taken together cultural diversity and socioeconomic diversity both “appear to increase with the size of community within a given country but not among different countries” (p.33). In other words, different nations have different levels of cultural

189 and socioeconomic diversity but these differences are not attributable to population size but rather history. However, when looking at within nation diversity, larger communities are more likely to have both higher degrees of cultural and socioeconomic diversity than smaller communities who are a part of the same polity.

Based on the above associations Dahl and Tufte argue that when looking at within system variation we should expect small communities to be more involved in agricultural production, have less cultural variation between members, and exhibit fewer differences in occupational specialization. Conversely, in larger communities we should expect members to be more involved in non-agricultural production, have more cultural variation, and exhibit greater differences in occupational specialization. However, they warn that the degree of socioeconomic diversity is much more strongly related to the size of a community (i.e. population size) than is cultural diversity; area appears to be more important in increasing cultural diversity

(1973, p.34).

When looking at the effects of area on diversity Dahl and Tufte make two observations; the first is obvious and the second less so. The first relationship they point out is that “A big country is more likely to be highly heterogeneous than a small country. Conversely, a little country is much more likely to be highly homogenous than a big ” (1973, p. 34). The second relationship they point out is much more interesting: “Cultural diversity [tends to be] greater in less densely populated and less highly urbanized countries” (p.34). The reason; low density nations have low levels of socioeconomic development with spatially and

190 culturally separated communities that tend to be small. Therefore, while differences in socioeconomic diversity are low within and between each small community, differences in cultural diversity are greater between communities because they are not economically integrated and do not share lived experiences. What this suggests is that large nations with low density composed of many small communities are more likely to be socioeconomically less diverse but culturally more diverse than small nations with high density and large communities. Therefore, population size and density act together to determine the degree of categoric diversity (i.e. cultural and socioeconomic diversity) in predictable ways.

While the above relationship between diversity and size is abstract, less abstract are the relationships between organizations and size. Dahl and Tufte accept the notion that modernizing factors (particularly technology) impacts how many and how complex organizations are within polities. However, they also argue that population size and density play key roles in dictating complexity within and between organizations.

Perhaps the most important relationship observed by Dahl and Tufte concerning the effect of area on organizations is the impact increasing area has on a regime’s ability to control organizations and sub-units. As they state:

Whatever the structure the authority may be, whether democratic or hierarchical, the search for effective means of communication and control seem to produce a powerful tendency within any organization to break down into subunits as it grows in size. There are limits to the “span of control” that impose imperious requirements on all organizations: if an organization does not recognize them its formal structure will cease to parallel its informal structure. Some degree of decentralization is required, therefore, even in organizations dominated by hierarchical authority, such as military units or political dictatorships. (1973, p. 36)

191

With this being stated, they offer two important hypotheses concerning the relationship between area and organization: The first suggests that net of level of development the larger the polity (in terms of both population and area) the greater the number of organizations and subunits it will be composed of. Second, if all other variables are held equal, the larger the polity (again, in terms of its population and area) the more governmental subunits it will have. Because the effects of population size and area affect all organized population, Dahl and Tufte suggest that together the above hypotheses indicate that the larger a nation is the more decentralized its government will be. They find that this relationship appears to be driven more by the size of a nation’s area than by the size of its population (1973, p.

38).

Given these observations concerning the relationship between size and categoric diversity, as well as the relationship between size and decentralization (i.e.

“span of control”), Dahl and Tufte make two final assessments that support one very important hypothesis concerning the relationship between the size of political systems and the number of organizations it contains. The first observation is that

“every organization or subunit tends to develop its own interests or goals, along with leaders and other members willing to invest time, energy, and wit to achieve them. To this extent, organizations and subunits tend to become interests groups”

(1973, p.39). The second observation is that there is lower threshold with respect to the size of an organization’s likelihood of generating a separate interests group.

Therefore, if a community is too small (in terms of its area and population) the chances of generating a single organization (i.e. interest group) remains improbable

192 and with fewer than 3 impossible, but these chances increase with size up to a certain point. With these two observations Dahl and Tufte make the following hypothesis: “The larger the size of a political system (whether size is measured by population or area), the greater the number of organized interests or interest groups” (1973, p.39).

Even though Dahl and Tufte do not extend their analyses of size dimensions to urban areas, their logic fits squarely within the purview of supporting complimentary assertions concerning the effect urban “size” has on the degree of diversity and number of interest groups that arise within nations. That is, the “size” of a given urban system (in terms of people and the number of cities in the system) should be related to political outcomes since urbanization is directly tied to the redistribution of modern populations and thus dictates how many, where, and what types of organizations (i.e. interest groups) will be formed.

If we start with the realistic assumption that urban populations are a proxy for modern populations, then we can make several assertions concerning the effect the size-dimensions of urban populations have on how centralized/decentralized a polity will be. Generally speaking, urban populations have more political and economic clout than rural peoples—especially in a global economic context. Thus, the size and arrangement of urban populations within nations should be important for determining the type of political system in place since “size” will be associated with level of heterogeneity and heterogeneity with decentralization.

For instance, use the logic of Dahl and Tufte (1973) we can assert that the larger the urban population (in terms of absolute size) the more modern (i.e. highly

193 specialized) interest groups a nations will have, and thus the more political power should be divided (i.e. decentralized). Similarly, the more dispersed a nation’s urban population is throughout its urban system (i.e. the more area a nation’s urban population covers) the more interest groups there should be. That is, the more distinct urban areas there are the more city-specific interests there will be both between cities and within cities. Therefore, both the size and the (spatial) distribution of urban populations should have a direct impact on the number of interest groups within a nation, and thus it should directly affect the degree of decentralization within the urban system and nation.

What the above means is that a nation’s level of urbanization (percent urban), its urban structure (i.e. urban hierarchy), level urban concentration (i.e. urban primacy) as well as the absolute size of a nation’s urban population should have a direct effect on political complexity (i.e. the number of cross-cutting affiliations and degree of economic interdependency). That is, we should expect large urban populations, regardless of where they are located, to be linked to more complex societies which require the decentralization of economic and political power. In addition, the more dispersed an urban population is across its urban areas (i.e. the less concentrated or less primate the urban system) the greater heterogeneity between those urban populations and the more diffused interests and power should be. Conversely, nations with small urban population that are highly concentrated into one area (i.e. exhibit urban primacy and concentration simultaneously) should be associated with greater homogeneity. Thus, nations with

194 a small urban population located in a primate city should be less complex, more homogeneous, and exhibit a higher degree of political centralization.

In short, using the size components of Dahl and Tufte’s arguments, one can logically conclude that urban populations and their distribution have a direct effect on the type of political system in place. Large decentralized urban populations should favor democratic political systems, while small highly concentrated urban population should favor autocratic political systems. In Part III, these and other hypotheses concerning the effect of urban population on polity are empirically tested.

195

PART III

196

INTRODUCTION

Up to this point the preceding chapters have been aimed at establishing two central points: First, there is a rich theoretical tradition in sociology of tying population pressures to social change. And second, this rich tradition has not been applied to the topic of political change as directly or as often as it should within cross-national comparative studies of development.

In Part II I detailed how cross-national comparative research has failed to account for the impact urban populations have on political change within nation- states as they modernize. Specifically, I presented evidence from the literature that this oversight has persisted for several decades across several disciplines despite the wide-spread attention given to parallel processes and relationships (i.e. economic growth and political change, see Figure 3.1). Not surprisingly, my review of the literature did not reveal any direct empirical support for the notion that urban population pressures affect political systems. Nonetheless, in Chapter 5 I explored how the theoretical used to support development studies in related areas can applied to such a topic. With this in mind, Part III of this dissertation is dedicated to offering empirical evidence that both urban population structure and the modern population pressures of urbanization play a crucial role in dictating the type of political systems that persist within nation-states. It is also aimed at testing several fundamental hypotheses derived from the theoretical perspectives discussed in the first two parts.

197

In most cases, when applying the logic of modernization theory, ecological theory, and dependency/world-system theory to the above topic each predicts concentration of urban populations into one urban area favors autocratic rule or systems where political power is concentrated into the hands of the few (see

Chapter 5). Where they differ is with respect to the overall effects urbanization and urban population structure has on a nation’s inclination toward one type of political systems over another—autocratic or democratic.

From the modernization perspective, early stages of economic development require a centralized urban population because centralization fosters economic efficiency when industrial production is labor intensive (see Figure 3.5 in Chapter

3). However, centralization also favors autocratic (i.e. centralized) rule and thus promotes high levels of economic and political inequality. At later stages of development, economic growth brings about the need for greater system complexity to maintain efficiency in the production and distribution of goods and services.

Eventually economic growth results in complex economies which must decentralize in order to maintain peak efficiency (see section 3.2.2 in Chapter 3 for a discussion).

Economic decentralization results in a more balanced urban system which is made possible by advancements in production, transportation, and communication technologies. A balanced urban system begins to favor decentralized political systems as power and wealth become more evenly dispersed among a more specialized, interdependent, and educated population. Eventually, centralized rule becomes incompatible with and counterproductive to modern economic systems and democracies begin to flourish.

198

For dependency and world-system theory urban population structures reflect the hierarchy in the world-economy brought about by capitalist production.

The balanced urban systems found in core areas reflects those nation’s ability to exploit the periphery while the distorted urban structures found in the periphery proxy for how deep the core has penetrated the economies of weaker states. Thus, from the dependency/world-system perspective, increased urbanization has opposing linear effects on political outcomes because how and where urbanization unfolds is dependent upon a nation’s position in the world-economy. In core regions urbanization unfolds more or less into a balanced urban system because core nations use the excess resources and labor from the periphery to maintain steady economic growth and develop articulated economies. These excesses permit them the privilege of being able to sustain balanced urban systems and democracy.

In contrast, urbanization in the periphery unfolds in a highly distorted manner due to the core’s exploitation of the periphery’s resources and people. As capitalists penetrate deeper into periphery economies, their actions promote long- term underdevelopment and excessive urban concentration as well as overurbanization (see Chapter 3). Underdevelopment and high levels of urban concentration permit dictatorial and repressive regimes unique opportunities to seize power (a situation that is especially true post-colonialism). Moreover, capitalists find it profitable to directly support dictators both financially and militarily because dictators are often willing to ignore citizen rights to help capitalists’ secure cheap labor, grant foreign investors access to the nation’s natural

199 resources in return for rents, and guarantee relaxed regulation over production—all of which increase capitalists’ profits at the expense of the masses.

In light of the above, the most fundamental question that needs to be addressed (aside from whether or not urbanization and urban population structure even affect political systems) is whether or not centralized urban systems are more likely to support autocracies as both perspectives argue. However, if such a relationship exists one must then determine if it is monotonic as dependency/world-system theory suggests or more dynamic as modernization theory argues.

Although the first two hypotheses are straight forward, the last is not as easy to assess since either a curvilinear or threshold model could be used reject a monotonic relationship. However, for this dissertation the best way to test whether or not the relationship is monotonic (should it exist), is to consider how percent urban affects the relationship since how “modern” or how urban a population is should prove to be an important control for both modernization/ecological and dependency/world-system theories. In short, controlling for the percent urban is important for this reason: from the modernization perspective centralization of the urban population is a temporary condition of development as is the autocratic rule that accompanies it. As nations begin to develop and transform their economies from traditional to modern production systems, they inevitably become more urbanized—a key element for democratization (see Chapter 4). Therefore, from the modernization perspective the percent of the population that is urban, regardless of where it is located (i.e. centralized or not), is perhaps a more important indicator of

200 how ripe a nation is for democracy than the balance of its urban population structure. In contrast, from the dependency/world-system perspective the degree of imbalance within the urban system is more important. In fact, any gain in percent urban within primate systems in non-core nations is likely occurring within the primate city. Thus, if anything, gains in the percent of the population that is urban should only intensify the ill effects of highly centralized urban population structure on both economic and political inequalities within non-core nations. In short, interacting percent urban with measures of concentration/urban structure should offer a key test as to whether or urban population and its structure has a monotonic or dynamic effect on political systems. In sum, the three central hypotheses to be tested include:

H0: Ceteris paribus, urban population structure has no effect on political systems.

H1a: Ceteris paribus, centralized urban systems tend toward autocratic governance.

H1b: Ceteris paribus, centralized urban systems tend toward autocratic governance, but this affect attenuates as urbanization rises.

While modernization theory and dependency/world-system theory support arguments for why urban population structure should impact political systems via its effects on economic systems, ecological theory (as well as the classical theories) can be used to make a more direct argument concerning the effect modern population pressures have on political outcomes. That is, these theoretical perspectives can be used to make the argument that modern population pressures brought about by the urban transition should matter in addition to urban population structure, and in some cases population pressures should matter more

201 when in the context of specific urban population structures. For example, as

Durkheim and Spencer made clear, absolute population size should matter for how societies meet their needs and thus absolute city size should affect how nation- states meet their regulatory needs as well (i.e. the structure of their political systems).

If we recall in the first two chapters I explored classical and contemporary theoretical understandings of the effect population increases had on changes in social organization. In Chapter 1, it was pointed out that Spencer argued population increases necessitated parallel increases in specialization within the axes of differentiation (see Figure 1.1). Not surprisingly Spencer grouped the rise of cities, markets, and transportation/communication systems with the rise of local governments, bureaucracy, and the political state. His concept of compounding made it clear that change in economic and political systems overlapped because they are both tied to population pressures brought about by increases in population size (and to some extent density). Similarly, Durkheim equated population growth, both in size and density, with the rise of modern (organic) societies. From his perspective population increases placed pressure on individuals to differentiate in order to integrate into society. Differentiation simultaneously strained social institutions which were forced to accommodate new social relationships that traditional

(mechanical) social institutions could not accommodate. Population growth and increased density began to force societies to base integration on complementary differences rather than on similarities, thus changing the basis of social

202 relationships and the institutions that supported them, including political institutions.

Ecological theory also recognizes that larger more dense populations require more efficient and differentiated key functions (systems of energy production) as well as more efficient supporting functions (systems that distribute energy) than smaller populations. In addition, ecological theory acknowledges that larger populations cover larger areas, increasing the probability of populations spanning more diverse environments. Thus, larger populations inevitably require a more diversified knowledge of its total environment (i.e. technology) in order to extract the energy needed to support its size. As a result, larger populations are more likely to develop more complex interactions between functional units resulting in increased functional differentiation and interdependency. Together the pressure to develop a more diversified and efficient connection to environment (obtained via increased knowledge of environment) and more efficient distributional system

(achieved though the expansion of communication and transportation networks) ultimately forces larger populations to decentralize its key functions (i.e. in this case economic and political institutions). Given the above we should expect the absolute size of the urban population in a nation to have a direct effect on political change with respect to the type of regime in place. Not only does urban population lead nations in their economic endeavors they also lead their political interests as well, especially when concentrated into one large urban area (Jefferson 1939). Thus, cities with very large urban populations (regardless if they are primate or not) should be associated with movements toward democracy.

203

Indeed, most of the evidence concerning the reverse argument (i.e. that autocracies cause concentration/primacy) makes the case that autocracies promote the centralization of urban population into one city because it reduces the costs associated with controlling/appeasing people (Ades and Glaesner 1995, Davis and

Henderson 2003, Henderson and Wang 2006). However, these costs are surely dependent on the absolute size of the urban population being regulated. While it may be true that autocracies implement policies that affect where urban population growth is likely to occur through taxes and investments, and further, autocrats are successful at directing growth (i.e. causing primacy), even dictators have little say as to what effects modern population pressures associated with the urban transition have on social interactions/relationships. Thus, in line with Davis and Henderson’s

(2003) logic and findings that there is an “optimal” level of urban concentration for economic growth, so too might we expect an optimal level of urban population in terms of absolute size (and perhaps concentration) for autocratic rule. Small to intermediately sized primate cities may help economize power for autocrats by giving them control over citizens, but efficient control may be dependent upon the size of the urban population being governed regardless of where they are located.

Thus, once a nation’s urban population reaches a certain size the changes in the types of interactions needed to maintain modern population (as outlined above) and the diversity of interests/goals that arise from modern arrangements are likely to make it increasingly difficult for any autocrat to govern effectively via a centralized authority. In other words, the larger modern population become the more difficult

204 is it to maintain compatibility between an autocrats’ interests (absolute control of exchange) and those developing within his citizenry (freedom of exchange). Thus:

H2a: Net of the level of urban primacy, the larger the absolute size of the urban population in a nation the more democratic that nation should be.

In the Chapters to follow I put the above hypotheses to test. I also present a handful of informal tests of ecological theory, modernization theory, and dependency/world-system theory in order to assess the effect urbanization, in terms of modern population pressures and urban population structure, has on political systems within nation-states.

205

CHAPTER 6

MEASURES, DATA, AND METHODS

6.1 Analytical Variables and their Measures

In this chapter I outline the operational definitions of the measures used to test the hypotheses stated in the introduction to Part III. I identify the sources my dependent and independent variables, discuss their construction, and offer a brief explanation as to how and why each variable is useful for testing the theoretical perspectives discussed throughout this dissertation. I also discuss the strengths and weaknesses of the statistical method I have chosen as well as the diagnostics used to detect the presence of known statistical problems with time-series cross-sectional data.

6.1.1 Measures of Political Systems

Measuring change within political systems, and in particular for democratization, has been quite contentious over the years. Several cross-nation studies have used a variety of indices of political democracy in an attempt to capture key aspects of democratic political systems (e.g. Lipset 1959, Cutright 1963,

Neubauer 1967, Jackman 1973, Bollen 1979, 1980, Arat 1988, 1991, Alvarez,

Cheibub, Limongi and Przeworski 1996, and Przeworski 2000). While a majority of democracy measurers are highly correlated due to their reliance on common data

206 sources (e.g. Arthur Banks), Munck and Verkuilen (2002) argue all fail in some way in their attempts at conceptualizing, estimating, and aggregating the concept of democracy and/or autocracy (see Table 3 p.10 of their article for a list of the most used datasets and their attributes).

If one accepts Munck and Verkuilen’s (2002) argument, it follows that choosing any measure of political systems should be based more on one’s methodological needs than on the basis of the data’s validity (i.e. the high correlation between the data sets indicates that they are reliable and internally valid but not necessarily externally valid). Out of all the available datasets which attempt to measure political regimes, the “best” for the present empirical analysis is the

Polity IV dataset. This is because the Polity IV dataset offers extensive coverage and approximations for how centralized/decentralized political regimes are due to its use of scaled measurement (more on this below). However, it is not without a few shortcomings that must be pointed out.

Dahl (1971) discussed two key dimensions of political democracy which have become the conceptual basis for measures of it. The two dimensions are the degree of participation and the level of competition within the political system.

When conceptualizing democracy, Polity IV does not fully address the dimension of participation because it fails to measure the expansion of political/civil rights—a key feature of democratization in the late twentieth and early twenty-first centuries

(Munck and Verkuilen 2002, p.11). Less detrimental, but still noteworthy, is that the Polity IV aggregation rules do not have a theoretical justification. However, its open coding scheme and publicly available composite measures allow researchers

207 to reconstruct the variable when and if theory requires greater specification/ restrictions (Munck and Verkuilen 2002).

Despite these two flaws the Polity IV dataset offers a few benefits over the other datasets available for testing the hypotheses laid out in the previous chapter

(again, refer to page 10 of Munck and Verkuilen’s article for a complete list and comparison). First, with the 2007 revision now completed, composite polity scores are available for all nations and all years for the urban data I collected (from 1960-

2005—more on this below). This means that using Polity IV in conjunction with my annualized primacy dataset provides a near-balanced time-series cross-sectional

(TSCS) sample. Such complete coverage for a dependent variable is a rarity in TSCS analyses.

Second, the Polity IV data set “focuses specifically on the more or less institutionalized authority patterns that characterize the most formal class of polities, that is, states operating within the world’s state system” (Polity IV: Dataset

Users’ Manual 2007 revision, p.1). Moreover, political regimes assessed on the basis of the properties of their political systems rather than absolutes. Therefore, “there is no ‘necessary condition’ for characterizing a political system as [either] democratic [or autocratic]” (Polity IV: Dataset Users’ Manual 2007 revision, p.14).

Instead, a variety of measures are used to estimate the degree to which a political system is democratic or autocratic which fits nicely with the hypotheses tested in the next chapter. This is especially true since the dimensions used to construct the combined composite polity score attempts to capture how centralized political power is among the political elite (i.e. can it be contested) and how open

208 participation is to citizenry (i.e. who can participate). This approach fits nicely with the theoretical discussions in previous chapters which are more concerned with how centralized and decentralized political power is within the total system than it is concerned with the specific structure of the political system in question.

In addition to the above, and contrary to Munck and Verkuilen’s (2002) critiques, the weights assigned to the polity scores support the research design in this dissertation. Although the weights may not have theoretically underpinnings, they are arranged in terms of intensity or degree of openness, lending some justification to their ordering (see Appendix A for details). Thus, while the dimensions may not be weighted to accommodate one specific theory (a task that only individual researchers should do when required), there is a consistent logic and a consistent application of the composite measures over time.

In sum, it can be logically inferred that the measures used to assign autocracy scores approximate how centralization or concentrated power is among the political elite and conversely how decentralized or diffused power is among the political elite in democracies. This is the key to the strength and appeal of using the Polity IV dataset for this dissertation.

Instead of looking at either democracy or autocracy, it was decided that since the focus of most of the hypotheses stated earlier are concerned with how open/closed or centralized/decentralized political systems are the revised combined polity score what the appropriate measure to use as the dependent variable. This score is derived from the democracy and autocracy Polity IV measures which are based on a eleven point scale ranging from 0 to 10. A 10

209 indicates the highest degree of either democracy or autocracy for that polity’s political system. The democracy score is based on four dimensions while autocracy is based on five dimensions. Each index is weighted such that the sum of the maximum weighted dimensions equals 10 while the sum of the minimum is 0 (see

Appendix A for a detailed summary of these weights and their composite measures).

Nations can score above zero on both scales which is used to calculate the combined polity score.

The revised combined polity score, which is the dependent variable used in this dissertation, uses combined polity score which ranges from -10 (most autocratic) to +10 most democratic (see Appendix A for details). The revised combined polity score (i.e. Polity2) replaces -77 (cases of interregnum/anarchy) and -88 (cases of transition) with estimated values but codes -66 (cases of foreign interruption) as missing. However, because societies technically are never without some form of political structure (even in times of crises), and not all interruptions completely alter political systems, I decided that -66 values would be interpolated where possible.

Three additional points about the alterations made to the dependent variable needs to be discussed: First, nations with -66 “missing” periods at the start and end of entry into the sample remained coded as missing since interpolation could not be calculated (no extrapolation was conducted at any point). Lebanon was excluded from interpolation (and from the sample) after 1974 because interpolation would have taken place over 29 years. It made much less sense to interpolate these values than any other nation in the sample (where missing values occur for only a few

210 years). Second, and most important, the altered revised combined polity score used was led five years. A five year lead was chosen for several reasons: First, political change, whether from elections, civil unrest, or wars generally do not occur on a year to year basis but rather over the course of several years. For instance, most

(but certainly not all) civil wars, wars, transitions, and coups are settled/completed within a five year period. Thus a five year lead gives ample time for macro changes like these to “set-in.” Second, elections in democracies (and in some autocracies, e.g.

Iran) generally take place within five year intervals. Third, population growth and migrations usually take several years to begin to have effects on social institutions and social interactions. Thus, from a theoretical standpoint the lagging nature of macro demographic and political change made a five year lead appropriate.

Moreover, empirically a five year lead also made sense since political change is a slow process, which from year-to-year, offers little variance to be explained by statistical models.

Finally, several democratic nations did not experience any political change over the forty-five year period. Therefore, it was decided to exclude the most developed nations who had experience little or no change in the observed time period and that were also the most economically, culturally, and politically stable.

Specifically, twenty of the thirty OECD (Organization for Economic Co-operation and

Development) nations were excluded (i.e. those that held core status throughout the years being observed).35 Excluding these nations from an analysis of political

35 Core nations that were excluded: , Austria, Belgium, Canada, Denmark, Finland, , Federal Germany, Germany, Ireland, , Japan, Netherlands, New Zealand, Norway, Portugal, Spain, , Switzerland, , United States.

211 change made theoretical sense from both a modernization perspective (i.e. the most advanced industrial nations) and from a dependency/world-system perspective (i.e. core nations) because both theoretical perspectives are aimed at explaining transitioning populations not stable or developed peoples.

6.1.2 Measures of Urban Population and its Structure

It is important to acknowledge from the start that measures of urban population are influenced by subjective operational definitions of what constitutes

“urban.” Definitions of “urban” are not standardized across national contexts which can present problems in cross-national comparative research (Gibbs 1966; Jones

1967; Rogers 1982; Kasarda and Crenshaw 1991). For instance, the United Nations recognizes each nation’s standard of what they determine “urban” is regardless of how large or small the population threshold used. This makes comparisons across national boundaries questionable. Similarly, population thresholds tend to vary from nation to nation with some being as low as 2,000 people (Gibbs 1966) and others as high as 100,000 people (Schnore 1964; Rogers 1982; Kasarda and

Crenshaw 1991).

In terms of data collection the UN has defined urban localities as 20,000 population and cities at 100,000. While this standard has offered some uniformity to the UN datasets, a related issue of determining city limits (spatial incongruities) presents other definitional problems that cannot be easily addressed. In fact, the

UN’s Urbanization Prospects publication series and Demographic Yearbooks have been inconsistent in reporting “” versus “agglomeration/metropolitan”

212 data over the years making figures inconsistent over time.36 Furthermore, any estimation of urban populations beyond five years remains a serious point of contention in UN data (Brockerhoff 1999; Cohen 2003, 2006).37

Although data collection and definitional issues cannot be resolved easily, determining which statistical measures to use to proxy for urban population structure is somewhat clearer. Since demographers have taken interest in measuring the urban transition two basic measures of urban population systems have remained important focal points within cross-nation comparative studies; the degree of urbanization and the rate of urban growth. Rogers (1982) offers a simple but insightful distinction between these two aspects of the urban transition. He states, “Urban growth refers to an increase in the number of people living in urban settlements. Urbanization, on the other hand, refers to a rise in the proportion of a total population that is concentrated in urban settlement.” Further, he points out that “urbanization…is a process that has a beginning and an end… [While] urban growth…has no such inherent limit inasmuch as cities can continue to grow…”

(p.486). In other words, once a society becomes one-hundred percent urban (e.g.

Hong Kong) the process of urbanization is complete but urban growth (in terms of both absolute population size and density) can continue to increase indefinitely.

The urban transition has an end but the continued effect of urban population growth on social organization does not. Therefore, the impact of urban population

36 It is important to note that the UN has updated estimates of agglomeration areas going back to 1960. 37 While some of the urban population data that was collected was form projections, to projects past five years were used and in a majority of the cases the projections were 1 to 2 years or current at the time of colleciton.

213 pressures on a society’s organization and integration remains an important causal variable even at one-hundred percent urban. With this in mind let us look to the various measures which attempt to capture urban population structure.

As a statistical measure urbanization refers to the percentage of the population that resides in urban areas (i.e. Total Urban Population / Total

Population), and therefore, it is as much a function of rural areas as it is the “urban” population (Davis and Golden 1954). Percent urban is the most basic measure of how “urbanized” a nation is since it attempts to capture the degree to which a politically and geographically unified population resides in areas designated as

“urban.” Urban growth is measured in terms of the total addition of urban population over two periods of time within the same political and geographic boundary. It has two relevant empirical measures; the rate of growth (or how fast urban populations increase in size over two points in time, i.e. [UPt-UPt-1]/UPt-1), and the size of growth (or how many people are added to urban areas over two points in time, i.e. UPt-UPt-1).

Even though urbanization and urban growth are useful summary measures of urban populations and a nation’s urban transition, they have two limitations when assessing the structure of urban population: First, as Gibbs (1966) points out

“countries having the same degree of urbanization may be quite different with regard to urban size structure…” (p.171). Thus, two nations with seventy-five percent urban population may have the same proportion of its population living in urban areas but this tells us nothing of the urban population’s distribution within those nations. One nation may have one large city containing all the urban

214 population and the other nation may have fifteen or thirty urban areas containing their total urban population. Similarly, two nations may have the same urban growth rate but growth may be unfolding more or less evenly between the two nations depending on where the growth is occurring (i.e. in one city compared to multiple cities). Thus, while the degree of urbanization and urban growth rate are useful comparative summary measures of the urban transition, they fail to capture the structural component of urban populations (i.e., the distribution and concentration of urban populations) (Gibbs 1966, 1967; Jones 1967).

Luckily, observing the distribution and concentration of urban populations has garnered a great deal of interest since the urban transition has been linked to economic development (Kelley and Williamson 1984). In particular, measures of urban concentration have been used to proxy for the structure of a nation’s urban population. Unlike the two measures above, concentration statistics require three additional pieces of information that are lacking in summary measures when comparing urban population structures between nation-states. The first is the size of the nation containing the urban population (e.g. most often a politically unified area in km2), the second is the size of all the urban localities within the nation (i.e. the spatial/economic or political boundaries of the urban population—city limits), and the third is the number of distinct urban localities that comprise the urban system containing the urban population (i.e. how many urban areas are there?).

While the size of urban localities has been virtually impossible to track over time from a statistical standpoint, demographers, geographers, economists and sociologists have attempted to estimate urban concentration/distribution and map

215 urban systems without knowledge of economic or administrative boundaries. From these efforts three measures have been used to approximate the degree to which urban populations are concentrated and distributed within a nation.38 These measures include; the ‘H’ concentration measure or sum-of-squared-shares measure

(Adelman 1969; Wheaton and Shishido 1981), Zipf’s (1941,1949) rank-size rule, and a variety of urban primacy indexes from simple measures of concentration to more complex approximations (as we will see there are multiple measure of urban primacy which originated from Jefferson’s [1939] article). Let us look at each of these measures of urban population structure/concentration in greater detail.

The ‘H’ concentration measure or the sum-of-squared-shares measure (see

Wheaton and Shishido [1981] p.18), estimates the degree to which a nation’s population is concentrated in all urban areas relative to the total population of that country. It is calculated as the sum-of-the-square-shares of each city (or urban/metro area) relative to the total population. In order to calculate this measure every city’s population in a given nation’s urban system is divided by the total population and then summed. Mathematically the ‘H’ concentration measure is defined as:

Where n is the number of cities in the city-system of a certain size being observed, H is a number from 1 to n with 1 indicating absolute concentration and n complete

38 The has been considered as a fourth measure of urban concentration (Jones 1967). However, because it is a measure of relative urban concentration, when applied to the urbanization process in a cross national comparative context it is less useful (Jones 1968).

216 dispersion, P is the total population of the country, i is the rank of a city in a class of n cities, and Ci is the population of a city with rank i.

The rank-size rule, derived from Zipf’s law (1941, 1949), attempts to evaluate the distribution different sized market areas by examining the hierarchical nature of urban areas within nations. It is based on the observation that city size ranks within nations tend to follow a log-linear pattern with a slope of -1. The explanation for why urban systems tend toward Zipf’s observed log-linear pattern have been mostly attributed to central place theory which looks at market sizes as the key determinant of urban hierarchies (Christaller [1933]; also see location theory see Losch [1938, 1954]; Beckmann [1958]). However, it has been argued that this theory has severe limitations due to its unrealistic assumptions such as the assumption that there exists an equal distribution of resources over a given area (El-

Shakhs [1972]). In recent years a variety of other alternative explanations for this recurring pattern in urban systems has emerged (see Gabaix 1999a, 1999b for a review and extension). Despite the theoretical quibble, as a measure the rank-size rule remains useful because it offers an ideal type of urban population structure with which one can compare urban systems within nation-states to. Moreover, it is similar to measures of urban primacy in that the measure can be used to determine if a nation’s urban population tends towards primacy or a multi-centric distribution.

Mathematically the rank-size rule states that “the size of the nth city [should be] approximately one-nth the size of the largest city” (Backmann 1958, p.246).

According to Beckmann (1958) the rank-size rule approximates a Pareto

217 distribution39 with an exponent of 1. This function produces a log-linear pattern with a slope of -1 and with the intercept determined by the largest city whose rank is 1 (Fonseca 1989). When applied to urban hierarchies, rank-size for a system of cities is calculated as a within-system comparative measure in which an expected rank-size distribution is calculated first by using the population of the largest urban area with a rank of 1. From this intercept the rank-size distribution for the entire urban system is supposed to follow the unity of City Rank ∙ City Size = Size of Largest

Urban Area. Thus, a city with a rank of 4 should be four times smaller than the size of the largest city in that system (i.e. theoretically urban population hierarchies approximate an economic equilibrium).

Once a nation’s expected urban population hierarchy is calculated as a function of its largest city’s size, the actual rank-size of each city within the urban system can be plotted against the predicted rank-size slope. This allows one to determine how mature, uneven, or inefficient a country’s urban system is (see

Figure 6.1 for an example). In other words, theoretically the rank-size rule allows researchers to determine if a country’s urban hierarchy is more or less primate than it “should” be given the size of its largest city whose rank is 1. If the rank-sizes of the actual urban hierarchy fall below the predicted rank-sizes the system is said to be primate or underdeveloped because those cities with a rank below 1 are not at the size they “should” be given their rank (Smith 1985). If they fall above the slope then the urban system is said to be multi-centric, in that, cities below rank 1 are larger in size than they should be. In either case deviations from a log-linear

39 Pareto, V. Cours d’Economie Politique. Lausanne, 1896-1897.

218

284,276

Guatemala City

100,00 0

10,000

POPULATION 10 20 30 40 50

Predicted Rank-Size of Urban Area RANK Actual Rankings of Urban Areas Largest Urban Area/City

†Adapted from Smith (1985), p.90. Note: Not drawn to scale. For illustrative purposes only.

Figure 6.1: City-Size Distributions in Guatemala 1950 Indicating a Primate Urban

Hierarchy†

pattern with a slope of -1 are seen as signs of “uneven” development with primacy being more detrimental to economic systems (Smith 1985; Walters 1985). The closer a nation’s city system is to the predicted rank-size slope the more integrated and mature the urban system is supposed to be (Vapnarsky 1969).

So far I have discussed two measures which attempt to capture urban distributions; the ‘H’ concentration measure and the rank-size-rule. The problem, however, is that both of these measures technically require extensive knowledge about a country’s entire urban system from the largest to smallest urban area to be

219 effective in a cross-national comparative context.40 While extensive information on city size may be available for developed nations, for many developing nations this information is unavailable, making a true cross-national comparative analysis over time impractical—and in some cases impossible. For this reason many studies examining the relationship between urbanization and economic development have relied on a variety of versions of the urban primacy index (and in some cases simplified versions of the above measures). Although primacy measure offer simpler approximations of urban structure they nonetheless offer a glimpse into how urban systems are arranged.

Jefferson (1939) was the first to recognize that urban structure (i.e. urban primacy) was a feature of political and economic development when he devised “the

Law of the Capitols” (p.227). Yet he did not discuss urban primacy as a “measure” of urban concentration or urban structure. Instead he discussed primacy as a feature of economic and political power that was reflected “in 28 of the leading countries of the world [where] the largest city is more than twice as large of the next, [and] in 18 more than three times as large” (p.227). True to Jefferson, early studies measured urban primacy as the ratio of the largest city to that of the second largest city—a near approximation of rank-size. This version of the index became known as primacy 1-2 (Metha 1964; Linsky 1965). Later studies broadened the measure to include a larger sample of the urban system in an effort to better approximate the total urban population structure. Thus, primacy 1-4 is the ratio of the largest city to

40 One could argue that they would be more effective in a study examining one or a few nations whose urban areas are known and where extensive historical data is available (e.g. U.S., Australia, Japan, Britain, France, Italy, etc.). Still, even with limited knowledge of the urban system these measures can still be approximated and used effectively as will be shown in Chapter 7.

220 the sum of the next four largest cities (Owen and Witton 1973). Yet others have used an urban primacy index that simply takes the ratio of the largest urban area to that of the entire urban population (Wheaton and Shishido 1981, p. 19), what I refer to as Px.

The strengths and weaknesses of all the measures of urban population structure discussed above are more reliant on the availability of the data and the definition of “urban” used than they are on the formulas used to calculate them.

Since data on the total urban population and the largest urban/metro area have been readily available through the United Nations Statistical Division, most cross- national comparative research linking urban population structure to economic growth, and vice versa, have relied on some version of an urban primacy index. In particular, information on the largest (and often times second largest) cities as well as the urban population have been available through the UN’s World Urbanization

Prospects series, thereby favoring simple measures of urban primacy such as Px or primacy 1-2.

While the World Urbanization Prospects and the Demographic Yearbooks have been useful for assessing data on the world’s largest cities, most information on urban areas has been limited to cities with 100,000 population or more. What this means is that the urban population structure of many of the worlds least developed nations’ have been left unobserved by more complex primacy measures since they often have urban areas comprised of less than 100,000 population prior to the

1980s and even 1990s. For this reason I collected my own data on the five largest cities for every nation in the world from 1960-2005 whose largest city contained at

221 least 400,000 population by the year 2005.41 No minimum population size was placed on the fifth largest cities and indeed some of these “cities” had fewer than

1,000 people throughout the 1960s. Appendix A offers a detailed summary of the nations included in this dataset as well as the sources and methods used in data collection.

The benefits of having an extensive dataset on a nation’s largest fiver urban areas over forty-five years is that it allows one to approximate urban population structure using multiple measures, from simple measures of urban primacy to more complex approximations (e.g. El-Shakhs 1972). Thus, deciding on which measures to use as a proxy for urban population structure in order to test its impact on political systems involves a bit of scrutiny. In all I considered five measures of urban population structure to use as the independent variable. In the end I settled on using four measures to test both the general and theory-specific hypotheses. As mentioned at that time the primary focus of this dissertation is to offer empirical evidence that supports a link between urban population structure and political change. Therefore, it seems quite reasonable to test the effects of several well constructed measures rather than only one which may lead one to falsely accept the null hypothesis. With this in mind let us examine the four measures that are used.

The first is also the simplest; i.e., the largest urban area divided by the total urban population(Px). Px is essentially a simplified version of primacy measures with the exception that its population is contained in the denominator. Rather than considering individual elements of secondary cities it offers an approximation of

41 In a few instances I collected data on nations with an urban population just below that figure, e.g. 380,000.

222 how concentrated the urban population is into the largest city, while urban primacy is a measure of how much larger the largest city is than secondary cities in the urban system. While some have pointed out that Px completely ignores the smaller elements within the urban hierarchy (Wheaton and Shishido 1981), the primary benefit of Px is that it can be used to proxy urban population concentration, again an aspect of urban population structure that primacy indices do not completely capture. Viewed from this perspective Px becomes much more meaningful as a feature of a nation’s urban population structure since it presents an estimate urban concentration.

A second measure, primacy 1-4 (P14), is a more comprehensive measure of urban population structure because it captures the size differences between the largest element in the urban system while considering smaller elements and to some extent it is also a measure of urban concentration. In other words, P14 introduces a control for size while indirectly measuring the degree of concentration within the urban system, making it a more refined measure of urban population structure than Px.

The final two measures of urban population structure are taken from El-

Shakhs (1972); the first is a direct reproduction of his proposed primacy index and the second is a derivative of it. While his measure (what I call Pi) is similar to P14, it is more refined in the sense that it considers the relationship between the largest element and each of the smaller elements in the urban system on an individual basis. That is, his primacy index calculates the “average ratio of the differences in population size between [the largest city] and each smaller city” within the urban

223 system (p.18). In short, his measure uses the same information as P14 but it takes the size of the largest urban area in the system as a point of reference for the urban system and compares it to each of the smaller elements. The calculations for El-

Shakhs’ primacy index for the first five elements in an urban system is:

Where Pi (Primacy Index) is a ratio ranging from 0 to 1, C is the population of a city whose rank is i and n is the largest city whose rank is 1. For an urban system of 5 cities calculating the first city’s value equals 1 (a constant) and therefore it is left out of the equation.

While El-Shakhs’ measure is intuitive and useful for assessing urban concentration, it can be altered to construct a measure of urban system balance that includes an assessment of both centralization and decentralization within urban systems. Such a measure is needed since all the other measures of urban population structure tend to be optimal for assessing concentration or size differences. The measure I constructed can be used as an ideal typical approximation of Zipf’s rank- size rule. In short, this rank-size primacy index (D) measures an urban system’s average deviation from a log-normal distribution whose slope is negative one and whose slope is determined by the largest urban element in the nation. This measure, while highly negatively correlated with El-Shakhs’ Pi, introduces a system specific empirical “threshold” that approximates how much a given urban system

224 deviates from the log-normality either in terms of primacy or multi-centrism. The measure is calculated as follows:

Where D is the rank-size ratio coefficient for an urban system whose value is an integer between lim0< x

R=i). So, for example, calculating the distance for the fourth element in the system would be: ([Population of City 4 ∙ 4]/Population of City 1). According to the rank- size rule D should equal one since the fourth largest city should be four times smaller than the largest city. Therefore, when the sum of D=1 the rank-size of the urban system is perfectly log-normal (a balanced log-normal urban structure).

Values less than 1 indicate a primate urban structure and any value greater than1 indicates a multi-centric urban structure (i.e. two or more large cities of similar size in the urban system).

Although D is a useful measure, it becomes even more useful and intuitive if one considers the meaning of its logged values. That is, when D is not logged a value of one indicates that the urban system is perfectly log-normal (i.e. conforms to the rank-size rule). Any value above 1 indicates an urban system’s movement toward a multi-centric urban structure while any value below 1 a movement toward a primate system. This empirical threshold becomes more intuitive when D is logged

225 and allowed to take on both positive and negative values. Any logged values of D below 1 produce negative numbers while values above 1 produce positive numbers.

Moreover, zero indicates perfect log-normality within the urban system. Thus, the logged values of D retain all the valuable characteristics of its unlogged counterpart with the added benefit that negative values indicate centralization in the urban system and positive values decentralization.

In light of the available measures of urban population structure four measures will be used to test the general hypotheses with. These include; Px, P14, Pi and the log of D.

6.2 Other Important Independent Variables

Aside from dependent and independent variables discussed above there are other important controls to consider when testing theory-specific predictions made by the perspectives discussed in this dissertation. These include demographic, geographic, and economic properties of nations as well as a measure of education and others dummy variables.

6.2.1 Geographic Controls

The most important geographic control to test is the size effect on democracy as suggested by Dahl and Tufte (1973) is area (in km2). Data for area was taken from the World Bank Development Indicators (WDI) 2005 (but also from WDI on-line data 2007, and 2009). In some cases km2 of arable land were not available for 1960 but were for 1961, thus 1961 values were used for 1960; similarly the same was true for 2004 and 2005. In these instances WDI 2007 and 2009 on-line sources

226 were consulted to update the values.42 Most updated values were available for 2005 only. Therefore the average between 2003 and 2005 values were used for 2004.

When updates were not available values from the previous two years were used

(e.g. 2003 values were used for 2004 and 2005). Both variables were logged to correct for skewness.

Other important geographic controls, especially for testing the hypotheses of ecological theory, is geographic climate and its impact living opportunities. While percent arable land certainly plays a role in dictating living space, climate often dictates where arable land is located, how much is accessible, and when it can be used to support population—especially urban populations. Thus, in order to control for the effect hostile climates and inaccessible geographic regions have on urban populations, data on climatic zones and geographic regions that are useless for urbanization were obtained from the Center for International Development (CID); specifically, data was taken from the Mellinger, Sachs, and Gallup (2000) dataset. A variable was then constructed to estimate the percent of a nation’s land that was in the Koeppen-Geiger “bs”, “bw”, “h” and “e” zones (i.e. arid, semiarid, polar and highland regions).

According to ecological theory arid and semi areas, as well as mountainous and polar regions, should force populations to form fewer but denser urban settlements with very little transportation and communication networks connecting urban areas to rural areas. Arid and semiarid areas, as well as mountainous and

42 Note that some values available on-line were drastically different from the values obtained from the WDI 2005 CD. In those instances the 2003 values were used for 2004 and 2005. In other instances the 2005 values equaled the 2003 values and thus the 2003 values were used for 2004 and 2005.

227 polar regions, should also have a much less developed rural economy since these climates are hostile to stationary peoples. Thus, there should be a relationship between the percent of a nation’s land that is in these regions and the centralization of key functions making these nations more prone to centralized autocratic rule.

Restated more formally using climate controls to proxy for living opportunities, one may state it hypothetically as: The higher the percent a nation’s area is located in an arid or semiarid as well as mountainous and polar climates the more likely the nation is to be autocratic.

6.2.2 Urban Population Controls

For size of the urban population as well as percent urban, figures were obtained from the Food and Agricultural Organization of the United Nations

(FAOSTAT) 2006 revision. Note that figures for urban population were not available for 1960 using the FAO data, for this year WDI 2005 estimates were used.

Moreover, data were missing for Belgium from 1960-1999. As a result, data for

Belgium’s urban population were replaced with data from WDI 2005. As for annual growth, figures were obtained from the World Urbanization Prospects: The 2007

Population Database made available on-line.

6.2.3 Kilometers of Paved Roads

One variable of great concern to modernization theory is economic networks—most often these networks have been discussed in terms of transportation and communication systems. To proxy for these networks it was necessary to collect data on kilometers of paved roads. Unfortunately, data on

228 paved roads is scarce and scattered. To overcome this problem, data were collected from three sources and then merged to maximize coverage. First, data from

Canning (1998) were used as a base since his measure of paved roads is published and well known. Although his data have the greatest coverage for earlier years it lacks in years past the early 1990s. Thus, WDI 2005 data on percent of paved roads and kilometers of all roads was used to construct a variable that gave an estimate of kilometers of paved roads for later years (i.e. WDI percent paved times WDI total km of roads). These figures were then appended to Canning’s (1998) measure. It is important to mention that even in cases where there was a large jump or drop between the sources, missing data was still replaced with the WDI data where possible, but all WDI data was subordinate to Canning’s (1998) estimates (i.e.

Canning’s data took precedence over all other data in the construction of this variable). In most cases there were no significant differences. Once these two data sources were combined, missing values were then replaced with data from the CIA

World Factbook on-line.43 Interpolation was used to fill missing values where possible. Interpolated numbers were rounded to the neared kilometer. The final constructed variable was a measure of road density. It consisted of dividing total kilometers of paved roads by the total land area (see above). The measure covers the years 1960-2005 but most estimates for nations cover only a portion of those years.44

43 The latest information on paved roads for a given nation was taken from both https://www.cia.gov/library/publications/the-world-factbook/ and http://www.nationmaster.com/graph/tra_roa_pav-transportation-roadways-paved. 44 In some instance WDI 2005 data was available for total roads years prior (usually 1 year prior) but the percent of paved roads was not available. In those instances the percent of paved roads for the next year was used to calculate the prior year but no more than 5 years prior. Also, in some instances

229

6.2.4 Real Gross Domestic Product per Capita

In addition to paved roads, data on real gross domestic product per capita

(RGDP/c) was collected to proxy for level of development. Data on RGDP/c in constant prices based on the Chain Index were taken from two closely related sources: The Penn World Tables 2004 and Penn World Tables 2000. In one instance, for Angola, data were mostly missing for the entire period. The only data available was for the year 2000. Therefore, the estimates for Angola were taken from the PWT 2000 (which covered from 1960-1996). Once appended, values for

Angola between 1997-1999 were constructed via interpolation using the PWT 2000

1996 estimate and the PWT 2004 2000 estimate.

6.2.5 Education

One key control throughout most democracy studies is education. As Lispet

(1959) and other modernization theorists have argued, an educated population should be more democratic because they are more open to political and cultural difference. To control for education attainment, estimates of secondary school enrolment from Barro and Lee (2000) were used. Although their coverage does not include the entire years of interest in this study, their dataset is the most comprehensive available. It covers portions of 142 nations from 1960-2000 with

106 complete for the entire time period.

The data are in five year intervals and thus it was interpolated to maximize coverage (see Barro and Lee [2000] for a description of their collection and

WDI estimates conflicted with CIA Factbook data where the CIA data made more sense. In those cases the CIA data were favored.

230 estimation methods as well as the improvements made to their data from earlier versions). Estimates for secondary school enrolments are available for two age groups and for two populations. The two age groups are 15 and up and 25 and up.

The two populations are those who have completed secondary school and those who have attained secondary school status (i.e. those enrolled but not completed).

For the purposes of this dissertation it was decided that the age group and population that best approximated the theoretical effects of education on political regime (i.e. democratization) was the 15 and up who had attained secondary status.

There are three related reasons for using this population and this age group to proxy for the effects of education instead of those 25 and up who had completed secondary schooling.

First, as Barro and Lee (2000) point out about their own estimates the “over-

15 age group corresponds better to the labor force for many developing nations” (p.

2). On a similar note, and especially for this study, it makes more sense to be over- inclusive when deciding which threshold to use to determine who is “educated” and who is not. If we restrict our population to those who are 25 and up and who have completed secondary education, our measure favors Western democracies and wealthy democracies whose populations are more likely to fall into this category now and for past generations as well. Therefore, the effects of education may be an artifact of the standard used for its proxy and not the education itself (e.g. education levels proxy for wealthy democracy—they do not cause them).

Second, given the wide-ranged sample of nations included in this analysis it makes more sense to be more inclusive as to who and how education is attained.

231

This is especially true for developing nations whose population may not need as high of levels of education as developed nations since their economies are often driven by low skilled labor. Moreover, these nations are more likely to have events that may keep populations from completing school formally (e.g. war, civil-war, lack of funding etc.) but the population may be as educated as those who received a formal degree.

Third, developing populations are more likely to have fewer formally certified educated people. However, this standard tells us nothing about the type of education they receive or that their “incomplete” education is any less effective for building tolerance toward others. Indeed, one problem with Lipset’s (1959) original argument and modernization theory’s treatment of education in general, is the assumption that education is universally liberalizing in its outcome in the same way as Western nations have used it. It seems reasonable to assume that if education can be used to teach tolerance then it could also be used to teach intolerance (e.g.

Nazis). No formal verification standard (completed vs. attained) or any given age restriction (15 and up or 25 and up) can measure such a difference, making it difficult to justify any real cut-off or restriction. Therefore, using a broader more inclusive education threshold seems to offer a more conservative test of the modernization hypotheses that a more educated population should result in democratization.

6.2.6 Colonial History

In order to test the dependency/world-system hypotheses that history with core regions has lasting effects of political systems in periphery regions, no other

232 controls is as important as a nation’s colonial history and whether or not a nation had a communist history.

Dummy variables for colonial and communist histories were constructed using the CIA World Factbooks and Wikipedia (when the CIA Factbook was sourced).45 For colonialism, of particular interest were the French, Spanish,

Portuguese and Belgian histories. As Lyman (1992) argued, all four of these nations governed using centralized government structures (“direct rule”) which may have promoted primate urban systems. Thus, a preliminary hypothesis suggests that nations who have experience with direct rule should be more likely to remain autocratic since both the political and urban structures left behind favor centralization (i.e. urban primacy and centralized political authority). In order to test this hypothesis a dummy for French, Spanish, Portuguese, and Belgian history was constructed.

6.2.7 Exports, FDI, and Primary Commodity Exporters

Following Breedlove and Armer (1997) I use exports as a share of GDP to proxy for export dependence. This measure is used to test the relationship between economic dependency and political change when controlling for the effects of urban population structure. As discussed earlier, according to dependency/world-system theory nations that have export oriented economies should have economic systems and an urban population structure that serves the interests of both capitalist and political elites. This means that the more a nation exports goods and services the

45 Thanks to Kristorpher K. Robison at Northern University for providing me with these variables.

233 larger its largest city should be (i.e. primate), and the more autocratic (centralized) its political system should be. Data for exports as a share of GDP was taken from

WDI 2005 with coverage ranging from 1960-2003.

Still, another way to test the effects of urban primacy on political systems is to control for primary commodity exports. Primary commodities are raw or unprocessed materials (e.g. minerals, ores, timber, fruit, fuels, etc.) that are extracted and/or harvested with very little processing prior to being used.

According to the dependency/world-system perspective nations whose economies rely heavily on the export of primary commodities to other nations should be more likely to have an underdeveloped urban system as well as a primate urban system

(i.e. one large overurbanized city that is used to export resources to the core but does little to help economic growth). Moreover, economic wealth and political power should be concentrated into this one semi-developed urban area, making it easier for autocrats to maintain power and control over the exploited population.

Thus, when controlling for the effect of urban primacy, primary commodity export nations should be more likely to be autocratic.

While it is difficult to quantify primary commodity exports, especially over a forty-five year period, it is possible to identify those nations whose economies have historically relied on these exports in order to construct a dummy variable. Dummy variables were constructed using Chapter 5 in the International Monetary Fund’s

2008 World Economic Outlook Update. In all three dummy variables were generated: one for non-fuel primary commodity exporters, one for fuel primary commodity exporters, and one for both. In the end I used the combined dummy

234 variable to proxy for the effect of primary commodity exporting on a nation’s political system. Nations identified as primary commodity exporters of non-fuels included Burkina Faso, Chad, Chile, Democratic Republic of the Congo, Guinea,

Malawi, Mauritania, Mongolia, Mozambique, Tajikistan, Uzbekistan, Zambia, and

Zimbabwe. Nations identified as primary commodity exporters of fuels included

Algeria, Angola, , Republic of the Congo, Ecuador, Iran, Kazakhstan,

Kuwait, Libya, , , Saudi Arabia, Sudan, Syria, Turkmenistan, United

Arab Emirates, Venezuela, and Yemen. Finally, to test whether or not the economic interests of core nations affects political systems when controlling for urban population structure data on foreign direct investment (FDI) was taken from World

Development Indicators 2005. Coverage is scattered from 1970-2003 and does not include all nations in the full sample.46 FDI is measured as the net inflows of foreign investment as a percent of GDP (current USD).

6.3 Statistical Methodology

Now that the measures used have been identified, I offer a brief overview of the statistical method used to test the hypotheses: OLS regression with panel- corrected standard errors (Beck and Katz 1995). I discuss four common problems with time-series cross-sectional (TSCS) analysis; panel heteroskedasticity, contemporaneous correlation, serial correlation, and heterogeneity. While OLS with panel-corrected standard errors correct for heteroskedasticity and contemporaneous correlation, it does not address serial correlation or

46 Nations not included: Afghanistan, Cuba, Czechoslovakia (Former), Ethiopia (Former), Federal Germany, German Democratic Republic, Iraq, Libya, , Saudi Arabia, Serbia and Montenegro, USSR, United Arab Emirates, and Yugoslavia TFR.

235

Table 6.1: Simple Statistic Measures of Dependent and Independent Variables.

heterogeneity directly. Therefore, I explore whether or not implementing the suggested techniques for dealing with serial correlation (the inclusion of a lagged version of the dependent variable) and heterogeneity (the use of least square

236 dummy variables) is appropriate for my analysis.

6.3.1 OLS with Panel-Corrected Standard Errors

OLS with panel-correct standard errors (PCSE) has become the predominant statistical method for testing hypotheses using time-series cross-section data (TSCS) among cross-national comparative researcher in the social sciences. TSCS analyses are usually conducted on several fixed units (e.g. 10 or 100 nations) that have multiple observations over multiple time periods (e.g. ten or fifty years). Thus, is important to emphasize from the start, as Beck and Katz do in several of their articles, that the conventions of OLS with PCSE do not apply to other types of panel regression techniques, especially those that have fewer observations (less than 10) and whose analyses cover shorter periods of time (less than 20). In such cases GLS and FGLS estimates may be more appropriate estimators or even fixed effect models with a lagged dependent variable (Beck and Katz 2004, p. 3). Given the size and scope of the TSCS dataset for testing the hypotheses set forth in this dissertation,

OLS with PCSE is the most appropriate statistical method to use since contemporaneous correlation and panel heteroskedasticity are often serious problems with TSCS analyses.

OLS with PCSE is a statistical technique developed by Beck and Katz (1995,

1996) which attempts to overcome two of the four issues that plague TSCS analyses

( panel heteroskedasticity, contemporaneous correlation, serial correlation, and heterogeneity). Panel heteroskedasticity occurs when “the error variance differs across cross-sectional units due to characteristics unique to the units.”

Contemporaneous correlation occurs when “the errors across cross-sectional units

237 are correlated due to common shocks in a given time period,” while serial correlation occurs when “the errors within units are temporally correlated”

(Kristensen and Wawro 2003, p.2). Panel heterogeneity is when the phenomenon under observation is not generated by the same process which makes pooling units

(e.g. nations) inappropriate for statistical analysis, especially with respect to generating intercepts in multiple regression. OLS with PCSE attempts to correct for two of these four TSCS problems (Beck and Katz 2004, p.4). In order to correct for the third issue, serial correlation, it has been argued that one should include a lagged version of the dependent variable in the regression model (Beck and Katz

1996, 2004; Kristensen and Wawro 2003). The use of a lagged dependent variable to correct for serial correlation will be discussed further in the section to follow. To correct for heterogeneity among units the most fashionable method is to use a fixed effects (FE) model. However, because TSCS data must deal with panel heteroskedasticity and contemporaneous correlation as well, using a FE model leaves these issues unresolved. Therefore, as Beck and Katz (2004) note, it is more appropriate to use the least squares dummy variable (LSDV) methods (i.e., including a dummy variable for all but one unit in the sample in order to deal with heterogeneity). While these two issues are important, for now let us briefly review why Beck and Katz developed PCSEs.

As stated above, OLS with PCSE errors is intended to deal with the first two problems associated with TSC data—panel heteroskedasticity and contemporaneous correlation. Prior to Beck and Katz (1995), the accepted method of dealing with error structure issues within TSCS data was to use the Parks (1967)

238 correction method. The Parks method relies on FGLS (feasible generalized least squares) to estimate standard errors which are used to determine the significance of regressors, and thus play a crucial role in statistical analyses. However, according to Beck and Katz (1995, 1996) the Parks correction method does not sufficiently correct for contemporaneous correlated errors or serially correlated errors when dealing with TSCS data. This results in underestimating standard errors which leads to overconfidence when determining the true impact of independent variables on the dependent variable.

Beck and Katz (1995) trace the standard error underestimates of the Parks method for TSCS data to two sources: The first is the incompatible assumptions

FGLS makes about the error structure of TSCS data which is a problem that plagues

OLS as well (more on this below). FGLS assumes that the researcher has prior knowledge about the error process which the researcher never really has (1995, p.634). While making this assumption is less risky when estimating the standard error for only a few parameters, when estimating the standard errors for a large number of parameters (as is typically done in cross-national research), such an assumption is much harder to accept. Indeed, as Beck and Katz (1995) show using

FGLS to estimate multiple parameters in TSCS data results in the underestimation of sample variance.

Second, the Parks method uses FGLS to simultaneously correct for panel heteroskedasticity, contemporaneous correlation, and serially correlation of the errors. The problem is that when correcting for these problems using FGLS, TSCS data requires that N be as large a T and ideally that T be several times larger than N

239

(Beck and Katz 1995). While this is not a problem for panel data in which N is often several times smaller than T, in most cross-national research N is often many times larger than T (e.g. following 100 nations over 25 or even 50 years). Moreover, the

Parks method for correcting for serial correlation is also problematic because it applies a one-size-fits-all approach when controlling for the unobserved error structure. That is, the Parks method assumes that the “serially correlated errors…follow a unit-specific first-order autoregressive (AR1) process” (1995, p.637). According to Beck and Katz this assumption is at odds on a theoretical level with what most cross-national analysts assume about the structure of their data (p.

638). Indeed, the analyses conducted here examine a much longer lag period.

So what then are the properties of OLS with PCSE that make it better suited to deal with the issues of panel heteroskedasticity as well as contemporaneous and serial correlation for TSCS data? To answer this question it is necessary to understand why OLS alone cannot properly deal with TSCS data. It also requires keeping in mind that while OLS with PCSE corrects variability estimates that address the issues of contemporaneous correlation and panel level heteroskedasticity, it does not deal with serial correlation which “must be eliminated before the panel-corrected standard errors are calculated” (1995, p.368) or issues of unit heterogeneity.

The problem with applying OLS to TSCS data (without PCSE) is that OLS is based on several assumptions about the random nature of the error structure. Such assumptions are perfectly legitimate for cross-sectional data but quite inappropriate for TSCS data that pool cross sections over time. Specifically, the temporal and

240 spatial properties of TSCS make it difficult to believe that “errors for a particular unit at one time are unrelated to errors for that unit at all other times (no serial correlation) and that errors for one unit are unrelated to the errors for every other unit (no spatial correlation)” (1995, p.636). It also makes it difficult to believe that pooling without consideration for heterogeneity is a non-issue in TSCS data (Beck and Katz 2004). If the above were true, and TSCS data had a simple error structure and units shared the same histories, cultures, and social systems, then applying OLS would produce unbiased estimates of the standard errors. However, making these

OLS assumptions is risky given the complex error structure which is common to most, but not all, TSCS data and given that most nations have some level of heterogeneity which must be accounted for on a unit to unit basis.

The Beck and Katz (1995) correction, which deals with contemporaneous correlation and panel heteroskedasticity only, is implemented by first running an

OLS of all the parameter estimates. The model also typically includes a lagged dependent variable to control for serial correlation prior to making the panel corrections (Beck and Katz 1995, p.645).47 Once the OLS standard error estimates are obtained they are “corrected” using estimates that assume contemporaneous correlation and heteroskedasticity. Monte Carlo simulations show that the panel- corrected standard errors result in estimates that are much less biased than FGLS estimates (for details on how standard errors are estimated see Beck and Katz 1995, or Kristensen and Warwro 2003; for a detailed comparison between the accuracy of the Parks estimates of the SEs and the PCSE see Beck and Katz 1995).

47 Whether or not one corrects for heterogeneity has no effect on the panel corrected standard errors.

241

6.3.2 “To lag or not to lag?”

While OLS with PCSE helps researchers deal with the issues of panel heteroskedasticity and contemporaneous correlation, it does not address serial correlation or heterogeneity. Beck and Katz (1995) state that in order for OLS with

PCSE to be efficient and BLUE serial correlation should be removed prior to calculating the PCSEs. Their recommendation is to use a LDV as a correction for serial correlation (Beck and Katz 1995, 1996, 2004)—but this recommendation is not an absolute (Achen 2000; Kelly 2004; Keele and Kelly 2006; De Boef and Keele

2008). In fact, Achen (2000) has shown (using Monte Carlo simulations) that inserting LDVs results in biased coefficients regardless if the specification is theoretically sound or not (Keele and Kelly 2006). So how does one proceed to determine whether or not to include a LDV, either as a theoretical control or as a correction for serial correlation? To answer these questions let us again define serial correlation and explore the literature on role of LDVs in TSCS data analyses.

Serial correlation (or autocorrelation) is a temporal case of correlation of errors in which residuals at time 1 are correlated with the residuals at time 2. When present, this type of correlation violates a fundamental assumption of OLS (i.e. no autocorrelation). While this violation does not bias OLS coefficient estimates, it does affects standard error and t-score estimates. In fact, positive autocorrelation, which is the most common form in TSCS analysis, leads to an underestimation of the standard errors and an overestimation of t-scores. Thus, determining significance levels as well as whether to accept or reject hypotheses becomes problematic. From

242 a hypothesis testing perspective serial correlation is perhaps more serious than the biased coefficients produced by LDVs.

In most research situations, when present autocorrelation follows an AR1 process after which point the “memory” or “shocks” begin to weaken (i.e. as time the between observations increases the correlation between errors weakens). Because newer time-series data cover more cases over a greater number of consistent time periods (e.g. years—1950, 1951, 1952….etc.), the impact of serial correlation on regression results (especially as an AR1 process) has become more and more of an issue.

In the past when datasets were much smaller (in terms of both N and T), including a LDV (i.e. Yt-1) was recommended as the preferred correction because the bias introduced by the LDV was thought to be trivial (Kelly 2004; Keele and Kelly

2006). It was generally accepted that the consequences for not controlling for serial correlation (underestimation of the standard errors and overestimation of t-scores) outweighed the consequence of doing so (efficient but biased coefficient estimates)

(Keele and Kelly 2006). However, recent empirical scrutiny of the LDV correction has been gaining ground among time-series methodologists who suggest that the bias introduced by LDVs may be more serious than first acknowledged (Achen

2000; Kelly 2004; De Boef and Keele 2008). Therefore, automatically including a

LDV without justification draws skepticism. As De Boef and Keele (2008) state:

More typically, however, the justification [to include a LDV] doesn’t appeal to theory at all, as the lagged dependent variable is included to clean up serial correlation. When this restriction is invalid, β0 and α1 will be biased. The degree and direction of bias are a function of the covariance of Xt and Yt-1, respectively, with Xt-1. Because Xt tends to be highly autocorrelated, the bias in β0 will tend to be large. (p.187)

243

There are several known problems with including LDVs even in the presence of serial correlation. The first is perhaps the most important to consider here since it violates a fundamental assumption of OLS—that being, regressors cannot be correlated with the error (Achen 2000; Kelly 2004; Keele and Kelly 2006). Error is supposed to be random and some have argued that including an LDV, while it may help with serial correlation (i.e. temporal error), results in correlation between unobserved non-temporal error (Kelly 2004, see p. 4). Unnecessarily introducing this type of correlated error leads to biased coefficient estimates and thus estimates are no longer BLUE (De Boef and Keele 2008). Therefore, as Kelly (2004) argues the question as to whether or not including an LDV produces biased estimates is not really a question since the answer is that it does; the real question for researchers interested in using LDVs is whether or not the bias introduced is large enough to warrant an exclusion of an LDV or if it is small enough to include one without concern for biased coefficient estimates.

So given the above uncertainty about whether or not to include a LDV when modeling TSCS data what is one supposed to do? Does one include a LDV or not?

While answering this question requires a review of the techniques for determining if serial correlation is present (see below), Keele and Kelly (2006), De Boef and Keele

(2008) and even Beck and Katz (2004, 2009) all agree on the importance of using a theory-first approach when modeling temporal dynamics before introducing temporal corrections/restrictions. In fact, De Boef and Keele (2008) suggest that the best approach is to start with a general model that poses no restrictions on

244 temporal dynamics because “working backward from a restricted model is unlikely to get us the best model” (p. 189).

Still, one should be aware if serial correlation is present and how much of a problem it is before and after introducing a correction. If there is enough serial correlation to warrant a correction then its inclusion will not do harm since it will not dominate the results of regression (Beck and Katz 2004). However, if there is serial correlation and the correction fails to correct for it, then it may be best to take a more aggressive approach for dealing with serial correlation and use an error correction model (see De Boef and Keele 2008 for a discussion and alternative correction models). Thus, what is needed is a reliable way to determine if and how much serial correlation is present. In all there are three such tests: the Durbin-

Watson, the Breusch-Godfrey Lagrange Multiplier (LM) test, the Wooldridge (2002).

The latter applies to FE and random effects (RE) models and will not be discussed.

6.3.3 Testing and Correcting Serial Correlation

The Durbin-Watson is the oldest of the tests for serial correlation. Its benefits are that it can detect both positive and negative forms of autocorrelation.

The test statistic is based on a four point scale with a value of 2 indicating no serial correlation, below 2 positive serial correlation, and above 2 negative serial correlation. Values below 1 and above 3 are said to be cause for alarm. However, the critical values for the Durbin-Watson statistic vary and depend on sample size

(N), the number of regressors used (k), and the level of significance (generally .05 and .01 levels). Thus, the values listed above are general guidelines. In order to

245 properly decide if the Durbin-Watson statistic obtained is cause for alarm, one should reference a Durbin-Watson critical value table.

There are several limitations to the Durbin-Watson test statistic when applied to TSCS models that make it inappropriate. First, the statistic only estimates serial correlation for first-order autoregressive processes which is a major limitation if one expects long run time-dependent effects in the phenomenon of interest. Second, if a LDV is used to correct for serial correlation the test cannot be performed to determine whether or not any serial correlation remains once it is introduced—that is, the Durbin-Watson cannot be used with LDVs. Thus, the

Durbin-Watson is quite limited for TSCS analyses.

An alternative to the Durbin-Watson which allows one to include a LDV and test for serial correlation across more than one time period (i.e. lags) is the Breusch-

Godfrey Lagrange Multiplier (LM) test. This test of serial correlation which uses residuals to determine if serial correlation is present is advocated for TSCS analyses by Beck and Katz (1995, 2004). The Breusch-Godfrey is a test statistic based on the null hypothesis of no serial correlation. Generally the null is not rejected unless the statistic’s chi-square probability test of significance is greater than .17 (Watson and

Teelucksingh 2002, p. 189) although most use a one-tail chi-square threshold of .05.

When the null hypothesis is rejected, and serial correlation is present, analysts can use a LDV in an attempt to correct for serial correlation. The Breusch-Godfrey LM

246 chi2 test statistic can then be calculated again to determine if including the LDV removed the serial correlation.48

In conclusion, it appears from the literature that the (current) accepted way to deal with serial correlation and LDVs in TSCS analyses is to first determine if serial correlation is present. If it is present and highly significant one should then specify a model using theory without imposing restrictions to the general ADL

(autoregressive distributed lag) model (see De Boef and Keele 2008 for a list of common ADL models). Once the model is calculated one should determine if serial correlation remains using the Breusch-Godfrey LM test (Beck and Katz 2004). If serial correlation remains one may have introduce the appropriate number of lags or use an error correction model which deals with serial correlation more aggressively since it accounts for both short run and lung run processes that may be generating serial correlation (De Boef and Keele 2008).

6.3.4 Heterogeneity in Time-Series Cross-Section Data

The final issue to be addressed here is unit heterogeneity. As discussed earlier panel heterogeneity is when the phenomenon under observation is not generated by the same process among units in the sample which makes pooling units (e.g. nations) inappropriate. The most efficient way to determine if the

48 Some statistical packages do not have the Breusch-Godfrey LM test or a version of the test that is appropriate for panel data. Despite this the test is rather simple to calculate using the residuals. In the case of using panel-corrected standard errors the steps are as follows: First, one runs the model with panel-corrected standards errors. Once residuals are estimated they are used in a regular OLS regression. That is, the estimated residuals become the dependent variable and the same independent variables are used with the addition of the lagged residuals. Once the regression is run the number of time observations, n, is multiplied by the adjusted R2 obtained from the residual regression. The number obtained is the chi-square test statistic. This statistic is used to test the null hypothesis of no autocorrelation using 1 degree of freedom. See Appendix B for an example using Stata syntax.

247 pooled units’ intercepts are significantly different (heterogeneous) is to use the F- test (or Chow test). In most statistical packages (e.g. Stata) F-tests can be obtained by running a fixed-effects model with a LDV. If the F-statistic is significant the null hypotheses of no heterogeneity can be rejected and using fixed-effects is appropriate. If the null cannot be rejected it is advisable to model the relationship using a random or between effects model.

Although fixed-effects models do not calculate panel-corrected standard errors, one can approximate fixed-effects in OLS by using least-square dummy variables for every nation in the sample since fixed-effects models are akin to this approach (Beck and Katz 2004). The problem with using LSDVs, however, is that other dummy variables and other time-invariant measures (e.g. area in km2) cannot be included in the regression equation without producing biased estimates. Thus, in order to control for all four problems that typically plague TSCS analyses, the most restrictive models risk limiting what types of relationships can be assessed with the data available.

248

CHAPTER 7

RESULTS

7.1 Main Findings

In this section I examine the main empirical findings aimed at testing the first three hypotheses stated in the introduction to Part III. Specifically, Table 7.1 presents a series of regression models that examines the effect of multiple measures of urban population structure (i.e. primacy, concentration, and hierarchical balance) on political systems for non-core nations (i.e. non-OECD nations) from 1960-2004.

It introduces controls mostly tied to modernization and ecological theories. Table

7.2 presents controls aimed at testing dependency/world-system theory. Both sets of regression models are quite statistically restrictive, and thus offer a highly conservative evaluation of the effects urban population structure has on political systems. The models include controls for panel heteroskedasticity, contemporaneous correlation, heterogeneity, and serial correlation. And as mentioned in the previous chapter, the regression models are based on OLS with panel corrected standard error estimates. They include a lagged dependent variable

(LDV) as well as nation dummy variables to approximate fixed-effects. The dependent variable for all models is a five-year lead of the revised combined polity score.

249

The decision to include a single five-year LDV, as opposed to several (i.e. one for each year of the DV’s lead for a total of five), was determined using a chi-square statistic based on the Breusch-Godfrey (B-G) LM test. This test uses residuals obtained from the regression models to estimate a chi-square test statistic49 which determines whether or not to accept or reject the null-hypothesis of no autocorrelation. Because omitting a LDV from any of the models produced a test statistic that indicated a very serious problem with serial correlation some sort of correction was warranted. Although it was plausible that five separate lags would be required to “correct” the serial correlation, for the most part the B-G tests revealed that the inclusion of a single five-year LDV was enough to limit, and in some case eliminate, serial correlation.

Still, even with a single LDV some of models in Table 7.1 produced chi-square statistics that were significant at the .05 level indicating a moderate level of concern for serial correlation. In other models, specifically those in Table 7.2 and in

Appendix B, the chi-square statistic reached significance at the .01 level indicating a high level of concern for serial correlation. Despite these findings, however, it was decided that since the same LDV included in a majority of the models eliminated or minimized the impact of serial correlation, there was little reason to add additional lags at the risk of overspecification. As a precaution, and for the sake of full

49 The statistic is obtained by running a regular OLS using the residual as the DV and a LDV of the residual at the control for serial correlation. Once run the adjusted R2 is multiplied by NT. This produces the chi-square test statistic which is evaluated using one degree of freedom for every lagged dependent variable included in the regression. For my test I had one degree of freedom and thus values larger than 3.841 were significant at the .05 level. 250 disclosure, the chi-square test statistics for serial correlation are listed in all tables that include a LDV.

Aside from the LDV and the use of panel corrected standard errors, the models also include dummy variables for every nation in the sample to approximate fixed-effects (FE).50 Justification for including fixed-effects estimators was obtained by running fixed-effects models in place of PCSE and assessing the F-test obtained from the Stata output.51 The F-test in Stata (sometimes referred to as a Chow test) is a test that the time independent errors associated with individual units in the sample (e.g. nations) are not significantly different from one another and thus not correlated with the independent variables. When the F-test is not significant a random effects model can be used instead of a fixed effects model (i.e. there are no unit-specific effects). However, in every case the F-test was highly significant (as shown in all the tables with FE estimators) and thus the inclusion of nation dummies was used to control for panel heterogeneity among nations.

Finally, studentized residuals were calculated to detect outliers. Three standard deviations were used as the outlier threshold since political change is a rare event even for a sample including the most politically unstable nations of the world. The residual analysis, while not perfectly consistent from model to model, identified at most 36 nation years as outliers for movements toward democracy (i.e. an r-student above 3) and 20 nation years for movements toward autocracy (i.e. an r-student below -3). Removal of these cases slightly improved the final results in

50 Except for Afghanistan which was the referent nation. 51 It should be noted that these estimates were obtained from a GLS model with fixed-effects and not an OLS model as used in PCSE. 251 some of the models (e.g. urban population growth found significance at the .05 level in a few of the models, particularly with models that included Pi). However, the outliers were included in the final models since their removal could not be justified with consistent empirical support.

The first model in Table 7.1 presents a basic test of the modernization hypothesis regarding the relationship between development and democracy (see

Chapter 4 for a review and discussion of this research). While it is often customary to use a logged version of real gross domestic product per captia (RGDP/c) to correct for skewness, in this case using the logged measure produced inconsistent results in which the signs of the coefficients were reversed (i.e. negative then positive). This runs contrary to past findings and theory. Because the sample excludes developed nations (i.e. OECD nations) and the signs of the coefficients were reversed, the likelihood that multicollinearity was problematic for the logged measures was apparent. In order to assess the level of multicollinearity the VIF was obtained for the log of RGDP/c and its square. The VIF indicated multicolinearity was a serious problem.52 In its place the unlogged measure was use to proxy for level of development which produced significant findings with coefficients in the expected direction.53

52 The VIF statistic for the log of RGDP/c was 179 and 178 for its square. This is compared to the VIF of the untransformed version and its square which had VIF statistics of 4.8 and 4.5 respectively. Given that a VIF above 5 indicates a high level of multicollinearity, it is apparent that the logged version poses a much greater concern. 53 It is also worth noting that the log of development and percent urban (a key measure in these models) was much more highly correlated with each other (r=.83) than the unlogged measure of development and the logged version of development were (r=.80). Moreover, the unlogged version of RGDP/c and percent urban were correlated at a much lower level (r=.66) than the logged version was with percent urban. Although these simple correlations are not too serious, they offer another reason why using the logged version of RGDP/c could have lead to problems of colinearity. 252

253

The second model offers a very simple test of the central argument of this dissertation; that urbanization is an important determinant of political change. The fact that percent urban finds significance and is positive (indicating that increases in percent urban are associated with movement toward democracy) is not all that surprising given that modernization theory has always acknowledged but not stressed this relationship (Lipset 1959). What is surprising, however, is that its inclusion nullifies the effect of level of economic development. Given the statistical scrutiny found in democratization studies over the years this finding was sure to be due to underspecification.

The third model introduces important controls that add greater specification to what role urbanization and urban population structure play in eliciting political change. These specifications include; the urban population growth rate, paved road density, and the log of the total urban population.

Paved road density offers a control for how economically and culturally integrated a nation’s urban areas are. That is, the more interdependent a nation’s population becomes the more transportation networks are needed to maintain efficiency between parties. Although non-significant, the coefficient in model 3 indicates that for every logged unit increase in road density there is a 1.378 point increase in the polity score.

The inclusion of the log of the absolute size of the urban population serves a dual purpose. First, it offers the simplest measure of how large a nation’s “modern” population is. In other words, people living in urban areas are more likely to be involved in economic and social relationships that are supported by modern

254 production systems and economies. Thus, the log of the urban population is a population estimate (as opposed to an economic estimate) of how “modern” a nation is. Second, it offers a test of the classical predictions of Spencer and

Durkheim as well as ecological theory which all stipulate that the size of a population matters for how it connects to its environments. And indeed, model 3 lends support to these hypotheses which predict size forces political change. For every 1 logged unit increase in the log of the urban population there is a 2.526 increase in the polity score. Not only is this coefficient statistically significant at the

.01 level, it is a relatively large increase given the range and standard deviation of this variable (see Table 6.1).

In addition to the size of the urban population, the inclusion of urban growth is an important control for all development theories. It serves as a proximate control for a nation’s experience with the urban transition, especially for newly urbanizing nations. That is, nations with a shorter history of urbanizing are more likely to have much higher rates of urban population growth than those who have a longer history with the urban transition. From a theoretically standpoint, all perspectives expect rapid urban population growth to be problematic for the economic and political systems of developing nations which struggle to integrate new population into the current economic, political, and cultural systems (see

Chapter 3). In fact, every theory discussed in this dissertation suggests that rapid population growth should have a negative effect on democratization because the strain it creates on social institutions and economic/political inequality it promotes

255 favors oppressive regulation. Thus, its inclusion is necessary to make sure that the other urban variables are not being driven by the effects of rapid urban growth.

The results in the third model show how further specification of urbanization’s contribution to development parallels economic controls and does not nullify them. This is apparent in the development coefficients which returns to a high level of significance and in the expected direction (positive then negative). The main coefficient for RGDP/c suggests that for every $1000 increase, the revised combined polity score moves 1 point toward democracy. However, this effect is slightly weakened at higher levels of RGDP/c as the negative squared coefficient indicates. In model 3, the inflection point suggests that RGDP/s stops contributing to movement toward democratization when it reaches $38,038. This is a somewhat high (but not unexpected) level to reach for this sample of nations since the average

RGDP/c is $3,954. Thus, for only a few of the most economically developed nations in these samples a threshold exists. For the majority of nations RGDP/c continues to be associated positively with democratization which is what theory and past findings would predict.

In addition to RGDP/c, percent urban remains significant and positive even in the presence of other controls.54 The positive and significant coefficient suggests that for each additional percent increase, the combined polity score raised .265 points. In other words, for every 4% increase in percent urban there is a 1 point increase in the combined revised polity score.

54 One key variable from the political modernization perspective that has found support throughout democratization studies is secondary education. This measure was removed from the analysis because it lacked coverage and never achieved significance. See Appendix B Table 7.1.1 for the results of its inclusion.

256

As for the other variables, urban growth produces a coefficient that is in the expected direction but it fails to reach significance. The log of the urban population is positive and strongly significant giving support to notion derived from classical and ecological theory that absolute size of a nation’s “modern” population should have positive effects on democratization.55 Indeed, this control remains a key component for all models in all tables and remains highly significant throughout.

Finally, road density produces a positive coefficient but never reaches significance.

The fourth and fifth models introduce the first proxy for urban concentration

(Px) which is a key dimension of urban population structure. Although Px captures how concentrated the total urban population is within the largest city, it ignores the share of percent urban in other elements within the same system (e.g. the second, third, fourth, etc., largest cities in the urban system). Therefore, it remains a weaker test of the null hypothesis since it ignores the structure of the urban hierarchy. However, it does offer a simple test of H1a as well as a good starting point for understanding the effects of urban concentration and primacy on political systems. And indeed, in the fourth model Px produces a negative and highly significant coefficient indicating that increased urban concentration contributes to

55 Note that the log of total population and log of the largest city produced similar positive highly significant results suggesting that all three were keying in on a size effect that was positively related to democratization. And while each is a slightly different measure, they are all highly correlated. The log of the largest city and the log of the total population had the lowest correlation (r=.80), while the log of the largest city and the log of the total urban population the highest (r=.94), with the log of the urban population and total population in between (r=.89). Because the log of the urban population correlated with the log of the total population at such a high level, it was decided that the log of the urban population was more meaningful as a measure of size in these models than log of the total population. However, it should be noted that the log of the total population performed about as well and thus contradicts Dahl and Tufte’s (1973) assertion that absolute size of the population has no effect on democratization.

257 movements toward autocracy or away from democracy (at least for this sample of nation for these years) lending support to both schools of thought

The fifth model offers the first test of the competing predictions between modernization/ecological theory and dependency/world-system theory. If we recall H1b stated that net of percent urban the effects of centralization within the urban system should still favor movement toward autocratic governance. Thus, when interacted with percent urban a negative and significant coefficient of the interaction would lend support to dependency/world-system theory’s monotonic prediction. In contrast, a positive and significant coefficient lends support to modernization/ecological theory. As we see the coefficient of interaction is positive and significant. Even so this means little if it takes unrealistic levels of urbanization to reverse the negative effects of urban concentration captured by Px. Therefore an interpretation of the interaction is in order.

As Allison (1999) notes; “When you estimate a model with an interaction term, the first thing you should look at is the p value for the product of the two variables… [If significant, one] can conclude that there is strong evidence in [the] data that the effect of [one variable] depends on the [other]” (p. 167). The main- effect coefficients have a limited but useful meaning (i.e. the effect of X1 when X2 is 0 and vice versa) while the interaction effects tell us how many units of X1or X2 it takes to generate a one-unit increase/decrease in the effect of X1X2. In this case the interaction between percent urban and Px is highly significant and can be used to determine how urbanized a nation needs to become before the negative effect of Px on movements toward democracy is nullified.

258

Based on the coefficient of the main effect, the fifth model shows that for each percentage increase in percent urban the negative effect of Px on the revised combined polity score decreases by .232. At 0% urban one can expect Px to contribute -9.88 points to the predicted revised combined polity score. This means that when a nation's population becomes 42.6% urban the negative effect of Px on democratization is nullified (i.e. the combined interaction and main effects equal 0), while at levels above 42.6% urban Px begins to positively contribute to democratization. Surprisingly this is a low percentage to reach even for this particular sample of non-core nations since the mean percent urban is 42% and the median 40%. This finding presents the first empirical evidence that the effects of urban concentration on political outcomes are not monotonically negative as predicted by dependency/world-system theory.

In an effort to introduce controls for other key dimensions of urban population structure (i.e. hierarchy and system balance) models 6 and 7 use primacy 1-4 (P14) in an attempt to proxy for the effects of size disparity (i.e. hierarchy) within the urban hierarchy. Moreover, models 8 and 9 use El-Shakhs’

(1972) urban primacy index (Pi) which accounts for both concentration and urban system balance simultaneously.

Despite its somewhat simplistic construction, when compared to Px, P14 is not a measure of concentration but rather of size differences within the urban system.

Moreover, unlike Px , P14 considers smaller elements in the urban system. Yet even

P14 is not as refined as El-Shakhs’ urban primacy index which is perhaps the best approximation for both concentration and size bias as one will find. In calculating

259 the average size difference between the largest city in a nation and each of the smaller elements, Pi allows the size of the third, fourth, and fifth elements in the system to affect the final ratio.56 Furthermore, as a measure Pi produces a more refined estimate of urban concentration than Px and does so while simultaneously permitting size imbalance within the urban hierarchy to impact ratio values. For instance, a nation with equally sized primate cities and an equally sized secondary city population will produce exactly the same Px and P14 values. However, for Pi two nations with equally sized primate cities and equally summed secondary cities will not produce the same value when the distribution of the secondary city population is not the same. This is because differences in the sizes of each nation’s secondary cities affect levels of Pi since Pi is an estimation of the average difference between the largest city and each smaller city in the urban hierarchy. Nations with secondary cities closer in size to one another and with the largest city will produce values smaller than nations whose secondary city population is unbalanced (i.e. have greater disparity between population sizes). In short, for Pi the greater the disparity in size between secondary cities and the largest city the higher its value will be. Thus, in terms of how complex each measure of urban population structure is, Px remains the simplest measure, P14 a somewhat more detailed but different estimation, and Pi the most complex since it combines aspects of both Px and P14.

Model 6 show that when P14 is entered into the equation by itself high levels urban primacy contributes to movements away from democracy as indicated by the highly significant negative coefficient of -.356. Given that the highest value of P14 is

56 This is as opposed to P14 which simply sums the second through fifth elements thus eliminating any differences between them which would allow them to affect the measures individually. 260 above 12 and the average is 2 for this sample, this main effect can have an important impact at even moderate levels of primacy (i.e. the average negative effect is -.712 which is almost a full point on the scale and can be as high as -4). But like Px the total urban context in which urban primacy is embedded plays an important role in how balance within the urban hierarchy (here in terms of urban primacy) affects political systems. Thus, interacting primacy 1-4 with percent urban allows one to estimate the effect of size disparity within urban systems.

In model 7 we see that the interaction between P14 and percent urban produces a positive and highly significant coefficient indicating that the effects of urban primacy are conditioned by how urbanized a nation’s population is. An interpretation of the coefficient reveals that, for each percentage increase in urban population, the negative effect of primacy 1-4 on the revised combined polity score is .0198 units. This means that when a nation’s population reaches 50% urban the negative effects of urban primacy are nullified, and at percentages above this urban primacy begins to contribute positively to movements toward democratization.

Using Pi to assess this relationship reveals that for each percentage increase in urban population the negative effects of Pi on the revised combined polity score are decreased by .562 units. This means that when a nation’s population reaches 40.1% urban the negative effects of urban primacy are nullified, and at percentages above this, primacy contributes to movements toward democratization.

Finally, models 10 and 11 offer an alternative to measures of urban primacy and concentration used in the first three sets of interactions. The log of D as discussed in Chapter 6, offers a measure of how log-normal or balanced an urban

261 system’s population is. It is not a measure of size differences like primacy nor is it a simple measure of concentration, although both elements certainly affect values of the log of D. Instead, the log of D is an approximation of how close the first five elements in the urban system approximate an urban system with a log-normal rank- size distribution whose slope is -1. At a value of 0 an urban system is perfectly log- normal. Above 0 the system moves towards a multicentric urban structure (i.e. one with several large cities) and at values below 0 the system moves towards a primate urban structure. As the summary statistics show (Table 6.2), the average urban system for this sample is slightly primate which is in line with the other measures used so far.

Model 10 presents the results when the log of D is included instead of a measure of urban concentration/primacy. Not surprisingly the coefficient is positive and highly significant. Also note that the effects of economic development, size of the urban population, and percent urban all find significance and produce coefficients in the expected direction. Since the log of D is a new measure these consistent findings suggest that it performs in-line with other tested measures of urban population structure.

On this note, model 11 includes the interaction between the log of D and percent urban in an attempt to capture the notion that log-normal urban systems contribute more to movements toward democracy when a higher percent of a nation’s population lives in urban areas. The interaction, however, produces a negative coefficient suggesting that there might be a point at which a population

262 becomes so urbanized that log-normality or a multicentrism begins to work against movements toward democracy. How is this so?

A closer inspection of the coefficient reveals that in order for the above to occur a nation would have to be roughly 730% urban. Obviously this is an extremely out-of-bounds estimate which indicates that there is little reason to interact D with percent urban to capture its effect on political outcomes. It appears that at all levels of urbanization log-normal and mutlicentric urban population hierarchies contribute positively to movements toward democracy.

In conclusion, the results of Table 7.1 offer strong evidence that urban population structure is an important determinant of a nation’s political system.

Economic development generally produced, and confirmed, that its effects are curvilinear in the sense that there is a threshold for how much it can contribute to movements toward democracy. Another consistent finding is that size of the urban population has an impact on positive movements toward democracy. No matter what controls for the urban transition were introduced, the results here clearly indicated that for this sample of nations in these years using very conservative models the size and structure of a nation’s modern (urban) population does matter.

Despite the above findings the effects of road density and urban population growth never reached significance although the coefficients they produced were in the right direction. Finally, Table 7.1 shows that when one considers how percent urban conditions the effects of urban population structure on political outcomes, the results produced are consistent and highly significant no matter what proxy for urban population structure (or urban concentration/primacy) is used.

263

7.2 Alternative Models

The models in the previous section rest heavily on a modernization/ ecological framework. Given this, some may wonder about their robustness when controlling for dependency/world-system variables that are deemed to be important contributors for the centralization of economic and political power within periphery nations. In this section I look to two additional controls that are important for testing the dependency/world-system argument. Specifically, I examine the claim that greater core penetration into the economies of periphery nations distorts political development. The two variables used to proxy for capitalist penetration are exports of goods and services and foreign direct investment (FDI), both measured as a percentage of GDP.

In each case the dependency/world-system framework predicts increases in either should be associated with greater centralization of economic production and political power. The centralization of the means of production and labor within periphery nations should result in the suppression of worker’s rights as well as the concentration of wealth into the hands of the few, making mobilization difficult. In short, the more integrated a periphery nation becomes into the world economy the greater control capitalists from core nations should have over developing economies and their political systems resulting in the monopolization of economic and political power.

Table 7.2 presents the results which test the hypothesis that greater export dependence is associated with movements towards autocratic rule. A complimentary test using FDI as a percent of GDP can be found in Table 7.2.2 in

264

Appendix B. Before reviewing the results, however, it is important to note that the log of D and its interaction with percent urban was omitted from Table 7.2 because the estimated VCE was not positive definite and could not be run. It is also important to note that the number of nations included in both tables is lower than those in Table 7.1. This is largely due to that fact that measures of exports of goods and services as a percentage of GDP are available for fewer nations and for FDI the years covered are only from 1970-2003.57

In light of these sample size differences, and in order to allow for broader comparison, the regressors from Table 7.1 were run using the smaller sample of nations from 7.2.58 The results appear in Table 7.2.1 in Appendix B. The results from the regression in Table 7.2.1 are somewhat different from those in Table 7.1.

These same differences are reproduced in Table 7.2 and therefore are worth exploring further before interpreting the results in Table 7.2.

The most important difference between Table 7.1 and Table 7.2 is that road density produces a negative coefficient instead of a positive one in Table 7.2 but it never reaches significance. Because it was possible that one of the four excluded nations could have been an outlier and responsible for this subtle difference, a third table (Table 7.2.3) omitting the missing nation of Afghanistan, Albania, Former

Ethiopia, and Iraq was run using the same regressors from Table 7.1. Table 7.2.3 in

Appendix B shows that the omission of these nations has no bearing on the results

57 For a complete list of nations for each set of models see Appendix B. 58 The sample size for 7.2b was used because it produced significant coefficients and coefficients that were closest to those in 7.1.

265

from Table 7.1, and indeed road density returns to producing positive non- significant coefficients with this sample of nations. This suggests that it is the

266 specifications used Table 7.2 and not the sample that produces the negative coefficients for road density.

Finally, aside from this minor discrepancy, one must look to the results in both Table 7.2 (export dependence) and Table 7.2.2 (FDI) with some level of skepticism concerning significance levels. This is because serial correlation is shown to be an issue of concern even with the inclusion of a LDV. However, how much of a role serial correlation plays in affecting standard errors and t-scores in these particular models cannot be determined since the B-G LM test used is a detection tool and not as a judge.

The results from Table 7.2.2 in Appendix B clearly shows that FDI never reaches significance, although the coefficients are in the hypothesized direction.

Like Table 7.2.1 road density turns negative but never reaches significance.

Moreover, percent urban alternates between positive and negative values. Despite these key differences the effects of economic development, the log of the urban population, and the interactions between the measures of urban population structure and percent urban remain significant while producing coefficients and percent urban thresholds similar to those in Table 7.1. The major difference is that the threshold for Px lowers to 33.7%, P14 lowers 43.3% urban, and the threshold for

Pi lowers to 38.5% urban. Aside from these few observations very little can be said about or concluded from Table 7.2.2 since FDI does not appear to impact political regime type.

As for Table 7.2, the results are much more consistent and interesting although a minor difference with Table 7.1 remains and serial correlation appears to

267 be an issue (see above). Besides the different sample sizes, like Table 7.2.2, in Table

7.2, road density produces negative coefficients. However, because they never reach significance they do not appear to be a major area of concern for interpreting the rest of the model. This is because the key controls (economic development, the log of the urban population, and every interaction between percent urban and measures of urban population structure) remain significant and produce coefficients that are in the same directions as those in Table 7.1. Given these key similarities and the fact that exports consistently reaches significance at both the .05 and .01 levels, Table 7.2 offers a much clearer evaluation of dependency/world- system predictions. However, the effect of exports on the combined revised polity score is very modest. If we average the coefficients from these tables, the mean coefficient is roughly -.021. This means that for every percent increase in the percent of a nation’s GDP that comes from a nation’s exports, that nation’s polity score moves .021 points toward autocracy. In short it takes about 50% of a nation’s

GDP to come from exports before it moves the nation one point toward autocracy.

When compared to the effect the urban coefficients had in the previous models, these movements are very small, very modest, and very rare even for periphery nations.

When included in the models from Table 7.1, exports as a percent of GDP proves to be an important and robust predictor of political movements toward autocratic government. This indicates that there is some evidence to support dependency/world-system theory which asserts export oriented economies favor autocratic rule, and perhaps due to the symbiotic relationship formed between

268 foreign economic elites (core capitalists) and domestic political elites (dictators, chiefs, kings, etc.).

While the above finding is important, at the same time it is worth pointing out that despite the inclusion of this particular proxy the positive and significant effects of the urban controls do not change in any substantive way, and therefore the support of these findings for dependency/world-system theory may be limited. For instance, export oriented economies may favor autocratic regimes as dependency/world-system theory predicts, but the effects of urban population structure and modern population pressures appear to impact political outcomes regardless of export dependence and FDI. Therefore, while export dependence may favor autocratic governments, contrary to the dependency/world-system arguments, this effect is not monotonic and indeed appears to be limited by the very process that makes it possible. In other words, the manufacturing of goods and services in modern economies requires urban areas for production and export into the world economy. As the production of export goods and services become an important component of a given periphery economy, both modernization/ecology and dependency/world-system theories expect increased urban primacy, higher degrees of urban concentration, and larger urban populations. The problem for dependency/world-system theory is that each of these factors has been shown to be linked to movements towards democratization over time (see Table 7.1). Moreover, the strength of the interactions between percent urban and measures of urban population structure maintain significance and low threshold levels (for Px 51% urban, for P14 66% urban, and for Pi 41.5% urban). Therefore, despite the finding

269 that export dependence contributes to movements toward autocracy, emergent properties associated with modern population pressures stemming from the urban transition (a process that is necessary for the continued increase of exports as a percent of GDP) appears to override core capitalists’ interests. In other words, autocrats may prosper from capitalist penetration into developing nations early on by producing disarticulated or primate urban systems, but as those urban systems mature/grow the emergent properties stemming from the size and distribution of urban populations begin to work against a centralized autocratic government’s ability to regulate complex urban populations.

7.3 Exploratory Models

In this section I present several “relaxed” models with the goal of exploring other intervening and contributing relationships that could not be explored in previous models due to the inclusion of country fixed-effects estimators. When used in regression models, fixed-effects estimators make it impractical to include

(other) dummy variables as well as time-invariant controls (e.g. land area) because such measures have a tendency to exhibit perfect colinearity when placed into the same equation (Frees 2004). However, even when fixed-effects dummies are removed another problem arises—at least for an analysis of the dependent variable used in this study.

In the absence of fixed-effects the variance being analyzed in the regression models moves from an assessment of within nation variance to both within and between nation variance (i.e. random effects). In most cases an assessment of both between and within unit variance is not problematic and may even be desirable.

270

However, as mentioned above for this particular study, assuming random effects presents a problem for the inclusion of a LDV. This is because change in political systems (the dependent variable) is a near-rare event and thus when a LDV is included it alone explains a majority of the between nation variance. This makes it very difficult to assess the impact of other independent variables have with respect to between nation variance, a key component of random effects analyses.

The problem of including a LDV holds true even with a 5 year lead of the dependent variable is used. In fact, running a random effects regression that includes the LDV as the lone independent variable results in a between R2 of .93 and a within R2 of .29.59 As one can tell from this result, the inclusion of an LDV leaves very little room for other independent variables to contribute to the explained variance in the model. Therefore, in order to maximize the analysis of dummy and time invariant variables, the models presented in this section omit both LDV and FE estimators. However, since panel-corrected standard errors have no ill effect on the estimates (and the properties of OLS are desirable for TSCS data over GLS—see Beck and Katz [1995,1996,2004]) the models maintain this approach to control for panel heteroskedasticity and contemporaneous correlation. Still, the reader should be aware that the models presented in this exploratory section do not control for heterogeneity (i.e. country fixed-effects) or serial correlation. As a result, findings should be viewed with greater caution than those presented in first two sections.

59 It should be noted that these estimates were obtained from a GLS model with random effects and not an OLS model. Therefore, the results of the R2s are not directly applicable to the OLS models with panel-corrected standard errors that are used in this section. However, the estimates do provide a close approximation of what the within and between explained variance is for the OLS models when a LDV is included in the equation.

271

This is because they are more prone to underestimating standard errors leading to overconfidence in significance levels (see Chapter 6 for a discussion).

As discussed already, one of the benefits of relaxing the restrictions on the panel-corrected OLS models is that it allows one to test dummy variables and time- invariant measures. In Chapter 5 there were several relationships discussed which either requires a time-invariant measure (e.g. size of the nation) or a dummy variable to approximate (e.g. colonial history). The two most important of these measures for dependency/world-system theory are a nation’s history in the world economy, especially with core nation penetration, and a nation’s position in the world-economy. Each can be approximated with dummy variables. In all there are three dummy variables and two complimentary variables which can be used to further assess the dependency/world-system theory predictions concerning historical effects. These variables include a dummy variable for colonial history, a dummy variable for communist history, a dummy variable for direct rule (i.e. nations who were colonized by either France, Spain, Portugal or Belgium; Lyman

1992), a measure of colonial duration (i.e. how long a nation was colonized) and a measure of communist durations (i.e. how long a nation was under communist rule).

In order to better assess position in the world economy I introduce three dummy variables for primary commodity export nations. Primary commodity export (PCE) nations tend to hold a lower position within the world economy due to their export-dependent economies which provide core regions with raw materials

(e.g. African nations). Thus, a dummy for PCEs further approximates export

272 dependence and position within the world system. The first PCE dummy is for fuel- only exporters, the second for non-fuel exporters, and the third is a dummy for both.

In terms of predictions, dependency/world-system theory argues that each measure of colonialism should be associated with autocratic governance. This is because nations with colonial histories should have undeveloped and disarticulated economies and be more dependent on core economies. Both historical circumstances should have resulted in greater economic and political inequality (see

Chapter 3). Similarly, the longer a nation was colonized the more distorted its economic and urban systems should be—both conditions that favor autocratic rule.

Moreover, both conditions place the masses in periphery nations at a disadvantage for establishing self-rule since they do not have the resources to mobilize. Similarly, nations with a communist history should be more autocratic since they have already provided nations with a history of centralized economic and political institutions as well as political elitism. Finally, nations that are primary commodity exporters should have a very low position in the world economy, and thus have the most underdeveloped economic and urban systems due to an on-going history of exploitation by the core. Such exploitation should contribute to the monopolization of economic and political power thus favoring centralized urban/economic systems as well as centralized political rule (i.e. autocracy).

As for the effect of direct rule (Lyman 1992) there are two competing hypotheses which can be established. From the dependency/world-system perspective, if what Lyman observed holds true (i.e. nations with direct rule aided in the creation of primate urban systems), then nations with direct rule histories

273 should tend toward autocratic governance. Conversely, from the modernization/ecological perspective, nations with a history of direct rule should have a longer history with primate urban systems. Primacy may be a pre-condition for establishing a mature economic system (Crenshaw and Oakey 1998). Thus, these nations may have an “urban advantage” in that they have developed mature urban systems whose interdependent economies eventually promote economic decentralization and democratization. In short, from the modernization/ecological perspective there is a real possibility that direct rule, over time, has indirectly contributed to economic and demographic conditions that are favorable to democracy via its effect on urban structure.

Aside from introducing several dummy variables to make further assessments of dependency/world-system theory, the relaxed models also allow one to assess the effects of size (in terms of area) on democratization. Finally, to better assess ecological theory I constructed a composite measure of climate/geography to observe the impact that ecumenic environments have on political forms. The measure used, what I call “no urban” which is the sum of the percent of a nation’s area that consists of arid, semi-arid, polar, and mountainous

(highland) regions. Such regions are hostile to urbanization because they present populations with several problems for maintaining high levels of density, most of which have to do with a dense population’s ability to connect to environment and distribute energy efficiently enough to support itself.

From the ecological perspective the more a nation’s areas is composed of these hostile regions, the more concentrated its population should be and thus the

274 more centralized its economic and political systems should be. This array of concentrations, in the absence of opportunities to expand, may present populations with a state of near permanent centralized rule. In fact, at very high percentages of these climates it may be difficult if not impossible for a population to decentralize its production. In short, some degree of autocratic rule may be inevitable in climates with highly restrictive living opportunities. In addition, because living opportunities are fewer and more homogenous, we should expect regions dominated by hostile climates to have more ethnic and cultural diversity but less economic diversity— both due to the lack of resources in environment. Thus, it should be very difficult for disparate people to integrate economically keeping them from forming the cross-cutting affiliations that foster democratization. In short, nations with a high percentage of “no urban” lands should tend heavily toward autocratic rule because they have disintegrated economic systems and tend toward urban primacy without the possibility of “growing” out of it.

Finally, in terms of area one is inclined to expect large nations to lean democratic for the reasons discussed by Dahl and Tufte (1973) which were reviewed at the end of Chapter 5. However, because this sample excludes developed nations and especially large resource-rich nations that are very democratic, it may be that the largest nations left in the sample are composed of large resource poor nations. In such an instance one would expect size to be associated with autocratic rather than democratic governance.

Table 7.3 presents the results of the inclusion of PCE dummies, the log of area, and the percent of a nation’s area composed of hostile urban climates. Table

275

7.4 presents the historical variables and then combines them with the controls in

Table 7.3. As for Table 7.3 it is important to note that the controls for urban population structure/urban primacy are limited to Pi and the log of D. These limitations were mainly due to the fact that Px and P14 performed almost identically to Pi. It is also important to point out that economic development and its square never achieve significance in any of the models contained in this section. As shown in Table 7.3.1 in Appendix B, this is not due to Pi or the log of D and it is not due to the time invariant or dummy variables. Rather, it appears to be the result of the shift in analysis of the variance from within nation to within and between. As evidenced in Table 7.3.2, when the square is removed economic development becomes negative and significant—a contradictory finding. While the removal of the square has no bearing on the outcome of the other variables, it does diminish the R2 indicating that the inclusion of the square is a better “fit.” This too is consistent with past findings. Given this, it was decided that for the sake of comparison across models the inclusion of the square is a better “fit” and more consistent with past findings and therefore should be retained. With this being said let us turn to the results in Table 7.3.

Model 1 in Table 7.3 reproduces the base model used in all the previous models. As one can see the key difference produced, which remain fairly consistent throughout, is that economic development is not significant but its square is. Meanwhile, both urban population growth and road density are highly significant and in the hypothesized directions. That is, increases in urban population growth are associated with movements toward centralized governance while increases in road

276

density are associated with decentralization in polity (i.e. democratization). The fact that these two measures maintain such a high level of significance regardless of other controls seems to indicate that the very conservative estimates in the other analyses may have been either suppressing these relationships or they are mainly

277 driven by between nation variance. In either case it appears that in future considerations the effects of both population growth and road density should continue to be considered when exploring political regime change.

As for the other familiar measures, the two controls for urban population structure are less robust and somewhat inconsistent, at least for Pi. Not only does the interaction between Pi and percent urban lose significance in a few instances, the threshold for the interaction is much higher than in past models, in most cases it is above 85% urban. However, the log of D performs well and remains fairly consistent with previous findings. Still, given the introduction of between nation variance and in the absence of a control for serial correlation very little should be concluded about these inconsistent findings when compared to those of Table 7.1 and 7.2.

Looking to the new controls introduced in Table 7.3 several interesting patterns emerge. For instance, the log of area consistently performs well producing a negative coefficient that remains highly significant indicating that large nations (at least in this sample) tend toward autocratic governance. Similarly, the control for hostile urban climates performs even better achieving high levels of significance in both Table 7.3 and 7.4. Finally, controls for PCEs offer mixed results, although the dummy for fuel primary commodity exporters consistently achieves significance indicating that fuel production has an important effect on how primary commodity exporting affects political systems (i.e. state controlled wealth with little to no diversification in the economic system). In fact, in model 5 it nullifies the interaction between Pi and percent urban but fails to effect the interaction between

278

279 the log of D and percent urban. Thus, aside from offering some empirical support for dependency/world system theory as well as ecological theory, not much else can be concluded from Table 7.3.

Table 7.4 introduces several measures aimed at exploring the effects of colonial and communist histories. Again, because all measures of urban population structure performed similarly or as expected (in the case of the log of D) Pi was used as the lone control for brevity. Models 1 through 5 look at colonialism and communism independent of the controls introduced in Table 7.3. What they show is that each obtains significance but not always in the expected direction. For instance, the dummy for direct rule is positive and significant. Moreover, in contrast to dependency/world-system theory, both the dummy for colonialism and the measure of duration are highly significant and positive. Not surprisingly, the dummy and measure of duration for communist histories produce significant and negative coefficients.

When combined with the previous measures in Table 7.3, the measures of communist history and duration both maintain their significance. As for the colonial measures the dummy turns negative but it is not significant while the colonial duration measure remains positive and highly significant. The same is true of the direct rule dummy which remains positive and highly significant.

Finally, looking to model 11 we see the results when all the controls are considered. While there are only a few changes to what has been previously discussed, what changes arise are worth pointing out. First, communist duration becomes positive but not significant. Second, colonial history becomes negative and

280 significant. Thus, from the models presented here whether or not colonial history promotes democracy or autocracy is not all that clear once other important controls are accounted for. However, as stated repeatedly throughout this section, the models are intended to be exploratory not explanatory and therefore the effects of colonialism are neither ruled out nor accepted. However, both climate and area appear to be consistent performers indicating that their effects are robust and meaningful.

281

CHAPTER 8

CONCLUSION

This dissertation presents empirical evidence that population pressures associated with urban transitions have direct and meaningful effects on political systems within nations. Specifically, it presents strong evidence that net of other variables urban concentration, urban primacy, and urban hierarchical structure are associated with within-nation political regime change. With this in mind, I conclude my dissertation with a summary followed by a discussion of the main empirical findings as well as what they mean for future cross-national comparative studies of political change and development.

8.1 Summary

My dissertation started with a simple question: Why have cross-national development studies ignored the role of urban population when explaining political change? In seeking an answer, in the first chapter I looked to classical explanations of social change which focused on the role population growth plays in shifting societies from traditional to modern forms of organization. However, as pointed out in Chapter 1, an important exception to this approach was Marxian theory. Despite recognizing the existence of population pressures (and contrary to Spencer and

Durkheim), Marx and Engels emphasized the role of economic elites and the means

282 of production as the primary sources of social change. Thus, instead of emphasizing population as a cause for emergent social change it was emphasized as an outcome of capitalists’ interests.

In Chapter 2, I extended my focus on population to contemporary understandings of modern development within nation-states and within a global economy. Because modernization and ecological theory both view population growth and population pressures similar to Spencer and Durkheim (i.e. as central mechanisms of macro social change), it was argued that both theories could be grouped together. Similarly, the dependency and world-system theories were both grounded in the Marxian view that economic elites in a global capitalist economy shape development. Moreover, both tend to view modern population arrangements within nations as outcomes of capitalist interests which also reflect the economic hierarchy in the world economy. Thus, it was concluded that dependency/world- system theory could be grouped together.

In Part II of this dissertation I looked more closely at how comparative development studies have overlooked the relationship between urban population and political outcomes. This was shown to be the case despite paying close attention to key processes (e.g. economic development) which parallel urban transitions. Specifically, in Chapter 3 I looked to the theory and empirical literature which has focused on the reciprocal relationship between urbanization and economic development. I explored how urban transitions are intimately connected to, and perhaps inseparable from, modern production systems. Further, I discussed how modernization/ecological theory and dependency/world-system theory have

283 been used to explain urbanization as an outcome of the rise of modern production systems. Moreover, I explored how each perspective explains urban imbalances that arise during the development process. Modernization/ecological theory tended to explain urban imbalances as temporary situations that accompany the total development process while dependency/world-system theory tended to see urban imbalances as inevitable outcomes of a world economy driven by capitalist production.

In the fourth chapter I continued the exploration of parallel processes tied to development. Specifically, I looked to democratization studies which linked level of economic development to political systems and vice versa. I looked at those studies which argue economic development causes democracy and those which argue democracy causes economic growth. Moreover, I explored how democratization studies have removed the urban component from conceptions of development, and thus, have failed to account for the unique effect urban populations have on political change.

In Chapter 5 I reviewed what literature exists that has tested or considered the relationship between urbanization and political systems. Not surprisingly, what

I found were only a handful of studies which explored how political systems affected urbanization, urban growth, and urban concentration and most ignored reciprocal causation (i.e. urban systems affecting political outcomes). I then turned to an examination of a few empirical studies that could be used to support the notion that urban population structure and urban population pressures have a direct impact on political outcomes.

284

Finally, in Part III of this dissertation I used theory and empirical findings to compose hypotheses which were aimed at testing whether or not urban population structure and modern population pressures affected political systems. In Chapter 6

I discussed the measures and methods that were used for the statistical analyses, and in Chapter 7 I presented the empirical findings. In the section below I offer a discussion and conclusion of these findings.

8.2 Discussion

This dissertation set out to answer the question: Does urban population structure and/or urban population pressures play a role in macro political change within nation-states? Now that empirical evidence has been generated it is possible to answer this question with a “yes.” While this is the simple answer, the data analysis offer a more telling and interesting response that is worth elaborating upon.

As for the null hypothesis, that urban population structure has no effect on political systems, the evidence clearly indicates a strong and confident rejection.

Not only did the results show a consistent and persistent relationship between urban population structure and political change, it did so using a variety of urban measures. For instance, Table 7.1 tested three measures of urban population structure (i.e. concentration, primacy, and hierarchical structure) with each producing highly significant coefficients. In addition the log of the urban population performed even more consistently, showing that urban population pressures generated from population size contribute to political change, specifically to democratization, in a meaningful way. And while this finding confirms the fourth

285 hypothesis (H2a), it needs to be explored further in future analyses to establish exactly how urban population size contributes to collective behaviors that evoke political change (e.g. coups, riots, demonstrations, terrorism, protests, etc.).

However, the strength of this initial finding should be a clear signal to those interested in macro cross-national political change that urban population matters.

As for the second and third hypotheses aimed at testing whether or not centralization within the urban system is monotonically related to movements towards autocracy the results are clear. The empirical evidence indicates that contrary to dependency/world-system theory continued growth (either in terms of size or concentration within the largest urban area in the urban system) does not indefinitely contribute to autocratic rule. When considering the context in which the concentration of urban populations occurs (i.e. percent urban), it is apparent that there is a real threshold for how much concentration and/or primacy can support autocratic rule before its emergent properties begin to act against centralized and oppressive political systems. Indeed, the empirical findings consistently showed that at around 40-50% urban, urban concentration/primacy starts to work in favor of democratization. This means that an urban system can remain primate but contribute to democracy if a large enough percentage of the population resides in urban areas (i.e. engaged in modern economic systems which promote interdependency and interest group formation). In addition, the results show that at all levels of urbanization log-normal urban systems promote democracy.

286

What these findings suggest, along with the consistent finding that the absolute size of the urban population contributes to movements toward democracy, is that maturation of a nation’s urban system is an important determinant of democratization. That is, as a nation’s largest city achieves primacy, its modern economic system becomes so overpowering that is places strain on rural populations to either contribute to its dominance or become part of it. As a greater percentage of a nation’s population becomes “modern,” and in the processes begin to contribute to economic, cultural, and political plurality, a nation’s urban population begins to create the social institutions and economic relationships that are necessary to support democracy and that undermine autocracy. The importance of such a finding for future assessments of democratization cannot be overstated. The evidence clearly supports the modernization/ecological argument that a maturation effect is a key part of the development/democracy relationship.

In sum, this dissertation presents strong evidence that changes in urban systems and urban population are an important part of this maturation effect.

Although the tests of the main hypotheses strongly supported modernization/ecological theory, this does not mean the basic claims of dependency/world-system theory can be dismissed. The empirical results also produced evidence that a nation’s role in the world-economy impacts the type of polity it has; although these effects did not override the impact of urban population.

For instance, Table 7.2 consistently showed that export dependent nations were more likely to be autocratic. This finding supports to the notion that a nation’s place in the world-economy affects political relations within that nation. And although

287 less precise, the exploratory analyses also consistently showed that primary commodity exporters were more likely to be associated with autocratic governments—especially those that exported fuels. Less clear, however, were the findings concerning the effects of colonialization; for they tended to contradict the dependency/world-system perspective. Having a colonial history appears to be associated with democratic governance and the longer the history the stronger that association appears to be. Still, given the lack of restrictions placed on these models nothing conclusive can be taken from them.

In conclusion, what can be taken from this dissertation is the following: First, there has been a clear tendency within comparative development studies, from a variety of disciplines, to ignore the role of urban populations and urban systems in political development; particularly with respect to the impact urban populations play in the determining the type of government in place. As was argued and illustrated in Part II, cross-national comparative development studies have tended to over-emphasize the role of the level of economic development when explaining political change (perhaps due to data availability) and under-emphasized other key factors of the total development process such as the role of urban populations and their distribution within nation-states. Second, when looking at the effects of urban population structure and urban population pressures on macro political change, it is clear that each plays an important and undeniable role in eliciting change within political systems—even when one controls for level of economic development. In fact, for the sample of nations used in the empirical analyses urban population controls performed as good or better than level of economic development. With this

288 being said, it is hoped that the contents of this dissertation and the empirical evidence generated will be used to awaken comparative development scholars in all disciplines to the importance of urban population for explaining political change in an increasingly urban world.

289

APPENDICIES

290

APPENDIX A

DATA AND MEASURES

Nations and Years of Urban Dataset*

UNCC Nation Formerly Years

4 Afghanistan 1960-2005 8 Albania 1960-2005 12 Algeria 1960-1998 24 Angola 1960-2005 31 Azerbaijan USSR 1990-2005 32 Argentina 1960-2005 36 Australia 1960-2005 40 Austria 1960-2004 50 /East 1970-2005 51 Armenia USSR 1990-2005 56 Belgium 1960-2005 68 Bolivia 1960-2005 76 1960-2005 100 Bulgaria 1960-2005 104 Myanmar/Burma 1960-2005 112 Belarus USSR 1990-2005 116 1960-2005 124 Canada 1960-2005 140 Central African Republic 1960-2005 144 Sri Lanka 1960-2005 148 Chad 1962-2005 152 Chile 1960-2005 156 1960-2005 170 Columbia 1960-2005 178 Congo Republic 1961-2005 180 Democratic Republic of the Congo 1960-2005 188 Costa Rica 1963-2005 191 Croatia Yugoslavia 1991-2005 192 Cuba 1964-2003 200 Czechoslovakia, Former United 1960-1989 203 Czech Republic 1990-2005 204 Benin 1960-2005 208 Denmark 1960-2005 214 Dominican Republic 1960-2005

291

(cont.)

218 Ecuador 1960-2003 222 El Salvador 1960-2005 230 Ethiopia, Former 1960-1991 231 Ethiopia 1992-2005 232 Eritrea Ethiopia 1992-2005 246 Finland 1960-2000 250 France 1960-2005 268 Georgia USSR 1990-2005 276 Germany (United) 1990-2005 278 German Democratic Rep (East) 1960-1989 280 Federal Germany (West) 1960-1989 288 1960-2005 300 1961-2001 320 Guatemala 1964-2002 324 Guinea 1960-2005 332 Haiti 1960-2005 340 Honduras 1960-2005 344 Hong Kong (China) N/A 348 Hungary 1960-2005 356 1960-2005 360 1960-2005 364 Iran 1960-2003 368 Iraq 1960-2003 372 Ireland 1961-2005 376 Israel 1960-2005 380 Italy 1960-2005 384 Ivory Coast 1960-2005 392 Japan 1960-2005 398 Kazakhstan USSR 1990-2005 400 Jordan 1960-2004 404 Kenya 1962-1999 408 North Korea N/A 410 South Korea 1960-2005 414 1960-2005 417 Kyrgyzstan USSR 1990-2005 422 Lebanon 1960-1974, (2005) 428 Latvia USSR/Germany 1989-2005 430 Liberia 1961-2005 434 Libya 1960-2003 440 Lithuania USSR 1989-2005 450 Madagascar 1960-2005 454 Malawi 1960-2005 458 1960-2000 466 Mali 1960-2005 478 Mauritania 1960-2005 484 Mexico 1960-2005 496 Mongolia 1960-2005 498 Moldova USSR 1990-2005

292

(cont.)

504 Morocco 1960-2005 508 Mozambique 1960-2005 524 Nepal 1960-2005 528 Netherlands 1960-2005 554 New Zealand 1960-2005 558 Nicaragua 1960-2005 562 Niger 1960-2005 566 Nigeria 1960-2005 578 Norway 1960-2005 586 Pakistan 1960-2005 591 Panama 1960-2005 600 Paraguay 1962-2005 604 Peru 1961-2005 608 1960-2000 616 1960-2005 620 Portugal 1960-2005 642 1960-2005 643 Russian USSR 1990-2005 682 Saudi Arabia 1960-2005 686 Senegal 1960-2005 703 Slovakia Czechoslovakia 1990-2005 704 Viet Nam 1960-2005 706 Somalia 1960-2005 710 South Africa 1960-2005 716 Zimbabwe 1960-2005 724 Spain 1960-2005 736 Sudan 1960-2005 752 Sweden 1960-2005 756 Switzerland 1960-2005 760 Syria 1960-2005 762 Tajikistan USSR 1990-2005 764 1960-2002 768 Togo 1960-2002 784 United Arab Emirates 1970-2005 788 Tunisia 1960-2004 792 1960-2005 795 Turkmenistan USSR 1990-2005 800 Uganda 1960-2005 804 Ukraine USSR 1990-2005 807 Macedonia 1991-2005 810 USSR 1960-1989 818 Egypt 1960-2005 826 United Kingdom 1960-2005 834 Tanzania 1960-2002 840 United States of America 1960-2005 854 Burkina Faso 1960-2005 858 Uruguay 1960-2004 860 Uzbekistan USSR 1990-2005

293

(cont.)

862 Venezuela 1960-2005 887 Yemen 1960-2005 889 Yugoslavia, Former United 1961-1991 890 Yugoslavia Federal Republic 1992-2002 891 Serbia and Montenegro Yugoslavia N/A 894 Zambia 1960-2005

*Note that the above sample reflects the maximum number of nations and year spans for the urban data and the Polity IV data (described below). It does not reflect the actual nations and years observed in the analyses found in Chapter 7. For a complete list of nations used in the statistical analyses, see Appendix B.

294

Polity IV Dataset (Dependent Variable)

Note: The excerpt below is taken directed from Polity IV Project: Political Regime Characteristics and Transitions, 1800-2007, Data Users’ Manual by Monty G. Marsha and Keith Jaggers (2009), pages 13-16.

2.1 DEMOC (all versions) Institutionalized Democracy: Democracy is conceived as three essential, interdependent elements. One is the presence of institutions and procedures through which citizens can express effective preferences about alternative policies and leaders. Second is the existence of institutionalized constraints on the exercise of power by the executive. Third is the guarantee of civil liberties to all citizens in their daily lives and in acts of political participation. Other aspects of plural democracy, such as the rule of law, systems of checks and balances, freedom of the press, and so on are means to, or specific manifestations of, these general principles. We do not include coded data on civil liberties. The Democracy indicator is an additive eleven-point scale (0-10). The operational indicator of democracy is derived from codings of the competitiveness of political participation (variable 2.6), the openness and competitiveness of executive recruitment (variables 2.3 and 2.2), and constraints on the chief executive (variable 2.4) using the following weights:

Authority Coding Scale Weight

Competitiveness of Executive Recruitment (XRCOMP): (3) Election +2 (2) Transitional +1

Openness of Executive Recruitment (XROPEN): only if XRCOMP is Election (3) or Transitional (2) (3) Dual/election +1 (4) Election +1

Constraint on Chief Executive (XCONST): (7) Executive parity or subordination +4 (6) Intermediate category +3 (5) Substantial limitations +2 (4) Intermediate category +1

Competitiveness of Political Participation (PARCOMP): (5) Competitive +3 (4) Transitional +2 (3) Factional +1

This "institutional democracy" indicator follows a logic similar to that underlying the Polity I analyses. There is no "necessary condition" for characterizing a political system as democratic, rather democracy is treated as a variable. For example, the scale discriminates among Western parliamentary and presidential systems based on the extent of constraints on the chief executive. Charles de Gaulle as president of

295 the French Fifth Republic operated within slight to moderate political limitations. Thus the early years of the Fifth Republic have lower Democracy scores than the United States or the Federal Republic of Germany, where constraints on the executive approach parity. Similarly, the onset of "cohabitation" in France during the second phase of the first Mitterrand presidency is marked by a shift toward parity on the Executive Constraints scale and a concomitant increase in France's Democracy score. If the composite indicator of institutionalized democracy is inappropriate for some conceptual purposes, it can be easily redefined either by altering the constituent categories and weights, or by specifying some minimum preconditions. A mature and internally coherent democracy, for example, might be operationally defined as one in which (a) political participation is unrestricted, open, and fully competitive; (b) executive recruitment is elective, and (c)constraints on the chief executive are substantial.

2.2 AUTOC (all versions) Institutionalized Autocracy: "Authoritarian regime" in Western political discourse is a pejorative term for some very diverse kinds of political systems whose common properties are a lack of regularized political competition and concern for political freedoms. We use the more neutral term Autocracy and define it operationally in terms of the presence of a distinctive set of political characteristics. In mature form, autocracies sharply restrict or suppress competitive political participation. Their chief executives are chosen in a regularized process of selection within the political elite, and once in office they exercise power with few institutional constraints. Most modern autocracies also exercise a high degree of directiveness over social and economic activity, but we regard this as a function of political ideology and choice, not a defining property of autocracy. Social democracies also exercise relatively high degrees of directiveness. We prefer to leave open for empirical investigation the question of how Autocracy, Democracy, and Directiveness (performance) have covaried over time.

An eleven-point Autocracy scale is constructed additively. Our operational indicator of autocracy is derived from codings of the competitiveness of political participation (variable 2.6), the regulation of participation (variable 2.5), the openness and competitiveness of executive recruitment (variables 2.2 and 2.3), and constraints on the chief executive (variable 2.4) using the following weights:

Authority Coding Scale Weight Competitiveness of Executive Recruitment (XRCOMP): (1) Selection +2

Openness of Executive Recruitment (XROPEN): only if XRCOMP is coded Selection (1) (1) Closed +1 (2) Dual/designation +1

Constraints on Chief Executive (XCONST): (1) Unlimited authority +3 (2) Intermediate category +2

296

(3) Slight to moderate limitations +1

Regulation of participation (PARREG): (4) Restricted +2 (3) Sectarian +1

Competitiveness of Participation (PARCOMP): (1) Repressed +2 (2) Suppressed +1

The logic of this "institutionalized autocracy" scale is similar to that of the institutionalized democracy scale, below, and it is subject to the same kinds of operational redefinition to suit different theoretical purposes. Note that the two scales do not share any categories in common. Nonetheless many polities have mixed authority traits, and thus can have middling scores on both Autocracy and Democracy scales. These are the kinds of polities which were characterized as "anocratic" and "incoherent" in the Polity I studies. As a group they proved to less durable than coherent democracies and autocracies (see Gurr 1974, Harmel 1980, Lichbach 1984).

2.3 POLITY (all versions) Combined Polity Score: The POLITY score is computed by subtracting the AUTOC score from the DEMOC score; the resulting unified polity scale ranges from +10 (strongly democratic) to -10 (strongly autocratic).

2.4 POLITY2 (p4 and p4e only) Revised Combined Polity Score: This variable is a modified version of the POLITY variable added in order to facilitate the use of the POLITY regime measure in time- series analyses. It modifies the combined annual POLITY score by applying a simple treatment, or ““fix,” to convert instances of “standardized authority scores” (i.e., -66, -77, and -88) to conventional polity scores (i.e., within the range, -10 to +10). The values have been converted according to the following rule set:

-66 Cases of foreign “interruption” are treated as “system missing.” -77 Cases of “interregnum,” or anarchy, are converted to a “neutral” Polity score of “0.” -88 Cases of “transition” are prorated across the span of the transition. For example, country X has a POLITY score of -7 in 1957, followed By three years of -88 and, finally, a score of +5 in 1961. The change (+12) would be prorated over the intervening three years at a rate of per year, so that the converted scores would be as follows: 1957 -7; 1958 -4; 1959 -1; 1960 +2; and 1961 +5.

Note: Ongoing (-88) transitions in the most recent year (2007) are converted to “system missing” values. Transitions (-88) following a year of independence, interruption (-66), or interregnum (-77) are prorated from the value “0.”

297

Urban Population (Primacy) Dataset Sources

Oxford Atlas of the World 13th Edition Europa World Yearbooks: 1960, 1964, 1965, 1968, 1970, 1973, 1975, 1978, 1980, 1983, 1985, 1988, 1990, 1993, 1995, 1998, 2000, 2003, and 2005. World Gazetteer Online (www.world-gazetteer.com) United Nation’s Demographic Yearbooks from 1960-2006 Chase-Dunn (1985) http://www.irows.ucr.edu/cd/courses/10/readme.html City Population (www.citypopulation.de) GeoHive: Global Statistics Online (www.geohive.com) Population Statistics (www.populstat.info); see website for list of cited sources. World Gazetteer (www.world-gazetteer.com) World Urbanization Prospects: The 2007 Revision Population Database

How Urban Population Data was Collected

Data were collected in the summer of 2007 through the fall of 2007 with the aid of a research grant provided by the Department of Sociology at The Ohio State University. Data collection began by using the Oxford Atlas of the World 13th Edition which was used to identify nations that had at least one urban area (city, , or agglomeration) that was 400, 000 population or larger. After identifying a broad sample of nations that met the initial requirement the Europa World Yearbooks were consulted (for years consulted see above) to generate a “base” dataset. Specifically, population statistics were gathered for the top seven to eight largest cities from 1960 to 2005 when and where possible. In some instances the top five largest city data could not be collected and were excluded from the final sample (e.g. North Korea). After the Europa “base” was completed in late 2007 the UN’s World Urbanization Prospects: The 2007 Revision Population Database was used to replace population figures for the largest urban areas. While the Urbanization Prospects provided the most consistent information for the largest urban areas it did not provide much information urban areas with populations below 750,000 by 2007 (see Panel 2 for details). As a result, there were many nations with only one or two urban areas that were included for the time period between 1960 and 2005. This left a considerable gap in the data for the third through seventh largest cities. In order to capture more city data the UNs Demographic Yearbooks were consulted to acquire total population counts and metropolitan/urban agglomeration figures for urban areas with populations 100,000 or more (in earlier years it was 20,000 and more). Once these data were placed into the dataset Christopher Chase- Dunn’s 1985 primacy data was used to supplement missing information on cities with less than 100,000 population. However, there were still gaps in the data. Thus, the remaining sources cited above were used to collect data on at least the five

298 largest “urban” areas for each nation from 1960-2005 (some figures ended up being less than 2,000 in the 1960s and thus the reason for the quotes around urban). Urban agglomeration data from the UN were favored as were data that were reported by the UN or the nation as originating from the census. When it was brought to my attention that a revised census figure was available those figures were used. Although efforts were made not to include or areas, it is important to note that some nations report those data as agglomeration data and in a few instances it may be included in the “uncleaned” data set. Moreover, in some instances it made little sense to treat two cities as separate when they clearly “grew” into one (e.g. Tokyo agglomeration). Therefore, the population of the complimentary urban areas were totaled into the agglomeration (this occurred only in a few places as is noted in the dataset—e.g. for Bolivia the city of El Alto was included in Greater ’s figures). Most of the data for small cities of the developing nations comes from the World Gazetter, City Population, GeoHive, and Population Statistics. All identify the source of their data. In many cases these four cites had conflicting figures. In such cases if two had the same or similar numbers those were favored. When all four conflicted I favored Population Statistics due to the comprehensive effort made to cite sources. Where unique issues existed I made an effort to document them in my dataset. Once all the data were gathered they were cleaned in the following manner: First, in order to maintain accuracy it was decided that UN agglomeration data would be used every five years since it was readily available and provided several data points providing the most accuracy. However, if census was available and it was found to be more accurate than the UN figures, the census data were used. On this note, for all other urban areas census data were ALWAYS favored and they comprise the majority of the data points in the dataset. To this end non-census data were used only if the following conditions applied.

1. There was no census data available for a nation at either the start or end of its years in the sample. 2. There was no census data available for over 10 consecutive years. 3. There was no census data available for the nation at any time from 1960- 2005.

In such cases figures were sought from primary print sources first, and if no information was available then on-line sources were used if they had a citied source. The final dataset includes a series of data points for each nation that have both census and non-census figures. Official estimates were used in some cases with the most recent revisions taking precedent. Upon verifying the included non-census numbers with at least two sources, interpolation was used to generate population estimates for each nation’s five largest “urban” areas for the years 1960-2005. In many cases nations entered the dataset a few years late or exited a few years early (see above for a complete list of year).

299

Position in the World-System (Snyder and Kick [1979])

Core Nations: Tajikistan Libya Australia Turkmenistan Madagascar Austria Ukraine Malawi Belgium USSR Malaysia Canada Uzbekistan Mali Denmark Yugoslavia (Former United) Mauritania Federal Germany Yugoslavia Federal Mexico Finland Republic Morocco France Mozambique Germany Periphery: Myanmar Greece Afghanistan Nepal Ireland Algeria Nicaragua Italy Angola Niger Japan Argentina Nigeria Netherlands Bangladesh Pakistan New Zealand Benin Panama Norway Bolivia Paraguay Portugal Brazil Peru Spain Burkina Faso Philippines Sweden Cambodia Saudi Arabia Switzerland Cameroon Senegal United Kingdom Central African Republic Somalia United States Chad South Africa Chile Sri Lanka (Former) Communist: Colombia Sudan Albania Congo Republic Syria Azerbaijan Costa Rica Thailand Armenia Democratic Republic Congo Togo Bulgaria Dominican Republic United Arab Emirates Belarus Ecuador Tunisia China Egypt Turkey Croatia El Salvador Uganda Cuba Ethiopia (Former) Tanzania Czechoslovakia (Former) Ethiopia Uruguay Czech Republic Eritrea Venezuela Georgia Ghana Yemen German Democratic Guatemala Zambia Republic Guinea Zimbabwe Hungary Haiti Kazakhstan Honduras Kyrgyzstan India Latvia Indonesia Lithuania Iran Macedonia Iraq Mongolia Israel Moldova Ivory Coast Poland Jordan Romania Kenya Russian Federation Korea, South Serbia & Montenegro Kuwait Slovakia Lebanon Viet Nam Liberia

300

Colonial History

French Colonies: Senegal Nicaragua Algeria Syria Panama Benin Viet Nam Paraguay Burkina Faso Tunisia Peru Cambodia Philippines Cameroon Spanish Colonies: Uruguay Central African Republic Argentina Venezuela Chad Bolivia Congo Republic Chile Portuguese Colonies: Guinea Colombia Angola Haiti Costa Rica Brazil Ivory Coast Cuba Mozambique Lebanon Dominican Republic Madagascar Ecuador Belgian Colonies: Mali El Salvador Congo Republic Mauritania Guatemala Dem. Rep. of the Congo Morocco Honduras Niger Mexico

Communist History

Afghanistan Federation Albania Serbia and Montenegro Angola Slovakia Armenia Tajikistan Azerbaijan Turkmenistan Belarus USSR Bulgaria Ukraine Cambodia Uzbekistan China Viet Nam Croatia Yemen Cuba Yugoslavia, FDR Czech Republic Yugoslavia (Former United) Czechoslovakia (Former) Ethiopia Ethiopia (Former) Georgia German Democratic Republic Hungary Kazakhstan Kyrgyzstan Latvia Lithuania Macedonia Moldova Mongolia Mozambique Poland Romania Russian

301

APPENDIX B

ALTERNATIVE MODELS

302

302303

304302

302305

302306

302307

List of Nations by Table

Tables 7.1, 7.3, 7.3.1, 7.3.2, and 7.4

Afghanistan Kenya Viet Nam Albania Kuwait Yemen Algeria Kyrgyzstan Yugoslavia FDR Angola Latvia Zambia Argentina Liberia Zimbabwe Armenia Lithuania Azerbaijan Madagascar Bangladesh Malawi Belarus Malaysia Benin Mali Bolivia Mauritania Brazil Mexico Bulgaria Moldova Burkina Faso Mongolia Cambodia Morocco Cameroon Mozambique Central African Republic Nepal Chad Nicaragua Chile Niger China Nigeria Colombia Pakistan Congo Republic Panama Costa Rica Paraguay Croatia Peru Cuba Philippines Czech Republic Poland Democratic Republic Congo Romania Dominican Republic Russian Federation Ecuador Saudi Arabia Egypt Senegal El Salvador Slovakia Eritrea Somalia Ethiopia South Africa Ethiopia (Former) South Korea Georgia Sri Lanka Ghana Sudan Greece Syria Guatemala Tajikistan Guinea Tanzania Haiti Thailand Honduras Togo Hungary Tunisia India Turkey Indonesia Turkmenistan Iran Uganda Iraq Ukraine Israel United Arab Emirates Ivory Coast Uruguay Jordan Uzbekistan Kazakhstan Venezuela

302308

Table 7.1.1

Afghanistan Greece Panama Albania Guatemala Paraguay Algeria Guinea Peru Angola Haiti Philippines Argentina Honduras Poland Armenia Hungary Romania Azerbaijan India Russian Federation Bangladesh Indonesia Saudi Arabia Belarus Iran Senegal Benin Iraq Slovakia Bolivia Israel Somalia Brazil Ivory South Africa Bulgaria Coast South Korea Burkina Faso Jordan Sri Lanka Cambodia Kazakhstan Sudan Cameroon Kenya Syria Central African Republic Kuwait Tajikistan Chad Kyrgyzstan Tanzania Chile Latvia Thailand China Liberia Togo Colombia Lithuania Tunisia Congo Republic Madagascar Turkey Costa Rica Malawi Turkmenistan Croatia Malaysia Uganda Cuba Mali Ukraine Czech Republic Mauritania United Arab Emirates Democratic Republic Congo Mexico Uruguay Dominican Republic Moldova Uzbekistan Ecuador Mongolia Venezuela Egypt Morocco Viet Nam El Salvador Mozambique Yemen Eritrea Nepal Yugoslavia, FDR Ethiopia Nicaragua Zambia Ethiopia (Former) Niger Zimbabwe Georgia Nigeria Ghana Pakistan

302309

Tables 7.2 and 7.2.1

Albania Malawi Algeria Malaysia Angola Mali Argentina Mauritania Armenia Mexico Azerbaijan Moldova Bangladesh Mongolia Belarus Morocco Benin Mozambique Bolivia Brazil Nepal Bulgaria Nicaragua Burkina Faso Niger Cambodia Nigeria Cameroon Pakistan Central African Republic Panama Chad Paraguay Chile Peru China Philippines Colombia Poland Congo Republic Romania Costa Rica Russian Federation Croatia Saudi Arabia Cuba Senegal Czech Republic Slovakia Democratic Republic Congo Somalia Dominican Republic South Africa Ecuador South Korea Egypt Sri Lanka El Salvador Sudan Eritrea Syria Ethiopia Tajikistan Georgia Tanzania Ghana Thailand Greece Togo Guatemala Tunisia Guinea Turkey Haiti Turkmenistan Honduras Uganda Hungary Ukraine India United Arab Emirates Indonesia Uruguay Iran Uzbekistan Israel Venezuela Ivory Coast Viet Nam Jordan Yemen Kazakhstan Zambia Kenya Zimbabwe Kuwait Kyrgyzstan Latvia Liberia Lithuania Madagascar

310302

Table 7.2.2

Albania Madagascar Algeria Malawi Angola Malaysia Argentina Mali Armenia Mauritania Azerbaijan Mexico Bangladesh Moldova Belarus Mongolia Benin Morocco Bolivia Mozambique Brazil Nepal Bulgaria Nicaragua Burkina Faso Niger Cambodia Nigeria Cameroon Pakistan Central African Republic Panama Chad Paraguay Chile Peru China Philippines Colombia Poland Congo Republic Romania Costa Rica Russian Federation Croatia Senegal Czech Republic Slovakia Democratic Republic Congo Somalia Dominican Republic South Africa Ecuador South Korea Egypt Sri Lanka El Salvador Sudan Eritrea Syria Ethiopia Tajikistan Georgia Tanzania Ghana Thailand Greece Togo Guatemala Tunisia Guinea Turkey Haiti Turkmenistan Honduras Uganda Hungary Ukraine India Uruguay Indonesia Uzbekistan Iran Venezuela Israel Viet Nam Ivory Coast Yemen Jordan Yugoslavia, FDR Kazakhstan Zambia Kenya Zimbabwe Kuwait Kyrgyzstan Latvia Liberia Lithuania

302311

Table 7.2.3

Algeria Liberia Zimbabwe Angola Lithuania Argentina Madagascar Armenia Malawi Azerbaijan Malaysia Bangladesh Mali Belarus Mauritania Benin Mexico Bolivia Moldova Brazil Mongolia Bulgaria Morocco Burkina Faso Mozambique Cambodia Nepal Cameroon Nicaragua Central African Republic Niger Chad Nigeria Chile Pakistan China Panama Colombia Paraguay Congo Republic Peru Costa Rica Philippines Croatia Poland Cuba Romania Czech Republic Russian Federation Democratic Republic Congo Saudi Arabia Dominican Republic Senegal Ecuador Slovakia Egypt Somalia El Salvador South Africa Eritrea South Korea Ethiopia Sri Lanka Georgia Sudan Ghana Syria Greece Tajikistan Guatemala Tanzania Guinea Thailand Haiti Togo Honduras Tunisia Hungary Turkey India Turkmenistan Indonesia Uganda Iran Ukraine Israel United Arab Emirates Ivory Coast Uruguay Jordan Uzbekistan Kazakhstan Venezuela Kenya Viet Nam Kuwait Yemen Kyrgyzstan Yugoslavia FDR Latvia Zambia

302312

BIBLIOGRAPHY

Achen, Christopher. 2000. “Why Lagging Dependent Variables Can Suppress the Explanatory Power of Other Independent Variables,” Presented at the Annual Meeting of the Society for Political Methodology, UCLA.

Adelman, Morris. 1969. “Comment on ‘H’ Concentration Measures.” Review of Economics and Statistics 51:130-131.

Ades, Alberto F. and Edward L. Glaeser. 1995. “Trade and Circuses: Explaining Urban Giants”, The Quarterly Journal of Economics, 110(1):195-227.

Allan, Kenneth. 2005. Explorations in Classical Sociological Theory. Thousand Oaks, CA: Pine Forge Press.

Alonso, William. 1980. “Population as a System in Regional Development,” The American Economic Review 70(2):405-409.

Alvarez, Michael, Jose Antonia Cheibub, Fernando Limongi, and Adam Przeworksi. 1996. “Classifying Political Regimes,” Studies in Comparative International Development 31(2):1-37.

Arat, Zehra F. 1988. “Democracy and Economic Development: Modernization Theory Revisited,” Comparative Politics 21(1):21-36.

Armstrong, W. and McGee, T.G. 1985. Theatres of Accumulation: Studies in Asian and Latin American Urbanization. New York: Methuen.

Bairoch, Paul. 1988. “The Birth of Urbanism and the Economy,” Chapter 1 in Cities and Economic Development. : Press.

Banks, Arthur S. 1974. “Industrialization and Development: A Longitudinal Analysis,” Economic Development and Cultural Change 22(2):320-337.

Beck, Nathaniel and Jonathan N. Katz. 1995. “What to do (and not to do) with Time- Series Cross-Section Data,” The American Political Science Review 89(3):634- 647. ______. 1996. “Nuisance vs. Substance: Specifying and Estimating Time-Series-Cross-Section Models,” Political Analysis 6:1-36.

302313

______. 2004. “Time-Series-Cross-Section Issues: Dynamics, 2004,” Working Paper

______. 2009. “Modeling Dynamics in Time-Series- Cross-Section Political Economy Data,” Working Paper

Beckmann, Martin J. 1958. “City Hierarchies and the Distribution of City Size,” Economic Development and Cultural Change 6(3): 243-248.

Berry, Brian J. 1961. “City Size Distributions and Economic Development,” Economic Development and Cultural Change 9(4):573-588.

Berry, Brian J. and John D. Kasarda. 1977. “The Ecological Approach,” Chapter 1 in Contemporary . New York:MacMillian.

Bertinelli, Luisito and Eric Strobl. 2007. “Urbanisation, Urban Concentration and Economic Development,” 44(13)-2499-2510.

Blalock, Hubert M. Jr. and Ann B. Blalock. 1968. Methodology in Social Research. New York: McGraw-Hill.

Bollen, Kenneth A. 1979. “Political Democracy and the Timing of Development” American Sociological Review, 44:572-587.

______. 1980. “Issues in Comparative Measurement of Political Democracy” American Sociological Review, 45(3):370-390.

______. 1993. "Liberal Democracy: Validity and Method Factors in Cross- National Measures," American Journal of Political Science 37:1207-30.

Bornschier, Volker, Christopher Chase-Dunn, and Richard Rubinson. 1978. “Cross- National Evidence of the Effects of Foreign Investment and Aid on Economic Growth and Inequlaity,” The American Journal of Sociology 84(3):651-683.

Bornschier, Volker and Chase-Dunn, Christopher. 1985. Transnational Corporations and Underdevelopment. New York: Praeger.

Boserup, Ester. 1965/2005. The Conditions of Agricultural Growth: The Economics of Agrarian Change under Population Pressure. London: G. Allen and Unwin.

______. 1981. Chapters 1-3 in Population and Technological Change: A Study of Long-Term Trends. Chicago: University of Chicago Press. Bradshaw, York. 1985. “Overubanization and Underdevelopment in Sub-Saharan Africa: A Cross-National Study,” Studies in Comparative International Development , Fall:74-101.

314302

______. 1987. “Urbanization and Underdevelopment: A Global Study of Modernization, Urban Bias and Economic Dependency,” American Sociological View 52(2):224-239.

Breedlove, William L. and J. Michael Armer. 1997. “Dependency, Techno-Economic Heritage, Disarticulation, and Social Development in Less Developed Nations,” Sociological Perspectives 40(4): 661-680.

Brockerhoff, Martin and Ellen Brennan. 1998. “The of Cities in Developing Regions,” Population and Development Review 24(1): 75-114.

Brockerhoff, Martin. 1999. “Urban Growth in Developing Countries: A Review of Projections and Predictions,” Population and Development Review 25(4):757- 778.

Burkhart, Ross E. and Michael S. Lewis-Beck. 1994. “Comparative Democracy: The Economic Development Thesis” The American Political Science Review, 88(4): 903-910.

Chase-Dunn, Christopher. 1975. “The Effects of International Economic Dependence on Development and Inequality: A Cross-National Study,” American Sociological Review 40(6):720-738.

Chirot, Daniel and Thomas D. Hall. 1982. “World-System Theory,” Annual review of Sociology 8:81-106.

Christaller, Walter. 1933. Central Places in Southern Germany. Prentice Hall.

Clayton, Elizabeth and Thomas Richardson. 1989. “Soviet Control of City Size,” Economic Development and Cultural Change 38(1):155-165.

Cohen, Barney. 2003. “Urban Growth in Developing Countries: A Review of Current Trends and a Caution Regarding Existing Forecasts,” World Development 32(1):23-51.

______. 2006. “Urbanization in Developing Countries: Current trends, future projections, and key challenges for sustainability,” Technology in Society 28:63-80.

Crenshaw, Edward M. 1991. “Foreign Investment as a Dependent Variable: Determinants of Foreign investment and Capital Penetration in Developing Nations 1967-1978,” Social Forces 69(4):1169-1182.

______. 1992. “Cross-National Determinants of Income Inequality: A Replication and Extension using Ecological-Evolutionary Theory,” Social Forces 71(2):339-363.

302315

______. 1995. “Democracy and Demographic Inheritance: The Influence of Modernity and Proto-Modernity on Political and Civil Rights, 1965-1980” American Sociological Review, 60 (5): 702-718.

Crenshaw, Edward M. and Doyle Ray Oakey. 1998. ““Jump-Starting” Development: Hyperurbanization as a Long-Term Economic Investment” Sociological Focus, 31(4):321-340.

Crenshaw, Edward M., Matthew Christenson, and Doyle Ray Oakey. 2000. “Demographic Transition in Ecological Focus,” The American Sociological Review 65:371-391.

Crenshaw, Martha. 1981. “The Causes of Terrorism,” Comparative Politics 13(4):379-399.

Cutright, Phillips. 1963. “National Political Development: Measurement and Analysis” American Sociological Review, 28(2):253-264.

Dahl, Robert A. 1971. Polyarchy. New Haven, CT: Press.

Dahl, Robert A. and Edward R. Tufte. 1973. Size and Democracy. Sanford, CA: Sanford University Press.

Davis, Kingsley. 1945. “The World Demographic Transition,” Annals of the American Academy of Political and Social Science, 237:1-11.

______. 1955. “The Origin and Growth of Urbanization in the World,” The American Journal of Sociology 60(5):429-437.

______. 1963. “Theory of Change and Response in Modern Demographic History,” Population Index 29(4):345-366.

Davis, Kingsley and Hilda Hertz Golden. 1954. “Urbanization and the Development of Pre-Industrial Areas.” Economic Development and Cultural Change 3(1):6-26.

Davis, James C. and Vernon Henderson. 2003. “Evidence on the political economy of the urbanization process,” Journal of Urban Economics 53:98-125.

De Boef, Suzanna and Luke Keele. 2008. “Taking Time Seriously,” American Journal of Political Science 52(1):184-200.

De Cola, Lee. 1984. “Structural Determinants of the Population of a Nation’s Largest City,” Economic Development and Cultural Change 33(1):71-98.

302316

Duncan, Otis Dudley. 1959. “Human Ecology and Population Studies,” Chapter 28 (p.678-716) in The Study of Population edited by Philip M. Hauser and Otis Dudley Duncan, Chicago: University of Chicago Press.

______. 1961. “From Social System to Ecosystem,” Sociological Inquiry 31:140-149.

Durkheim, Emile. 1893/1998. From: De la Division dur travail social, in Readings from Emile Durkheim. Edited by Kenneth Thompson. New York: Routledge.

Easterlin, Richard A. 1967. “Effects of Population Growth on the Economic Development of Developing Countries,” Annals of the American Academy of Political and Social Science 369:98-108.

______. 1968. Population, Labor Force, and Long Swings in Economic Growth. New York: Press.

El-Shakhs, Salah. 1972. “Development, Primacy, and Systems of Cities,” Journal of Developing Areas 7(1):11-35. Evans, Peter. 1979. Dependent Development. New Jersey; Princeton University Press.

Feng, Yi. 1997. “Democracy, Political Stability and Economic Growth,” British Journal of Political Science 27:391-418.

Fonseca, James W. 1989. Urban Rank-Size Hierarchy: A Mathematical Interpretation Monograph #8, Institute of Mathematical Geography: Ann Arbor, Michigan.

Firebaugh, Glenn. 1979. “Structural Determinants of Urbanization in Asia and Latin America, 1950-1970,” American Sociological Review 44(2):199-215.

Frees, Edward W. 2004. Longitudinal and Panel Data: Analysis and Applications in the Social Sciences, New York: University Press.

Freidmann, John. 1973. “A Theory of Polarized Development,” (p.41-64) in Urbanization, Planning, and National Development. Edited by John Freidmann. Beverly Hills: Sage Press.

Gabaix, Xavier. 1999a. “Zipf’s Law for Cities: An Explanation,” The Quarterly Journal of Economics 114(3):739-767

______. 1999b. “Zipf’s Law and the Growth of Cities,” The American Economic Review 89(2):129-132.

Gibbs, Jack P. 1966. “Measures of Urbanization,” Social Forces 45(2):170-177.

317302

Gonick, Lev S. and Robert M. Rosh. 1988. “The Structural Constraints of the World- Economy on National Political Development,” Comparative Political Studies 21:171-199.

Gugler, 1982. “Overurbanization Reconsidered,” Economic Development and Cultural Change 31(1):173-189.

Hansen, Niles. 1990. “Impacts of Small and Intermediate-Sized Cities on Population Distribution: Issues and Responses,” Regional Development Dialogue 11(1):60-76.

Hawley, Amos H. 1944. “Ecology and Human Ecology,” Social Forces 22(4):398-405.

______. 1950. Human Ecology: A Theory of Community Structure New York: The Ronald Press Company.

______. 1963. “Community Power and Success” The American Journal of Sociology, 68(4):422-431.

______. 1968. “Human Ecology” (p.328-337) in The International Encyclopedia of the Social Sciences Vol.4, New York: MacMillan.

______. 1984. “Human Ecology and Marxian Theories,” The American Journal of Sociology 89(4):904-917.

______. 1986. Human Ecology: A Theoretical Essay. Chicago: University of Chicago Press.

______. 1992. “The Logic of ,” Annual Review of Sociology 18:1-14.

Helliwell, John F. 1994. “Empirical Linkages between Democracy and Economic Growth,” British Journal of Political Science 24:225-248.

Henderson, Vernon, J. 2002. “Urban Primacy, external costs, and quality of life,” Resource and Energy Economics 24:95-106.

______. 2003. “The Urbanizaiton Process and Economic Growth: The So-What Question,” Journal of Economic Growth 8:47-71.

______. 2004. “Urbanization and Growth,” Draft Chapter for Handbook of Economic Growth Volume 1. P. Aghion and S. Durlauf (eds), North Holland.

Henderson, J. Vernon and Hyoung Gun Wang. 2007. “Urbanization and city growth: The role of institutions,” Regional Sciences and Urban Economics 37:283-313.

302318

Hoselitz, Bert F. 1955. “Generative and Parasitic Cities,” Economic Development and Cultural Change 3(3):278-294.

Heo, Uk and Alexander C. Tan. 2001. “Democracy and Economic Growth: A Causal Analysis,” Comparative Politics 33(4):463-473.

Inkeles, Alex. 1960. “Industrial Man: The Relation of Status to Experience, Perception, and Value,” The American Journal of Sociology, 66:1-31.

______. 1975. “Becoming Modern: Individual Change in Six Developing Countries,” Ethos 3(2):323-342.

______. 1981. Chapter 1 “Convergence and Divergence in Industrial Society,” (p. 3-38) in Direction of Change: Modernization Theory, Research, and Realities. Boulder, CO: Westview Press.

Jackman, Robert W. 1973. “On the Relation of Economic Development to Democratic Performance” American Journal of Political Science, 17(3):611-621

Jefferson, Mark. 1939. “The Law of the Primate City,” Geographic Review 29(2):226- 232. Jones, F. Lancaster. 1967. “A Note on ‘Measures of Urbanization,’ with a Further Proposal.” Social Forces 46(2):275-279.

Junius, Karsten. 1999. “Primacy and Economic Development: Bell Shaped or Parallel Growth Cities,” Journal of Economic Development 24(1):1-22.

Kamerschen, David R. 1969. “Further Analysis of Overurbanization,” Economic Development and Cultural Change 17(2):235-253.

Kasarda, Jack D. and Edward M. Crenshaw. 1991. “Third World Urbanization: Dimensions, Theories, and Determinants” Annual Review of Sociology 17:467- 501.

Keele, Luke and Nathan J. Kelly. 2006. “Dynamic Models for Dynamic Theories: The Ins and Outs of Lagged Dependent Variables,” Political Analysis 14:186-205.

Kelly, Nathan J. 2004. “The Nature and Degree of Bias in Lagged Dependent Variable Models,” Working Paper

Kelley, Allen C. and Jeffrey G. Williamson. 1984. “Population Growth, Industrial Revolutions, and the Urban Transition,” Population and Development Review 10(3):419-441.

Kentor, Jeffrey. 1981. “Structural Determinants of Peripheral Urbanization: The

319302

Effects of International Dependence,” American Sociological Review 46(2):201-211.

Kerr, Clark, John T. Dunlop, Frederick H. Harbison and Charles A. Myers. 1960. Chapter 1 and Chapter 3 in Industrialism and Industrial Man, Ney York: Oxford University Press.

Kristensen, Ida Pagter and Gregory Wawro. 2003. “Lagging the Dog?: The Robustness of Panel Corrected Standard Errors in the Presence of Serial Correlation and Observation Special Effects,” Working Paper presented at 2003 Summer Methods Conference.

Lenski, Gerhard. 1966/1984. Power and Privilege: A Theory of Social Stratification. United States: The University of North Carolina Press.

______. 2005. Ecological-Evolutionary Theory: Principles and Applications. Boulder, Colorado: Paradigm Publishers.

Lenski, Gerhard, and Patrick D. Nolan. 1984. “Trajectories of Development and Technoeconomic Heritage: A Test of Ecological-Evolutionary Theory” Social Forces, 63:1-23.

Lerner, Daniel. 1958/1964. The Passing of Traditional Society: Modernizing the . London: The Free Press of Glencoe.

Linsky, Arnold S. 1965. “Some Generalizations Concerning Primate Cities” Annals of the Association of American Geographers, 55(3):506-513.

Lipset, Seymour M. 1959. “Some Social Requisites of Democracy: Economic Development and Political Legitimacy”. The American Political Science Review, 53(1):69-105

Leblang, David. 1996. “Property Rights, Democracy and Economic Growth,” Political Research Quarterly, 49:5-26.

______. 1997. “Political Democracy and Economic Growth: Pooled Cross- Sectional and Time-Series Evidence,” British Journal of Political Science, 27:453-472.

Lipton, Michael. 1977. Why Poor People Stay Poor: Urban Bias in World Development. Cambridge, MA: Harvard University Press.

London, Bruce. 1987. “Structural Determinants of Third World Urban Change: An Ecological and Political Economic Analysis,” American Sociological Review 52(1):28-43.

302320

London, Bruce and David A. Smith. 1988. “Urban Bias, Dependence, and Economic Stagnation in Noncore Nations,” American Sociological Review 53(3):454-463.

Lösch, August. 1938. “The Nature of Economic Regions,” Southern Economic Journal 5(1):71-78.

______. 1954. The Economics of Location. Yale University Press: New Haven, CT.

Lyman, Brad. 1992. “Colonial Governance in the Development of Urban Primacy,” Studies in Comparative International Development 27(2):24-38.

Marx, Karl and Friedrich Engels. 1848, (1986) The Communist Manifesto, in Karl Marx: A Reader. Edited by Jon Elster. New York: Cambridge University Press.

McCrone, Donald T. and Charles F. Cnudde. 1967. “Toward a Communications Theory of Democratic Political Development: A Causal Model,” The American Political Science Review 61(1):72-79.

McKenzie, R.D. 1924. “The Ecological Approach to the Study of the Human Community,” The American Journal of Sociology 30(3):287-301.

______. 1927. ‘The Concept of Dominance and World-Organization,” The American Journal of Sociology 33(1):28-42.

Metha, Surinder K. 1964. “Some Demographic and Economic Correlates of Primate Cities: A Case for Revaluation” Demography, 1(1):136-147.

Mera, Kochi. 1973. “On the Urban Agglomeration of Economic Efficiency,” Economic Development and Cultural Change 21(2): 309-324.

Meyer, John. W., John Boli-Bennett, and Christopher Chase-Dunn. 1975. “Convergence and Divergence in Development,” Annual Review of Sociology 1:223-246.

Midlarsky, Manus I. 1992. “The Origins of Democracy in Agrarian Society: Land Inequality and Political Rights,” The Journal of Conflict Resolution 36(3):454- 477.

Moir, Hazel. 1976. “Relationships between Urbanization Levels and the Industrial Structure of the Labor Force,” Economic Development and Cultural Change 25(1):123-135.

Muller, Edward N. 1988. “Democracy, Economic Development, and Income Inequality” American Sociological Review, 53(1):50-68.

302321

______. 1995. “Economic Determinants of Democracy” American Sociological Review, 60(6):966-982.

Munck, Gerardo L. and Jay Verkuilen. 2002. “Conceptualizing and Measuring Democracy: Evaluating Alternative Indicies,” Comparative Political Studies 35(1):5-34.

Mutlu, Servet. 1989. “Urban Concentration and Primacy Revisited: An Analysis and Some Policy Conclusions”, Economic Development and Cultural Change, 37(3):611-639.

Namboodiri, Krishnan. 1994. “The Human Ecological Approach to the Study of Population Dynamics,” Population Index 60(4):517-539.

Neubauer, Deane E. 1967. “Some Conditions of Democracy” The American Political Science Review, 61(4):1002-1009.

Notestein, Frank W. 1945. “International Population Readjustments,” Proceedings of the Academy of Political Science 21(2):94-102.

Olsen, Mancur. 1993. “Dictatorship, Democracy, and Development,” American Political Science Review 87:567-576.

Owen, Carol and Ronald A. Witton. 1973. “National Division and Mobilization: A Reinterpretation of Primacy” Economic Development and Cultural Change, 21(2):325-337.

Park, Robert Ezra. 1936. “Human Ecology,” The American Journal of Sociology 42(1):1-15.

Parks, Richard. 1967. “Efficient Estimation of a System of Regression Equations When Disturbances are both Serially and Contemporaneously Correlated,” Journal of the American Statistical Association 62:500-509.

Peterson, William. 1988. “Marxism and the Population Question: Theory and Practice,” Population and Development Review 14:77-101.

Przeworski, Adam. 2000. Democracy and Development: Political Institutions and Well-being in the World 1950-1990. New York: Cambridge University Press.

Przeworski, Adam and Fernando Limongi. 1997. “Modernization: Theories and Facts” World Politics, 49(2): 155-183.

Quinn, James A. 1940. “Human Ecology and Interactional Ecology,” American Sociological Review 5(5):713-722.

322302

Rueschemeyer, Dietrich, Evelyne Huber Stephens, and John D. Stephens. 1992. Capitalist Development and Democracy. Chicago: University of Chicago Press.

Rogers, Andrei. 1982. “Sources of Urban Population Growth and Urbanization, 1950- 2000: A Demographic Accounting,” Economic Development and Cultural Change 30(3):483-506.

Rogers, Everett M. 1962/2003. Diffusion of Innovations. Fifth Edition. New York: The Free Press.

Rostow, W. W. 1960. The Stages of Economic Growth. New York: Cambridge University Press.

Rubinson, Richard. 1976. “The World-Economy and the Distribution of Income Within States: A Cross-National Study,” American Sociological Review 41(4):638-659.

Rustow, Dankwart A. 1970. “Transitions to Democracy: Toward a Dynamic Model,” Comparative Politics 2(3):337-363.

Sawers, Larry. 1989. “Urban Primacy in Tanzania”, Economic Development and Cultural Change, 37(4):841-859.

Schnore, Leo F. 1964. “Urbanization and Economic Development: The Demographic Contribution,” Journal of Economics and Sociology 23(1):37-48.

Shandra, John M., Bruce London, and John B. Williamson. 2003. “Environmental Degradation, Environmental Sustainability, and Overurbanization in the Developing World: A Quantitative, Cross-National Analysis,” Sociological Perspectives 46(3):309-329.

Sica, Alan. 2005. Social Thought: From the Enlightenment to the Present. , MA: Pearson Education Inc.

Sirowy, Larry and Alex Inkeles. 1990. “The Effects of Democracy on Economic Growth and Inequality: A Review,” Studies in Comparative International Development 25:126-57.

Smith, Arthur K. Jr. 1969. “Socio-Economic Development and Political Democracy: A Causal Analysis,” Midwest Journal of Political Science 13(1):95-125.

Smith, Carol A. 1985. “Theories and Measures of Urban Primacy: A Critique,” Chapter 6 in Urbanization in the World-Economy. Edited by Michael Timberlake, Charles Tilly, and Edward Shorter: Academic Press.

323302

Sovani, N.V. 1964. “The Analysis of ‘Over-Urbanization’,” Economic Development and Cultural Change 12(2):113-122.

Timberlake, M. 1987. “World-system theory and the study of comparative urbanization.” (p.37-65) in The Capitalist City: Global Restructuring and Community Politics (ed. E. P. Smith and J. R. Feagin) Oxford: Basil Blackwell.

Timberlak, Michael and Jeffrey Kentor. 1983. “Economic Dependency, Overurbanization, and Economic Growth: A Study of Less Developed Countries,” The Sociological Quarterly 24(4):489-507.

Thompson, Warren S. 1929. “Population,” The American Journal of Sociology 34(6):959-975.

Todaro, Michael P. 1969. “A Model of Labor Migration and Urban Underdevelopment in Less Developed Countries,” The American Economic Review 59(1):138-148.

______. 1970. “Labor Migration and Urban Unemployment: Reply,” The American Economic Review 60(1): 187-188.

Turner, Jonathan H., Leonard Beeghley, and Charles H. Powers. 2002. The Emergence of Sociological Theory. Fifth Edition. Stamford, CT: Wadsworth Thomson Learning.

Vapnarsky, Cesar A. 1969. “On Rank-Size Distribution of Cities: An Ecological Approach.” Economic Development and Cultural Change 17(4):584-595.

Wallerstein, Immanuel. 1974. Chapter 7 “Theoretical Reprise.” (p.346-357) in The Modern World System I: Capitalist Agriculture and the Origins of the European World Economy in the Sixteenth Century. New York: Academic Press.

______. 1976. “Semi-Peripheral Countries in the Contemporary World Crisis,” Theory and Society 3(4):461-483.

Walters, Pamela Barnhouse. 1985. “Systems of Cities and Urban Primacy: Problem of Definition and Measurement,” Chapter 5 in Urbanization in the World- Economy. Edited by Michael Timberlake, Charles Tilly, and Edward Shorter: Academic Press.

Walton, John. 1982. “The International Economy and Peripheral Urbanization,” Chapter 6 (p.119-134) in Urban Policy Under Capitalism. Edited by Norman I. Fainstein and Susan S. Fainstein. Beverly Hills, CA: Sage Publications.

Watson, Patrick and Sonja S. Teelucksingh. 2002. A Practical Introduction to Econometric Methods: Classical and Modern. University of the West Indies Press.

302324

Weeks, John R. 2002. Population: An Introduction to Concepts and Issues. Eighth Edition. Belmont, CA: Wadsworth Thomson Learning.

Wheaton, William C. and Hisanobu Shishido. 1981. “Urban Concentration, Agglomeration Economies, and the Level of Economic Development,” Economic Development and Cultural Change 30(1):17-30.

World Bank, The. 2008. World Development Report 2009: Reshaping Economic Geography. Chapter 4 “Scale Economies and Agglomeration.”

Zipf, George K. 1941. National Unity and Disunity, Bloomington, Indiana. ______. 1949. Human Behavior and the Principle of Least-Effort. Addison- Wesley.

302325

DATA SOURCES

Barro, Robert J. and Jong-Wha Lee. 2000. “International Data on Educational Attainments: Updates and Implications” (CID Working Paper No.42) http://www.cid.harvard.edu/cidwp/042.htm

Canning, David. 1998. "A Database of World Stocks of Infrastructure: 1950-1995," The World Bank Economic Review 12(3), pp 529-548. http://www.worldbank.org/html/dec/Publications/Workpapers/WPS1900 series/wps1929/canning1.xls

CIA World Factbook: https://www.cia.gov/library/publications/the-world-factbook/

Center for International Comparisons, Penn World Tables: http://pwt.econ.upenn.edu/php_site/pwt_index.php

Center for International Development (CID): http://www.cid.harvard.edu/ciddata/ciddata.html

Food and Agriculture Organization of the United Nation: FAOSTAT 2006 revision, http://faostat.fao.org/

International Monetary Fund 2008 World Economic Outlook Update, Chapter 5 “, Commodity Prices, and Developing Countries,” see Appendix 5.1. Data and Methodology.

POLITY IV Project: http://www.systemicpeace.org/polity/polity4.htm

Unite Nations Statistics Division. The World Urbanization Prospects: The 2007 Revision Population Database, http://esa.un.org/unup/index.asp?panel=1

World Development Indicators 2005 (CD-Rom), 2007 and 2009 on-line database: The World Bank, http://web.worldbank.org/data

1 No were harmed in the writing of this dissertation.

326302