1 URBAN DEVELOPMENT AND VARIATIONS OF CLIENTELISM IN LOCAL GOVERNMENT JURISDICTIONS IN KENYA By LEVY CHARLES ODERA A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2013
2 2013 Levy Charles Odera
3 To my father Philip Onyango, my mother Margaret Onyango, my brothers and sisters, and my wife Eric a Oder a
4 ACKNOWLEDGMENTS This dissertation would not have been possible without the physical, mental, and emotional support from so many helpful people. More people than I can mention deserve to be thanked. I would like to assure those that I will not b e able to mention here, in Kenya and the United States, that I am deeply grateful for all the forms of support and assistance they have accorded to me. It all started with my parents who nurtured the desire for higher education in me when I was still a you ng scholar. With very minimal resources they did everything they could to ensure that my intellectual foundation was stable. They sacrificed so much and worked tirelessly to ensure that my siblings and I received quality primary and secondary school educat ion, which has been a bedrock for the pursuit of my PhD degree. They have never ceased to continue encouraging and supporting me throughout the time that I have been conducting this study. The extensive fieldwork I conducted in Kenya would not have been po ssible without the comfort they provided whenever I was in the country. Throughout my studies they have always ensured that I have a comfortable room to conduct my research and extended all the support I needed beyond my expectation. I also thank my brothe rs and sisters for all the support they have given me during my fieldwork and my studies in the United States. Their generosity is immeasurable. The intellectual ability to complete this dissertation would not have been possible without the constant dedi cation and commitment of my advisor Staffan Lindberg. He has been committed to academic rigor and excellence through the process, which I will keep with me throughout my career. The in depth and prompt feedback he has given me has not only made it possible to finish my PhD in a timely manner, but also significantly increased the quality of the dissertation. His wisdom has been a source of inspiration for
5 me. He has consistently supported and shaped my academic endeavors ever since I started my PhD. He has g iven his time generously and treated my studies with utmost priority. I am deeply grateful to him for significantly improving my scholastic abilities. I also thank the members of my committee: Michael Martinez, Bryon Moraski, Conor and Brian Child. I am very thankful that they accepted to serve in my committee and have always been available to advice and encourage me. They have invested a substantial amount of their time to review my work and made very insightful suggestions on how t o improve the dissertation. Their passion for my study has been a significant inspiration to me and has given me encouragement to continue with the study. Their dedication to my success through their prompt feedback and advice has been highly helpful in th e timely completion of my PhD. I would not have been able to enter the PhD program at the University Florida were it not for the financial support I received through the Almuni Fellowship from the University of Florida. I am deeply grateful to the Univers ity for the opportunity to accomplish my academic dreams with minimal worries about financial issues. I also thank the Center for African Studies at the University of Florida, the Department of Political Science, and the Africa Power and Politics Programme (APPP), which was supported by the Overseas Development Institute, for their generous financial support of my fieldwork. The support was priceless. I am deeply grateful for making it possible to leave for fieldwork promptly and frequently enough to extens ively collect the data that I needed for this study and the writing grant from the APPP that enabled me to start working on the dissertation immediately I returned from Kenya. The extensive nature of the fieldwork I conducted was made possible by my dedica ted and hardworking
6 research assistants: Daniel Warui, Christine Nyagaya, Samuel Otieno, Pauline Nanjala Lubia, and Nancy Ruth. I thank them for the dedication and commitment to perfection in their work that has provided a high quality of data and informat ion for the dissertation. I am very thankful to my good friend Dr. Dawit O. Woldu for his intellectual and technical support in the development of the dissertation. I would like to also express my deep gratitude to the hardworking business owners in the in formal sector in Kisumu, Nairobi, and Mombasa who agreed to be interviewed and surveyed for my dissertation research. They have been both a source of knowledge and inspiration for this study. It is my sincere desire that my research be used one day to posi tively impact their business success. I owe the utmost debt to my wife Erica. As the closest person throughout the development of this dissertation, she has extended extensive support since the time I started my fieldwork. She has sacrificed so much perso nally to ensure that I had whatever I needed to complete the dissertation. She has always ensured that my working environment is conducive for me to focus on my research with minimal distractions. Her intellectual insight has always provided helpful critiq ue of my ideas and sharpened them. I am deeply grateful the amazing helper and motivator that she has been.
7 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ .......... 10 LIST OF FI GURES ................................ ................................ ................................ ........ 12 ABSTRACT ................................ ................................ ................................ ................... 14 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 16 1.1 Urban Develop ment and Clientelism: The Missing Link ................................ .... 19 1.2 Structure Agency Approach ................................ ................................ .............. 21 1.3 Clientelism ................................ ................................ ................................ ........ 23 1.4 Urban Development ................................ ................................ .......................... 27 1.5 Mechanisms of Urban Development and Clientelistic Strategic Considerations ................................ ................................ ................................ ..... 30 1.6 Low Socioeconomic Status Attracts Clientelism ................................ ............... 34 1.7 Urban Development Improves Socioeconomic Status ................................ ...... 42 1.8 Hypothe ses and Research Questions ................................ ............................... 49 1.9 Possible Objections and Responses ................................ ................................ 50 1.10 The Structure of the Dissertation ................................ ................................ .... 53 2 CASE STUDY METHOD AND CASE SELECTION: WHY KENYA? ...................... 57 2.1 Why a Case Study? ................................ ................................ .......................... 57 ................................ ................................ ................ 59 3 THE INFORMAL SECTOR IN KENYA: ITS GROWTH AND PERMANENCE ........ 67 3.1 What is the Informal Sector? ................................ ................................ ............. 67 3.2 The Growth and Significance of the Informal Sector ................................ ......... 70 3.3 State Development Strategies and Growth of the Informal Sector .................... 76 3.4 Permanence of the Sector: Outcome of MFIs Expansion and their Strategies ................................ ................................ ................................ ............ 80 4 URBAN DEVELOPMENT IN LOCAL GOVERNMENT JURISD ICTIONS ............... 91 ................................ ......... 91 4.2 A Short History of the Local Government System ................................ ............. 93 4.3 Main Features of the Local Government System ................................ .............. 98
8 4.3.1 The Governing Rules: Local Government Act of 1978 ............................ 98 4.3.2 Local Authority Urban Development Responsibilities and the Political Challenges ................................ ................................ ................................ .. 100 4.3.3 Local Authority Revenues and Expenditures ................................ ......... 101 4.3.4 Recent Local Government Reforms: Stimulating Urban Development .. 102 ..................... 105 4.4.1 Nairobi ................................ ................................ ................................ ... 106 4.4.2 Mombasa ................................ ................................ ............................... 112 4.4.3 Kisumu ................................ ................................ ................................ .. 114 5 .... 121 5.1 Party Clientelistic Linkages in Kenya: The Roots ................................ ............ 121 ............................... 130 5.2.1 Infrastructural Development ................................ ................................ .. 130 5.2.2 Market Development ................................ ................................ ............. 131 5.2.3 Property Development ................................ ................................ ........... 132 5.2.4 Business Development ................................ ................................ .......... 132 5.3 Urban Development and Variation of Socioeconomic Status .......................... 133 5.4 Socioeconomic Status and Variation of Political Party Clientelism ................. 145 6 TRADITIONAL VERSUS MODERN CLIENTELISM: CONCEPTUALIZATION AND OPERATIONALIZATION ................................ ................................ .............. 153 6.1 Differentiating Modern and Traditional Client elism ................................ .......... 154 6.1.1 Scholastic Distinction ................................ ................................ ............. 154 6.1.2 A Modernization Perspective of the Two Modes ................................ ... 156 6.1.3 Distinguishing Relations ................................ ................................ ........ 158 6.1.4 Distinguishing Transactions ................................ ................................ ... 160 6.1.5 The Role of Bro kers ................................ ................................ ............... 162 6.1.6 The Role of Group Membership ................................ ............................ 163 6.2 Operationalization and Key Empirical Expectations ................................ ........ 164 6.2.1 Dependent Variables: MPs Versus Councilors Business Assistance .... 167 6.2.2 Key Independent Variables: ................................ ................................ .. 169 184.108.40.206 Participation in elections ................................ .............................. 169 220.127.116.11 Campaign versus non campaign (business)visits ........................ 170 18.104.22.168 Frequency of contacts (level of access to politician) .................... 170 22.214.171.124 Two way link with politician ................................ .......................... 171 126.96.36.199 Lobby group membersh ip ................................ ............................ 171 188.8.131.52 Demographic and business aspects ................................ ............ 172 7 EXAMINATION OF TRADITIONAL VERSUS MODERN CLIENTELISM .............. 174 7.1 Data ................................ ................................ ................................ ................ 174 7.2 Methods ................................ ................................ ................................ .......... 176 7.2.1 Binary Regression Models ................................ ................................ ..... 176 7.2.2 Ordinal Regression Models ................................ ................................ ... 179
9 7.2.3 Plotting Probabilities ................................ ................................ .............. 180 7.2.4 Diagno stic Tests ................................ ................................ .................... 181 7.3 Analysis and Findings ................................ ................................ ..................... 183 7.3.1 Comparing the Occurrence of Clientelism: Urban Development and Modern Cliente lism ................................ ................................ ..................... 184 7.3.2 Urban Development and Traditional Clientelism ................................ ... 185 7.3.3 MPs versus Councilor Usage of Modern Clientelism (%) ...................... 187 7.3.4 MP versus Councilor Usage of Traditional Clientelism (%) ................... 190 7.3.5 Logistic Regression Analysis ................................ ................................ 192 ................................ .. 192 7.3.7 The Effect of Campaign Visits ................................ ............................... 196 7.3.8 The Effect of Participation in Elections ................................ .................. 198 ......................... 202 7.3.10 The Effect of Getting Business Visits ................................ ................... 208 7.4 Overview of Findings ................................ ................................ ...................... 214 8 CONCLUSION ................................ ................................ ................................ ...... 216 8.1 Overview of Data and Methods Utilized ................................ .......................... 217 8.2 Summary of Chapters ................................ ................................ ..................... 221 8.3 Distinguishing Clie ntelism and Increasing Focus on the Traditional Mode ..... 230 8.4 Beyond Kenya ................................ ................................ ................................ 243 APPENDIX A DATA COLLECTION METHODOLOGIES ................................ ............................ 249 B QUESTIONS AND VARIABLES ................................ ................................ ........... 256 C SUPPLEMENTARY TABLES AND FIGURES ................................ ...................... 262 REFERENCES ................................ ................................ ................................ ............ 273 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 315
10 LIST OF TABLES Table page 4 1 A summary of ................................ 119 5 1 Measuring capital intensity ................................ ................................ ............... 136 5 2 Urban development and capital intensity ................................ .......................... 137 5 3 Urban development and business size (having employees) ............................. 138 5 4 Larger business size and financial success ................................ ...................... 139 5 5 Urban development and level of education attained ................................ ......... 141 5 6 Urban development and attainment of business education .............................. 142 5 7 Urban development and attendance of business workshops ........................... 142 5 8 Urban development and attendance of business seminars .............................. 143 5 9 Attaining business education and achieving financial success ......................... 143 5 10 Attending business seminars and achieving financial success ......................... 144 5 11 Urban development and political party clientelism ................................ ............ 146 6 1 Comparison of features of modern and traditional clientelism .......................... 173 7 1 ........................... 189 7 2 ........................ 190 7 3 ................................ ......... 194 7 4 istance ............ 195 7 5 access to MP and campaign visits by politicians ................................ .............. 197 7 6 ................................ 203 7 7 ............................ 205 7 8 .................. 207 7 9 order ed logit model ................................ ................................ ........................... 210
11 7 10 for the ordered logit model ................................ ................................ ................ 211 8 1 The relationship between clientelism and urbanization in Africa ....................... 247 C 1 Descriptive statistics ................................ ................................ ......................... 262 C 2 Variance infla tion factors (VIF) and tolerance values for binary regression .................. 263 C 3 Variance inflation factors (VIF) and tolerance values for b inary regression .......... 264 C 4 Nairobi and Mombasa) ................................ ................................ ..................... 265 C 5 variables for Nairobi and Mombasa) ................................ ................................ 266 C 6 Logit a variables for Mombasa and Kisumu) ................................ ................................ 267 C 7 variables for N airobi and Mombasa) ................................ ................................ 268 C 8 variables for Mombasa and Kisumu) ................................ ................................ 269 C 9 Dependent and independent variables ................................ ............................. 270
12 LIST OF FIGURES Figure page 1 2 The effect of urban development on clientelistic transaction s ............................ 33 2 1 Average rate of urbanization in Africa, 2000 2010 ................................ ........... 65 2 2 Average GDP growth rate in Africa, 2000 2012 ................................ .................. 65 2 3 Average level of corruption in Africa, 2007 to 2010 ................................ ............ 66 2 4 Average level of democracy in Africa, 2000 2010 ................................ ............ 66 3 1 Growth of informal sector in Kenya, 1975 2010 ................................ ............... 71 3 2 Growth of rural and urban informal sector in Kenya, 1986 2010 ...................... 72 3 3 Urban and rural population distribution in Kenya, 1960 to 2001. ........................ 73 3 4 Growth of the informal sector by industry in Kenya, 1986 2010 ...................... 75 3 5 Economic growth by decade, 1963 2000 ................................ ........................... 81 3 6 MFIs loan portfolio and Active Borrowers, 1997 2009 ................................ ..... 85 3 7 Increase in urban poverty incidence. ................................ ................................ .. 87 3 8 African populations living in urban areas (%), by region: 1990 2025. ................. 89 4 1 ................................ .......... 105 5 1 Clientelism is more prevalent where urban development is low ....................... 147 5 2 Clientelism is prevalent where the level of poverty is high ................................ 148 5 3 Women are more likely to be targeted for clientelism ................................ ....... 149 5 4 Variation in infrastructural development based on capital expenditures in the three cities ................................ ................................ ................................ ........ 150 5 5 Variation in market development based on market fees ................................ ... 151 5 6 Variation in property development based on property rates ............................. 151 5 7 Variation in business development based on single business permit f ees ....... 152 7 1 Occurrence of modern clientelism (%) ................................ .............................. 184
13 7 2 Occurrence of traditional clientelism (%) ................................ .......................... 185 7 3 ................................ 188 7 4 MP versus councilor usage of traditional clientelism (%) ................................ .. 190 7 5 with regards to age and getting campaign visits ................................ ............... 197 7 6 Probability of with regards to age and frequency of participation in elections (voting) ........... 199 7 7 el of voting in elections varies with regards to capital intensity and direct access to the MP ................. 200 7 8 the counci lor varies with regards to getting business visits .............................. 208 7 9 business visits vary with regards to lobbying and duration of ownership .......... 212 7 10 business visits vary with regards to lobbying and low education ...................... 213
14 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy URBAN DEVELOPMENT AND VARIATIONS OF CLIENTELISM IN LOC AL GOVERNMENT JURISDICTIONS IN KENYA By Levy Charles Odera May 2012 Chair: Staffan I. Lindberg Major: Political Science This study is about clientelism, but not in the usual sense. Most studies on clientelism focus on explaining its effect on democracy and development. They tend to ask: how does clientelism affect democratic representation? How does clientelism affect electoral outcomes? How does clientelism affect programmatic politics? How does clientelism affect economic development? These are usually follow up questions to more overarching questions that ask: why are developing countries slow at democratizing? Why has development stalled in developing countries? This study, by contrast, seeks to explain not how clientelism affects democracy and develo pment but how urban development in local government jurisdictions affects the different clientelistic strategies and the usage of clientelism by politicians. It seeks to shape the way politicians use different clientelistic strategies. When I initially started this research I wanted to find out the way politicians government councils. The goal was to probe the nature of problems that small business owners in the informal sector encounter and whether politicians address the problems.
15 Most of the business owners mentioned that politicians never address their problems apart from some getting some h elp during election periods. Most are visited only for campaigning purposes. However, there were a few business owners who have been visited frequently and offered assistance frequently. These responses varied widely across t he three cities I was studying. understand the variations across the cities with regards to some important issues: why politicians visit some business owners during election seasons only and visit some during non election seasons; why some business owners have managed to frequently get access to politicians and frequent visits from them and some do not, and how these factors vary between different levels of pol iticians; why citizens in one city complained about the quality of local government services more than citizens in another city, and how all these factors are related. These are the main issues that this dissertation addresses.
16 CHAPTER 1 INTRODUCTION Urba nization is progressing rapidly in most African countries (Cohen 2006, 71; Kessides 2005, ix; Montgomery 2008; 761; Owusu and Otiso 2008; UN Habitat 2010). One of the most important consequences is a significant increase in the populations that politicians must reach out to in urban areas. This should fundamentally change clientelistic relationships, that a large literature consistently find to be the most common way leaders relate to citizens in African societies (e.g. Bratton and van de Walle 1997, 65 67; Hyden 2004; Lemerchand 1972; Van de Walle 2007; Vicente and Wantchekon 2009, 293). Yet, there is a dearth of studies of the effect of urbanization on clientelism on the continent. This is the gap that this study seeks to fill. The focus is on the informal sector, where a vast majority of the urban populations in Africa earn a living (Meagher 1995, 259; Potts 2008, 156; Gilbert and Gugler 1992; Potts 2008, 155; Werna, 2001; Becker 2004, 3). These are the citizens that politicians must establish linkages wit h since to garner the votes needed in elections. It is therefore imperative to establish if, and if so in which ways urbanization transform the nature of clientelistic relationships informal business people have with politicians. A key contribution of the study is that it systematically distinguishes two types of based on long version defined by short term, quid pro quo, and non personalistic relationships such as the vote buying during election periods. Secondly, this approach challenges and improves the way modernization theory links development and clientelism. That theory argues that clientel ism in general should disappear with the modernization of societies
17 (Kitschelt 2000, 867; Papakostas 2001, 34), suggesting a linear relationship between development and traditional norms such as clientelism (Jones 2005, 989 990). By systematically distingu ishing types of clientelism, this study makes it possible to demonstrate that the overall phenomena does not necessarily disappear but that it takes on different forms as urbanization progresses. One form of clientelism can be rare while another type is pr evalent. It also shows that there is a non linear relationship between clientelism and urbanization when the two types are compared. In a case where there are three levels of urban development low, moderate, and high modernization theory would suggest that clientelism should be high in a low development context, moderate in the moderately developed context, and low or absent in a highly developed context. This study shows that both modes are present in one context and with different levels. The traditio nal mode still exists even with increasing development. The modern mode takes a different pattern from the one modernization theory would predict, whereby clientelism is high in the moderately developed context, moderate in the low development context and low in the highly developed context. A third contribution consists of studying the types of clientelistic relationships citizens have with local and national politicians simultaneously. Our current understanding of political clientelism in Africa is to a l arge extent based on examinations that focus on countries as a whole and political representatives lumped in one broad 2011; Whitehead 2012; 1087; Ichino and Nathan 2012 ; Calvo and Murillo 2004; Eisenstadt and Roniger 1980; Fatton 1986; Fox 1994). This insight was originally provided at a time when governments were highly centralized and the lens has not
18 changed over the years. Yet, the wave of decentralization sweeping across most African countries (Boone 2003, 355; Bratton 2012, 516; Kauzya 2007, 3; Ribot et al. 2010, 35; Wunsch 2001, 277) is changing the political landscape in fundamental ways. Hence, there is need to improve our understanding of clientelism and also m ake it relevant for future policy making. This study examines how the two types of clientelism change in different urban levels of development and compares how politicia ns at the national and the local level differ across urban developments in their use of the traditional and modern types respectively. The analysis demonstrates the effects of decentralization by examining the role that local governments play in the transf ormation of clientelism. The occurrence of clientelism in the local government jurisdictions differs from the occurrence commonly shown in studies that focus on the national government level. These studies have shown the existence of a linear relationship between development and the modern mode of clientelism. On the contrary, this study shows a non linear relationship at the local government level. In the latter studies, as the level of development increases the level of modern clientelism correspondingly decreases, usually plotted in a straight declining line. In this case, clientelism is low where development is high, moderate where development is moderate, and high where development is low. This study shows a different pattern. It shows that in the three local government areas the highest level of clientelism is in the moderately developed context, moderate where development is low, and low where development is high.
19 1.1 Urban Development and Clientelism: The Missing Link How is the increasing developm ent of urban areas in African countries affecting clientelism? This is a timely and important question that is rarely asked. Almost all urban areas in Africa are undergoing several infrastructural and real estate developments. As these developments occur, citizen politician clientelistic interactions are also undergoing various transformations. From city to city, political and socio economic factors that feed and sustain clientelism are changing, such as levels of income, levels of education, employment opp ortunities, demands from citizens to politicians, and provision of public services. In spite of these changes we still do not know how the levels and modes of clientelism vary with regards to the increasing development of urban areas. This study aims to co ntribute to our understanding of how urban development infrastructural and property developments within local governments affects clientelism. Urban development in this case is approached from two perspectives. First, it is approached from the local go infrastructural projects meant to increase and maintain service provision. Second, it is activities and property development. Viewed from a governance perspective, the argument that this study is making is that clientelism is a societal problem associated with lack of adequate urban development initiatives at the local government level. Scholars who write a bout clientelism tend to take as a given the economic and the relationship between poverty and clientelism (e.g. Auyero 1999, 2001; Calvo and Murillo 2004; Michie 1981; Powell 1970; Scott 1972) but never pause to first ask how poverty varies with regards to variations in infrastructural and real estate features, and
20 how these variations subsequently determine the clientelistic strategies that politicians decide to u se in one urban area over another. Some analyze the effect of clientelism on public service provision at national government levels (e.g. Diaz Cayeros et al. 2001; Gibson 1997; Johnston 1979; Keefer 2007; Kurer 1993; Medina and Stokes 2002 Robinson and Ver dier 2002; Tullock 1980) but fail to investigate the effect on public service at local government levels. Some scholars provide accounts of how clientelism varies at a loca l and central government level (e.g. Morse 1962; Silverman 1965) but there are hardl y scholars that analyze how it varies geographically, especially across urban areas. These are the gaps in the existing literature that this study seeks to fill. One major reason for these shortcomings in the political science literature on clientelism is the oversight of theories from urban economics. As a branch of microeconomics, urban economics provides tools for explaining a wide range of urban variations in city develo pment with regards to size and location, the root causes of economic growth and decline, and most importantly for this study, how local governments affect urban economic growth. Scholars who study clientelism implicitly employ ideas from these theories whe n they use economic factors as variables, such as government expenditures (e.g. Calvo and Murillo 2004; Grzymala Busse 2008; Keefer and Khemani 2004; Kurer 1993; Wantchekon 2003), economic policy (e.g. Keefer 2007; Medina and Stokes 2002; Robinson and Verd ier 2002), and levels of income (e.g. Auyero 1999; Michie 1981; Powell 1970; Scott 1972). While urban economic theories explain how urban development transforms these factors, scholars of clientelism investigate the relationship that exists between them an d clientelistic practices.
21 The gap that exists between the two theories suggests a need to systematically connect them so as to appropriately address the central question guiding this study: How do the changes in urban development within local government jurisdictions affect the levels of clientelism and the modes of clientelistic strategies used by politicians? 1.2 Structure Agency Approach This study utilizes an interactional structure agency approach to connect the two theories. Such an approach links the behaviors of actors with social structures in an insightful way (Carlsnaes 1992, 246). It facilitates a better understanding of the conditions that transform the behaviors of political elites (Adeney and Wyatt 2004, 6). Structures can be defined as b road social factors that constrain and limit human agency (Simmonds 1989, 187). They consist of rules and resources that are cont inually applied in reproducing social systems (Giddens 1984, 377; Steen et al. 2006, 306). Agency is an action by actors who ac t independently to create and implement decisions that with others and against others, to form collective projects, to persuade, to coerce, and to monitor the simu 21). At a personal level, agency is concerned with the extent to which individuals feel that they can proactively transform their own lives (Woolley 2009, 10). Structures and agents are in terconnected through a reciprocal (Goldspink and Kay 2007, 3), recursive (Lewis 2002, 18) and mutually dependent relationship (Hays 1994, 65; Imbroscio 1999, structural fa ctors (Goldspink and Kay 2007, 10) and these factors subsequently affect political outcomes (Imbroscio 1999, 46). They shape how political leaders respond to
22 citizens preferences (Imbroscio 1999, 48) by influencing the utility of available options (Koopman s 2005, 20). Conventionally, structural factors have received most of the recognition for social and political outcomes (Adeney and Wyatt 2004, 1; Goffman 1983, 8; Hay and Wincott 1998, 952; Imbroscio 1999, 46). The shortcoming of structural deterministic approaches, where broader social forces are viewed as transcendent to individual behaviors, is that they are too rigid (Hays 1994, 61) and unreliable (Imbroscio 1999, 48) since they downplay the significance of human actions (Imbroscio 1999, 46; Jasper 200 4, 6; Sewell 1992, 2). Purposive and innovative actions of individuals play a significant role in reproducing and modifying structures (Alexander 2001, 252; Blumer 1969; Douglas 1970; Giddens 1984; Goffman 1974; Mehan 1990). Such actions can also constitut e structures (Sewell 1992, 4). This inter related characteristic of agents and structures makes it impossible to account for one without alluding to the other (Adeney and Wyatt 2004, 6; Carlsnaes 1992, 246; Jasper 2004, 7; Lewis 2002, 19). According to Gid structural properties of social systems are both the medium and the outcome of the political outcomes, therefore, it is essential to use an interactive analytical approach that takes into account the reciprocal effect of agential and structural factors (Jasper 2004, 4; Sibeon 1999, 18; Wharton 1991, 373). In this regard, structural factors should be viewed as outcomes of human actions (Cerny 2000, 437; Stone 1989, 10) and then viewed as constrainers (Lewis 2002, 17; Sewell 1992, 20), enablers (Giddens 1976, 161), and enhancers (Adeney and Wyatt 2004, 7; Hays 1994, 61) of these actions.
23 The study employs the interactional agency structure approach in three steps. First, it explains the role that human individual and collective initiatives have played in stimulating urban development. Second it analyzes and interprets patterns of urban development resultin economics theories. Finally, it uses a combination of theories on clientelism to analyze the effect of urban development on citizen politician clientelistic linkages in the three cities. Bridging this disconnect between the theories of urbanization and clientelism through such an approach can provide a more complete and insightful understanding of how urban development shapes clientelism. Such a move can also enlighten us on the political importan ce of urban development in African countries that are known for clientelism, with important policy making implications. 1.3 Clientelism Clientelistic transactions are usually characterized by four key aspects: they are dyadic, personal and long lasting, as ymmetrical, and reciprocal (Muno 2010, 21). This aspects are the basis of the distinction made between the two modes. The earliest studies that introduced the concept, especially from cultural and social anthropology (Kobayashi 2006, 1), highlighted the st rong, long term relational bonds that existed between patrons and clients as the underlying foundation that sustained the exchange of goods and favors between them (e.g. Eisenstadt and Lemarchand 1981; Eisenstadt and Roniger 1984; Gellner 1977; Gellner and Waterbury 1977; Mason 1986; Weingrod 1968). But in the recent years, particularly after the trends of democratization in developing countries, the growing prominence of formal institutions has led to a different application of clientelism. It is mostly hi ghlighted as a problematic informal institution
24 that undermines democratization because it is used as a mechanism for illegally buying votes during elections (Hopkin 2006, 3). Following leading scholars on clientelism (e.g. Boissevain 1966; Eisenstadt and Roniger 1980, 1984; Gellner and Waterbury 1977; Hyden 2004; Kaufman 1974; Kitschelt 2000; Lemarchand 1972; Piatonni 2001a; Scott 1972; Tarrow 1967; Wantchekon 2003; Weingrod 1968), this dissertation develops a conceptualization of clie ntelism that has two dimensions: The different forms it exists in and the time period they are used by politicians With the time they are used being the basis for the dimensions, a distinction should be made between clientelism during election and non election seasons (Scott 1972, 99). This dichotomy of clientelism is a common feature in the works of scholars from anthropology and political science. Weingrod (1968, 380) highlights the existence of this dichotomization by pointing out that clientelistic logists is an enduring relationship, while in the political science clientelism is based on enduring relationships it can be referred to as relational and when it is only enunciate d during elections it is non relational. The latter tends to have transactions that occur during election seasons only, while the former tends to have transactions that occur during non election seasons also. In recent studies (e.g. Keefer 2009; Lindberg a nd Morrison 2008; Van de Walle 2007; Vicente and Wantchekon 2009; Young 2009) this dichotomy is, however, not strictly observed although clientelistic interactions still have this character. An important contribution of this dissertation is to introduce a new way of thinking about the occurrence of clientelism around the two
25 seasons. This study differs from other studies by the way it distinguishes this occurrence using two distinct modes. What stands out from recent studies on clientelism in newly democra tizing regions (e.g. Keefer 2007; Lindberg 2010; Van de Walle 2007; Vicente and Wantchekon 2009), such as Africa, is the weak and short term nature of bonds between patrons and clients that are established for the purpose of buying votes (Wantchekon 2003, 400). The contradicting image in the early studies versus the one in the recent studies is one with no attachment to elections but due to the strong relational bonds (Eisenstadt and Roniger 1984, 48 49; Lande 1977; Mason 1986, 489), whereas in the new form of clientelism the responsiveness is tightly linked to elections and strong relational bonds are not always existent (Hopkin 2006, 5). This different characteristics of clientelism is what leads to the two modes compared in this study: traditional and modern. A useful distinction that is related to elections can be made between the two modes of clientelism. The traditional mode of clientelism is characterized by the occurrence of private interactions and transactions between citizens and politicians 1 throughout non election seasons. In this mode of clientelism, politicians selectively visit citizens that they have strong, long term bonds with and offer them personal a ssistance in exchange for their loyalty at the polls. These features are commonly highlighted in studies that approach clientelism from a relational perspective (e.g. Eisenstadt and Lemarchand 1981; Hopkin 2006, 3 4; Kaufman 1974; Piatonni 2001b, 9). On th e contrary, the modern mode of clientelism is characterized by the occurrence of 1 Such exchanges also take place between other actors in a society. But the focus of this study, which compares clientelism during electio n and non election seasons, is on citizens and politicians.
26 momentary interactions between citizens and politicians that are restricted to election seasons. Politicians only visit citizens to campaign and secure votes by distributing p rivate favors. In this mode of clientelism, the relational bonds between citizens and politicians are weak. Hence, scholars who study its occurrence during elections (e.g. Allum 1997; Hopkin 2006, 8; Kitschelt 2000; Kurer 1993; Lindberg 2010; Vicente and W antchekon 2009) tend to highlight its non relational features, in the sense that they pay more attention to the nature of transactions than the nature of relational aspects. Another important dimension that this study adds to distinguish the two modes of c lientelism is the presence or absence of a reciprocal link between a citizen and a politician. This type of link is evident when a citizen successfully visits a politician several times and is also visited by a politician. As this study will show later, th is link exists in the traditional mode but not in the modern mode. The above distinction between the modes of clientelism is not always clear in the literature on clientelism. This is because scholars fail to systematically separate them. Usually, in stud ies on clientelism and its effect on democratization (e.g. Keefer and Vlaicu 2008; Van de Walle 2007; Young 2009) or development (e.g. Hyden 2004; Medina and Stokes 2002; Roniger 2004) scholars overlook the distinction. The modern mode is however more prom inent. Scholars also never show how the mode of clientelism politicians choose to use varies with regards to their hierarchical position. This is another gap that this study seeks to fill. I argue that the hierarchical position of a politician, that is, wh ether it is an M.P., a Councilor, a Mayor, or a Chief, has varying implications on the mode chosen. The variation of the political offices that these politicians hold local versus national suggests that they have different clientelistic
27 strategies, yet most studies (e.g. Chappell 1989; Kitschelt 2000; Paul 2008) treat them in a unitary fashion. Scholars also tend to lump clients (e.g. Auyero 1999; Fatton 1986; Keefer 2009) but never find out how politicians employ the two clientelistic modes with regard among the poor. My contention is that the mode that is chosen is preferred because of its effectiveness with regards to two factors: the resources available to the politician, based on the hierarchical position, and the effect of urban development on the socioeconomic status targeted clients. 1.4 Urban Development Most investigations of the relationship between development and clientelism in African countries focus on national level aggregate variables rather than local government level variables: for example, gross domestic product (GDP), gross national product (GNP), and foreign direct investment (FDI). The shortcoming of national level aggregate variables is that they are una ble to depict important micro level developmental processes occurring at the local level. The present study focuses on the variables usually merged within the aggregate variables (GDP, GNP, and FDI), that is, local government expenditures on infrastructura l projects, citizen investments in small businesses and real estate. It is important to study the effect of local level aggregate variables so as to better understand micro level factors that sustain clientelism. Because national level variables mask thes e factors they are unable to offer substantive ideas for the development of policies that deal with the undesirable outcomes of clientelism, such as the lack of democratic accountability and democratic representation (Keefer 2007, 804; Keefer and Khemani 2 004, 372; Kurer 1993, 261; Lindberg 2010, 125). Thus, whereas trying to understand general patterns of clientelism across
28 countries has proven to be very useful due to its ubiquitous nature (Van de Walle 2007, 2), limiting our theory building efforts to e xplaining the effects of national level development variables risks leaving us without a better understanding of the relationship between local level development variables and clientelism. A goal of this dissertation is to introduce a new and supplementary approach that focuses on local government urban development initiatives to explain clientelism. Urban development schemes usually lead to urban growth, which has two components: economic growth and employment growth (Black and Henderson 1999, 253 Sullivan 2009, 90). Economic growth occurs when the average income in a city increases and employment growth occurs when the total workforce in a city capital deepening (Burmeister and Turnovsky 1972, 842), increase in human capital (Ito and Fukao 2005, 7; Storper and Scott 2009, 164), and agglomeration economies (Broersma and Dijik 2008, 183). Capital deepening occurs when the amount of physical capital in an area incre ases; human capital increases when the knowledge and skills workers have increases due to their education and work experience, and agglomeration Agglomeration economies ca n occur within one industry localization economies, or across different industries urbanization economies (Moomaw 1983). With regards to agglomeration economies, this study focuses on urbanization economies, which is more relevant because it applies to this study focuses on. The combination of capital deepening, human capital, and
29 90). This occurs through the exp ansion of physical structures that increase per capita income and total work force in an urban area (Bertinelli and Strobl 2007; Black and Henderson 1999; Cohen 2004; Storper and Scott 2009; Waddell 2007). Physical structures include features such as roads buildings, drainage systems, and electric features affects the clientelistic mode politicians consider to be effective. Urban development also increases the rate of ur banization, which is different from urban development but inter related to it. Whereas urban development refers to infrastructural and property developments, urbanization is the process of growth in the urban proportion of citizens relative to rural propor tion (Hope Sr. 1986, 41). Urbanization is characterized by two major factors: an increase of the urban population and rural urban migration (Hope Sr. 1986, 42; Todaro and Smith 2009, 320). Urban development The positive association between countries rural populations have been increasingly seekin g work in services and manufacturing sectors located in the urban areas because of the clustering of firms in cities (Davis and Henderson 2003, 99). The demographic outcome is highly dense urban areas, which in turn affects citizen politician linkages. Thi s dissertation thus views chosen not as the outcome of random decisions but as the product of intentional decisions meant to maximize the payoffs of clientelistic goods distributed, which are usually in the form of votes (Kurer 1993, 265; Medina and Stokes 2002, 3; Robinson
30 and Verdier 2002, 2). A particular strategy is preferred because of the challenges and opportunities present in an urban context. Framing the ro le of urban development this way highlights its constraining effect on politicians, in the sense that an urban context ceases being just a political territory and becomes a set of economic structures that influence their behaviors. Urban development theref ore becomes a mechanism that affects the way politicians choose to establish linkages with citizens. 1.5 Mechanisms of Urban Development and Clientelistic Strategic Considerations Figure 1 1 illustrates the process in which mechanisms of urban development affect clientelism. Figure 1 1. The effect of urban development on c lientelism 3 2 1 Local Government Urban Development Initiatives L.G. as a territory where citizens conduct projects L.G. as an administrative unit initiating development projects Urban Growth and Urbanization Economies Urbanizati on and growth of businesses in the informal sector Transformation of socio economic factors: level of incomes and level of education. Transformation of Clientelism Occurrence of clientelism Modes of clientelism: traditional versus modern
31 This dissertation shows how the two components of urban development economic growth and employment growth affect the level of clientelism an d the mode of clientelism employed by politicians in two stages. In the first stage it shows how variations in urban development lead to variation in socioeconomic status. In the second stage it shows how politicians, depending on hierarchical position, ch oose one clientelisti c mode over the other. Figure 1 1 above illustrates how mechanisms of urban development affect clientelism. The argument depicted is based on one simple proposition: politicians usually target the poor and uneducated for clientelistic transactions (Auyero 1999, 311; Dixit and Londregan 1996; Hopkin 2006, 8; Kaufman 1974, 300; Kitschelt 2000, 849; Penfold Becerra 2007, 67). Everything else being equal, it should be cheaper to buy the loyalty of the poor than the rich. The proposition th suggests that clientelistic politics can be viewed in economic terms and that the choice of clientelistic mode can be seen in terms of the quest to use a strategy that will be effective in li ght of economic conditi ons within an urban area. To the extent that the successfulness and effectiveness of one mode is determined by the cost of clientelism, the preferred mode for a politician will be the one that is less costly and still offers appealin g clientelistic contracts to citizens. In order to select the mode that is politically and economically efficient they will, therefore, consider three important factors: the value of goods they have to offer their targeted citizens to convince them to sel l their votes, the total population they have to select from, and the resources available to them. In most African countries access to abundant resources is determined by the hierarchy of the politician (Baylies and Szeftel
32 1982, 196; Holm 1987, 27; Mwendw a and Tangri 2005, 459; Ronning 2009, 161; Webster 2007, 50). This should lead to variations in the preferred clientelistic mode. The effect of these resources can be either increased or decreased by the extent of urban development. With reference to the argument depicted above, urban development matters in three ways. First, where it is high the socioeconomic status of citizens may undergo higher improvements. In such a context, clientelistic goods are less attractive and less useful as a basis for seekin g votes. But where there is low change in urban development clientelistic goods are more attractive and useful for seeking votes. Second, it matters because when higher levels of urban development lead to higher levels of urbanization it increases the pop ulation of citizens politicians have to choose clients from. When this population is high the total number of demands from citizens can also be high. In order to respond to these demands they must take into consideration whether it is strategic to respond through private goods or public goods (see Lindberg 2010). If they respond through private goods, in exchange for votes, in a highly populated area the strategy may be ineffective. This is because the total cost of recruiting a meaningful ratio is too high Using this strategy, it is only possible to recruit a few citizens, and this ratio compared to the overall population can have an insignificant impact at the polls. On the other hand, if they distribute public goods they can cast a wider net and capture more citizens, leading to a significant impact at the polls. It is a more efficient way for them to win votes in highly urbanized contexts since the congestion makes it possible to appeal to a large population with a few public goods.
33 A third way that urb an development matters, which is associated with 165). In the past 50 years thi s trend has been a common trend in developing countries (Bertinelli and Strobl 200, 2500). This outcome undermines that ability to successfully execute the traditional mode of clientelism, which depends on frequently visiting and monitoring clients during non election seasons. This is an impossible task in high density areas. In such a context the modern mode of clientelism, based on one time strategic visits, can be more successful. However, the modern mode can also be successfully executed in a low densit y area. Thus along all three channels higher urban development should lead to lower incidence of traditional clientelism and increase in either modern clientelism or public goods provisions. Fi gure 1 2. The effect of urban development on clientelistic t ransactions Urban D evelopment: Expansion of infrastructure (e.g. roads, electric grids) and real estate in local government jurisdictions Moderation of Clientelism: Politicians are less likely to buy votes from citizens socioeconomic status: Dri ving factors: capital deepening, increase in human capital, technological progress, and agglomeration economies
34 Viewed as separate structural and agential components, under the interactional structure agency approach that this study employs, the factor that links the effects of urban development to the changes in clientelist ic practices is the change in citizens s ocial economic status. Figure 1 2 illustrates this linkage. The following discussion elaborates on this connection and then derives the hypotheses and research questions that guide this study. 1.6 Low Socioeconomic Status Attracts Clientelism The vast literature on clientelism points out that clientelistic politics is mostly attractive in conditions of low economic productivity (Fox 1994, 154; Wantchekon 2003, 400; Robinson and Verdier 2002, 4; Calvo and Murillo 2004 742). Hence, clientelism has been one of the most prevalently used concepts for studying the nature of politics between politicians and the poor in such contexts (Auyero 1999, 297; Medina and Stokes 2002, 18). At its core, clientelism is undergirded by s ocioeconomic factors. on clientelism directed scholarly focus on the role of socioeco nomic development by pointing out that material insecurity largely accounts for the origin of clientelism (Lemarchand and Legg 1972, 155). This observation has continued to hold true in most cases where clientelistic relations are prevalent. Generally, cl ientelistic relations are undergirded by hierarchical structures of dependence based on inequality (Auyero 1999, 298). These relations structure both interpersonal relations and the flow of resources between politicians and citizens. Based on this understa nding, scholars have used it extensively to explain why the poor choose certain political leaders for both electoral as well as non electoral reasons on the basis
35 wi concept of patronage gained prominence as a framework for analyzing peasant societies (Weingrod 1968, 377). The aspect that contributed to the prominence of this concept is the socioeconomic underdevelopment of such contexts, which made peasants reliant on patrons for survival and patrons reliant on low cost of services from the poor. Some of the early studies on clientelism in political science that focused on peasant soc ieties (e.g. Powell 1970, 411) alluded to the low productivity of their economic activities as the main cause of their vulnerability to several livelihood threats and hence their subsequent reliance on patrons for survival. This is still the same logic tha t we can say sustains clientelism in less urbanized contexts today, where economic activities of the poor are less productive than in more urbanized contexts. These economic variations shape how actors evaluate the value of votes (Hopkin 2006, 5; Piattoni 2001a, 17). In urban contexts where levels of income are high the bidding price for votes can be high as well. This discourages patrons from using clientelistic strategies to seek votes. On the contrary, where the level of income is low the bidding price can be low enough to encourage patrons to use such strategies. Urban contexts with fewer resources for the poor to improve their socio economic status makes them vulnerable to targeted incentives during election periods in exchange for their votes (see Kit schelt 2000, 849; Robinson and Verdier 2002, 3). On the contrary, in contexts where development is high there are more resources that provide a basis for improving socio economic status. Hence, the distribution of targeted incentives during elections in
36 ex change for votes is likely to be avoided by politicians because the overall cost of incentives can be too high. Clientelism can be viewed from several perspectives depending on the purpose of each study. This study uses a rational choice and a functional p erspective to show how the level of urban development generates different clientelistic outcomes. From a rational choice perspective, clientelism is an exchange undergirded by economic principles whereby those with higher economic status patrons use th eir resources to purchase support and assistance from those with lower economic status clients (Scott 1972, 92). From a functional perspective, clientelism is a mechanism that is used by the poor to advance their socio economic status (Mair 1961, 6). Oth er scholars (e.g. Fatton 1986, 7) perceive patron dependence that persist for long periods (Auyero 2001, 213; C lapham 1993; Le Vine 1980; Lewis 1996). A rational choice logic for why clientelism thrives in underdeveloped chains, and prize instant advantages such that the appeal of direct, clientelistic On the contrary, citizens who are in developed contexts, who for the most part tend to be educated and affluent, face higher opportunity costs whe n choosing clientelism and so they demand rewards that are more expensive (Kitschelt 2000, 857). In turn, their expectations make clientelism increasingly expensive for politicians. Because politicians have limited resources they will either choose to rely less on clientelistic strategies or
37 with different skills and private market expectations is critical to understand the electoral benefits that political parties can e I argue that when development occurs through the increase of private and public economic activities in urban areas it changes rati onal calculations for both citizens and politicians and subsequently weakens clientelistic linkages. This is more likely to be the case at when clientelism is examined at the aggregated level. At a disaggregated level, the relationship between urban develo pment and clientelism may vary depending on other trade offs made by politicians between cost and population size. According to developmentalist arguments, clientelism is a prevalent form of linkage mechanism between parties and citizens in poor countries, but once develop ment occurs these mechanisms lo se prominence (Kitschelt 2000, 857). This is because affluent citizens are more likely to defect from clientelistic contracts. The less affluent, on the other hand, can be mobilized through targeted personal incentives (Wilson 1973, 72). Political parties in developing countries usually take into consideration these factors when mobilizing citizens (Kitschelt and Wilkinson 2007, 24). The lack of private economic activities leads to monopolistic parties that ci tizens heavily rely on for goods and services; low productivity of private economic activities also makes clientelism an attractive strategy for parties seeking to buy votes from citizens (Robinson and Verdier 2002, 24). In developing countries, lethargic economic development makes it possible for clientelism to serve as a mechanism for parties to monopolize resources that citizens do not have (Medina and Stokes 2002). Hence, an increase in private economic activities that avail these resources and an incre ase in public services can undermine
38 argued that modernization leads to programmatic parties. To be clear, although this study may sound similar it is not making the claim that the lower levels of clientelism in more urbanized areas are associated to programmatic parties. Instead, it only highlights The relationship between political parties in clientelistic societies has traditionally been characterized by the exchange of votes for personal favors, such that during election periods clientelism usually involves the exchange of votes for cash (Vicente and Wantchekon 2009, 293). For this matter, analysis of clie ntelism usually focuses on the efforts that parties make to buy votes during election periods and the vulnerability of the poor to clientelistic transactions. Economic deprivation and dependency on political parties usually sustains the grip that clienteli stic politics has on the poor (Auyero 1999, effect on party patronage (Warner 1997). It affects their decision on whether to use clientelism or programmatic appeals. As Lem archand and Legg (1972, 155) note which subsequently makes them politically mon opolistic (Medina and Stokes 2002). Political monopoly over resources makes it possible for political parties to have a grip on clients through clientelistic contracts (Park 2 008, 117; Weitz Shapiro 2007, 6 7). Party monopoly over resources is usually a fe ature of less urbanized areas, whereas in more urbanized areas party oligopoly over resources may be the case due to intense multiparty competition with several patrons having control over resources (Muno 2010,
39 6). This intensity is what may lead to an inc rease in the cost of clientelism as politicians compete to offer the best deal to clients. The resultant effect of such competition may be a withdrawal from clientelism by politicians who cannot afford to offer best deals and benefit from patronage, an iss ue that is also associated with party budgets. Budget constraints force parties to take into consideration a wide range of economically related factors that affect the success of clientelism, such as level of employment, level of skills, and market expec tations of citizens (Calvo and Murillo 2004, 751). Following from this, the expectation is that in contexts where levels of income and skill are low, parties can maximize the benefits from the goods they are distributing because they are able to use lesser resources to recruit several clients. On the contrary, when levels of income and skill are high, the benefits are minimal. The empirical implication is that we should expect to see a high percentage of citizens with low incomes and low skills in an urban area to be associated with higher levels of clientelism. Conversely, a high percentage of citizens with high incomes and high level of skills should be associated with lower levels of clientelism. This is because the budget constraints that parties face fo rces them to consider the tradeoff of offering alluring goods to only a few citizens with high incomes or to a larger group of low income citizens. Local government development initiatives play an important role in the transformation the socioeconomic fac tors that sustain clientelism. From this standpoint, clientelism indirectly depends on the way local governments effectively manage their ell managed urban areas facilitate economic activity
40 and make it possible for citizens to gain necessary access to utilities and services, subsequently leading to income generation (Rakodi 1991, 542). Urban management in ies which together shape and guide the social, physical, responsibility of [local] government and [it] is concerned with all aspects of urban development, both public and private. It is no way confined to the services operated by through employment creation. They play an important role in addressing scarcities of infrastructural development that hinder b usiness investments (Hansen 1990) by the urban poor is to enhance their limited access to the full range [local government] Inadequacy of these services can lead to high levels of economic insecurity for citizens. As a result, citizens become dependent on parties for personal favors for such services or rely on other privately provided services like those provided through community organiza tions. When opportunities availed by political, economic, and administrative institutions increase, individuals act upon them and this undermines patron client systems (Michie 1981, 21), especially through the development of market economies (Kessinger 19 more extensive the penetration of government into the periphery and the wider scope of s infrastructural development and extension of service provision can moderate the occurrence of clientelism. An increase in sources of needed
41 products and services by clients makes them less dependent on clientelistic networks (Lemarchand and Legg 1972, 16 2; Michie 1981, 21). According to Michie (1981, 24), the active involvement of the national level of government in the local government level forces patron client network structures or systems to respond to new circumstances. The responsiveness is mainly t riggered by an influx of alternative resources that shifts better public services the marginal utility of clientelistic goods offered to them decreases (Keefer 2009, 27). The availability of alternative sources of goods and services due to local government development can therefore have an effect of increasing the bidding price for votes. An increas ed price for votes and a need for a high ratio of votes can make clientelism too costly for parties in highly more developed urban areas. Just like in any other economic transactions, the volume of clientelistic transactions is shaped by the economic chara cteristics such contexts, as well as the needs of clients (Lemarchand and Legg 1972, 155). Access to governments and market networks can provide a wider societies schola rs (e.g. Powell 1970) have found that the lack of access to government services leads to intense patron traditional village patron client relat urban development initiatives have the potential of weakening the level of clientelism.
42 1.7 Urban Development Improves Socioeconomic Status This section lays out the theoretical framework that guides the examination of the effect of urban development on clientelism in Kenya. It provides a logic for how urban development transforms the socioeconomic factors that tend to attract clientelism and introduces the variables that are empl oyed in this study. The ideas employed in this section borrow largely from urban economic studies that have been done on formal sectors and attempts to appropriately apply them in the informal sector, where this study was done. The rapid rate of urbanizati on in developing countries and the inability of the formal sector to create employment opportunities has led to increasing attention on this sector as a solution to the increasing rates of unemployment (Todaro and Smith 2009, 335). This sector is, therefor e, an important source of income for those who migrate from rural to urban areas (Becker 2004, 9; Todaro and Smith 2009, 335) and has generated high incomes for the urban workforce (Todaro and Smith 2009, 337). New informal sector business owners increase the productivity and survival rates of their businesses by locating in urban contexts with several suppliers and extensive markets (Henderson 2002, 97). While most of the studies that look at the relationship between clientelism and economic development te nd to focus on macroeconomic factors, this study uniquely focuses on microeconomic factors. These factors play an important role in the development of urban areas (Alonso 1968, 4). In essence, the discussion in this section is about drivers of socioeconomi c development in local government urban jurisdictions. include facilitation of small business es by making premises for operating businesses
43 available through the construction of markets (Rogerson 1999, 517; Stren and Gombay 1994). Government policies significantly influence the rates of urbanization through structuring sectoral composition, which entails a transformation of agricultural and manufacturing industries (Davis and Henderson 2003, 99). A higher level of manufacturing and services industry leads to clustering of industries, hence increasing the rates of urbanization (Davis and Henderson 2 003, 117). These policies tend to affect the degree of urban concentration (Ades and Glaeser 1995, 2; Henderson 2002, 107), especially through infrastructural investments (Renaud 1981, 62). An increase in such investments is a sign of successfully managed urban areas (Mabogunje 1993, 3). The economic impact of infrastructural provision usually translates to economic advancement of business enterprises (McGill 1998, 466). For example, a more developed transport infrastructure increases urban concentration, w hich in turn Urban development moderates clientelistic practices by economically empowering the poor through improving several social and economic aspects. Urban areas experienc e self sustaining economic growth if they have an extensive variety of economic activities, a large pool of skills, large local markets that attract suppliers, a large demand for services, and major infrastructural developments (Alonso 1968, 7). The expans ion of urban economies usually improves socioeconomic development by increasing employment opportunities, better learning opportunities, and higher social high productiv ity of labor because of better economic opportunities (Frank, Jr. 1968,
44 to credit. Econom ic activities of the poor in more developed urban areas tend to be more productive than for those who are in less developed urban areas because of availability of credit from financial institutions (Chandavarkar 1985). The level of credit rationing in more urbanized areas tends to be relatively low compared to less urbanized areas because of low transaction costs (Fay and Opal 2000, 8). Dwellers in more urbanized contexts can, therefore, access credit easily. As a result of availability of credit, businesse s increase at a higher rate, which in turn leads to diverse economic activities. The diversification in goods produced and traded within a city is what contributes to its economic growth (Glaeser et al. 1991). This is mostly the case in large cities becaus 77). In these cities there is a large consumer base that supports the production and Scholars of urbanization have shown th at urban development is positively associated with improvements in human conditions (e.g. Njoh 2003, 170; Fay and Opal 2000; Todaro and Smith 2009, 321; Polese 1997, 1). This is because income levels in more urbanized areas tend to be higher than in less u rbanized areas (Njoh 2003, 173). Urban development stimulates growth in income per capita by making it possible for firms to benefit from large markets, inputs for their products, and better social services (Miro and Potter 1980, 124). The supply and deman d for public services also increases (Frank, Jr. 1968, 250). Studies have found that in the early stages of development, urban concentration 2 can lead to positive economic outcomes in developing countries 2 Urban concentration refers to the fact that the distribution of the population is highly concentrated in one city.
45 (e.g. Bertinelli and Strobl 2007, 2501; Davis and Henderson 2003, 99). As Alonso developed centers, which enjoy the existence of overhead facilities, external weal thy in the hinterlands to manufacturing and service industries in more developed urban areas foster positive externalities, such as knowledge spillovers and human capital accum ulation (Bertinelli illovers are critical for growth efficiency because they lead to higher returns in private human capital (Black and Henderson 1999, 253; Stroper and Scott 2009, 148). Most of the times it is the educated citizens that experience these higher returns in urb anized areas (Fay and Opal 2000, 14). The combined effect of these factors leads to concentration of economic activities. Concentration of citizens in one city also leads to economic development due to the accumulation of knowledge and spillover of informa tion (Henderson 2003). In this regard, cities where business owners have high levels of education and better access to learning opportunities are more likely to have more business owners with higher socio economic status. It is well known that firms in de veloped countries tend to benefit from agglomeration economies 3 due to clustering (Davis and Henderson 2003, 117; Henderson 2002, 97). When firms that are from different industries cluster together they 3 The economic forces that cause firms to locate close to one another in cluste rs.
46 9, 81). Urbanization economies 43). Urbanization economies cannot be experienced in th e informal economic sector, which is the context of this study, in the same way that they are experienced in the formal sector. But it is still possible that the effects of urbanization economies exist in the informal sector also. In the formal sector, sc holars talk about different industries clustering in one area, such as automobile and banking industries clustered in one area carpenters, tailors, etc. cluster in on e area. 4 Just like the way firms in the formal sector learn from other firms involved in the same activity, so do these small industries in the informal sector (Bosire and Gamba 2003). In the formal sector learning mostly takes place through formal intera ctions (Todaro and Smith 2009, 327). In the informal sector learning takes place through informal interactions. 5 Although economies of agglomeration have been mostly associated with firms in the formal sector, some of the factors that are experienced by s uch firms can also be experienced by businesses in the informal sector. Just as the productivity of firms in a city is enhanced when other firms are nearby (Henderson 2002, 91), so can the business productivity of businesses in the informal sector enhanced when other businesses are located nearby. In highly urbanized areas, firms benefit from urbanization economies, associated with the overall scale of different industries in a city. The manner in which businesses in the urban 4 Based on observations made during my fieldwork. 5 Based on in depth interviews with business owners during my fieldwork.
47 informal sector share ideas an d resources used to produce a final product is analogous to the urbanization economies associated with firms in the urban formal sector. This is especially the case when there are several businesses with huge capital investments (capital intensity) in urba n informal sectors. Firms benefiting from urbanization economies also share ideas and specialized workers beyond their industry by interacting with specialists from other industries (Henderson 2002, 93). It is not uncommon to see the same interactions in t he informal sector, where a business producing large products that require various inputs will seek the expertise of other businesses specializing in various parts to help in completion of the final product. For example, consider a carpenter with a big or der for sofas. From my studies I have found that most of the times such carpenters do not have all the equipment and labor they need. So, a carpenter with such a big order will take the order through different businesses that specialize in the various part s needed to complete the sofas. This sharing of ideas and specialized workers is mostly successful in highly urbanized areas, which have a wide range of businesses that can benefit from each other. As a result, productivity and levels of income are higher in these urban areas as opposed to the less urbanized areas. In the less urbanized areas such sharing is mostly absent due to low urbanization economies. In highly urbanized areas businesses in the informal sector can also produce advanced and high profit products because of this sharing of ideas and specialized assistance from workers beyond their micro industry. Businesses can produce a wide range of products because urbanization economies make it possible to have access to other services that they cannot afford. Businesses that tend
48 to have the ability to produce high end goods and help other businesses usually have high capital intensity. In the formal sector, firms cluster in industrial districts, where they are able to contract out work when they have unusually large orders (Todaro and Smith 2009, 329). Based on my fieldwork observations, the same contracting happens quite frequently in the informal sector, especially in more urbanized areas where there is a variety of small industries. Clustering in th e informal sector follows similar patterns to those in the formal, where businesses involved in the same act ivity locate next to each other 6 Clusters of related small industries in the informal sector lead to spillovers from technological advancements amo ng businesses. For instance, if a carpenter whose business is doing well adopts a new machine for cutting wood faster, neighboring carpenters within the cluster will also benefit. It has been shown by scholars that when firms from the same industry locate next to each other they are able to handle unusually large work orders because they contract out some of it (Todaro and Smith 2009, 329). capacity, and arrangement tha 2009, 328). It is not uncommon to see small sized businesses contracting work to larger businesses in the informal sector. 7 These kinds of relationships are even more likely to be observed in more ur banized sectors. This interdependence can make it possible for businesses to end their reliance on politicians for clientelistic exchanges that they have used for survival in case of business failure. 6 This study was done in small business clusters in the informal sector. This makes it possible to use the theory of clustering t o explain business productivity outcomes in the sector. 7 I have been able to see these contracting activities during my field work.
49 Urban informal sectors in highly urbanized areas also d o well because they serve as suppliers of various inputs for firms in the formal sector (Chaudhuri 1989, 351; Portes and Sassen Koob 1987; Tokman 1978). Existing evidence from informal sectors in developing countries show that when they interact with the b usinesses in the formal sector they tend to experience high rates of productivity (see Mukherjee 1990 and Banerjee 1985). The earnings from the informal sector can, therefore, be competitive to earnings in the formal sector (Yamada 1996, 308), especially b ecause of input sharing. Input sharing is also one of the agglomeration economies that generate urbanization intermediate inputs such as banking, transportation services, and public infrastructure. It characteristics are present in the informal sector in more urba nized areas, where there is a dynamic interaction between the firms in the formal sector and businesses in the informal sector. Input sharing is advantageous because it generates higher business ). 1.8 Hypotheses and Research Questions The line of reasoning presented above, based on the ill ustration presented in Figure 1 2, leads to the following hypotheses: H1: Citizens in an urban context that has high degree of physical development, high level of human capital, and a great variety of businesses clusters are likely to have higher socio economic status than those in an urban context that has a low degree of physical capital, low human capital, and a low variety of businesses clusters. H2: Politica l parties and politicians are less likely to use clientelistic strategies to win votes in more developed urban contexts, where most citizens have high socio economic status. Instead they are more likely to use these strategies in
50 less developed urban conte xts, where the socio economic status of most citizens is low. If H 2 holds we should expect to see a low occurrence of clientelism in general in a highly developed urban contexts, moderate occurrence in a moderately developed urban context, and high occurr ence in an urban context that has low development. There is a logical connection between the three hypotheses. H1 is concerned with socioeconomic status conditions that, if present, should ensure that the observations hypothesized in H2 and H3 will follow. Once these hypotheses have been extensively examined confirmed the study examines the following additional questions in order to scrutinize the findings further: Q1a: How does the use of each clientelistic mode (traditional versus modern) by politicians v ary across urban contexts with regards to their different levels of development? Q1b: Do these two clientelistic modes have a different relationship with the degree of urban development. Q2: How do upper (Members of Parliament) and lower level politicians (Councilors) use the two modes in different urban contexts? In other words, which of the two politicians is more likely to use the modern mode, and which one is more likely to use the traditional mode? In which urban context is either politician likely to use one mode over the other? 1.9 Possible Objections and Responses The analysis presented here does not attempt to provide a comprehensive theory of how development shapes clientelistic strategies but attempts to address the oversight of urban economic fac tors that is missing in the literature on clientelism. At present, the literature on clientelism is also devoid of explanations on the local government conditions that transform clientelism. Due to the rapid urban development taking place in Africa (Cohen 2006, 71; Kessides 2005, ix; Montgomery 2008; 761; Owusu a nd Otiso 2008; UN Habitat 2010 ), as well as ongoing decentralization of governments (Boone
51 2003, 355; Bratton 2012, 516; Kauzya 2007, 3; Ribot et al. 2010, 35; Wunsch 2001, 277), we need to account for the effect of these factors. It may be objected that politicians do not always consider the rate of urban growth in their strategic calculations because they do not have precise statistics on how it transforms socio economic factors. One could also arg ue that politicians have difficulties in knowing the exact socio economic status of their targeted clients. Another point of contention could be that economic status and so politicians face a more complex and unclear situation. In addition, one could argue that there are other factors that explain clientelistic outcomes that this study has overlooked. These are important criticisms to consider, and I have reflected upon them and addressed them using various analytical approaches. am suggesting they simply take into consideration the general knowledge that it transforms socio economic factors and do not necessarily seek pre cise statistics. I also contend that because of the high stakes involved in clientelistic transactions, politicians are highly strategic (Calvo and Murillo 2004, 751; Fatton 1986, 63; Kitschelt 2000, 849; Lindberg 2010, 128; Robinson and Verdier 2002, 1) a nd at least seek general information on socioeconomic status of potential clients. What forces politicians to seek this knowledge are the risks inherent in clientelistic contracts (Medina and Stokes 2002, 4; Park 2008, 117; Weitz Shapiro 2007, 6 7) Without knowing their clients better they may offer goods to a client whose status makes them likely to be disloyal. Indeed, it may not always be easy to acquire exact information on socio economic status. But it is possible for politicians to conjecture based on observable conditions. For example, a
52 politician targeting poor business owners can make fairly accurate assumptions on their level of poverty by looking at the external features of their business premises. If a business is being conducted in a small make shift structure made out of card board high. On the contrary, a business being conducted in medium or large permanent structure made of concrete walls can suggest a lower level of poverty. Second, even if increasing urban development does not directly improve every citizens socio economic status as some skeptics will contend, we have to take into account that there are external social benefits and spill over effects that indirectly affect citizens that interact with those that have directly benefited from urban development (Black and Henderson 1999, 252; Cohen and Paul 2005, 216; Glaeser et al. 1991, 2; Rosenthal and Strange 2003, 377). These indirect effects make it possible for them to context where the increase in urban development is high those businesses that have not directly benefited are likely to strategize on how to take ad vantage of the fact that business location decisions (Alcacer and Chung 2007; Devereux et al. 2007; Head and Mayer 2004). For example, when the socio economic status o f a business owner in a rapidly developing context gets better his or her business input demands will likely increase, and as a result a poor small business owner supplying such inputs in that area is likely to benefit. The implication is that such interac tions lead to the reluctance towards accepting clientelistic offers by poor business owners who depend on them for survival.
53 Finally, although I primarily focus on urban development, I am also cognizant of existing studies that look at other explanatory f actors and I borrow from them in the chapters that follow. These studies have successfully shown how other cultural (e.g. Chappell 1989; Lemerchand 1972; Milne 1973), political (e.g. Johnston 1979; Keefer 2007, 2009; Keefer and Vlaicu 2008; Lindberg and Mo rrison2008;) and economic (e.g. Kurer 1993; Lemarchand and Legg 1972; Medina and Stokes 2002; Michie 1981; Powell 1970; Scott 1972; Robinson and Verdier 2002; Roniger 2004 ) factors affect clientelism, but what is missing in this literature is an analysi s of the relationship that exists between urban development and clientelism The goal of this is to present a new way of understanding why clientelism varies within a country. This insight is timely for both theoretic and practical reasons given the rapid changes in urban development taking place in Africa, a region that has been widely known for clientelism (Bratton and van de Walle 1997, 65 67; Hyden 2004; Lemerchand 1972; Van de Walle 2007; Vicente and Wantchekon 2009, 293). 1.10 The Structure of the Di ssertation The rest of this dissertation is organized in seven chapters. Chapter 2 provides a discussion on the case selection. It justifies the necessity of a single case study for the issue this dissertation address and provides details on the rationale used for selecting Kenya. Chapter 3 provides an analysis of the context in which the study was conducted. It provides a description and brief historical analysis of the informal sector in Kenya. It shows how the informal sector grew rapidly after independ ence and the role that the state and the mic rofinance industry played. C hapter 3 ends with a brief qualitative analysis of the nature of politics in the informal sector with regards to the relationship that exists between citizens and politicians. Chapter 4 introduces the local government
54 system in Kenya by providing a brief history of its evolution and the main components of the system that pertain to urban development. It then provides a brief history of urban development in the three cities as it attempt s to show the reasons why they vary today. Chapter 5 addresses the first and second parts of the central question of the dissertation. How does local government urban development vary across the three cities in Kenya? How are the differences in urban deve lopment related to variations in political party clientelism. Drawing on secondary data collected from the local governments of the three cities, it first examines the following variations: infrastructural development projects by the local governments, lev el of service provision, development of new markets by the local governments, level of property development by citizens, and new business development by citizens. Having established the nature of variations in urban development across the three cities it t hen addresses the second part of the central question : the variation of clientelism in the three urban areas. C hapter 5 employ s a broad variety of qualitative and quantitative data to analyze the variation in the level and modes of clientelism across the t hree cities. Drawing on primary data collected from business owners in the informal sector in Kenya using focus group discussions, interviews, original survey, and interviews with key informants who work with politicians, it analyzes the following issues: the effect of urban development on socioeconomic status and how political party clientelism across the three cities varies with regards to the differences in socioeconomic status. After Chapter 5 has shown the variation in political party clientelism with regards to level of urban development, Chapter 6 and 7 investigate how the mode of clientelism varies with regards to level of urban development and the type of politician (Member of
55 Parliament versus Councilor). Chapter 6 starts this investigation with a systematic conceptualization and operational ization of the two modes. C hapter 6 uses a wide selection of literature from political science, anthropology, and sociology to clearly show the features that distinguish the two modes and then provides the varia bles that were used to collect data based on these features. Once this exercise has been undertaken, Chapter 7 examines the following questions: In which urban area is the traditional mode of clientelism mostly practiced? In which urban area is the modern mode of clientelism mostly practiced? Which politician mostly uses the traditional mode of clientelism? Which politician mostly uses the modern mode of clientelism? In which urban context do the politicians use the traditional mode? In which context do the y use the modern one. contribution by showing the gaps that it has attempted to fill in the existing literature. It makes a case for the need to distinguish the two modes in future studies, as well as the need to pay more attention to the traditional mode. It ends with an exploratory study that attempts to make the case for the need to examine the relationship between urban development and clientelism in other Sub Saharan Africa countries. Due to the lack of comparable cross national data on clientelism this study uses corruption as a proxy, and show that there is a relationship that exists between urbanization and corruption that needs to be further explored The conclusio ns made in C hapter 8 also reflect on the role that taking the aforementioned recommendations may have on policies that attempt to undermine the negative outcomes of clientelism on democratic representation. This dissertation does not promise to offer teste d solutions for dealing with clientelism, but it highlights the importance of paying attention to the economic dynamics of urban
56 development within local government jurisdictions. Doing so will improve our understanding of clientelism better and sharpen po licies designed to address its negative outcomes, especially at a time when urban development is rapidly occurring in developing countries known for prevalent clientelistic practices.
57 CHAPTER 2 CASE STUDY METHOD AND CASE S ELECTION: WHY KENYA? To develop a theory of how urban development affects clientelistic outcomes within local government jurisdictions, this dissertation draws on data from a case study of Kenya. The question this dissertation addresses, however, applies t o most African countries. To develop a theory that responds to it I could draw on empirical evidence from any country or several countries. So, why a case study and not a cross national study? And why Kenya? 2.1 Why a Case Study? This study seeks to build a theory on an issue that has not been examined. Case studies are useful in such circumstances (Eckstein 1975, 80 cited in Flyvberg 2006, 229; Gerring 2007, 9; Lijphart 1971, 691; Munck and Verku ilen 2005, 10; Walton 1992, 129 ). They can produce new and h ighly valid theories with minimal biasness superior to multiple cases for creating high suited for building theory on new research topic s when there is no existing literature or empirical evidence (Eisenhardt 1989, 546). This method is also appropriate because it allows for testing of multitude hypotheses (Gerring 2007, 40 41). Since these hypotheses can create complex interactions that ar e difficult to decipher, they are best analyzed using a intimate interaction with actual evidence often produces theory which closely mirrors reality (Eisenhardt 1989, methods they can be firmly grounded empirically (Odell 2001, 170).
58 The theoretical question driving a research project usually determines the appropriate unit of analysis (Culpepper 2005, 1). The issue in question clientelism is not suitable for a cross case approach due to lack of valid and reliable cross national data (Lande 1983, 439). Due to the different meanings attached to the concept (Kawata 2006, 4), it is difficult to measure it cross nationally while ensuring that comparisons are appropriately contextualized (Hilgers 2011). As Hilgers (2011, 567) aptly points out, he question driving this study can be appropriately addressed by focusing on a single case. Case studies have a high potential for achieving high conceptual validity than large N studies (George and Bennett 2005, 19; Munck and Verkuilen 2005, 10 ). Practic al considerations also play an important role in the choice of study. In studies that demand the collection of original data, such as this one, cross unit analysis involving multiple cases may be too costly and may also pose significant difficulties in ide ntifying cases within a short time frame. In the absence of sufficient time and resources for gathering a large amount of evidence, as was the case in this study, the appropriate research method that is easier to execute successfully is the case study for mat (Collier 1993, 107; Gerring 2007, 59). In order to gain a complete understanding of the link that this study seeks to make between urban development and clientelism, it is necessary to grasp the intricate details embedded in the mechanisms that stimul ate urban development. Such details are better understood through focusing on single cases rather than several cases (Flyvberg 2006, 327; Kennedy 1979, 663). One of the most important merits of studying a single case is the great depth of analysis that ca n be conducted on a subject which
59 little is known about (Gerring 2004, 345), especially at a sub national level (Culpepper 2005, 3). Essentially, within case analysis make it possible to asses several intervening variables, unexpected causal mechanisms and the conditions that activate them (George and Bennett 2005, 21). According to Munck and Verkuilen (2005, 8), A related strength of case and small N studies is that they offer a basis for assessing causality (internal validity). This is done, most fundamen tally, through a within case form of analysis that uses empirical evidence about causal mechanisms as a way to check expectations concerning the direction of causal processes, to eliminate potential third variables, and to verify the assumption of unit ind ependence and unit homogeneity. A practical advantage of single case studies is that they allow for triangulation of evidence (Gerring 2004, 353), a goal that is almost impossible to accomplish in a large N study. It possible to capture important nuances u sing single case studies because they make it possible to be closer to the situation on the ground (Flyvberg 2006, 223). Single cases have a higher degree of cross unit comparability unit homogeneity y to be comparable to one make heroic assumptions about the comparability of concepts and causal relationships 2.2 Kenya a s a Whether a single case study can be generalized to a broader population of cases or not depends on the type of case and the way it was chosen (Flyvberg 2006, 225). Single case studies are more insightful when the findings can be generalized to other cases (Gerring 2007, 91). The most generalizable type of case is the typical case cross case relationship. By construction, the typical case may also be cons idered a
60 case selection approach is usually meant for selecting a case that is representative of a larger number of cases (Gerring 2007, 91). Essentially, one that has typical values on th e phenomenon being studied. For this study Kenya was selected as a typical case from a population of other African countries. While Kenya, as a single case, has the advantage of holding many variables constant, more leverage for inference is provided throu gh a multi level analysis conducted at three different levels. At the city level there are three cases, 1 at the market level there are nine cases (three from each city), and at the individual level there are 360 cases (120 business owners from each city). A typical case can be identified using the mean, median, or mode of the main provided using an analysis of frequency distribution of its main independent variable urbanizat ion and three other variables that are closely related to the relationship between urbanization and clientelism: change in gross domestic product (GDP), level of democracy, an d level of corruption. Figure 2 1 presents the distribution of urbanization in Africa from 2000 to 2010. Verification of Kenya as a typical case using the histogram above, and those that follow, is based on comparing mean values and two tests that ensure the distribution is normal and not skewed. The purpose of the tests is to ensure that the distribution is symmetrical. When a distribution is symmetrical the mean is an appropriate measure for understanding what typically happens. The distribution above shows that from 2000 to 2010 most African countries (56%) had an average rate of u rbanization ranging from 3 1 At this level, all cases were selected for this study since Kenya has a total of three cities Nairobi, Mombasa, and Ki sumu.
61 to 4% with a skewness value of 0.04. Since the skewness is almost 0, the distribution is not skewed towards either side. The kurtosis is 2.39, which means that this is a normal distribution. A kurtosis value for a normal distri bution is usually around 3 (Acock 2006, 91). The mean rate of change for the all A frican countries was 3.7% while K considered as representative of most African countries. Fur distributions, starting with a variable that is closely linked to both urbanization and clientelism, GDP growth rate (Annez and Buckley 2008, 1). The main drivers and outcomes o f urbanization are usually captured in the composition of this variable since it measures the effect of labor and property on goods and services produced by a country (Jones 2001). Urbanization is related to GDP since it is regarded as a phenomenon that o ccurs as a result of increase in labor and property in an area (Black and Henderson 1999, 252; Cohen 2003, 27 29; Hope Sr. 1986, 45; Miro and Potter 1980, 124). GDP, as an indicator for development, is also related to clientelism in the sense that low leve ls of development are usually linked to the occurrence of clientelism and vice versa (Calvo and Murillo 2004, 742; Fox 1994, 154; Robinson and Verdier 2002, 4; Roniger 2004, 355; Vicente and Wantchekon 2009, 303; Wantchekon 2003, 400). In developing countr ies lethargic economic development makes it possible for clientelism to serve as a mechanism for parties to monopolize resources that citizens do not have (Medina and Stokes, 2002). According to Fatton (1986, 7) patron
62 clientelism make it an appropriate variable for evaluating the representativeness of Kenya. The distributi on presented in Figure 2 2 shows that from 2000 to 2010 most African count ries (40%) had an average GDP growth rate ranging from 4% to 5% with a skewness value of 0.6 and Kurtosis value of 3.7. This indicates that it is symmetrical and the values near the mean are typical. The mean rate of GDP growth for the all African countri es was 4.1 % % can therefore be considered as representative. The next two variables corruption and democracy are closely related to power or authority for 4) is interrelated to clientelism in the sense that narrowly self seeking public officials tend to use resources accessible to them for clienteli stic purposes (Harriss White 1996; Philip 1997; Robinson 1998, 4). Some scholars associate it with vote buying (Kawata 2006, xii), which is a main feature of clientelism (Wantchekon 2003, 400). It also tends to occur in places where clientelistic exchanges are deeply rooted (Kawata 2006, xii; Manzetti and Wilson 2007, 949). Corruption can, therefore, be used as a proxy for clientelism (Muno 2010, 14; Persson et al. 2003; Singer 2009, 9). These attributes make it suitable for examining whether Kenya is repre sentative with regards to the incidence of clientelism in Africa. The dis tribution presented in Figure 2 3 shows that from 2007 to 2010 most African countries (83%) had a high level of corruption, with 48 % of them getting a score of 3 % and 35 % getting a sco re of 2. The skewness value of the distribution is 1.1 and
63 Kurtosis value of 4.6. Although this values are slightly higher than the ones for the previous distributions, they still indicate that it is symmetrical and the values near the mean are typical. Th e symmetry of a distribution is usually questionable when the kurtosis value is greater than 10 (Acock 2006, 91). The mean score for corruption for % The level of corruption in Kenya is theref ore typical of that in most African countries. compare its level of democracy with that of other African countries. In most developing countries liberal democracy tends to be undermined by clientelistic interactions (Keefer 2007, 814; Kitschelt and Wilkinson 2007, 24; Robinson and Verdier 2002, 3). In these societies clientelism is viewed as the prevalent linkage between citizens and politicians (Cain, Ferejohn and Fiorina 1987 ; Keefer and Valicu 2008; Piattoni 2001aa, 202). In African countries clientelism has continued to occur as a form of democratic accountability (Beck 2008; Bratton and van de Walle 1997; Lindberg 2010, 133; Wantchekon 2003, 400). The close relationship bet ween these two variables in Africa accomplish this task I use data from Freedom House to analyze the average level of democracy in Africa between 2000 and 2010. The dis t ribution presented in Figure 2 4 below shows that from 2000 to 2010 most African countries (52%) were rated as partly free. According to the Freedom House ratings, these are countries with a rating that ranges from 3 to 5. The skewness value of the distrib ution is 0.2 and the Kurtosis value is 2.1. This indicates that it is symmetrical and the values near the mean are typical. The mean rating for all African
64 countries was 4.2 % % that period is t herefore representative of most African countries. With the large number of cases that can be considered typical in the distributions presented above, there is a wide range of options for selecting a case to study. Any of them can be reasonably selected bu t Kenya was the country of choice based on some important practical reasons. In a large N sample with several cases that have typical scores, it is strategic to select a case using a practical criteria that takes into consideration costs and convenience (G erring 2007, 94). Practical considerations such as time, money, and access to data play a major role in case selection (Seawright and Gerring 2008, 295). An in depth prior knowledge of the case is also a reliable practical basis for selecting a case (Gerr ing 2007, 147). Preliminary knowledge of the case is advantageous in the sense that it strengthens the research design (George and Bennett 2005, 24). I selected Kenya based on three factors: cost, time availability, and prior knowledge of the case. Given t hat I needed to collect original data on clientelism with limited resources, it would have been more costly to execute this task in several cases. Also before conducting this study I already had prior knowledge of the case, which I had acquired through a small business project tha t I designed and implemented. 2 2 Prior to beginning my doctoral studies I had conducted a research project on small business in Kenya that was funded through a grant from the business school in my former college (Valdosta State University). Throughout 2007 and part of 2008 I was i n charge of designing, implementing, and evaluating this project. Through this process I acquired basic knowledge on the relationship between local governments and small businesses since I had to get business permits from them and ensure the business comp lies by their by laws.
65 Figure 2 1. Average rate of u rbanization in Africa, 2000 2010 ( Source of data: United Nations Department of Economic and Social Affairs/Population Division. 2008. World Urbanization Prospects : The 2007 Revision. New York. ) Figure 2 2. Average GDP growth r ate in Africa 2000 2012 (Source of data: United Nations Department of Economic and Social Affairs/Population Division. 2008. World Urbanization Prospects: The 2007 Revision. New York.)
66 Figure 2 3. Average level of c orruption in Africa, 2007 to 2010 (Source of data: United Nations Department of Economic and Social Affairs/Population Division. 2008. World Urbanization Prospects: The 2007 Revision. New York.) F igure 2 4. Average l evel of d emocracy in Africa, 2000 2010 (Source of data: United Nations Department of Economic and Social Affairs/Population Division. 2008. World Urbanization Prospects: The 2007 Revision. New York.)
67 CHAPTER 3 THE INFORMAL SECTOR IN KENYA: ITS GROWTH AN D PERMANENCE C hapter 3 presents the context in which this study was done the informal sector. It lays the foundation for the argument presented in C hapter 1 by accomplishing two objectives. It explains the economic significance of the sector, and shows h ow it has grown to be a vast and permanent economic feature where most Kenyans earn their living. C hapter 3 is divided into four sections. The first section conceptualizes and operationalizes the informal sector. The second section shows the significance o f the informal sector by explaining the role it plays in offering a source of livelihood to most citizens in developing countries. In the third section, I show how state policies in Kenya have contributed to its growth. Specifically, I show how it created incentives for most citizens to seek employment opportunities in the informal sector by channeling its minimal resources towards a failed attempt to develop modern small industries in the formal sector during the immediate post independence period. And the n I show how its attitude towards the sector changed from the beginning of the 21st century with an extension of substantive resources towards the development of the informal se ctor. The final section of C hapter 3 shows how the growth of a formal microfi nance industry has contributed to the growth and permanence of the informal sector through formalizing loaning procedures. In the same section I also show how these microfinance institutions ive support to the sector. 3.1 What is the Informal Sector? 1973 (Hart 1973). Throughout the duration of its usage there have been extensive
68 discussions and debates on how to define the sector but no consensus on a single definition (House Soremekun 2009, 1445; Livingstone 1991, 651; Muraya 2006, 128). The first task that needs to be undertaken, therefore, is to clearly conceptualize and operationalize it. Defining the sector has always been a controversial issue because of its rapid and unpredictable growth. After many years of debates concerning the difference between formality and informality, two characteristics emerged as the widely preferred operational criteria for ident ifying informal sector enterprises: small business sizes and whether an enterprise is regulated and taxed (Lubell and Zarour 1990, 11; Gerxhani 2004, 269). In recent studies (e.g. Pedersen and McCormick 1999; Muuka 2003; Blaauw 2005; Simon and Birch 1992; Muraya 2006; Komollo 2010, 6), the most commonly used one is the size of businesses in the sector. The definition of the informal sector used in this dissertation borrows from two sources that employ this criteria: 1) It employs the categorization used by pioneered research on the sector in a study that was conducted in Kenya in 1972 (Gerxhani 2004, 269). The ILO (2007) categorizes the se ctor into four broad categories: 1. Own account workers in survival type activities 2. Paid domestic workers in households 3. 4. Self employed workers in micro enterprises with or without contributi ng family workers or employees. This dissertation focuses on the fourth category microenterprises. In Kenya, a business with ten or fewer workers is categorized as a micro enterprise (Gray et al. 1997, 67; Livingstone 1991, 652; Mitullah 2003, 2). The majority of businesses in this category are in the informal sector (Akoten et a l 2006, 930; Livingstone 1991, 651; Mitullah 2003, 2; World Bank 2006, 32). But they are also regarded as a component of
69 the formal sector (e.g. Daniels 1999; Gray et al. 1997; Atieno 2001). Due to the dominance of microenterprises in the sector, the term has been used interchangeably What uniquely distinguishes the micro enterprises in the forma l and informal sector in Kenya is that those in the formal sector pay income taxes, while those in the informal sector do not (Karingi et al. 2005, 32; Macharia 1997; Moyi and Ronge 2006, 4). This is because the enterprises in the informal sector are consi dered to be below the official value added tax rate (Karingi et al 2005, 55). A physical aspect that has distinguished the informal sector from the formal sector in Kenya is their locational decisions. This was a commonly used criteria for defining the i nformal sector in Kenya when it started gaining prominence in the late 80s and early 90s (e.g. McCormick 1988; Onyango 1992; Ouma 1990; King and Abuodha 1991). Most of the informal micro enterprises locate in markets owned by local authorities (Muraya 2006 8), which were originally established under the former 1 for the purpose of allocating business sites to street hawkers (Kinyanjui 2010, 27). Although businesses are not required to operate in those markets most of them have traditi onally done so because premises are inexpensive to rent (McCormick and Kinyanjui 1997, 1106). The other commonly used alternative is leasing shops in privately owned buildings but these have always been costlier (McCormick and Kinyanjui 1997, 1106). In mos t cases the formal sector businesses opt for this alternative. This variation in locational decisions makes it possible to physically distinguish between the formal and informal sector (Gray et al. 1996, 6). 1 President Moi was in power from 1978 2002.
70 3.2 The Growth and Significance of the Informal Sector The informal sector rapidly grew in Africa during the economic crisis of the 1970s and 1980s as an important source of livelihood for many (Meagher 1995, 259; Potts 2008, 156; Sethuraman 1977, 344). At the beginning of the 1990s the United Nations s peculated that in the decade of the 1990s the informal sector will be the most important openings (United Nations 1996). From the 1990s scholars (e.g. Gilbert and Gugl er 1992; Potts 2008, 155; Werna, 2001), as well as institutions (UNCHS and ILO 1995), were acknowledging that growth of the sector in developing countries was the solution to unemployment. By the end of the 1990s, it was clear that the informal sector had achieved, surpassed the speculated 1990 targets, and was even absorbing more workers from the formal sector and those that were unemployed (Mofokeng 2005, 3; Potts 2007). At the beginning of the 21st century it was clear that the informal sector in Africa (Charmes 2000) and other developing countries (Todaro and Smith 2009, 336 ) had absorbed more than a half of the working population in different countries. For example, by 2005, the proportion of workers absorbed by the informal economy in Burkina Faso was 80%, in Ghana 90%, in Kenya 72%, in Mali 80%, in Tunisia 50% and in South Africa 53% (Mofokeng 2005, 5). In Kenya, from 2003 to 2007, the total recorded employment in the informal sector rose from 7.5 million to 9.5 million workers, and by the beginning o f 2008 it constituted about 89 % of total employment outside small scale agriculture and pastoralist activities (Kenya National Bureau of Statistics 2008). Currently, The total recorded employment in the informal sector in Kenya constitutes 80.6 % of the ov erall employment (Kenya National Bureau of Statistics 2011). Figure 3
71 1 below show the growth trend o f the informal sector in Kenya. 2 As the F igure 3 1 shows, the informal sector has steadily increased since the mid 1970s. But the growth rate is rather mi nimal from 1975 to 1990. During this period less than one million people joined the sector. It particularly starts increasing at a rapid pace from 1990 onwards, with approximately one million people joining the sector after every two years. Figure 3 1. Growth of informal s ector in Kenya, 1975 2010 ( Source of data: Kenya National Bureau of Stat istics, Economic Survey Reports, 1975 2010 Nairobi, Kenya. ) The informal sector is a major economic component of the urban economy in most developing countries (Potts 2008, 153). The growth of the sector in urban centers has been fueled by a combination of rural to urban migration and the demand for cheap 2 The trend does not include data for the years 1995, 2000, and 2001. This data is missing in the archival data collected from Kenya National Bureau of Statistics. The same issue applies to Figure 3 2 and Figure 3 3.
72 sector generates surpluses [that] could provide an impetus to growth in the urban notable at the beginning of the 21st century when the informal sector in most African countr ies was contributing as high as 50 % of the non agricultural GDP (Charmes 2000). By the mid 2000s the informal sector had about 70% of the urban employment in sub Saharan Africa (Becker 2004, 3). In Kenya, the informal sector has consistently grown in both rural and urban areas. Figure 3 2 below shows the growth trend of the informal sector in these areas. 3 Figure 3 2. Growth of rural and urban informal s ector in Kenya, 1986 2010 (Source of data: Kenya National Bureau of Stat istics, Economic Survey Reports,1975 2010. Nairobi Kenya.) 3 The data starts from 1986 since this data does not exist prior to 1986. The same issue applies to Figure 3 3.
73 The goal of Figure 3 2 is not to compare the urban and rural informal sectors but to show how they have both grown, especially since the early 1990s. Although both the rural and urban informal sec tors have steadily grown, it shows that this trend is higher in rural areas. This is could be an outcome of return migration from urban to rural areas in the 1990s (Francis 2000) and the existence of few urban centers in Kenya (Agesa and Agesa 2007, 43; Ot iso 2005, 123). Figure 3 3. Urban and rural population d istribution in Kenya, 1960 to 2001 ( Adapted from Legovini A. 2000 Kenya Macro Economic Evolution Since Independence (Page 57, Figure 29 ) United Nations Development Pro gram, Nairobi, Kenya). Figure 3 2 shows the decline in urban informal sector growth during the late 1990s and an increase in the rural areas. Generally, the total population in the urban areas has also not been as high as it has in rural areas (Otiso 2005 117; Richardson 1980, 97; Rural Poverty Portal 2012; UNICEF 2012). In 2009, the percentage of the population that lives in rural areas was 68 % while that of the urban areas was 32%
74 (Kenya National Bureau of Statistics 2009, 3). However, as shown in Figur e 3 3 below, since 1960 the urban population has consistently increased from about 8% to 32% of the t otal population Figure 3 3 indicates that urbanization in Kenya has consistently occurred since the 1960s. This phenomenon has been regarded as a major co ntributing factor to the expansion of the urban informal sector (Fay and Opal 200, 24; House 1984, 278; Todaro and Smith 2009, 336). For years the informal sector had been regarded as a backward economic feature that will be eventually formalized (Potts 20 08, 154 158) but its consistent growth with increases in productivity changed this perspective (Lubell and Zarour 1990, 9; Meagher 1995, 262; Potts 2008, 155). It has been considered a significant contributor of economic output because of labor intensive o perations that the formal sector finds unprofitable to engage in (Blunch et al. 2001, 6; Nelson 1997). The shift in attitude is also a result of the advancements in business quality, from dominance of makeshift business operations to an increase in stable small and microenterprise (SME) operations, which have traditionally been considered as the upper tier of the sector (Fields 1990; Pyke and Sengenberger 1992; Daniels 1999, 55). Compared to the typical low quality business activities, this tier tends to h ave businesses that have higher capital intensity and are operated by more skilled and educated workers (Bosire and Gamba 2003, 4). In Kenya, these SMEs have significantly contributed to employment and economic growth (Atieno 2009, 13; Bird and Ryan 1998, 9; Livingstone 1991; Widner 1991; King, 1996, 2001; Daniels 1999, 63; Kabecha 1999). They have created more jobs than those in the formal sector (Livingstone 1991, 657; Muraya 2006, 130; Mead
75 1994; Mead and Liedholm 1998, 69). By 2010, they constituted abo ut 96 % of all the businesses in Kenya (Newa 2010, 93). Figure 3 4 shows the variation in types of jobs created in the sector. It provides a general idea of the composition of products and services produced in the sector. As shown in Figure 3 4, the two in dustries that have experienced the most growth in the sector are the wholesale and retail trade, and the hospitality industry. The evidence presented above suggests that the two industries could be the main bases for the clientelistic transactions that tak e place in the sector since most of the poor engage in those activities (Odhiambo and Manda 2003, 11). It is usually citizens who are poor that are targeted by politicians for clientelistic bids (Auyero 1999; Robinson and Verdier 2002, 3). Figure 3 4. Growth of the informal sector by i ndustry in Kenya, 1986 2010 (Source of data: Kenya National Bureau of Statistics, Economic Survey Reports,1975 2010. Nairobi Kenya.) The impact of the informal sector has transformed it from b eing viewed as a temporary feature to a permanent feature (Gerxhani 2004, 294; Hosier 1987). This is
76 mostly because of the employment opportunities it provides. Todaro and Smith (2009, 335) note that, with the unprecedented rate of growth of the urban popu lation in developing countries expected to continue and with the increasing failure of the rural and urban formal sectors to absorb additions to labor force, more attention is being devoted to the role of the informal sector in serving as a panacea for the growing unemployment problem. The expansionary effect that urbanization has had on the informal sector in African cities (Bradshaw 1987, 236; Chen 2001, 7; Coquery Vidrovitch 1991, 47 48; House 1984, 277 278) is likely to continue. The next two sections d escribe how two major factors state development strategies and microfinance institutions that continue to fuel the growth of the sector in urban areas in Kenya have done so in the recent past. The goal is to provide an appropriate contextual background for this study by showing how the main venue of citizen politician interactions the informal sector has acquired a sense of permanence in Kenya. 3.3 State Development Strategies and Growth of the Informal Sector The informal sector is an outcome of sh ifting state development strategies in two development policy was centered around getting rid of the informal sector and producing a formal sector with small modern indu stries (Hosier 1987, 383; Muraya 2006, 130; Potts 2008, 158; Ronge et al. 2002, 49). But the state led development of a modern sector failed (Himbara 1994a; Meagher 1995, 262). This led to the growth of unregulated and untaxed businesses that were not unde r state supervision (Ronge et al. 2002; ). Second, it has shifted its development policy from the beginning of the 21st century by extending substantive financial support towards sustaining the existence of the sector (Muraya 2006, 132). This section brief
77 towards the informal sector has shifted from the post independence period to the beginning of the 21st century. The post independence era is an appropriate starting point because it is the time when the state decided to ov erhaul its economic policy with a goal of developing domestic manufacturing industries (Himbara 1994b, 9 10). Since Kenya gained its independence in 1963, the analysis in this section starts from the late 1960s. This insight also provides the background ne cessary to understand the role growing needs (House Soremekun 2009, 1444). The lack of state responsivenes s towards the needs of most Kenyans during the post independence period (Murunga 2007, 118) forced them to earn their living in the informal sector, which it initially wanted to get rid of (Potts 2008, 159). It perceived the sector as an unplanned economic feature and a hindrance to its development plans (Ikiara 1991, 11). The apathetic attitude towards the informal sector was inherited from the colonial state. During the colonial period the state passed by laws that prohibited hawking in the capital city Nairobi (Macharia 1993, 5). The only form of employment that was authorized by the colonial state in the city was employment by Asian proprietors and by its agencies (Ikiara 1991). The post colonial state continued implementing these laws as it was only interested in expanding the urban formal sector through industrial activities (Ronge et al. 2002). For the state, the effective strategy for advancing the economy and reducing dependence on agriculture as a source of employment was to develop modern small industries (Ronge et al. 2002, 11). To implement this strategy, it established development finance institutions to provide financial support to potential entrepreneurs for the formal sector (Himbara 1994b, 472; Moyi and Njirain i 2005, 29; Njenga et al.
78 200 6 ). Its industrial development programs engaged in what was known as supervision of the state (Ikiara 1989, 12). Businesses in the informal sector were considered as a hindrance to industrialization projects and frequently demolished (Ronge et al. 2002, 11; Ikiara 1991, 11). In spite of the significant investments, the state was experiencing extremely high levels of unemployment (International Labor Organization 1972). This crisis provoked a change of attitude towards the informal sector, which started in parliament. In 1970 a parliamentary select committee consulted the International Labor Organization (ILO) on what to do with the informal sector. It recommended that f inancial support be extended to the sector because it was a highly needed source of employment (ILO 1972). The state accepted these recommendations in 1973 but it did not extend any financial support (Ronge et al. 2002). In the mid 1980s, declining politi cal legitimacy forced the state to affirm its desire to extend support to the sector through the industrialization programs (Macharia 1993, 2). But these intentions were significantly undermined by the initiation of structural adjustment programs (SAPs), w hich called for an end to the programs (Ikiara 1991, 28). The state however continued providing minimal support that only targeted established firms in the formal sector (McCormick 1997; Ronge et al. 2002, 16; Republic of Kenya 1989b), leaving the informa l sector to grow unmonitored. SAPs also recommended the retrenchment of several civil servants, leading to an upsurge of microenterprises in the informal sector (Akyuz and Gore 2001, 282; Ikiara 1991, 30; Pedersen 2005; Potts 2008, 159). This incremental e ffect from the decrease in public sector hiring was further compounded by stagnation of productivity in the formal sector (Parker and Torres 1994,
79 education subsidies (Rono 2 002, 92 93). As a result, the cost of education increased and so did the number of students dropping out from school and entering into the informal sector (Ikiara 1991). School dropouts were unable to secure wage employment (Rono 1992). At the beginning of changes with the initiation of financial support programs. In 2003, it acknowledged the permanence of the informal sector by considering microenterprises a potential breeding ground for employment history it backed this acknowledgement with the initiation of two loan programs Youth erprise Development Fund (WEDF) 4 This was a shift from its previous policies that acknowledged the sector without financial support. Since 1986, the state recognized the role of the informal economy on paper through five key policy documents (Republic of Kenya 1989a,b; Republic of Kenya 1992; Republic of Keny a 2002; Republic of Kenya 2003a, and Republic of Kenya 2003b). The current policy document (Republic of Kenya 2003b) focuses on reviewing the former rules and regulation that affect the operation and growth of the sector (Mitullah 2003, 4). This policy hi ghlights its economic strategy for wealth creation through investing in the informal sector. Through this policy document it intends to address the high cost of capital for the informal sector and the lack of supportive services (Mitullah 2003). To ensure their longevity, the current loan programs we re 4 parliame ntary Acts that passed them make various provisions that make the loans open to any business owner.
80 passed as parliamentary acts 5 so as to predetermine by law the amounts that the state would give to the programs without reneging on its obligations. These loans are being channeled through microfinance inst itutions (MFIs), which played a significant role in microenterprises to put pressure on the state a nd local governments to provide various forms of support that have continued to sustain the existence of the sector. It mainly focuses on the origins of these formalization strategies and the growth of MFIs that continue to use them. 3.4 Permanence of the Sector: Outcome of MFIs E xpansion and their Strategies Microfinance is largely known as the provision of a broad range of financial services such as deposits, loans, payment services, money transfers, and insurance to the poor and low income households and their microenterprises (Alila 1994; Atieno 2009, 10; Basu et al. 2004, 3; Kitchin 1980; Ochanda 2012, 2). By the end of the 1990s Most of the influential MFIs that operate in Kenya were initiated by international donor agencies in the mid 1980s (Hospes et al. 2002, 14 15; Institute of Economic Affairs 2002, 1; Ochanda 2012, 1). As Figure 3 5 shows, t his was a time when the economy was stagnating (Akyuz and Gore 2001, 281; Holmquist et al. 1994, 91; House Soremekun 2009, 1445; Legovini 2002, 3; Mosley 1986, 107 108 ). Figure 3 5 below shows that the economic growth rates dropped from 8% between 1970 and 1980 to an average of 2% to 4% from 1980 to 2000 (Legovini 5 YEDF Parliamentary Act and WEDF Parliamentary Act.
81 2002, 3). Between 1990 and 1999, the average annual growth rate was around 2.5% (Bedi et al. 2004, 15). Throughout that period the level of unemployment was rapidly rising (Buchmann 1999, 102; Kimalu et al. 2001; Legovini 2002, 6). From 1989 to 1997, the national unemployment rate rose from 6.5% to 18% (Kimalu et al. 2001, 55 57). Figure 3 5. Economic growth by d ecade, 1963 2000 ( Adapted from Legovini A. 2000. Kenya Macro Economic Evolu tion Since Independence (Page 29, Figur e 1 ) United Nations Development Program, Nairobi, Kenya ) With the decline of GDP rates shown above, unemployment rates were higher in urban than in rural areas (Kimalu et al. 2001, 56) since there was a growing urban population that could not secure wage employment in both the private sector and public sector (Buchmann 1999, 110; Miller and Yeager 1994, 154; Odhiambo and Manda 2003, 9; Republic of Kenya 1991). In the late 1980s the average unemployment rate in urban areas was 13 % compared to 4.9% in rural areas (Kimalu et al. 2001, 56). From 1986 to 1999, this rate increased by 15 % (Odhiambo and Manda 2003, 9). For the
82 urban unemployed the informal sector was the place to seek for livelihood by starting their own businesses (Akyuz and Gore 2001, 281; Bedi e t al. 2004, 15; Odhiambo and Manda 2003, 15) but they did not have the required capital (Adera 1995; Atieno 2001, 3; Daniels et al. 1995). The lack of capital increased the demand for loans but commercial banks, being the main source of business loans, wer e hesitant to issue them to informal sector businesses as they considered them risky (Buckley 1997, 1083;Ochanda 2012,4). This banking industry was however unreliable as it was shrinking and on the verge of collapse (Chao Beroff et al. 2000, 19). The combi ned lack of support from the state and commercial bank loans created a niche for MFIs, which responded through a wide range of loan products (Koopman 2001, 1). The formal processes embedded in these products was what led to the current permanence of the se ctor, which in turn provoked state support. Prior to the emergence of the current MFI industry in Kenya, their services were provided through traditional and informal ways in the 1970s and 1980s (Buckley 1997; Fowler and Kinyanjui 2004, 22). They were mos tly informal credit systems run by individual money lenders and shopkeepers (Rogerson 2001, 127). This early systems were instrumental in providing knowledge to the current industry (Basu et al. 2004, 7; De Aghion and Morduch 2003). With this insight, MFIs used advanced financial methods to formalize informal loaning procedures (Basu et al. 2004, 8). The main goal was to facilitate sustainability of their programs through addressing various problems encountered when loaning to the poor, such as lack of info rmation and weak collateral (Basu et al. 2004, 8). Within this new approach their clients, who are mainly from the informal sector (Atieno 2009, 10), have been required to apply for loans through a process that required evidence of legal documents, mainly business permits and leases
83 for their premises. For the informal sector, compliance has meant seeking state services in the form of business registration, business premises to conduct operations, and essential utility services (Macharia 1993, 15). MFI for malization schemes were pioneered by a microfinance program known as Kenya Rural Enterprise Programme (KREP) (Bird and Ryan 1998, 1; Chao Beroff et al. 2000, 19; Ochanda 2012, 3; Nyerere 2004, 139), which was established in 1984 through the United State Ag ency for International Development (USAID) funding (Rosengard et al. 2000, 1; Chung 1996, 184; Fowler and Kinyanjui 2004, 20). This is industry (Rosengard et al. 2000, 7; Hosp es et al. 2002). It was the first microfinance institution in Kenya and Africa to transition from an NGO to a licensed commercial bank (Nyerere 2004, 141). It was also responsible for the development of a policy that has institutionalized microfinance inst institutions (Nyerere 2004, 139 140). This section focuses on the USAID KREP partnership due to the significant role KREP has played in the expansion of other formal MFIs (Bird and Ryan 1998, 1; Institute of Economic Affairs 2002, 2) and growth of microenterprises (Bird and Ryan 1998, 4). A significant percentage of MFI programs that were started in the 1980s were run by NGOs (Institute of Economic Affairs 2002, 2; Chao Beroff et al. 2000, 19; Ochanda 2012 9). This was a time when aid agencies were seeking alternative avenues other than commercial banks and the state to channel development funds in Kenya (Fowler and Kinyanjui 2004, 35). These agencies were already extending funds for the development of sm all scale entrepreneurs in the informal sector in the late 1970s (United Nations 1996) through commercial banks but their scope was narrow (Fowler
84 and Kinyanjui 2004,33; USAID Kenya 1989) because they considered businesses in the informal sector too risky to lend money (Hoff and Stiglitz 1990, 237). Substantive donor support for the informal sector was subsequently extended through MFIs (Basu et al. 2004, 12; Bird and Ryan 1998; Craig and Goodwin Groen 2003; Hospes et al. 2002), with USAID being one of the pioneers (Rhyne and Christen 1999, 10). The main beneficiaries were domestic NGOs that already had a microfinance program in place, even if rudimentary (Fowler 1989). The overarching goal was to empower these NGOs to become major non state actors in steeri ng development (Fowler and Kinyanjui 2004, 36). USAID executed this initiative by partnering with KREP (Rosengard et al. 2000, 2; Barbara 1987), which also served as an experiment that would provide insight on how to execute such programs in other African states (Fowler and Kinyanjui 2004, 38). The goal was to redesign MFI policies and procedures of loaning using new formal structures (Fowler and Kinyanjui 2004, 50 51; Chung 1996, 184). Under these structures borrowers from the informal sector were to unde rgo thorough screening before getting loans (Bigelow et al. 1987). The effect was felt country implementation of the new structures after undergoing training from KREP (Koopman 2001, 6) through g roups and NGOs in order to establish or expand capacity in micro enterprises By the end of the 1980s KREP had trained 40 small NGOs on microfinance loaning procedures and was actively facilitating the expans ion of five large NGOs to
85 have a country wide reach (Fowler and Kinyanjui 2004). It was also playing a major role in shaping the MFI industry in East Africa by offering training in credit administration (Chung 1996, 185). By the late 1990s, NGOs benefiti ng from KREP USAID collaboration had risen to 25, with about 3.3 million U.S. dollars allocated to them (Fowler and Kinyanjui 2004, 55). Figure 3 6 below shows the consistent growth in outreach of s ome of the major MFIs in Kenya. 6 Figure 3 6. MFIs loan p ortfolio and Active Borrowers, 1997 2009 ( Source of data: Mic rofinance Information Exchange 2012 www.mixmarket.org/profiles reports accessed 7 November, 2012 ) Figu re 3 6 shows that clients for the major MFIs in Kenya have steadily increased since the late 1990s. Most of these clients are in urban areas (Institute of 6 Due to lack of data on all MFIs the purpose of the evidence provided below is to only provide an idea of the significant effect of the industry. The success of MFI institutions is usually measured in terms of total clients reached (Institute of Economic Affairs 2002, 7).
86 economic Affairs 2002, 7). The increase in these clients was a combined outcome of the increase in bo th urbanization and poverty rates in urban areas in the late 1990 s. During that decade the amount drastically rose to about 80 million U.S. dollars, with the largest portion (USD 23 million) being allocated to KREP (Hospes et a l 2002, 23). Throughout the 1990s KREP was making major advancements and by 2001 it had graduated to a banking institution providing the largest number of small loans for microenterprises (Hospes et al. 2002, 116; Institute of Economic Affairs 2002, 2; Nyerere 2004, 139). During the same time NGOs offering client based microfinance loans influenced by KREP had increased to 130, with 10 of them having major microfinance programs (Hospes et al. 2002, 24). These organizations played a significant role in supporting the informal sector by making it easy for microenterprises to access credit (Bird and Ryan 1998, 10). This was especially critical at a time when poverty levels were high. Figure 3 7 below shows the drastic increase in poverty rates that occurred from 1992 to 1997. It compares overall rate in the country. The urban poverty rates increase significantly from about 30% in 1994 to 49% in 1997. MFI clients start increasing around 2001 and sharply increase starting from 20 05 (see Figure 3 6). I am not suggesting that they are directly correlated but tentatively showing the possible effect that increase in poverty had on increase in MFI clients. The outstanding client loans has immensely increased in billions of dollars sinc e the late 1990s. The estimated amount of outstanding loans for major MFIs increased from about U.S. $ 3 million in 1997 to U.S. $ 1.2 billion in 2010, which is almost a 308% increase. With most of the clients being in the informal sector (Atieno 2009, 10 ), these trends suggest that MFIs have increased the size of the sector over time and continue to sustain its existence (Atieno 2001).
87 Figure 3 7. Increase in urban poverty i ncidence ( Adapted from Legovini A. 2000. Kenya Mac ro Economic Evolu tion Since Independence (Page 65, Figure 37 ) United Nations Development Program, Nairobi, Kenya ) Some of the MFIs that started in the 1990s have progressed from NGO sponsored credit programs to quasi independent financial institutions by the end of the 1990s (Koopman 2001, 19). This rapid expansion of the MFI industry has stabilized business operations through the formal processes in their financial and non financial loan products that were designed for this purpose (Fowler and Kinyanjui 2004, 33; Institute of Economic Affairs 2002, 2; United Nations 1996, 41 42). They have particularly made the sector a physical permanent feature in local government jurisdictions by requiring them to show evidence that they are legally registered and have an existing business operation site before being issued loans (Mwaniki 1996, 6).
88 In order to fulfill MFI requirements, microenterprises have placed extensive government l aw assigns them the responsibility of registering and licensing micro enterprises (Komollo 2010, 11; Patrick et al. 2005, 19; Kinyanjui 2010, 11). To meet the growing demand for business sites, they have built several markets and developed new parcels of l and to lease to the informal sector (Muraya 2006, 132). These have been important investments for them since they are major revenue bases (Devas and Grant 2002, 3). A significant budget of local governments depends on revenues collected from rental, regist ration, and annual license fees (Kapila 2006, 32; Smoke 1993, 905). The financial significance of the sector has also encouraged other forms of assistance forms of state assistance targeting the sector (Amenya 2007, 4). In 1995 the states established a fu the informal sector (Smoke 2004, 225). In 2001, the state created an avenue for local government level, namely the Local Authorities Service Delivery Action Plan. In one year, a wide range of groups, including informal sector associations had identified about 3,092 projects that it funded (Patrick et al. 2005, 20). The state also established a minis try (Ministry of Technical Training and Applied Technology) charged with the responsibility of creating programs for improving the productivity of the informal sector (Amenya 2007, 4). The overall outcome of local rather permanent informal sector. The magnitude and permanence of the informal sector makes it an important place to study the nature of linkages between citizens and politicians. As the place with
89 the largest pool o f voters, politicians have a strong incentive to sustain linkages with citizens. The high level of poverty in this sector also makes it attractive for clientelistic transactions, since it tends to thrive where there are high levels of poverty (Robinson and Verdier 2002, 3; Auyero 1999). There is a need to understand how the ongoing urban development is shaping these transactions through shaping levels of poverty in this sector, especially if we take into consideration that the rates or urbanization are proj ected to continue increasing in Africa. As shown in Figure 3 8 below the percentage of urban populations have been increasing across all African regions. Figure 3 8. African populations l iving in urban areas (%), by r egion: 1990 2025 ( Source of data : W orld Urbanization Prospects 1990. New York, United Nations ). With the trends shown above we should expect to see the level of urban development increasing in Africa as people engage in developing new property and governments develop public infrastructure (Bairoch 1988; Bertinelli and Black 2004; Davis and Golden 1954; Huriot and Thisse 2000; Weng 2007, 341). C hapter 4 below
90 cities where this study was conducted.
91 CHAPTE R 4 URBAN DEVELOPMENT IN LOCAL GOVERNMENT JURISDICTIONS C hapter 4 presents a historical background of urban development in three local government jurisdictions. It is divided into six parts. The first part provides a short rnment system has evolved since independence, with a focus on the gradual centralization of the system. It is important understand this system since it is what gives the three cities a high degree of unit homogeneity. The second part introduces the main fe atures of the centralized local government system that affec t urban development outcomes. The third part provides a summary of the reforms that have been undertaken to decentralize the system and strengthen local government development capacity. In the fou rth part I provide a short history of urban development in the three cities starting from the time they were founded and also conduct a qualitative comparison of their urban development patterns. 4.1 Kenya has small geographicall of local governments (LGs), namely: municipal, town, county, and urban councils. There are a total of 175 LGs. Several of them were created for political expediency by the central governm three cities Nairobi, Mombasa, and Kisumu, which are the subject of this study, fall under the municipal councils category. These councils are considered more urban and are larger th an town councils. For example, the infrastructure in municipal councils is
92 usually better than that in town councils (Barasa and Eising 2010, 3). Nairobi, being the capital city, is governed by a more independent and autonomous council since it was establi shed under a different set of by laws (Koti 2000, 19 20; Lumumba 2004, 3). The local government system consists of elected councils and a non elected administration. The organ that makes laws is the full council, which is a combination of both elected and nominated councilors headed by a Mayor. The administration is services and utilities. Majority of the councilors are elected through the popular vote, a third are nominat ed by political parties that win most seats in the council elections, and the Mayor is elected through a secret ballot held during the inaugural meeting of the council (Menon et al. 2008, 20). Each councilor serves in one or more of the committees that mak e decisions implemented by administrative departments (Koti 2000, 10). Local governments (LGs) play a significant role in managing urban development (Evans 1989). They provide a wide range of services, such as building and maintaining roads, building marke ts, and garbage collection. But they take care of these tasks under the supervision of the Ministry of Local Government (Kenya Local Government Reform historically been affecte d by the financial limits that the national government has placed on local government capacity (Oyugi 1983; Smoke 1994). Before making any major decisions, especially budget related issues, LGs needed the approval of the Ministry of Local Government. This centralized system makes them instrumental in the management of urban centers because of the intermediary position they take between the central government and citizens (Akivaga et al. 1985). For instance, the provision of
93 public water services, which is a n important economic development stimulus, tends to be more reliable when provided by LGs than when provided by the central government (Lewis 1998, 142). They also supplement services that the public or private sector is unable or fails to provide. Usually and an underdeveloped private sector, [their] effectiveness in providing physical infrastructure and a conducive atmosphere for economic activity can have an important influence on the success of the (Koti 2000, 22). A critical aspect of local governance, apart from providing basic public services, is facilitating market led developments through the private sector (Barasa and Eising 2010, 1). In order to gain a full understanding of the way that LGs shape urban development in Kenya one has to first understand how the historical transformations that they have undergone. This knowledge is also important because it informs and justifies the compariso components under the same system. The next section provides a brief history of this system. 4.2 A Short History of the Local Government System tablished in 1903 by the British colonial government (Barasa and Eising 2010, 204). This system had two tiers: one that had Africans and another that had European settlers (Akivaga et al. 1985; Smoke 1994; Stamp 1986). The local authorities based in the ar eas that Europeans settled in had a higher degree of autonomy in political and administrative decision making than those in the areas inhabited by Africans (Akivaga et al. 1985). European areas were also well represented in the government than African ones (Koti 2000, 15). In order to effectively
94 control the African areas the colonial government disempowered and underdeveloped their local authorities (Muia et al. 2010, 13). To accomplish this goal it passed laws that established a highly centralized local g overnment system (Muia et al. 2010, 16). It believed that this was an effective model for controlling African countries at the local level (Stamp 1986, 18). Since then, Kenya has had a highly centralized government that overbearingly controls l ocal governm ents (Mwenda 2010, 10). The system was introduced when two of the three cities in Kenya, Nairobi and Mombasa, were established as townships and then later on made municipalities (Stamp 1986). Although most of the main aspects of the initial system still e xist, it has also gradually evolved over time (Muia et al. 2010, 16). Upon gaining independence, Kenya was under a relatively decentralized structure that was negotiated by the British colonial government. Under this agreement the first post independence government was to establish a regional structure devolving power to the regional assemblies and give local authorities autonomy (Akivaga et al. 1985; Smoke 1994). It was not long before the post colonial government reneged on it. The central government wan ted to ensure that it had a firm control over LGs by using the Ministry of Local Government to appoint all the high ranking staff and approve their budgets (Dege Consult 2007, 7). A new constitution was introduced in 1964 that officially abolished the reg ional assemblies (Oyugi 1983; Ochieng 1995, 107; Muia et al. 2010, 17) and established a parallel administrative system, known as the Provincial Administration, to usurp most of the functions of the local government. It passed laws that took away important political, administrative, and fiscal powers from them and gave them to its provincial administrative officers (Mwenda 2010, 11), who were essentially
95 establishing a un itary single party state by creating institutions that allowed the president to shape all public policies with no constitutional limitations (Kimenyi 2004). developme 1974, 27 53; Mueller 1984, 407). This centralized system has not been beneficial for local government development since most local government development decisions have been cont rolled and constrained at the central level (Muia et al. 2010, 16). In 1965 the central government enacted a new Local Government Act (LGA) delegating specific functions to local authorities with regards to the services they would be mandated to be in char ge of. This Act also specified the sources of local authorities revenues, which included a combination of grants it would allocate and those they would have to raise from local resources. Since each local authority is endowed with different resources, the quality of services was dependent on the revenues they could raise. Jurisdictions that had more resources were able to provide better infrastructural and educational services, such as high quality primary school education and durable roads. Those that had less resources depended on grants that the central government allocated for such services. In most cases, the development efforts of the latter authorities were undermined due to the unwillingness and inconsistencies of the central government in making gra nts available to them (Muia et al. 2010, 18). By the late 1960s, the central government, via the provincial administration, had taken over the tax collection role of the local authorities and effectively drained power from the local authorities (Goldsmith and Rothchild 1969, 428).
96 By 1969, it was clear that most local authorities were unable to provide adequate services without substantial assistance from the central government. This situation led to the enactment of a new Act (Transfer of Functions Act of 1969) that would transfer the provision of major services to the central government, such as primary school education, health services and road construction (Muia et al. 2010, 19). This action left them with minimal service delivery functions, mainly road maintenance, market construction, and other minor functions. Due to lack of financial resources following the take over by the central government, local authorities were unable to handle the few functions they were left with (Dege Consult 2007, 85). They lost the power to collect revenues and provide most public services (Patrick et al. 2005, 19). The stipulations of the Act did not favor Kisumu as they favored Nairobi and Mombasa since Kisumu had scarcer resources within its jurisdiction. The relationship that the central government and local governments had after passing the Transfer of Functions Act and authorities [were] autonomous legal entities established by an Act of Parl iament: the the reforms were justifiable because local authorities were facing challenges in raising revenues and maintaining qualified staff (Southall and Geoffrey 1996, 506). It also argued that the level of patronage at the local level was too high and, as a result, councilors were less concerned about providing public services (Gertzel 1970, 37 38). adm central government made significant changes that abolished their power to directly
97 collect taxes and replaced it with a centrally controlled sales tax (Stamp 1986, 27). Due to this change, local authorities had to seek loans and grants for their capital projects 132 in Stamp 19 86, 27). A major constitutional reform that further transformed the local government system was the 1977 Local Government Act Cap. 265 (Smoke 1994). Under this act the Minister for Local authorities was empowered to be in charge of almost all local governm ent operations (Koti 2000, 18). By the late 1970s, the ministry kept close track of council politics and administrative performance through the requirement that every municipal, county, and town council provide an annual report and the minutes of all counc il and committee meetings. Financial controls were exercised through audit of accounts by the controller and auditor general; regulation of local taxes, fees and charges; and approval of the annual estimates, to which a council was obliged to adhere (Stamp 1986, 27). The second post independence regime, under President Moi, usurped more power from the local authorities than the first post independence regime (Southall and Geoffrey 1996, 508). From 1978, when he assumed power, his regime tightened the contro l on local government through more rules. In 1983, the government strengthened its presence at the local government level by introducing a new system known as the Kenya District Focus for Rural Development. This system was particularly used to establish pa appointed District Commissioners [made] decisions on local development and resource strateg y was to divide and rule councilors. Hence, during the 1970s and 1980s, urban development was undermined by internal wars for power and control by councilors (Herbert 1974, Stren 1978, Murumba 1987). By the 1990s Kenya had been transformed
98 to a centralized clientelism, the centralization of power in an executive, the displacement of the party by a bureaucracy answerable to the head of state, and the downgrading of representative institutions (Allen 1992: 305 common feature (Crook 2003, 81). During this period most of the staff in local authorities were recruited based on patronage, hence leading to the misuse of revenues that were used for pe rsonal reasons rather than for financing the provision of public services (Eising et al. 2010, 63). The overall outcome of these transformations was a set of institutions that have been intentionally designed to sustain the centralized bureaucratic system. 4.3 Main Features of the Local Government System 4.3.1 The Governing Rules: Local Government Act of 1978 Up to 2010, Local governments in Kenya have been operating under the Local Government Ac t of 1978 (Action Aid 2006, 1). 1 This Act has significantly sh aped the role that they have played in governance and urban development, from being able to make no less than 186 central government controls and regulations over local au thorities (World Bank Report No. 8997 that the local government system has taken, as well as the legal powers and service provision responsibilities of the local authorities under the supervis ion of the Ministry of Local Government (MOLG). It mainly usurped their powers and gave them to the 1 at will establish a fully devolved system of government. This study was conducted prior to this reform when this Act was still operational. Currently, local governments still technically operate under the 1978 Act. The new constitution is expected to sta rt taking full effect after the 2013 national elections that will be held in March.
99 Minister of Local Governments. The MOLG has been in charge of staff recruitment and the supervision of local government budgeting (Dege Consult 2007, 16). This structure has always been a source of contention between the council and the administration should have the powers to appoint administration officers (Mwenda 2010, 3 2). Apart from administrative powers, the Minister has been given powers that determine the nature of local governance since he can create or dissolve local authorities (Wainaina and Okoth 2002, 40; Nguri et al. 2010, 118; Mwenda 2010, 31). that he has to approve virtually all matters resolved by local authorities (Barasa 2010, some council ors without consulting citizens. This issue has often created political tension between councilors and administrators. This is because appointed councilors rarely respect the authority of city managers since they are protected by the Minister (Muia et al. 2010, 25). In general, the Act merged civil and administrative functions. Civil office. The Local Government Act also stipulated how local authorities were to initiate and i mplement urban development projects. It granted them the power to acquire land and use it for any property that may benefit the public. Under this law, they also had the perso n shall carry out development, in most cases housing projects, within the area of a
100 (Lumumba 2004, 4 5). But the downside is that they have been required to provide more service s than they originally did at independence. Their main responsibility at independence was to provide services that would stimulate national development through a wide range of projects (Oyugi and Kibua 2006, Muia 2008). 4.3.2 Local Authority Urban Developm ent Responsibilities and the Political Challenges Although the government took over control of local government responsibilities it left them with most of the urban development ones that they had been assigned at independence (Stamp 1986, 26). Most of thes e functions make them managers of urban space (Koti 2000, 3). Among the several functions that were assigned to local authorities, the most important ones were those that facilitated the extraction of revenues, such as provision and supervision of business license and regulation or business premises (Muia et al. 2010, 22). But they also have the responsibility of providing a wide range of other services, such as fire protection, education, rubbish collection, sports and recreation, maintaining bus parks, an d water and sanitation (Mwenda 2010, 87). Over the recent years they have been unable to increase or maintain better service provision due to lack of capacity, which in turn has led to the deterioration of urban development in most of the local authoritie s (Barasa and Eising 2010, 1). In spite of the power to shape urban development, most have not had full authority over the projects within their jurisdictions (Barasa 2010, 58; Koti 2000, 77). Public projects that could have been implemented by local gove rnments are usually politically influenced and implemented by other Ministries (Eising et al. 2010, 69 70). The MOLG bolsters projects at the local government through its Department of Urban
101 Development through political favoritism. This has made the admin istrative component of local authorities highly politically inclined in making decisions because of the strong influence that bureaucrats at the national level have on the use of funds, as well as, on the allocation of land and other natural resources (Bar asa 2010, 55). Politicians tend to determine how the administrative offices use development funds, as well as how they monopolize the decision f Kenya, 2009, 3). Patronage by Councilors has constrained project implementation. In most cases they have been known to coerce contractors to bribe them or else they will terminate the projects; they also insist on the employment of their political client s, who may lack the required qualifications (Eising et al 2010, 79). 4.3.3 Local Authority Revenues and Expenditures The main revenue base for LGs is a combination of locally generated revenues and financial assistance transferred from the central governm revenues. Their revenues are also largely influenced by the authority that the central government gives them in collecting revenues from their main sources, (Menon et al. 2008, 12). LG revenues come from various taxes and fees collected within their jurisdiction. Some local governments also receive donor assistance for infrastructural projects. In the highly urbanized areas, such as Nairobi and Mombasa, a substantial amount of revenue comes from property rates and business permit fees (De ge Consult 2007, 11; Lewis 1998, 145). Among these major sources of revenue, they tend to have more autonomy over the business permit fees (Okoth and Odipo 2005, 6). It was introduced by the central government in 1999 to simplify the business registration
102 Odipo 2005, 6 7). LGs collect a significant amount of these fees from big businesses in th e formal sector (Muia et al. 2010, 192). Other sources of revenues that they directly collect are parking fees, market fees, and other minor fees. Another type of revenue that local authorities get from the central government is the Contribution In Lieu of Rates (CILOR). This is the tax that it pays to the LAs for the properties it owns within their jurisdictions. But they are usually viewed as unpredictable (Menon et al. 2008, 16). LGs have not always been able to successfully collect their revenues due to inadequate administrative capacity. Consequently, the national government introduced a more direct form of assistance known as the Local Authorities Transfer Fund (LATF) to strengthen the revenue base of local authorities in 1999 (Nguri et al 2010, 119). use its revenues effectively. But this action is always delayed because the MOLG has to approve them first, especially, those that heavily rely on central government grant s (Akivaga et al. 1985). Most of the LG expenditures go to road construction and maintenance (Mwenda 2010, 104). By the late 1990s it was evident that most of them lacked the administrative capacity required to maintain their projects and deliver public s ervices (Oyugi and Kibua 2006; Menon et al. 2008, 9). In response, the Ministry of Local Government decided to implement a series of reforms that would make them efficient in revenue collection and subsequently service delivery (Muia et al. 2010, 179). 4.3 .4 Recent Local Government Reforms: Stimulating Urban Development The 1990s was a period of transformation for the local government system. Starting from 1995 the MOLG initiated a number of reforms through a program known as the Kenya Local Government Refo rm Programme (KLGRP). This program has had
103 of an evolution of the local government system th at started with the reforms that immediately followed independence (Mwenda 2010, 30). During the last decade the program introduced a number of other programs that have improved financial management and service delivery in local authorities (Barasa and Ei sing 2010, 204; Menon et al. 2008, 9). One major area where improvements have been made is revenue collection (Action Aid 2006, 1). In order to facilitate the goals of KLGRP, in 1999 the central government established a fund known as the Local Authorities Transfer Fund (LATF) to transfer 5% of the national income tax revenues it collects annually to local authorities (Devas and Grant 2003, 312; Mwenda 2010, 32). In order to access the LATF funds all local authorities have to present annual budgets and fina ncial reports to the central gover nment (Nguri et al. 2010, 104). 2 With this assistance they have been in a better position to develop their jurisdictions and also provide better public services. y outstanding debts (Eising et al. 2010, 77). The amount of funds a local authority is awarded under the LATF is based on the total population of the jurisdiction, as well as the performance record with related financial issues, such as well prepared budge ts and financial reports. The LATF allocation formula tends to benefit highly populated areas since it mainly uses the population size to determine the amount of funds to ap portion (Mwenda 2010, 89). Hence, highly urbanized local authorities are more lik ely to get most of the funds. 2 These documents have been the sources of the analysis done in Chapter 4
104 Through the LATF process the KLGRP has also tried to strengthen the linkage between local authorities and citizens. Prior to the introduction of KLGRPs, local governments had no political space for citizens to actively partic ipate in the projects or even push for accountability. LATF changed this situation by making citizen participation a condition for receiving funds from the central government (Devas and Grant 2003, 307). Since 2001, local authorities have been required to involve citizens in the budgeting process by preparing a Local Authority Service Delivery Action Plan 3). Local authorities are required to submit their budgets to the public for review before being submit through the channels that lead to the Minister for Local Governments (Menon et al. 2008, 18). Collaboration and inclusion in the production of LASDAPs a lso includes working with local civil society organizations (Barasa and Eising 2010, 2). This initiative process (wells, clinics, road repairs) appear to reflect th e interests of ordinary citizens better than those formerly prioritized by the LAs themselves (vehicles, offices, Reduction and Local Government Support Programme (RPRLGSP), which targets poverty reduction through supporting reform initiatives that seek to improve service delivery and financial management in rural based local governments (Barasa and Eising 2010, 2). The main outcomes of the KLGRP were the initiation of a Singl e Business Permit, improvement of budgeting and financial management, and increased financial support from the central government. All these were meant to increase and strengthen
105 the revenue base of the local authorities. Put together, these reform program s have been successful in improving urban development (Patrick et al. 2005, 29), particularly in historical factors. 4.4 A Short hree Cities Figure 4 1 below shows the location of the three cities in Kenya that are the subject of this study. Figure 4 1. ities ( Source: Central Intelligence Agency, The World Factbook. 2013. Kenya. https://www.cia.gov/library/publications/the world factbook/geos/ke.html accessed 18 March 2013 ) 3 3 Mombasa is approximately 274 miles from Nairobi while Kisumu is approximately 164 miles from Na irobi.
106 Urban local authorities have played a significant role in stimulating Ke economic growth since the beginning of the 21st century (Menon et al. 2008, 2). More than 70 % of the GDP is generated from its three cities: Nairobi, Mombasa, and Kisumu (Menon et al. 2008, 2). This study focuses on these three cities. 4 They have bee n hailed as major centers of innovation (UN Habitat 2005, 4). Nairobi and Mombasa were the first townships to be established in Kenya. Kisumu, which is a more poorer city than Mombasa and Nairobi, was established much later. These cities have also experie nced different levels of urbanization, which is evident from the different levels of development in the three cities. In terms of how they rank based on the evidence provided in the sections that follow, Nairobi ranks first, Mombasa ranks second and Kisumu ranks third. 4.4.1 Nairobi capital city, has been the dominant city in Kenya because of an investment environment that attracts a wide range of businesses and workers (Otiso 2005, 117). Economies of ttracting a variation of enterprises, even though the government has attempted to create business incentives in smaller cities to attract these enterprises (Obudho 1997). It is the major contributor of 4 extensive information on it than on Kisumu and Mombasa. However, recently Kisumu has been extensively studied for economic development purposes and so its introduction is more extensive than HABITAT.
107 Nair obi w as founded in 1899 (Menon et a l 2008, 11). It was originally established as an administrative center for serving the needs of the colonial settlers by providing them with a wide range of urban services (Furedi 1973, 276; Macharia 1992, 227). Prior to nomads and pastoralists. The construction of permanent buildings began in 1899, when the British colonial government decided to make it their railway headquarters (Macharia 1992, 227) Initially the headquarters were in Mombasa. The expansion of this railway played a significant role in developing the city. It led to the growth of commercial and business activities in the city. By 1900, it was a large and booming place with an increase in buildings and residential homes for arriving European settlers and Indian have an industrial sector (Furedi 1973, 276). By 1938 the unemployed population had drast ically increased due to the rapid migration of rural inhabitants seeking jobs that were highly scarce (Furedi 1973, 277). Urban development in Nairobi experienced a significant boost in the 1940s when the colonial government officially initiated a well fu nded program meant to transform its encapsulated in the 1948 Master Plan for a Colonial Capital, which set out a design for the development of the urban infrastructure f 138). The colonial government planned to construct new administrative headquarters, a modern commercial center, an expanded industrial area, a modern transportation system, and new houses for Africans (Anderson 2001, 1 industry boomed in the period between 1942 and 1955, boosted by the proliferation of
108 municipal housing developments...The infrastructure of the city was hugely expanded by the mid 1950s, as new roads were laid out and electrici ty, water supplies, drains and In 1950, Nairobi was officially acknowledged as a city, and the authority of its city council subsequently expanded (Anderson 2001, 138). From 1954 the colonial administ ration limited the migration of Africans from rural areas to the city by banning public housing for Africans and only allowing squatter settlements (Amis 1984, 89; Obudho and Aduwo 1989). Once Kenya became independent in 1963, this restriction was lifted, leading to a rapid population growth, with most people encountering inadequate housing (Dafe 2009, 14). The City Council of Nairobi did not have the financial or administrative capacity to handle this growth since they were unable to regulate the growth, p rovide land or services to the increasing number of residents (Syagga et al. 2001, 1). It experienced its first major shortfall of housing in 1972, mostly because the government did not have the capacity to match the rapid migrant influx that was taking pl ace in the city (Stren 1978). Due to the lack of a housing policy, the city experienced a rapid expansion of slums (Anderson 2001, 154). By the beginning of the 21st century Nairobi was experiencing a higher rate of urbanization than the other two cities. (Otiso 2005). The high rate was driven by the increase in industries, real estate boom, and expansion of small businesses (Mundia and Aniya 2005, 106). Nairobi is now d commercial center (Menon et al. 2008, 11) and also the most populated city, with an estimated four million people (Kenya National Bureau of Statistics 2010). Its economy is based on industrial activities, local and
109 international trade, financial activiti es and government services. The city not only plays an important economic role in Kenya but also in Africa. It serves as a regional center for several commercial activities for East, Central, and Southern African countries because of its infrastructural ne Road corridor from Cape Town (South Africa) to Cairo (Egypt) passes through Nairobi. It provides the main access route for various markets in Africa, the Indian Ocean and 010, 4). It also serves as a home for several international and regional organizations, including the headquarters of two United Nations Organizations (Menon et al. 2008, 11), as well as a center of international diplomacy, finance, banking, and commerce ( Da Cruz et al. 2006, 9). It is now the largest city in East Africa, with a population of about 4 million people (Kedogo at al. 2010, 5). It is the fourth largest city in Africa with a population growth rate of about 5 % per year (Hendricks 2010, 63). It has the most established manufacturing industries in East and Central Africa (Omwenga 2010, 5). Based on the high quality of corporate services that Nairobi has, it has been ranked as the third global city in Africa (Hendricks 2010, 63). Nairobi is also influ ential in the region due to a high number of education and research institutions that operate from there (Omwenga 2010, 6). 4 88 plan period, it was allocated 42.1 % of total local authority development expenditure (compared to 11.4 % for Mombasa and 31.7 % 1996, 516). Since the 1980s, capital expenditures for serv ice provision have not been able to keep up with the demand (Southall and Geoffrey 1996, 516). This is because of
110 a rapid rate of urbanization that has always kept demand levels above the supply levels n from 120,000 in 1948 to 3,138,369 in 2009. The annual growth rate was about 5% between 1989 and 1999, and 4% during the 1999 2009 decade despite the fall in employment opportunities al. 2011, S186). This fall in employment has since led to a rapid growth of the informal sector (UN Habitat 2005, 19) since it drastically reduced employment in the formal sector (Lyons and Snoxell 2005, 1080). The informal sector has reproduced Nairobi a nd transformed its urban design by increasing informal settlements (Chege 1981, 74). This is also an outcome of a segregated residential structure left by the former colonial lation density [that] ranges from between 358 and 2,490 per km2 in high and middle income o live in highly dense informal settlements (UN Habitat 2005, 19) services for its residents has been very low because of poor local governance (Gulyani et al. 2006, 49 53). Si nce independence, its politics has been marked by rampant corruption in service provision (Greenstone 1966, 204). This has significantly increased the level of poverty and inequality in the city. Almost 60 % now slum dwellers (S yagga et al. 2001, 1). The income distribution in Nairobi favors higher income groups, whose income is almost ten times higher than the low income groups (Hendricks 2010, 63). The quality of housing is very low in the low income areas
111 with barely any servi ces provided by the local government (Da Cruz et al. 2006, 4). current and projected future population falling out of the existing formal frameworks, urban governance issue s are paramount (Kedogo e centralized local government structure has historically undermined the capacity of the city council of Nairobi to respond to issues that affect the city (Da Cruz et al. 2006, 8). Since the late 1980 demand for efficient city services since the rapid urban growth has outpaced the capacity of local authorities to provide and maintain infrastructure and basic services (Stren and White, 19 the 1970s most people living in unauthorized houses and operating in the informal sector have been sustained by patronage from the ruling party, which overlooked their illegal settlements as i t sought their votes (Macharia 1992, 288). Democratic representation at the City Council of Nairobi has for years been significantly affected by clientelistic political practices (Mitullah et al. 2004; Mitullah 2004). Political parties have undermined the efficiency of local authority operations by channeling the minimal resources towards their own patronage activities (Hendricks 2010, 64). Councilors in Nairobi are known for sustaining patronage by bringing and manipulating clients to support their procure ment and tendering projects during council of citizens and often themselves subject to manipulation. Especially in informal settlements this often goes together with politi cization, patronage relations and
112 4.4.2 Mombasa ). It is six times smaller than Nairobi (Hope Sr. 2012, 6). Its working population is three times workers (Otiso 2005, 123). It serves as the principal port for Kenya and ot her East and Central African countries (Hildebrand and Holst 2010, 70; Menon et al. 2008, 11), especially for landlocked countries such as Rwanda, Burundi, Uganda, South Sudan and Congo (Kithiia and Dowling 2010, 470; Awuor et al. 2008, 232; Musingi et al. 1999, pre colonial Kenya when it was used as a trading point for other port cities in Africa, Persia, Arabia, and India. But congestion, poor transportation network, as well as local government mismanagement, has hindered the operation of the port (Rakodi et al. urban migration (Hildebrand and Holst 2010, 72; Musingi et al. 1999, 420). The continuously increasing population of Mombasa has led to land struggles, and competition for the scarce resources available (Kiitha 2010, 471). a trading center between the East A frican Coast and Arabia (Musingi et al. 1999, 419). Since the 11th century, it has been a vibrant maritime trading center that has attracted migrants from all over the country as well as the East African region (Hildebrand and Holst 2010, 70). Since indepe ndence its extensive p ort facilities have provided ca s u al employment opportunities to unskilled laborers (Stren 1970, 38). Its industrial center is second after Nairobi and is linked to major East and Central African cities by a railway
113 and a road system ( established industrial sector, comprised of six large and over 400 medium and small scale manufacturing enterprises. Almost half were engaged in agro processing, with oil refining, cement production and texti 154). Its major industrial activity is oil refining (Musingi at al. 1999, 419). Its economy is however diversified around tourism, manufacturing, trade, and other commercial activities. Although Mombasa has some of the major manufacturing industries in Kenya, a quarter of its population falls under the absolute poverty line (Rakodi et al. 2002, 153). Estimates from the recent Kenya Household and Budge t Survey show that contrast sharply with those for Nairobi where the proportion below the poverty line dropped from 5 0.2% in 1997 to 21.3%14 in 2006 (Hildebrand and Holst 2010, 73). unplanned settlements spr urban poor, who constitute a third of its total population, inhabit about 55 slums spread out around the city (Hildebrand and Holst 2010, 74). Due to low levels of education and lack of training, th e poor in Mombasa have no acces s to wage employment (Rakodi et al 2002, 153). Hence, they engage in unskilled activities in the informal sector (Hildebrand and Holst 2010, 74).Food security and economic performance is very low (Awuor et al. 2008, 232).
114 Mo mbasa was formally included as part of the local government system in 1963 (Municipal Council of Mombasa 2009, 7). Initially, it was a town council that was run by a board established in 1928. This board was accorded Municipal status in 1959, which resulte municipal council which has a mandate to provide education, health and roads. It also provides government and public housing facilities, markets, as well as regulates land us e, handles public health issues, and business licensing (Rakodi et al. 2002, 162). Due challenges. It has been unable to develop a strategy for tackling rising poverty levels and improving service delivery (Hildebrand and Holst 2010, 82). Its revenue comes mainly from property taxes, followed by service charges, market fees, house rents, and business licenses but it has not been historically efficient in collecting these reven ues. Hence, it has had major service provision shortfalls; for example, less than 30% of the system only manages to serve 15% of the population (Hildebrand and Holst 20 10, 82). 4.4.3 Kisumu 223; UN HABITAT 2005, 13). It is located at the shores Lake Victoria. Like Mombasa, it also serves as a gateway to the land locked countries of East and Centr al Africa. Unlike Nairobi and Mombasa, Kisumu is a less developed commercial and administrative center (Menon et al. 2008, 11). Even though it is the poorest, it is also one of the fastest growing urban areas in Kenya (Huchzermeyer 2009, 275; KPMG 2008, 8) Urbanization in Kisumu began in 1898 with the construction of a railway terminus (UN HABITAT 2005, 13). This terminus significantly transformed the economy of the city by increasing
115 trade in commodities across the lake between Kenya, Tanzania, and Uganda This trade generated employment opportunities and increas ed wealth in the city (DA CRUZ et al 2006, 7). In 1930, the boundaries of Kisumu were shrunk by colonial settlers, with the intention of developing specific urban areas and excluding African resid ents from enjoying the same benefits (UN HABITAT 2005, 18). By the 1960s and 70s its economy was growing at a high rate as a major manufacturing center for sugar, cotton, and fish. It was highly prosperous until 1977, when the East African Community collap sed and severed the trading links it had in the region (Earth Institute 2007, 33). Following the collapse of the EAC, the city has struggled to recover from the reduction of shipping services and the decline of railway trade (UN HABITAT 2009, 13). Hence, i ts development story has been one of steep decline and very slow recovery. and 1999 (UN economic growth: e xposure to international competition combined with under investment in infrastructure caused decline in industrial activity; trade liberalization led to influx of cheap imports, exposing inefficiency of parastatal enterprises; cotton collapsed and brewing industry left Kisumu; uncompetitive sugar industry sustained through import tariffs, and literacy levels fell due to increasing poverty with urbanization becoming a significant problem (KPMG 2008, 9). cantly affected the city through an increase in collapsing industries and commercial enterprises. Its industries were not able to compete successfully with cheaper imports that flooded the market during the economic liberalization era of the 1980s and 90s (Aguilo et al. 2007, 18; KPMG 2008, 8). This decline was accelerated by the effect of World Bank structural adjustment programmes, which reduced the employment rate by retrenching civil servants (Da Cruz
116 et al. 2006, 23 24). In spite of the significant cha llenges facing Kisumu, it has been somewhat successful in developing an economic base (Otiso 2005, 127) that plays a location on the shores of the second largest fresh water lake in the world, Lake Victoria, overall agricultural output is about 25%, a lthough most of its agricultural activities are under subsistence farming (KPMG 2008, 10). Its economy is traditionally based on cotton, fish, and sugar, which have been declining in that past few years due to stagnation in trade (Aguilo et al. 2007, 4). increased at the fastest pace, and it has the highest levels of poverty (UN Habitat 2009, of the population below the food poverty line in comparison to 38.6% in Mombasa, and from the informal sector where monthly incomes are as low as 40 dollars (UN HABITAT 2005, 21; Wagah et al. 201). The low quality of its demographic characteristics has contributed to the expansion and persistence of the informal sector. It has a labor force that is highly unskilled, its adult literacy rate that is lower than the national average, and it has a high gender parity in education between boys and girls (Aguilo et al. 2007, 14 15). The existing educational facilities are inadequate for the increasing population, and there are great inequalities in teaching standards and services between government run
117 in Kisumu lack the skills needed for attracting high end jobs such as accountancy, engineering, and technical farming (KPMG 2008, 26). Kisumu furt her suffers from the problem of an inadequate shelter to accommodate its inhabitants. About 75 % of its population live in temporary and informal settlements (slums), with about 60 % of them lacking basic services (Aguilo et al. 2007, 14 lings are without electricity and occupants use paraffin lamps, (Aguilo et al. 2007, 14 15). The uncontrolled expansion of slums has contributed to the infrastructural pro and under developed transport infrastructure presents a financial and logistical contributed to this stagnation by increasing the cost of doing business and affecting farmers ability to deliver fresh produce to markets in a timely manner (KPMG 2008, 26). The physical infrastructure of the city is more developed in the high income residential areas, leavin g the slum areas with a very poor road network (UN HABITAT 2005, 19). Historically, infrastructural development has been biased towards the high income areas because of poor governance and limited council resources (UN HABITAT 2005, 20). The municipal gove rnment has unsuccessfully struggled to keep up service provision with the rapid population growth because of political and financial challenges (Aguilo et al. 2007, 4). Service provision has been largely undermined by the lack of administrative capacity to collect revenues. The Municipal Council of Kisumu only manages to collect about half of its potential revenue (Da Cruz et al. 2006, 9). This has limited its ability to generate funds for its urban development projects (UN HABITAT
118 2005, 21). Like Nairobi and Mombasa, it mainly collects its revenues from property taxes, land rents, business permits, as well as the financial support that comes from the Local Authorities Transfer Fund (Menon et al. 2008, 14). But the highly centralized decision making under the local government system has undermined its ability to access these funds because the disharmonious relationship that it has historically had with the central government (Aguilo et al. 2007, 4). This is because it has been home to antagonistic oppositio n parties, which the central government has traditionally punished opposition politics that once was a hallmark of the Kisumu region often pitted the local political l eadership against the central government. The upshot was a slowdown in HABITAT 2005, 24). The combination of the antagonistic relationship and the highly centralized local government system highly affected its dev leading to a steep decline in economic growth and an sharp increase in extreme levels of poverty (UN HABITAT 2005, 25; Da Cruz et al. 2006, 4). Since then, the city has struggled to reco move out of under development placed upon it during the one party system and the frequent fallouts this involved between the local party leadership and the central HABIT AT 2005, 24). In spite of the improved relationship with the central government, the municipal challenges. For example, it has registered an expansive freehold land bu t failed to allocate land for developing its infrastructural needs and public facilities (Huchzermeyer
119 2009, 276). It has also been biased in its response to the needs of the poor. Recent urban improvement programs have given low priority to the municipal rental houses that are mainly occupied by the poor (UN HABITAT 2005, 20). Slum dwellers, who live in rural areas, have almost completely excluded from decision making on urban development issues (Da Cruz et al. 2006, 4). One of the main contemporary challe nges the council is therefore facing, in light of the increasing urbanization, is the inclusion of about 80 % of its land that is predominantly rural in its urban de velopment plans (Da Cruz et al. 2006, 18). Table 4 1 A s ummar y of urban development in Ken three c ities Nairobi Mombasa Kisumu The principal manufacturing center in Kenya. The main commercial center and regional headquarters for numerous international and regional organizations. Nairobi international airport (JKIA) has assumed increasing importance as a regional hub for East and Central Africa and a major program for upgrading and improving the facilities is ongoing. Nairobi is the largest city in East Africa with a population around 4 million. Has the lowest poverty head count index amon g the three cities 43.9% 3.4 million people (2009) Area: 696 Km 2 The principal seaport of Kenya. A principal manufacturing center. It is the best equipped and most modern port in East Africa. Has two tanker berths for the bulk transfer of crude and ref ined petroleum. The oil refinery at Mombasa has an annual capacity of 4.2 million tons of crude oil. Is based in a province with a lower poverty index than where Kisumu is, but higher index 57.6%. 996,000 people (2009) Area: 295 Km2 The main Kenyan port on Lake Victoria. Not a principal manufacturing center; only has two main industries: a textile and a sugar refinery. It is the commercial center of western Kenya. Most of Kenya's sugar factories are located within a 50 miles ra dius of the city. Is based in the province with the highest poverty index in Kenya Nyanza 64.6%. 394,684 people (2009) Area: 417 Km2
120 C hapter 4 has demonstrated that Nairobi has always had the highest level of urban development, followed by Mombasa, and then Kisumu. The descriptive analysis of the trends in urban development, based on data collected from the local governments shows that this tendency has been further strengthened over the past decade. Another important goal of C hapter 4 was methodolo gical. The analysis establishes that there is unit homogeneity by showing that they have undergone similar changes under the former centralized local government system. The examination shows that since independence they have operated under the same laws an d administrative processes under the supervision of the Ministry of Local Governments. This is what makes them comparable. The larger issue that this study seeks to address is how the varying trends in urban development have influenced clientelistic linkag es in the three cities. Chapter 5 and Chapter 6 analyze how the level and modes of clientelism vary across the three cities.
121 CHAPTER 5 Chapter 4 provided a comparison of the level of urban developme three cities, thereby laying the foundation for examining the relationship between urban development and clientelism. C hapter 5 carries out this examination. It is divided into three sections. The first section provides a contextual backgroun d for the analysis by describing the historical roots of the clientelistic linkages that prevail in Kenya today. (Mueller 2011, 105). In essence, current clientelistic n etworks are mostly well maintained old networks. I attempt to show why and how they were formed, and how they have been sustained and strengthened over time. The section does not attempt to cover every historical aspect but the key transitional moments tha t provide an illuminating background for this study. The second section shows the variation of urban development in the three cities. The third section examines the relationship between urban development and socioeconomic status in the three cities in the three cities. It compares the nature of this relationship across the three cities. Having shown these variations, the final section conducts an analysis showing the moderating effect that socioeconomic development has on political party clientelism. 5.1 Pa rty Clientelistic Linkages in Kenya: The Roots (Murunga et al. 2006, 10). The modern day form of political clientelism has its roots in the first post independence regime that was led by President Kenyatta (Steeves 1997, 27). The form of patronage that has been dominant in the current multiparty system is also an outcome of dominance by a wealthy political class that was part of the first and
122 second post independence regimes (Na rman 2003, 344; Omolo 2010, 219). Their wealth was instrumental in sustaining the clientelistic linkages they formed with citizens. party system. Aside from the implementation of laws allowing for multiparty politics, there has not occurred a fundamental structural change One can get a clear picture of how the current citizen politician linkages in Kenya operate by analyzing their historical roots because they have not undergone any significant transformations (Murunga et al. 20 06, 10). This is the goal of this section. particular forms of patron client relations, particularly those in which the patron plays the role of a broker on behalf of clients tha t are distant from the administrative power their political allies to establish patronage networks between the ruling party and citizens at the local government level (Klop p 2001, 90). The prevalence of such (Mutua 2010; Mwangi 2008, 269). The absence of such a system has meant that leaders appeal to ethnic sentiments when seeking for supporters (Mwangi 2008, 269). Hence political parties have often built on the dominant ethnic cleavages in the country rather than ideological platforms (Elischer 2008, 3). The history of party organization and establishment of linkages with citizens afte r independence shows how old ethnic and patronage networks
123 have had a significant role in shaping the type of party citizen linkages that exist in Kenya today (Carey 2010, 58). The development of political parties and the types of linkages they establishe d with citizens after independence was largely shaped by inherited colonial governance strategies and institutions (Murunga et al. 2006, 10; Berry and regional links colonial political system (Steeves 1997, 27). These politicians were first elected in the colonial Legi slative Council in 1957. This is when two factions that later became Kenyan African National Union (KANU) and Kenyan African Democratic Union (KADU) began forming. At their infancy, they were only allowed to operate at the dis trict level following a 1953 ban on national parties (Carey 2010, 55; Throup 1993, 372). Under these restrictions, both KANU and KADU formed around loose coalitions of local ethnic leaders (Barkan 1987, 218). In 1959 the colonial government lifted the rest rictions and for the first time allowed parties to be formed at a national level. Because of the short period that was available for organizing and preparing for the elections that were to be held in 1962, party formation occurred along two different lines : one was at the behest of a single leader who provided intellectual and financial patronage and who drew a core of founders linked more by personal ties forged in the ethnic arena than by ideological commitment. The other track stemmed from the colonial p ractice of restricting African political activities to local areas, which were ethnically homogeneous (Ndegwa 1997, 604). Both KANU and KADU followed this structure (Carey 2010, 56). They particularly sought for political power by using patronage to seek c lients within the ethnic communities that were represented in the parties (Berman et al. 2011, 463).
124 the country based on ethnic composition (Fox 1996, 597). This division set the tone for (Fox 1996, 607). The 1962 elections revealed the political culture and political strategies that would come to dominate in Kenya, with voters mainly voting fo r candidates that advantage of this prevalent perception and strategized effectively around it. Due to organizational weakness within the opposition party (KADU), KANU won the elections. Meanwhile KADU experienced shifting alliances between leaders of different ethnic an et al. 2011, 478). Kenyatta gave loyal MPs access to state resources so that they could use them to secure electoral victory for KANU (Barkan 1987 cited in Klopp 2001, 82). The outcome of this strategy was an extensive use of the bureaucracy in banning political meetings and distribution of political positions to political leaders of the opposition party, KADU, with the goal of establishing one party rule (Mueller 1984; Oyugi 1992). KANU tightened its control on other opposition leaders as well. All oppo sition candidates seeking parliamentary and city council seats could only do so through KANU (Africa Research Bulletin 1966, 512). In 1964 the opposition party was informally dissolved and as most of its leaders joined KANU (Throup 1993, 373). erience with the multiparty system after independence only lasted for opposition parties, making Kenya a one party state (Carey 2010, 58; Throup 1993,
125 375). However, intra pa rty competition still continued between politicians in the ruling party, KANU. Under the single authority on elders who belonged to an upper income class and had been accorded the privilege of makin g decisions for their communities (Berman et al. 2011, 472). To be (Berman et al. 20 11, 474). Kenyatta not only encouraged the formation of patron client links but also used the government to monitor them (Carey 2010, 67). A more clear sign of his intention was a major constitutional reform that he made in 1966, stipulating that the gover nment was going to use the civil service to oversee such operations (Wanjohi 1997, 127). The main goal behind this strategy was to use the civil service to establish clientelistic linkages with citizens through various projects. He mainly established linka ges through self help organizations, whereby citizens organized projects at the help development pr parliament took advantage of this self help system and used it as a tool for building local sup port by contributing financially to the small projects; citizens also sustained this system by casting their vote based on the extent to MPs contribution (Carey 2010, 63). The rationale behind this self help system was to make the ruling party to appear mo re accountable and also responsive to the needs of the public (Carey 2010, 64).
126 Kenyatta cemented this patronage system by pushing citizens to seek their (Barkan 1992, 177 politicians devoted the time they spent in their constituencies to discussion of these 83). The linkages that MPs formed with voters were however multiparty movements were being formed (Widner 1994, 142). Throughout the late 1970s, therefore, Kenya had a p atronage network that was characterized by political parties that selectively responded to citizens demands at the local level (Tamarkin 1978, elite to rural communities client networks were highly developed (Tamarkin 1978, 314). This was the political context in 1978. But since he wa president he was quite aware of the patronage system and also a good disciple. client system. Moi continued the kind of ethnic politics that was perpetrated by Kenyatt a when he radically change his entire cabinet by filling almost all the positions with members of his ethnic group, the Kalenjin (Widner 1992, 165 166). He diverted state resources that hnic group) to his ethnic group (the Kalenjin) and regions where his allies came from (Throup 1993, 371). His administration was notably known for misusing state resources to sustain
127 rampant political patronage during campaigns, including bribing and buyin g out political local and regional clientelist organizations that embraced a diversity of interests, Moi came to view the party as a mechanism to control leaders with indepen (Barkan 1992, 177). Throughout the 1980s Moi intensified patron client networks by establishing a program known as District Development Focus, which was used for channeling patronage resources in areas that his party sought political suppo rt (Anangwe 1995, 72). While these projects encouraged local participation, they also forced MPs to establish linkages with citizens at the local level (Barkan and Chege 1989). In the 1990s, Kenya was marked by political turmoil as internal and external pr party politics (Berman et al. 2011, 463). Under so much pressure from civil society and the international community, Moi eventually ended the one party rule in 1991 and restored the m ulti party system. Numerous parties were formed and launched around the political leaders of different ethnic communities. These parties were mostly patron client based networks between ethnic leaders and their supporters as they lacked a cohesive ideolog y or a concrete organization structure (Berman et al. 2011, 481). In the first multiparty elections in 1992 the new opposition parties were unable to unseat President Moi because of extreme divisions and conflict of interests between leaders and followers (Berman et al. 2011, 481). These parties primarily had endless contests multiparty elections on 1997 (Oyugi 1997).
128 The restoration of multipartyism created competitio n between state elites that had both extensive networks and access to resources and those who have neither of these advantages. Hence, the re introduction of multiparty politics led to what Klopp mpetitive elections escalated Under this context, KANU was able to survive during t he multiparty era because it secured support from the extensive patronage networks it had built over the years (Van De Walle 2007, 19). While intensifying and using the same strategies, KANU set a template for a party structures that revolve around the whi ms of a powerful party leader they played significant patronage roles during its for ty year rule (Berman et al. 2011, 478). Most of them quit KANU and became civil society leaders that were agitating for the end of one party rule while also setting up their own parties (Mutua 2010, 113). But throughout the democratization period they have used patronage strategies. The current political parties in Kenya are driven by powerful and wealthy politicians from the previous regimes, who have made them vehicles for personal rather than national interests (Mutua 2010, 115). They are products of th e early 1990 ethnically based opposition parties (Elischer 2008, 24), but with different names. They have still not distinguished themselves on ideological or programmatic platforms utors of private rather than public goods (Mueller 2001, 105). They still distribute national resources to
129 their clients based on ethnic backgrounds (Mueller 2011, 105). The high degree of ethnic diversity in Kenya has made it difficult for parties to secu re victory without forming coalitions with parties representing other ethnic groups (Cheeseman 2008, 172). Hence parties now base their winning strategies around such coalitions (Elischer 2008, 3). The overall legacy of past party organization strategies i s political parties with weak democratic linkages to the society because of lack of mechanisms for supporters to articulate interests or shape how their parties operate (Barkan & Henderson 1997, 27 28; Mwangi 2008, 269). Previous and present parties have had an inherent failure of being unable to respond to the demands of the masses (Carey 2010, 65). Citizen politician linkages have also been tenuous because the party system in Kenya has always been in a state of fluidity because of frequent shifts in par ty affiliations among elected political representatives, especially Members of Parliament (Elischer 2008, 21). Shifting affiliations among MPs has been a critical factor because they have always by being the link between these bosses and the citizens (Berg traditionally advocated for the demands of the upper class through the administrative hierarchy but responded to those of the poor through patr on client networks (Tamarkin client network (Barkan 1987, 231). Moi effectively used them to create a network of patronage bosses to buy the loyalty of party members and citizens (Thr oup and Hornsby 1998, 331). Councilors have also been traditionally expected to support these networks or be ejected from their positions (Kassim 1999, 48 50). The goal of this study is to find out how the increasing rate of urban development is shaping th ese clientelistic linkages. It
130 begins by showing the variation in the level of urban development in the three cities in Kenya, it then compares the socioeconomic status of business owners across the cities and then the occurrence of client elism in those ci ties. Figure 1 2, in C hapter 1, illustrates the argument analyzed in this C hapter 5 5.2 This section conducts a descriptive analysis that compares their level of urban development. They focus on initi atives that have been undertaken by the respective local governments and citizens. 5.2.1 Infrastructural Development Infrastructural development is an essential component of urban development (Bertinelli and Black 2004, 81). The role of developing infrastr ucture usually lies with local governments (Black and Henderson 1999). They also conduct other physical development projects that increase the level of urban development. In order to compare the levels of infrastructural and other physical development init iatives that the three city councils have undertaken I used the data on their annual capital expenditures. This data captures the amount of revenues that they have spent on infrastructural projects and service provision (Appendix A provides a detailed desc ription on how this data was collected). In general, local governments in Kenya use most of their revenues on road construction and maintenance, construction and maintenance of education facilities, public lighting, solid waste, construction and maintenanc e of water and sanitation and health facilities. The bulk of their expenditures are therefore used in enhancing service delivery to citizens. Higher amounts in capital expenditures, therefore, reflect three things: urban development is high, there is bette r management of the urban area, and citizens have better access to public services. According to Figure 5 7 below, all three
131 cities had very low levels of capital expenditures from 2002 to 2006. From 2006 to 2009, the level of capital expenditures in Nairo bi rapidly increased, while that in Mombasa and Kisumu still remained generally flat, with a very small increase in Mombasa. This could indicate that access to public services rapidly increased for business owners in Nairobi. The findings from this study s how that in Nairobi most of the business owners rely on the local government for such services than in Mombasa and Kisumu. When asked whether they seek government services from their local governments, 52% of the business owners in Nairobi responded that t hey do, compared to 20% in Mombasa and 28% in Kisumu. This could be because these services are more available in Nairobi than the two other cities. Lack of such services has a negative effect on business productivity and subsequently on socioeconomic devel opment. 5.2.2 Market Development Another important feature of urban development, especially in developing countries, is the availability of affordable business premises for business owners to conduct their businesses (Devereux et al. 2007). The increase in these premises can be from private individuals or local governments (Harpham and Boateng 1997; McGill 1998). Thus, the physical construction of these premises is a major driver of urbanization and urban development. In this study, I used data on fees coll ected from city council owned markets to analyze the rate of urban development driven by the city council economic activities. The findin gs presented in Figure 5 5 show that the city of Nairobi has experienced a higher increase in new markets than both Mo mbasa and Kisumu. It also shows that the rate of increase in markets is slightly higher in Mombasa than in Kisumu. There is almost no increase in Kisumu.
132 5.2.3 Property Development Urban areas that are experiencing high levels of development are usually ch aracterized by high levels of property development (Weng 2007). In order to compare property development driven by citizen actions, I use data on fees that local authorities collect from property rates to get an idea of the development of new properties. T his is a form of property tax that local authorities impose on a wide range of buildings. An increase in property rates should indicate an increase in the development of new property. However, the increase could also be due to improvements in personnel ass igned to collect the fees. More information from local authorities can shed more light on this issue but it is missing in the reports. But the changes on property development suggests that the increase could be from property developments. The data shows th at Nairobi has generally experienced a higher increase in property development than Mombasa and Kisumu over the period 2002 to 2009. As Figure 5 6 shows, Nairobi tends to have the highest share of new properties over time. In Nairobi, property rates colle cted have generally increased at a higher rate than Mombasa and Kisumu, with the exception of 2005, when the rates collected plunged. Both Mombasa and Kisumu have generally had very low to almost no increase in property rates collected, apart from a slight increase in Mombasa in 2008. 5.2.4 Business Development Urban development increases when private economic activities increase (Alcacer and Chung 2007; Moomaw 1983). One way in which these activities increase is through 2009, 43). It is impossible to conduct a survey of proxy is instead the fees collected from business permits by city councils of Nairobi,
133 Mombasa, and Kisumu to anal yze the rate at which new businesses are emerging in the three cities. According to the reports, Nairobi has had a continuous increase in business growth over the period 2002 to 2009. 5.3 Urban Development and Variation of Socioeconomic Status Cross tabul ation has been used to analyze the relationship between a set of dependent and independent variables that show the relationship between urban development and socioeconomic status, which includes examining its relationship with business performance. I test for the strength of the relationships using four types of tests: chi ratio chi squared (X2)statistic examines the likelihood that the results occurring by chance. The significance levels test for the relationship between nominal variables. The strength ranges from 0 to 1, that are less than 20 indicate a weak relationship. Values that lie between .20 and .49 indicate a moderate relationship, and those that are more than .50 indicate a strong relationship. Gamma tests for the relationship between ordinal variables. Similar to strength of the relationship varies from 0 to 1, with 1 indicating a highly st atistically significant the null hypothesis is rejected because there is a relationship between the variables. The argument pursued in this study is that at an aggregate level clientelism mostly occurs in urban contexts that have low levels of urban develo pment, but when disaggregated the occurrence of different types of clientelism vary across urban areas.
134 This happens because of the transformation of socioeconomic status brought about by increase in per capita income and employment growth (Black and Hende rson 1999, 253 254). These are outcomes of urban development infrastructural and real estate projects (Bertinelli and Strobl 2007; Stroper and Scott 2009; Waddell 2007). There are four sources of economic growth that affect socioeconomic status in urban a reas. I have used them to come up with measures for analyzing the effect of urban development on clientelism in the informal sector, namely: capital deepening, Sullivan 2009 90 it Sullivan 2009, 90). In an urban economy, capital deepening is measured by the level of physical capita l available for production of goods and services. These are typically in the form of machines, equipment, and buildings constructed to facilitate production of goods and services. Businesses that have high productivity and high income are usually large in size (Todaro and Smith 2009, 328). Evidence of such productive clusters has been difficult to gather in the informal sector due to lack of data, but it is possible to use the estimate level of capital investment in a business and the number of employees as proxies for income and business size, respectively. Urban development improves economic status by improving these two factors (Henderson 2000; Njoh 2003). In urban areas that are more developed, clustering of large businesses leads to high er level of business productivity (Moomaw and Shatter 1996). The high quality of these businesses is associated with the high level of capital investments in the businesses (Burmeister and Turnovsky 1972, 842). The substantial capital investments
135 required for businesses that use higher levels of physical capital imply that business owners experiencing high levels of poverty are unable to own them. Form fieldwork that u se advanced machines and deal with bulky products. For example, furniture businesses, welding businesses, and car repair businesses. The nature of these businesses offer more employment opportunities due to their labor intensiveness. We should expect to se e these businesses in a more developed city (Nairobi). To examine the relationship between urban development and level of poverty I use a variable that is based on the estimated level of capital investment (capital intensity) in a business and the size of equipment owned by individual businesses. This variable was created using estimates that were provided by business owners and the size of equipment observed in their businesses. The business owners were asked to provide a financial estimate of the total a mount of money they invested in purchasing the equipment they use to conduct their business activities. These measures were used to create three capital intensity categories that were cod ed for each respondent. Table 5 1 below shows the factors that were u sed to create the categories. Table 5 2 shows the relationship between level of urban development and the capital i ntensity in a business. Table 5 2 shows that most (68%) of the businesses with high capital intensity are based in Nairobi, followed by Momba sa (29%), and then Kisumu (3%). This suggests that the socioeconomic status of most business owners in Nairobi is the highest. With regards to the factors that drive the costs of clientelis m, the findings shown above imply that the cost of clientelism is likely to be much higher in Nairobi than in
136 Mombasa and Kisumu. Nairobi has the highest level of development in infrastructure market s property, and business. Hence, it is appropriate to assume that the level of socioeconomic status of businesses is the highest When the cost of clientelistic strategies is too high politicians are likely to avoid using it or use it moderately. Table 5 1. Measuring capital i ntensity Capital Intensity (Level of poverty) Overall size of physical capital (as observed) Capit al investment estimates as provided by business owners (in Kenya shillings) Main types of businesses (as observed) Low (High) Small 2,000 5,000 Green grocer, Hair dresser, Restaurant, fish monger, grains and cereals, Hand crafts, barber, fruit stand, Moderate (Medium) Medium 6,000 20,000 Tailor, Electrician, cloths and fabrics seller, shoe seller, butchery, laundry, beauty products dealer High (Low) Large More than 20,000 Restaurant, Computer services, Hardware shop, Welding, Retail shop, Bakery, Carpenter, Car mechanic The size of physical capital entails the approximate surface area and the number of equipment used by the business. Those that fall under the small category tend to operate in make shift structures or lay their items on the ground and have a maximum area of about 50ft squared to 100ft squared. They have about three to five items and do not have any employee, apart from the business owner. They do not use electricity or running water to conduct their operations. Those that fall und er the medium size
137 operate in semi permanent to permanent structures that are walled, operate in an area of about 100ft squared to 300ft squared. They conduct their operations using fairly bulky equipment, own about five main machines for conducting their operations, and have 1 to 2 employees and the business owner. They tend to use electricity and running water. The businesses that fall under the large size operate in an area that is about 100ft squared and more than 300ft squared. Own more than five main equipment that are bulky in size, they use more advanced equipment than the medium sized ones, are well equipped for their operations, have more than three employees, and consume more running water and electricity than the medium sized ones. Table 5 2. Ur ban development and capital i ntensity Level of Urban Development N (%) Low Moderate High Capital Intensity Kisumu Mombasa Nairobi Total Low 73 (61) 60 (50) 38 (32) 171 (48) Moderate 45 (37) 39 (33) 32 (27) 116 (32) High 2 (2) 21 (17) 50 (41) 73 (20) Total 120 (100) 120 (100) 120 (100) 360 (100) p < .001, With regards to the factors that drive the costs of clientelism, the findings shown above imply that the cost of clientelism is likely to be much higher in Nairobi than in Mombasa and Kisumu since it has the highest (41%) number of businesses with high capital intensity. Kisumu has the highest number of businesses with low (61%) and moderate (37%) capital intensity. Mombasa has more businesses with moderat e capital intensity (17%) than Kisumu (2%). When the cost of clientelistic strategies is too high politicians are likely to avoid using it or use it moderately. Another measure that indicates the level of prosperity of a business owner is the size of the b usiness. Table 5
138 three cities. I use the number of employees a business has to estimate its size. Since most (67%) of the businesses do not have any employees apart from the bu siness owner, I consider a business with one or more paid employees as being larger in size. 1 Table 5 3. Urban development and business size (having e mployees) Level of Urban Development N (%) Low Moderate High Has Employees Kisumu Mombasa Nairobi Total No 85 (71) 91 (76) 66 (55) 242 (67) Yes 35 (29) 29 (24) 54 (45) 118 (33) Total 120 (100) 120 (100) 120 (100) 360 (100) p < .01 Table 5 3 shows that most of the larger businesses in Nairobi (45%), compared to 25% in Mombasa and 29% in Kisumu. Hence, we should expect to find more successful business owners in Nairobi. Due to lack of data on their revenues and profits, I have used the ability of a business owner to pay the rent of their premises, the ability to purchase business st ock, and the ability to acquire business loans as proxies for indicating the successfulness of the business. These are by no means accurate measures of business success but suitable proxies in this case. While some business owners did not have this data re adily available during the surveys, most of them did not because they do not have any financial records. These indicators were provided during in depth interviews when business owners were asked how they measure their business performance if they did not k eep any financial records. This information was then included in a survey question asking them which financial difficulty they mostly 1 A breakdown of the employees in the businesses shows that 67% of them have no employee, 12% have one employee, 11% have two employee s, 4% have three employees, and 6% have more than three employees.
139 encounter. In a related follow up study to this one, which investigated the relationship between microfinance institutions and the informal sector, I used survey questions to find out the percentage of business owners that measure their business performance by looking at their flow of stock and ability to pay rent. The study was based on 300 surveys conducted in the same thre e cities and the same markets. The findings showed that 88% of business owners measured their performance by only looking at the flow of stocks and 53% used their ability to pay rent. Using the ability to acquire business loans as a measure is strengthene d by the fact that most financial institutions scrutinize business owners based on their financial successfulness before approving them for a loan. In most cases they try to establish that the businesses are financial stable and have the potential of maint aining financial success if given a loan. The three proxies are therefore useful in getting a good idea of which city has more successful businesses. Usually, businesses that are doing well are able to meet the rental cost of their premises with less diff iculty, re stock frequently as needed, and acquire business loans with less difficulty from financial institutions. Table 5.4 shows a positive relationship between business size and the three proxies. Tabl e 5 4. Larger business size and financial s uccess Business Has Employees N (%) Business Challenges No Yes Total Difficulty paying rent 82 (34) 27 (23) 109 (30) Difficulty purchasing required stock 95 (39) 31 (27) 117 (33) Difficulty acquiring business loans 61 (25) 56 (47) 126 (35) Other 4 (2) 4 ( 3) 8 (2) Total 242 (100) 118 (100) 360 (100) p < .001, Th e findings presented in Table 5 4 show that it is mostly small businesses that experience difficulties in the issues that directly affect business operat ion. 34% have
140 difficulties in paying the rent of their premises and 39% in re stocking frequently, compared to 23% and 27%, respectively, of those that have employees. On the contrary, most of those that have employees have difficulties in the issue that i ndirectly affects business. 47% of them have difficulties acquiring loans compared to 25% of those that have no employee. Since business owners that have larger business sizes are mostly in Nairobi (high level of urban development), we can cautiously concl ude that they are more financially successful than those in Mombasa (moderate urban development) and Kisumu (low urban development). Higher levels of human capital can also have a moderating effect on clientelism since they enable one to effectively manag e business. Human capital increases when a f access to apprenticeship training and its low cost in the informal sector leads to increase in human capital. Studies done by urban economists (e.g. Glaeser, Scheinkman, and Shleifer 1995) have shown that human capital has a positive effect on business p roductivity and income in urban areas. formal and informal settings, and a worker w ith more human capital has more eloping countries completing secondary education has a positive effect on income growth (Mody and Wang 1997).
141 Urban development also has a positive effect on education outcomes because of the availability of more learning institutions and learning opportun ities. These educational outcomes usually translate into better business management skills and business productivity (Bosire and Gamba 2003). This study measures human capital by looking at the level of education and whether a business owner has attained e ducation on business management. The level of education ranges from lack of formal education to completing college education. Having college education and having business education indicates high human capital. We should expect to see higher levels of huma n capital in Nairobi. The findings presented in Table 5 5 and Table 5 6 below show that both of these aspects are higher in Nairobi than in Mombasa and Kisumu. Table 5 5. Urban development and level of e ducation a ttained Level of Urban Development N (%) Low Moderate High Level of Education Kisumu Mombasa Nairobi Total No formal education 4 (3) 1 (1) 2 (2) 7 (2) Primary school educated 27 (23) 49 (41) 21 (17) 97 (27) High school educated 59 (49) 54 (45) 40 (33) 153 (42) College educated 30 (25) 16 (13) 57 (48) 103 (29) Total 120 (100) 120 (100) 120 (100) 360 (100) p < .001, Table 5 5 shows that in Nairobi 48% of the business owners have attained college education compared to 13% in Mombasa and 25% in Kisumu. The kind of education one gets also matters for business management. Business owners who have had some form of formal business education are more likely to manage their businesses well and depend on them for their livelihood. Most of the business owners t hat have had formal business educa tion are in Nairobi. As Table 5 6 shows 43% of the business
142 owners in Nairobi have had formal business education, compared to 21% in Mombasa and 23% in Kisumu. Table 5 6. Urban development and attainment of business e duca tion Level of Urban Development N (%) Low Moderate High Attained Business Education Kisumu Mombasa Nairobi Total No 92 (77) 95 (79) 77 (64) 264 (73) Yes 28 (23) 25 (21) 43 (36) 96 (27) Total 120 (100) 120 (100) 120 (100) 360 (100) p < .01 s V=0.15 Human capital accumulation not only comes from getting formal education and formal business education but also other forms of learning opportunities. Learning opportunities that have an effect on business productivity are those that lead to acqu isition of knowledge on how to manage businesses (Bosire and Gamba 2003, 1). In order to examine these effects this study focuses on the number of business workshops and business seminars that business owners in the informal sector have attended. Table 5 7 and Table 5 8 show that business owners in Nairobi have benefited from such learning opportunities than those in Mombasa and Kisumu. As shown in Table 5 7, Nairobi 18% of the business owners have attended more than two business workshops compared to 3% in Mo mbasa and 7% in Kisumu Table 5 7. Urban development and attendance of business w orkshops Level of Urban Development N (%) Low Moderate High Workshops Attended Kisumu Mombasa Nairobi Total None 95 (79) 110 (92) 79 (66) 284 (79) 1 to 2 workshops 17 (14) 6 (5) 19 (16) 42 (12) More than 2 workshops 8 (7) 4 (3) 22 (18) 34 (9) Total 120 (100) 120 (100) 120 (100) 360 (100) p < .001,
143 Table 5 8. Urban development and attendance of business s eminars Level of Urb an Development N (%) Low Moderate High Business Seminars Attended Kisumu Mombasa Nairobi Total None 83 (69) 111 (92) 74 (62) 268 (74) 1 to 2 seminars 25 (21) 8 (7) 23 (19) 56 (16) More than 2 seminars 12 (10) 1 (1) 23 (19) 36 (10) Total 120 (100) 1 20 (100) 120 (100) 360 (100) p < .001, Table 5 8 shows that 19% of the business owners in Nairobi have attended more than two business seminars, compared to 1% in Mom basa and 10% in Kisumu These findings suggest tha t business owners in Nairobi have better knowledge on how to succeed in business than those in Mombasa and Kisumu. Subsequently, their socioeconomic status should be higher than for those in Mombasa and Kisumu. This is the case when we take into considerat ion the relationship between attaining business knowledge and financial successfulness. Table 5 9 shows a positive relationship between attaining formal business education and financial successfulness. Table 5 9. Attaining business education and achievin g financial s uccess Attained Business Education N (%) Business Challenges No Yes Total Difficulty paying rent 86 (33) 23 (24) 109 (30) Difficulty purchasing stock often 94 (36) 32 (33) 126 (35) Difficulty acquiring business loans 81 (30) 36 (38) 117 (33) Other 3 (1) 5 (5) 8 (2) Total 264 (100) 96 (100) 360 (100) p < .05, The findings presented in Table 5 9 show that most of the business owners that have not attained business education have difficulties in the two issues that directly affect business operation. 33% find it difficult to pay the rent of their premises and 36% find it difficult to re stock frequently, compared to 24% and 33%, respectively, of those
144 that have attained business education. On the other hand most of those that have attained education have difficulties in the issue that indirectly affects business. 38% of them find it difficult to acquire business loans compared to 30% of those that have not attain ed business education. Tables 5 10 also shows the relationship between attending business seminars and financial successfulness. Table 5 10. Attendi ng business seminars and achieving financial s uccess Business Seminars Attended N (%) Business Challenges None 1 to 2 More than 2 Total Dif ficulty paying rent 83 (31) 20 (35) 6 (17) 109 (30) Difficulty purchasing stock 96 (36) 21 (38) 9 (25) 126 (35) Difficulty acquiring loans 84 (31) 15 (27) 18 (50) 117 (33) Other 5 (2) 0 (0) 3 (8) 8 (2) Total 268 (100) 56 (100) 36 (100) 360 (100) p < .05, Table 5 10 shows that most of the business owners that have not attended any business seminar or attended 1 to 2 of them have difficulties in the issues that directly affect the operation of business when compa red to those that have attended more than two seminars. 31% of those that have not attended any seminar and 35% of those that have attended 1 to 2 have difficulty in paying the rent of their premises compared to 17% of those that have attended more than tw o. 36% of those that have not attended any seminar and 28% of those that have attended 1 to 2 have difficulty in purchasing stock compared to 25% that have attended more than two seminars. On the contrary, most of the business owners that have attended 1 t o 2 seminars have difficulties in the issue that indirectly affects business operation, acquiring loans. 50% of those them have difficulties in acquiring loans compared to 31% of those that have not attended any seminar and 27% that have attended 1 to 2 se minars. Since business owners that have attained business education and attended more than two seminars are in Nairobi (high
145 level of urban development), we can tentatively conclude that they are more financially successful in the issues that directly affe ct business operation than those in Mombasa (moderate urban development) and Kisumu (low urban development). Taking into consideration the direct effect of larger business sizes, attainment of education, and attainment of business education on capital inte nsity (level of poverty) and financial successfulness we can conclude that high level of urban development improves socioeconomic status. The next section examines how the occurrence of clientelism varies with regards to variation in socio economic status. 5.4 Socioeconomic Status and Variation of Political Party Clientelism In general, the socioeconomic factors examined above should have a moderating effect on the attractiveness of clientelism because they increase self reliance of business owners in the i nformal sector This attitude affects clientelistic calculations since politicians are typically concerned with reducing the uncertainty that clients will not vote for them after being given goods (Clapham 1993; Kitschelt and Wilkinson 2007, 10; Le Vine 19 80; Lewis 1996). For them, this can involve giving the client a good they the variation of socioeconomic status examined above, we should expec t the level of political party 2 clientelism to be lowest in Nairobi, moderate in Mombasa, and highest in Kisumu. The operationalization of political party clientelism borrows from the existing c or not 2 Unfortunately, this study does not compare the variat ion across specific parties. This is because the fieldwork was done a few months after the post election crisis in Kenya and in that environment asking respondents about specific parties was a sensitive issue. Pilot surveys showed that asking about specifi c parties had a high non response rate. When the names of all parties were eliminated from the surveys the response rate was very high.
146 is determined by finding out whether the party help that a citizen received was offered around election periods (Auyero 2001, 213; Brinkerhoff & Goldsmith 2002, 3; Hopkin 2006, 5; Keefer 2009; Kitschelt and Wilkinson 2007, 10; Sandbrook 1972, 108; Weingrod 1968, 380). This is widely used strategy for winning elections in clientelistic contexts (Grzymala Busse 2008, 643). Parties in these contexts, even ideological ones, have learned that in order be electorally successful they need to establish cli entelistic linkages with citizens at the grassroots level (Coppedge 2001, 176). The clientelism literature in political science (e.g. Van de Walle 2007, 3) often refers to these term electoral bargains between mass parties and voters (Hilgers 2011, 585). These types of bargains have been viewed as mostly occurring during competitive elections in nascent multi party systems (Muno 2010, 8). To compare the occurrence of clientelism in the three cities, business owners were asked whet her they have mostly received such assistance during election or non election periods. 3 Table 5 11. Urban d evelop ment and political party c lientelism Level of Urban Development N (%) Has mostly received party assistance during elections Low Kisumu Mod erate Mombasa High Nairobi Total No 50 (42) 102 (85) 109 (91 ) 261 (73) Yes 70 (58) 18 (15) 11 (9) 99 (27) Total 120 (100) 120 (100) 120 (100) 360 (100) p < .001, 3 The exact question asked was: Have you mostly received assistance from the party during election periods? Respondents recollect the exact number of times they have received assistance from the time they started their businesses. To avoid misinterpretation on their part, the interviewer clarifi ed that the question required them to consider the overall times they have received such assistance during the time they have been operating their business.
147 Table 5 11 above shows that most of the business owners that have receive d assistance during elections are in the city with the low level of urban development. The finding shown is based on party clientelism. Other measures of clientelism will be explored later. Figures 5 1, 5 2 and 5 3 show how these relationship varies with regards to the type of business ownership, level of poverty, and gender respectively. Figure 5 1 below shows the percentage of business owners that have mostly received assistance from political parties around election periods in the three cities. This i s the period when parties conduct campaigns and also undertake clientelistic transactions. Figure 5 1 Clientelism is more prevalent where urban development is l ow Figure 5 1 shows two categories of business owners: those who are in partnerships with others and those who are sole proprietors. Kisumu, the city with the lowest level of urban development, has the highest level of clientelism when compared
148 to Mombasa and Nairobi in both categories. The level of clientelism among sole proprietors is higher in Mombasa than in Nairobi. In Kisumu it is mostly the sole proprietors that are targeted for clientelistic transactions. When it comes to comparison based on lev el o f poverty, as shown in Figure 5 2 below in Kisumu it is mostly the very poor (low capita l intensity) that are targeted for clientelistic transactions. The level of clientelism is also high among the moderately poor and the least poor. In Mombasa, it is mostly the least poor (high capital intensity). Clientelistic practices among the moderatel y and least poor are much lower than in Kisumu. In Nairobi the level of clientelist transactions is low all three levels of poverty. This finding suggests that the price of clientelism is higher is Nairobi and Mombasa and cheaper in Kisumu. Figure 5 2 Clientelism is prevalent where the level of poverty is h igh
149 With regar ds to gender, shown in Figure 5 3 below, 4 in Kisumu it is mostly the female business owners that are targeted for clientelistic offers, whereas in Mombasa and Nairobi the differences be tween male and female involved in clientelis m are not significant (Figure 5 3 ). Figure 5 3 Women are more l ik ely to be targeted for c lientelism C hapter 5 has shown that urban development is highly correlated with clientelism. An important issue that n eeds examined further, using an appropriate method, is the causal effect of urban development and clientelism and not just the correlation. While the findings presented here are insightful, more can be learned by 4 In the overall sample size 56% of the respondents are women and 44% are men. In Nairobi: 64% wom en and 56% men. In Mombasa: 62% women and 58% men. In Kisumu: 76% women and 44% men.
150 understanding the relationship that exists between urban development and the clientelistic strategies that different politicians use. This is the objective of the Chapter 6 Figure 5 4 Variation in infrastructural d evelopment based on capital expenditures in th e three cities ( Source of Data: Local Authorities Transfer Fund. 2010. Annual Report and Review of Local Authority Financial Performance 2002 2009 Nairobi, Kenya. )
151 Figure 5 5 Variation in market d evelopment based on marke t fees (Source of Data: Local Authorities Transfer Fund. 2010. Annual Report and Review of Local Authority Financial Performance, 2002 2009. Nairobi, Kenya.) Figure 5 6 Variation in property d evelopment based on property rat es (Source of Data: Local Authorities Transfer Fund. 2010. Annual Report and Review of Local Authority Financial Performance, 2002 2009. Nairobi, Kenya.)
152 Figure 5 7 Variation in business d evelopment based on single business permit fees (Source of Data: Local Authorities Transfer Fund. 2010. Annual Report and Review of Local Authority Financial Performance, 2002 2009. Nairobi, Kenya.)
153 CHAPTER 6 TRADITIONAL VERSUS MODERN CLIENTELISM: CONCEPTUALIZATION AND OPERATIONALIZATION Cl with citizens vary in different contexts but the concept tends to be applied in a highly generalized form. This is because it has been used to accommodate a wide range of polit ical exchanges (Hopkin 2006, 2). The concept is useful for explaining the political interactions that take place between high and low level actors (Powell 1970, 423). This includes politicians, brokers, rural and urban voters, and other traditional leaders (Kaufman 1974, 302). Scholars tend to emphasize either the exchange and reciprocity aspects or the relational aspects of clientelism (Muno 2010, 3). Some scholars (e.g. Auyero 1999, 301) have called for a shift from the common approach that mechanisticall y views the poor as citizens who cast votes for a political candidate in exchange for favors and refocus on the relational links that are established for problem solving. Studies have shown that clientelism can take different forms in rural and urban areas (Smith 1960, 260). The form that clientelism takes is largely dependent on the context it exists in (Lemerchand and Legg 1972, 156). We can gain a better understanding of the different forms that it takes by identifying the forms that have been highlighte d in the vast literature on clientelism and then finding out how they vary in different cont exts. This is the goal of C hapter 6 Chapter 4 showed how, in general, its prevalence varies with rega rds to urban development. C hapter 6 goes further by identifyin g the different forms of clientelism, based on the insight offered in the literature on clientelism, and showing their variation in the three cities in Kenya. The first part of Chapter 6 analyses the variation in the occurrence of two types of clientelism : a traditional versus a modern mode. It also draws out variables that can
154 be used to examine their incidence. The second part elaborates on how these variables can be measured. The third part compares their incidence in the three cities on two levels. It first compares their prevalent use with the MP and Councilor lumped in one comparisons are done using two methods: descriptive and regression analyses. The fourth part co ncludes C hapter 6 6.1 Differentiating Modern and Traditional Clientelism 6.1.1 Scholastic Distinction The initial studies on clientelism were rooted in the sociological and anthropological studies that were done on traditional societies in the 1960s and 1 970s (Hilgers 2011, 568). Hence, our current understanding of clientelism is mostly based on ethnographic studies that have identified its characteristics at a microsociological level (Eisenstadt and Lemarchand 1981; Eisenstadt and Roniger 1984; Gellner an d Waterbury 1977; Piattoni 2001a). There have been two significant waves in the study of clientelism (Roniger 2004, 355). The first wave was in the late 1960s and early 1970s. This wave largely involved case studies done by anthropologists and political s cientists attempting to conceptualize clientelism (e.g. Graziano 1976; Lemarchand 1972a, Scott 1969). During this wave sociologists and anthropologists (e.g. Cornelius 1977; Weingrod 1977) originally framed the relationships between chiefs or landlords an d peasants as patron client relations (Brinkerhoff and Goldsmith 2002, 2; Hilgers 2011, 569). Their approach, which used the local or village level as the unit of analysis, has been regarded as the traditional approach (Eisenstadt and Roniger 1980, 54; Fat ton 1986, 61; Hilgers 2011, 585; Kaufman 1974, 290; Kitschelt 2000, 849; Muno 2010, 3; Paul 2008, 213; Roniger
155 2004, 355 356; Van de Walle 2007, 3; Scott 1972, 112). In this study, I refer to it as traditional. The second wave started in the late 1990s (Ro niger 2004, 355). This was at a time when the process of democratization was also taking off in Africa, marked by the rise of multi party competition (Van de Walle 2007). The view of clientelism in this wave reflects the rise in importance of elections an d a shift in the way clientelistic politicians used clientelism as a strategy for winning the votes of rural inhabitants that were the focus of the first wave. The integration of peasants in such societies in electoral political competition was accompanied by the use of clientelism as a political mobilization or persuasion strategy (Szeftel 2000, 430). This has been regarded as the modern approach (Hilgers 2011, 568; Hopkin 2006, 8; Kitschelt 2000, 849; Paul 2008, 213; Roniger 2004, 355 356; Schaffer 2007). I refer to it as the modern mode. In the first wave the concept was used to explain the nature of politics in villages but in the second wave it was broadened and used to explain national level politics (Brinkerhoff and Goldsmith 2002, 2; Hilgers 2011, 5 68). The attempt to broaden the use of clientelism to explain national level politics has led to a generalized definition that has downplayed some characteristics that still exist in the traditional mode (Hilgers 2011, 571). This has been driven by a desir e to compare the significance of behaviors observed in traditional clientelistic activities across historical periods and geographical as well as hierarchical space (Hilgers 2011, 568). What clientelism has meant for political scientists has diverged fro m what it has clientelism often refer to the divide between an anthropological clientelism describing
156 traditional relationships in the countryside and a political science one ref erring to the impersonal and short term electoral bargains between mass 2011, 585). Anthropologists have often emphasized unequal social relationships bound by kinship ties that exist in small communities in the countryside (Br inkerhoff and Goldsmith 2002, 2; Hilgers 2011, 568; Piattoni 2001a, 9; Weingrod 1968, 379) while sociologists and political scientists have emphasized the existence of these relations in bureaucratic structures (Eisenstadt and Roniger 1980, 42). Two ways of examining clientelistic practices within the social sciences therefore persist, studies that emphasize traditional patron client ties and studies that emphasize the distribution of resources in exchange for political support (Roniger 2004. 355). The lat ter is sometimes viewed as one based on anonymous interactions under political party machines (Kitschelt 2000, 849). A sharp contrast has been made between the two modes by some scholars (Roniger 2004, 355 356). But they are also viewed as a continuum from traditional to modern norms (Kitschelt 2000, 849). The two modes are rarely exclusive from each other (Hopkin 2006, 4; Lemarchand 1972, 77). Instead they can overlap and occur simultaneously as a nation develops (Scott 1972, 1145). But there can be instan ces where one is stronger than the other. 6.1.2 A Modernization Perspective of the Two Modes Modernization of political and economic systems has driven the examination of the two modes. Attempts to comprehend how clientelism evolves with regards to the mod ernization of state and society structures date back to the late 1960s and early 1970s when the emphasis was placed on peasant societies (e.g. Graziano 1973; Lemarchand and Legg 1972, 1972; Scott 1969, 1972). Social and economic changes engendered by moder nization were expected to weaken traditional clientelistic norms
157 but instead they persisted (Lemarchand 1972, 88). These norms were considered a predominant feature of non industrialized contexts, with their resilience constraining tra were also viewed as antagonistic to political development (Lemarchand 1972, 87 88), which is a key component of modernization (Huntington 1965, 393). Due to their persistence, they are viewed a sites modernization has transformed clientelism. According to Graziano (1973, 369), clientelism transforms fro m the traditional form to the transitional form. The difference between these two forms is that the latter form has clientelism incorporated as a strategy for organized political parties and is widely used for electoral competition while in the former it i s not commonly used for those reasons (Graziano 1973, 369). Lemarchand (1972, 72) frames it as a transformation from feudal to patrimonial forms. The traditional mode is the feudal one. It is characterized by a personalistic bond between a superior and sub ordinate person (Cohen 1966, 91 cited in Lemarchand 1972, 73). The modern mode i s the patrimonial one. It involves exchanging targeted goods for various political benefits, particularly votes (Lemarchand 1972, 72 73). Although modernization may have led to a decline of traditional dyadic, face to face ties, they are still an inherent part of several contemporary political systems (Lande 1983, 444
158 prominence since the late nineties because it has been widely used to distinguish the nature of electoral competition in new democracies and advanced democracies worldwide (e.g. Auyero, 2001; Brusco, Nazareno and Stokes, 2002; Estevez, Magaloni and Diaz Cayeros, 2002; Fox, 1994; Gay, 2001, 1998; Hag opian, 1996; Kitschelt et al., 1999; Levitsky, 2001, 1999; Mainwaring, 1999; Malesky, 2001; Robinson and Verdier 2002; Roniger 2004; Stokes, 1995; Szwarcberg, 2001; Van Loo and Taran, 2001; Wantchekon, 2002). This is because the provision of private, targe ted goods has been a common feature of younger democracies (Keefer 2007, 807). The modern form of clientelism is a strategy that is also commonly associated with organized parties, where favors are distributed by party members and agents (Hopkin 2006, 8). Hence, examinations of clientelistic relationships between parties and clients have focused on the exchange of votes for favors (e.g. Auyero 1999; Gunther 1996; Hoskin 1997; Kitschelt and Wilkinson 2007; Lindberg 2003, 2012; Lindberg & Morrison 2008; nell 1996; Piattoni 2001a; Robinson and Verdier 2002; Stokes 2007). The increasing attempt to distinguish clientelistic from democratic norms in the democratization era (e.g. Brinkerhoff and Goldsmith 2002, 4) has increased the focus of clientelistic trans actions as being based on exchanging votes for goods and services. This is different from the exchanges associated with the traditional mode. 6.1.3 Distinguishing Relations inadequately explores, the distribution of goods and services is a necessary, but not the 304). Clientelistic
159 transactions are usually sustained by complex bonds that can be either weak or strong (Brinkerhoff and Goldsmith 2002, 2). Contemporary political relationships broadly fiable as instrumentally rational and short lived, while others cross the lines among instrumental, value, emotional, and traditional he patron client dyad itself can be conceived in a large number of forms, all the way from the exclusive and monopolistic aspects of the traditional peasant landlord relationship to the more fluid and limited exchange process character istic of the political machine 289).The existence of strong long term face to face interactions in the traditional mode of clientelism differentiates it from the short term ones that characterize the modern mode (Hilgers 2011, 568). In the traditional mode, the establishment of a strong relationship takes precedence. Anthropology and political scien ce scholars have highlighted the different nature of relationships in the two modes. For anthropologists, patrons and clients first establish enduring interpersonal relationships then manipulate the relationship to continually achieve mutual benefits (Wein grod 1968, 379). For political scientists, politicians establish ephemeral relationships for the sake of exchanging favors for electoral support rod 1968, 380). In a traditional context, patron client dyads tend to last almost throughout the lifetime of both partners (Barnes 1968; Powell 1970, 413). Patrons form these traditional dyads because they provide long term security for their positions an d can be invoked when other factors they depend on are unreliable (Scott 1972, 102). Due to the affective
160 nature of the relationship between patrons and clients, they have been viewed as friends (Boissevian 1966, 32). Pitt societies where the traditional form of clientelism was observed (Eisenstadt and Roniger 1980, 54). On the contrary, this is not a necessary conditio n for the exchanges that take place in the modern mode where the recipient is only seen as a potential voter (Schaffer 2007). 6.1.4 Distinguishing Transactions or flow of re clientelistic interactions can lead to different exchanges (Auyero 1999, 323). Clientelistic res types of exchanges are associated with the modern mode and the sociational types wi th the traditional mode. This distinction characterizes a common divergence in long term interactions involving shows of personal concern and liking between the partie s involved; others label it an interest maximizing exchange of goods and services and apply it to incidents ranging from vote buying to pork This observation highlights the factors that distinguish the exchanges in the two m odes of clientelism. Rather than viewing patron client relations as mechanism for buying
161 Roniger 1980, 56 ). In the modern understanding of the concept, exchanges are conceived as reliant on the availability of a personal vote (e.g. Grzymala Busse 2008, 665). Citizens explicitly use their votes as a bargaining tool to get politicians to respond to their demand s (Kitschelt 2000, 851). Although votes are a primary good they are not the only basis of transactions (Gay 1990, 648 cited in Auyero 1999, 297). The dominant focus on votes as the key aspect of clientelistic linkages has masked traditional types of exchan ge. Traditional clientelistic norms are embedded in communitarian ethics of reciprocity (Van de Walle 2007, 3). In this case, community norms play an important role of anchoring private relationships (Kaufman 1974, 285). What binds patrons and clients is the strong interpersonal loyalty that strengthens their solidarity (Eisenstadt and Roniger 1980, 50). These bonds are established for the sake of increasing of obligati The patron primarily protects them and provides their needs because of the shared affective bo nds (Muno 2010, 3). Initial studies on the norm found that clientelistic exchanges were marked by patrons offering assistance and protection to clients who reciprocated by becoming loyal followers (Boissevian 1966, 18; Mason 1986, 489). Scholars (e.g. Lema rchand and Legg 1972; Scott 1969, 1972) emphasized that the against economic emergencies. For the patron, fulfilling these duties mattered because it provided a loyal following that secured his superior position (Camic et al. 2005, 167).
162 Patrons who employ the traditional mode do not seek a one time transaction but a long term obligation from clients (Scott 1972, 97). Hence, the obligation that the patron and the clie nt have between each other does not require immediate response, it can take a long time before the transaction is fully completed. As Boissevian (1966, 22) until the foll them to a debt of obligation that they may never be able to fully reciprocate (Scott 1972, 93). This implies that there is more than meets the eye in cases where exchange of votes fo for favors is seen from the inside in many different (and, sometimes, antagonistic) ways: manipulation versus caring, interested action (politics, calculative exchange) versus dis exchange of votes for goods and services is supported by a more deeper relationship between patrons and clients. There is also a specificity of time based on elections that mos tly applies to the transactions in the modern mode. In this mode the use of clientelism tends to increase as the intensity of electoral competition increases (Scott 1969; Van de Walle 2007). In the traditional mode the range of time when transactions take place is diffuse (Scott and Kerkvliet 1977; Blau 1964, 93). 6.1.5 The Role of Brokers Brokers play an important intermediary role in the modern mode due to the lack of interpersonal relationships between patrons and clients. They mediate between patrons an d clients that are either geographically or personally distant from each other by connecting complex networks (Eisenstadt and Roniger 1980, 43). They are able to establish mutually beneficial relationships between patrons and clients (Boissevian
163 1966, 24 2 for clients (Kettering 1988, 425). Their importance stems from the connections they have to a wider network of powerful and powerless people and their ability to facilitate negot iations and the trading of resources (Powell 1970, 412). In some cases brokers can be local patrons that are well connected to clients and so are recruited by politicians in the central government to establish links with clients (Kaufman 1974, 290). These patrons can also link political party elites that are detached from the local level politics to the masses (Lemerchand 1972, 86). 6.1.6 The Role of Group Membership There are almost no studies that have looked at how traditional and modern clientelism vary when it comes to distributing private goods to clients that are members of organized groups, yet apart from the distribution of goods to individuals clients, politicians can also distribute goods to groups (Hopkin 2006, 10). This study seeks to examine t his issue by including a variable that analyses the effect of clients that are members of groups versus those who are not. Although clientelism is mostly based on one to sec uring a political position when he has several of them. One way of increasing these relationships may be to distribute goods to groups (Robinson and Verdier 2002, 2). In groups, patron client linkages can exist in the form of a network of dyads formed by p eople who know each other (Hilgers 2011, 580; Lemarchand 1972, 75). The reasons why politicians may prefer to distribute private goods to citizens in groups over those b enefits from transactions. The literature on clientelism has shown that politicians are
164 respond to. They can strategically respond to the demands of citizens who belong to groups and those that they have a high level of contact with. This can especially be the case in urban areas where politicians do not usually trust urban residents to be loyal at the polling station (Lindberg 2010, 127). For clientelistic promises to be fulfilled close monitoring is necessary. In most cases politicians are unable to effectively monitor whether voters they have given goods vote as expected. This inability to monitor voters can make the distribution of goods a costly investment for them. Si nce groups are in a position to monitor each other successfully in ensuring all the members deliver their votes, they may reduce the cost of monitoring individual clients. Thus, for politicians it may be easier for people who have close interactions to mon politicians can credibly exchange with will be determined by the social network of individuals whose behavior they can observe relatively well...This network, a clientele, ...allows politician monitoring through groups is more likely to make it an attractive strategy for patrons who use the modern mode since they are more detached from clients, especially intimately. Table 6 1 summarizes the key differences in the two modes, with regards to the four aspects discussed above. 6.2 Operationalization and Key Empirical Expectations Based on the previous discussion of the two modes, this section introduces the variables that can be used to analyze how the two modes of clientelism vary with regards to urban development and level of politicians, as well as the empirical expectations. The survey questions that were used to construct these variables are listed in Appendix B. In order t o compare how the modes vary with regards to urban
165 development, this study examines how they play out in election and non election that clients offer patrons one sho uld separate electoral related help from non electoral help. There is a difference in the way transactions in the two modes play out during election periods. Clients can vote for patrons due to the strong relationship and sense of obligation in response to a good or service that had been delivered a long time ago (Hopkin 2006, 3), or due to a one time vote buying transaction that is devoid of a long term relationship and sense of obligation. The former interaction falls under the traditional mode and the la tter under the modern mode. Hence, we can be able to differentiate between politicians that use the traditional versus modern mode by finding out whether the goods delivered are related to campaign visits and voting patterns or not. In this case, visits m ade primarily during campaigns fall under the modern mode. Delivery of goods from patrons associated with high voting patterns also fall under this mode. The variables that can be used to conduct this analysis are those looking at the periods politicians m ostly visit citizens(election versus non election), the reason why politicians visit citizens (campaign versus non campaign), frequency of voting in elections, and the time when citizens mostly receive clientelistic goods (election versus non election). An other way of comparing the two modes is by looking at the effect of proximity. This is essential in the execution of traditional patron client transactions (Powell 1970, to face in teractions cemented by normative bonds between politicians and their clients occur
166 therefore, characterized by recurrent interactions between patrons and clients, allo wing the patron to closely monitor clients. Since clients must be closely monitored, patrons that employ the traditional mode have to invest a significant amount of time and energy in seeking and maintaining clients on t he ground (Roniger 1994, 212). These patrons also tend to supply durable goods to their followers so as to retain them for a long time (Hopkin 2006, 2); hence, the ties that bind them tend to be greater (Scott 1972, 100). So we should expect the provision of durable goods and frequent visits to be more associated with direct interactions and long term relationships during non election periods. In this regard, we should also expect a higher level politician to be associated with the modern mode and a lower level politician with the traditional mode. Key variables that can be used to analyze the effect of proximity are: frequency of visits from low and high level politicians, the durable or non durable nature of the goods distributed, and the level of direct, face to face, contact between politi cians and clients. The availability of vast resources can compensate for the lack of proximity needed in the traditional mode since patrons who have a large resource base tend to increase their instrumental or non affective ties (Scott 1972, 99). In this r egard, it should be expected that a politician who has access to more resources is more likely to use the modern mode. Because the analysis in this C hapter 7 is extensive, including many variab les, it is necessary to introduce the main variables here An extensive list of the other variables, with their descriptions, is provided in the appendix. It also elaborates further on the main variables that I briefly presented below.
167 6.2.1 Dependent Variables : MPs Versus Councilors Business Assistance The two types of politicians that this analysis compares are members of political representatives in Kenya. 1 Councilors play an important role of dealing with the city council on var ious matters that affect business owners. Since they represent the business owners in council meetings with the Mayor, who heads the local government, they can act as advocates for those who are unable to meet all the pre requisites for issuing business st alls, such as paying the full rent and license fee on time. Findings from focus group discussions and interviews show that that these are major problems in the informal sector. 2 For clients in these markets promising loyalty to Councilors has long term be nefit of getting protection from the harassments that business owners in the informal sector normally get from local government inspectors when they are late in paying rent or have violated any of the business permit requirements. These harassments usually include possible eviction from the stalls. Successful establishment of linkages with MPs also serves an important goal of accessing financial resources that make it possible to cover the costs of acquiring and maintaining a business stall. The main resou rce that business owners in Kenya seek from MPs comes from a development fund that was established by the state for the purpose of stimulating development in constituencies. This fund is known as the 1 of politicians will emerge, namely: governors, senators, and county representatives. 2 A number of business owners pointed out that they are usually harassed by the city council officers for rent and license; given five choices of problems they are facin g to pick from, most (33% ) of the business owners mentioned high rent charges from the city council as the major problem affecting their ability to conduct business. From the interviews one business owner in Kisumu informed me that they have been suggesting to the city council to allow them to pay a joint license fee for th e market to ease their inability to pay the fee on time.
168 Constituency Developme nt Fund (CDF) 3 The Fund is disbur sed to every constituency through the MP using guidelines that were established under a parliamentary Act. The covered under this Act shall be submitted by the member of Parliament for that businesses and so they can easily serve as mechanisms for politicians seeking long term clientelistic loyalties. The Act requires the government to gra nt the program a minimum of 2.5% of the national revenue for each financial year 4 The stability of the fund serves as a major incentive for business owners to seek linkages with politicians. Such linkages can increase the likelihood of accessing funds freque ntly, given that the CDF managing committee in each constituency is usually formed by the MP. It is comprised of the MP, the councilor and nine other members from the constituency. 5 Since Councilors are also part of this committee it also creates an incent ive for business owners to form linkages with them because they can facilitate access to get the funds. The literature on clientelism also shows the necessity of comparing their roles since it distinguishes the traditional and modern mode with regards to the local level and 3 The Constituency Development Fund (CDF) was created in Kenya in 2003 through an act of parliament ich being and to relieve members of parliament from the heavy demands of fund raising for projects which ought to be financed through the Consolidated Fund. The legal provision of the establish ment and operation of the Act suggests that the fund is essentially a model for decentralization of development planning and implementation. 4 Currently, there are calls by M.P.s to increase the CDF amount; several M.P.s have argued that the devolved fund has contributed to growth at the grassroots and lobbied to have the kitty tripled. 5 which shall be constituted and convened by the elected Member of Parli ament within the first thirty days of a new Parliament and shall have a maximum of fifteen members, comprising of 1) the elected member of Parliament ; 2) two councilors in the constituency; 3) one district officer in the constituency; 4) two persons r epresenting religious organizations in the constituency; 5) two men representatives from the constituency; 6) two women representatives from the constituency; 7) one person representing the youth from the constituency; 8) one person nominated from among
169 national level politics (Brinkerhoff and Goldsmith 2002, 2; Powell 1970, 423). As far as legislative issues are concerned, in Kenyan politics the Councilor is usually closer and the MP farther. It can, therefore, be expected that Counci lors are associated with the traditional mode and the MP the modern mode. To examine this distinction, I asked business assistance. This question is then followed by questions tha t try to find out the types of business assistance that they have received, as well as the frequency. 6.2.2 Key Independent Variables : The analysis conducted in this dissertation consists of several independent variables that cannot be detailed in full h ere. The following section highlights the main ones that the study is based on. 184.108.40.206 Participation in e lections The literature on clientelism has also highlighted the fact that transactions in the modern mode are mostly enunciated during elections while those in the traditional mode are not (Auyero 1999, 297; Graziano 1973, 369; Hilgers 2011, 577 578; Hopkin 2006, 3; Moran 2001; Muno 2010, 8; Wantchekon 2002; Weingrod 1968, 380; Szeftel 2000, 427). This implies that we can tell whether the MP and Council or employ either of the two modes by finding out whether the delivery of goods is associated with business measures: whether a business owner has voted in at least one electio n and how many elections he or she has voted in. associated with these two measures, we can cautiously conclude that the politician uses the modern mode. If not, then the politician uses the traditional mode.
170 6.2 2.2 Campaign versus non campaign (business)visits In order to investigate this issue I used a survey question that asked business owners the main reason why politicians have visited them. The list of choices they were given in the questionnaire included several reasons why they mainly visit business owners. These reasons were provided by business owners during focus group discussions and in depth interviews. They fall under two categories: campaign and non campaign visits. Campaign visits mostly take place during the election periods while non campaign visits take place during non election periods. The non campaign visits are visits related to delivering various forms of business assistance by politicians, such as helping business owners in gettin g stalls, providing them with funds to improve their business, and delivering government funds allocated to small businesses. I have labeled the non campaign visits as business visits. 6.2 .2 .3 Frequency of contacts (level of access to politician ) The natur e of bonds or ties between citizens and politicians vary from weak to strong. Hence, it is possible to tell whether politicians use the traditional or modern mode by comparing the relationships that they have with business owners. In order to examine these relationships business owners were asked about their frequency of contact with the politicians from the time they started operating their businesses The choices provided were: 0 times, 1 to 2 times, 3 to 4 times, and more than 4 times, which I have label ed as none, low, moderate and high respectively. Traditional and modern clientelism also vary in terms of face to face contact. As shown in the discussion on the bonds of affection between patrons and clients, the traditional mode is mostly associated with face to face contact while this is rare for the modern mode. This study compares three types of contact that were mentioned by respondents during focus
171 group discussions and interviews then included in the survey questionnaire: letters, phone calls, and o ffice visits. I then ranked them as low, moderate, and high direct contact respectively. In this case, the traditional mode is associated with the most direct contact office visits. Respondents were asked which of these forms of contact they have mainly used to gain access to the politicians. 220.127.116.11 Two way link with politician The review of the literature on the two modes has shown that the traditional one is based on patrons and clients having a reciprocal relationship. This not only implies exchanging goods but also casual interactions. In this mode the patrons tend to visit their clients more often than in the modern mode. Hence, one way that the two modes can be compared is by finding out whether a reciprocal relationship exists or not, in the sense that the client has visited the politician and vice versa. In this study I call this relationship the two way link. A politician who is associated with this l ink uses the traditional mode. 6.2.2 .5 Lobby group membership The effect of group membership in s ecuring private clientelistic goods has not been examined in studies on clientelism. Usually, scholars make a distinction between private goods issued to individuals and club goods issued to groups. But during the focus group discussions and interviews tha t were conducted for this study some business owners reported that they have received private benefits through their groups. This provoked further interest in examining the role that group membership plays in securing private goods. Since most of these gro ups are used to advocate for a wide range of goods and services for markets I consider them lobby groups. I used two measures: whether a business owner is a member of at least one lobby group and
172 whether a business owner is a member of more than one lobby group. Respondents the local government for better service to markets. Those who respond affirmatively are then asked other follow up questions that seek to find out th e number of lobby groups the frequency of their contribution. The goal behind asking about contributions is to find out the relationship between active participation i n the group and both modes. This variable is most likely associated with the modern mode because it is not based on long term personal relationships. In the traditional mode politicians maintain close relationships with citizens, hence they are less likely to interact with groups. 6 .2.2 .6 Demographic and business aspects In addition to the main variables above the analysis that follows also include some demographic variables and other important ones that are related to their businesses, namely: gender, leve l of education, the duration that the business owner has had the business, and the size of the business. Table C 1 in Appendix C provides the descriptive statistics of these variables, as well as the variables mentioned above. Using these variables, C hapte r 7 conducts an analysis of the variations in modern and traditional clientelistic practices.
173 Table 6 1. Comparison of features of modern and traditional c lientelism Features Modern Clientelism Traditional Clientelism Transactions Take place durin g election periods One time Most of the time the transacting parties (patron and client) do not know each other personally Mostly based on votes as the main good Take place during both election and non election periods Continuous Most of the time the t ransacting parties (patron and client) know each other personally Other goods and services apart from votes are exchanged Relations Less intimate and affective Short term Weak interpersonal loyalty Less reciprocation Less accountability More intimate a nd affective Long term Strong interpersonal loyalty More reciprocity More accountability Brokers Patrons mostly use brokers to establish connections with clients Patrons rarely use them to establish connections clients Groups Patrons are more likely to use groups so as to increase their ability to monitor them and Patrons are less likely to use groups as a way of monitoring clients since their interactions with clients are built at the individual level.
174 CHAPTER 7 EXAMINATI ON OF TRADITIONAL VERSUS MODERN CLIENTELISM C hapter 7 first compares the traditional and the modern mode based on two levels: their occurrence in the three cities, how the MPs and Councilors use the two modes across the three cities. Regression analyses ar e then done to find out how the voting behaviors of citizens and their relationships with politicians are related to getting assistance from the two politicians. The effect of these variables is compared with that of the demographic ones to find out how wh ich ones have a strong effect on the probability of getting assistance from MPs and Councilors. The regression analyses are also used to compare the effect of campaign and non campaign visits on the probability of getting assistance from the politicians. 7 .1 Data This study was conducted in nine markets in across the three cities compared in this study. Three markets were selected from each city. The data was gathered using four methods: focus group discussions, semi structured interviews, surveys, and key informant interviews from three people who have first hand knowledge of the relationship between politicians and the informal sector. Three research assistants (RAs) were recruited to help with the data collection. The first step in gathering the data was selecting markets on the basis of geographical location. The goal was to ensure that the markets were in political wards represented by different politicians. We then mapped all the markets and recruited focus group participants. In each city we recruited two participants from each market. All the focus group discussions were tape recorded, along with note taking. In total 3 focus group discussions were done.
175 In the next stage, I analyzed the information from the focus groups and used them to develop quest ions for the semi structured interviews. Semi structured interviews were more appropriate, as opposed to in depth interviews, since I was getting assistance from three RAs in different locations and it was important that the information from the interviews be comparable. Moreover, although semi structured interviews are different from unstructured interviews they still have much of the freewheeling quality of unstructured interviewing and require the same skills (Bernard 2006, 212). The participants for bot h the interviews and surveys were recruited using the systematic random sampling technique. In each city we conducted 20 interviews, leading to 60 interviews total. After the interviews were completed all the interviews were transcribed. In the transcribe d interviews I looked for common themes and common variables to create a survey questionnaire. Based on the information gathered on the citizen politician linkages, I developed a survey meant to scrutinize the linkages further. with the assistance of the R As, we were able to conduct 360 surveys (120 in each city) from nine markets. In the last stage I decided to interview three key informants so as to verify the findings from the politicians stand point, although in an indirect way. These key informants we re: 1) A personal assistant for the one of the current Deputy Prime lost in his bid for a parliamentary seat at the nomination stage and later worked as a strategist fo political parties (Orange Democratic Party). Appendix A provides a more detailed
176 discussion on the data collection. Clarification of the survey questions used for variables included in this study and coding of the responses is provided in Appendix B. 7.2 Methods Two types of regression analysis methods for categorical dependent variables have been used to examine the data: binary regression models and ordinal regression models. The depen dent variables business assistance have been measured as binary variables that differentiates those measured using two ordi nal variables indicating the number of times a business owner has received such help and the quality of goods they have received. The model that has been used for the binary dependent variables is the binary logit. The ordered logit has been used for the t 7.2.1 Binary Regression Models I generate regression models for binary outcomes to find out how the independent variables (Table C 1, Appendix C) affect the probability of a business owner gett is nonlinear, the magnitude of the change in the outcome probability that is associated with a given change in one of the independent variables depends on the level of all the indepe presents the challenge of interpreting them. This requires a simple way of summarizing the effect of change in the binary outcomes for the dependent variable for a change in the in dependent variables. For the logit models I fit different models with the two binary dependent variables based on the following equation:
177 1 x 1 k x k As an independent variable, x increases by one unit, the dependent variable, P(y=1), increases (or decreases) by units. To illustrate the findings, I use figures showing nonlinear models predicting the dependent variables y. Each of these models has an independent categorical variable x The lines shown in these models plot y as x changes. For the models, the probability of the business owners getting MPs or Councilors help is: Pr(y = 1|x) = F (x ) Where F is the logistic cumulative density function for the logit model. The probability P(y = 1) takes a S curve shape. indicates the direction (up or down) of the curve as x increases. can fully Freese 2006, 157). Several methods can be used to present them in a more functions of those probab a more understandable way I use three different approaches. The first method I use is showing the effect that each of the independent variables have on the predicted probability of an outcome using the following formula: Pr(y = 1 | x x k ) The equation above represents the probability to be computed when all other variables are set to their means, except the variable of interest x k. This variable is can be set at a
178 specified value, such as its minimum. Hence, to find out the effect that it has when it varies from its minimum to maximum the following calculation is done: Pr(y = 1 | x, max x k ) Pr(y = 1 | x min x k ) This method allows one of the independent variables to vary from its minimum valu e to its maximum value, while other others are held at their means. The second effe change for a change of some quantity in the independent variable is: x k = Pr(y = I | x, x k Pr(y = 1 |x, x k ) x k from x k to x k x k holding all other variables and Freese 2006, 171). This method is able to show the effect of a finite change in the independent variables. In the analysis of this change, I examine the change in predicted probability at the independent variables change from their minimum (x k ) to ma The third method I use to interpret the findings examines the odds ratio. This method transforms the coefficients ( ) of the independent variables in the logit models to show the factor chang e in the odds of business owners getting assistance from the MPs and the Councilors. The method takes the antilogs of both sides of a logistic regression equation, such that: Log (P(y = 1)/(1 which yields P(y = 1) / (1 P(y = 1) = e = e (e ) x
179 The exponential regression on the right side indicates that a constant value multiplied by another raised to the power of x increases the odds by e in a multiplicative manner. This means that for a unit change in an independent variable, the odds of an outcome are expected to change by a factor of e 7.2.2 Ordinal Regression Models In the analysis I also use two ordinal dependent variables for business assistance from the councilors. I analyze them using an ordinal logistic regression m odel. I use the following proportional odds model: 1 xi,l the set of factors or predictors of the model. The equation is an ordinal logistic model for k number of predictors with p 1 levels of response for each ordinal variable. In order to convert the out of the ORM into an interpretable form I examine the discrete change for the variables. This is the predicted probability for a change in a variable from the initial value (xs) to the end value xe, that is: Pr (y = m | x, xk = xs) where Pr (y = m | x, xk) is the probability that y = m given x, taking into consideration a from xs interpreted using odds ratios. With the ordered logit model defined as: The effect of a change in a dependent variable xk of 1 takes the form:
180 outcome being less than or equal to m is chan ges by the factor exp ( increase in xk, the odds of a lower outcome compared with a higher outcome are changed by the factor exp ( 2006, 217). 7.2.3 Plotting Probabilities To graphically present the findings from the binary regression models I plot the probabilities of an outcome over the change from minimum to maximum of an independent variable. Sinc e there are several independent variables I examine the effects of the main ones. For example, consider the effects of age and business owners voting in elections (BV) on getting assistance from a politician (GAP). The effect of both independent variables, is then plotted by holding all other variables at their means and allowing AGE and BV to vary. To do this, where x0 contains the mean of all other variables, BV and AGE will be allowed to vary. Then, is the predicted pr obability of getting assistance from a politician for business owners of a given age who have not voted and are average on all other characteristics. The predicted probability for those who have voted, Pr(GAP = 1 | AGE, BV = 1 similarly compu ted and the predicted probabilities plotted. The interpretations are then made. For instance, the plotted results of the example above can be interpreted as: for each additional n years of age, the probability of getting assistance from a politician increa ses by p, holding all other variables at their means.
181 To present the findings from the ordinal regression models I plot the cumulative nce, to examine the frequency at which a business owner gets assistance from a politician (never, rarely or often), the cumulative probability of one of the outcomes being less than or equal to m is Pr(y = m | x) ) = I n the example, the cumulative probabilities would be the probability of never getting ly getting uncover the three trends by showing the effect of an independent variable on the three categories. 7.2.4 Diagnostic Tests Before interpreting the results f rom the logit and ordinal logistic regression analyses, this study ensures they are valid by conducting the necessary diagnostic tests. For the binary regression models it test for multicollinearity, which occurs when there are correlations among the indep endent variables. In this case if an independent has been determined by the linear combination of these variables. The presence of multicollinearity makes the models inappropriate since the estimates of the coefficients cannot be uniquely obtained, hence they are unreliable. It also makes the model unidentifiable, in the sense that it is not possible to differentiate the effect of the collinear independent variables on the dependent variable.
182 To test for multicollinearity I examine the interrelationships that exist between the variables in each binary regression model using two commonly examined measures: variance inflation factors (VIF) and tolerance The VIF indicates the extent of inflation of the standard error that is caused by correlation between the independent variables. tolerance indicates the level of collinearity an analysis can tolerate. If all the variables are uncorrelated with each other both their VIF and tolerance values are 1; if they are correlated tolerance drops to 0 and VIF values increase. The rule of thumb is that multicollinearity is problematic when any of the VIF is greater than 10, or conservatively 5 (Heiberger and Holland 2004, 243). The ordi nal regression model is based on an assumption, known as the proportional odds assumption, that needs to be tested to ensure that the models are appropriate to use. The assumption made is that the effects of any of the independent variables on the differen t categories of the dependent variable are consistent or proportional. Essentially, we should be able to observe the same odds ratio at each threshold since it constrains them through the assumption. I perform a test that checks whether when different inde the coefficients for each threshold are equal. The null hypothesis to be tested is that 1 If the proportional same. If not, the null hypothesis is rejected and so it the proportional regression assumption. In this case the models are not appropriate and another approach should be considered. To cond uct the test I use an approximate likelihood ratio test. Under this
183 test, when the test statistic is significant the proportional regression assumption is violated. The binary logit models have been examined for goodness of fit using the Hosmer and Lemesho of fit test. This test examines how a model fits by finding out how the predicted frequency and observed frequency match. If they match closely, it is concluded that a model has a good fit. This test is computed as the Pearson chi square and t he p value used to determine whether the model fits the data or not. The null hypothesis is that the model fits the data. The alternate hypothesis is that the model does not fit the data. If the p value is statistically significant, there is a problem with the fit of the model since we have to accept the alternate hypothesis. If the test is not statistically significant, we fail to reject the null hypothesis and conclude that the model may fit the data sufficiently well. For the ordered logit models I exami ne the log likelihood p value to find out if the model as a whole is statistically significant. 7.3 Analysis and Findings The first part of the analysis is based on cross tabulations comparing the utilization of the two modes of clientelism in the three c ities. I start by presenting descriptive findings that provide evidence on the way the two modes of clientelism vary across the three cities. Comparisons on the variation have been done in two levels. First, with the MP and Councilors lumped together in on e category as politicians. At the second level, I compare how modes vary by each type of politician. The second part of the analysis is based on maximum likelihood estimations on the probability of the MP Councilor using the modern and traditional modes, r espectively.
184 7.3.1 Comparing the Occurrence of Clientelism : Urban Development and Modern Clientelism Figure 7 1 below presents the frequencies of the utilization of modern clientelism in the three cities Figure 7 1. Occurrence of modern c lientelism ( %) F igure 7 1 above shows the percentages of business owners that were visited mainly for campaign purposes during election periods. The modern mode is mostly used in the city with a moderate level of urban development (Mombasa). It is also used more of ten in least developed city (Kisumu) than the most developed city (Nairobi). 48% of the business owners in Mombasa have mainly 1 received business assistance when politicians visit for campaigning purposes, compared to 30% in Kisumu and 13% in Nairobi. 1 they started operating their businesses visits for the purpose of campaigning are the most common.
185 7.3 .2 Urban Development and Traditional Clientelism Figure 7 2 below present s the frequencies of utilization of the traditional mode. It shows the variation of non campaign visits that are made during non election periods. The traditional mode is prevalently used in Kisumu. 39% of the business owners in Kisumu compared to 17% in Mombasa and 9% in Nairobi reported that they have mainly received business assistance when politicians visit them for non campaign reasons, which is usually during non election periods Figure 7 2. Occurrence of traditional c lientelism (%) The traditional mode is rarely used in the most populated city (Nairobi), moderately used in moderately populated city (Mombasa), and widely used in the least populated city (Kisumu). There is a 9% prevalence level in Nairobi, 17% in Mombasa, and 39% in Kisumu. The low level of both modes in Nairobi reflects the findings presented in Chapter 5 showing that, in general, clientelism is used less often in Nairobi because of its costliness in highly u rbanized areas (See Figures 5 1, 5 2 and 5 3 ). In
186 C hapter 5 the study also demonstrated, using socioeconomic factors, that in such a context the value and magnitude of goods that politicians have to distribute to make clientelis tic transactions appealing to voters is too high. Similar observations have been made in other studies on clientelism (e.g. Fatton 1986, 69; Kitschelt 2000, 857). Another factor that seems to reduce clientelistic transactions in Nairobi is that most of the business owners directly seek public services from the local government rather than going through politicians. The findings from a question trying to find out the percentage of business owns that seek such services shows that in Nairobi, 52% of the busine ss owners directly access public services from the local government compared to 20% in Mombasa and 28% in Kisumu. There are two conditions that possibly increase the prevalent use of the modern mode in Mombasa. First, there are more poor business owners th an in Nairobi. This reduces the b idding price for votes. Table 5 2 in Chapter 5 (using capital intensity as a proxy for poverty) shows that 50% of the business owners in Mombasa are poor compared to 32% in Nairobi. This implies that the cost of modern clie ntelism in Mombasa is likely to be lower, hence making it more appealing for politicians to use. The lower level in the occurrence of modern clientelism in Kisumu can be explained by the high incidence of the tra ditional mode shown in Figure 7 2. It shows that this mode is used more often in Kisumu than in Mombasa and Nairobi. The low level of urban development in Kisumu provides fertile grounds for using the traditional mode. Hence, it is more appealing for politicians to use it rather than the modern mode It has been shown in other studies that t raditional patron client relations are a predominant feature of non industrialized contexts (Fatton 1986, 62). Modernization in Nairobi and Mombasa
187 can also contribute to the prevalence of the traditional mode in Kisumu as citizens seek alternative ways of catching up with the socio economic advancements in the those more urbanized cities. Other studies (Powell 1970, 425) have shown that as modernization takes place in more urbanized contexts, socio economic factor s create tensions that intensity the importance of traditional clientelistic norms in less urbanized contexts. A second plausible reason why the modern mode is prevalently used in Mombasa is that the high population density makes it challenging to success fully use the traditional mode. Mombasa has a population of about 0.9 million residents and Kisumu has about 0.4 million residents (Kenya Open Data, 2012). Hence, it is difficult for politicians to employ the traditional mode, which works effectively in a low population context because it needs frequent monitoring and interacting with clients (Warner 1997, 535). The same population factor explains why the traditional mode is also employed less frequently in Nairobi than in Mombasa. Nairobi has a much higher population that the other two cities, which makes the utilization of the traditional mode a difficult task to execute successfully. It has about 3.4 million residents (Kenya Open Data 2012). 7.3.3 MPs versus Councilor Usage of Modern Clientelism (%) To p rovide more insight on the utilization of the two modes, the next stage of analysis compares their variation with regards to the level of politicians. The higher level politicians are Members of Parliament (MPs) and the lower level ones are the Councilors. This differentiation is based on the level of the legislature they serve in and the political issues they are responsible for. Whereas MPs represent citizens in a higher lawmaking institution and mostly legislate on national issues, Councilors serve as r epresentatives in a lower institution and legislate on local issues (Infotrack East Africa,
188 2008). The findings show that, in general, MPs employ the modern mode while the Councilors employ the traditi onal mode. As shown in Figure 7 3 below their use of t his strategy also varies in the three cities. MP Councilor Figure 7 3. lientelism (%) The variation shown in Figure 7 3 raises an important question: why is the prevalence of the modern mode higher in Mombasa yet Kisumu has more poor business owners who make it cheaper to buy votes? The higher population density in Mombasa could be forcing MPs to make a trade off that lead to the higher usage of the mo de. Its population makes it difficult to build traditional clientelistic relationships. Thus to make up for this difficult situation, MPs engage in more vote buying than in Kisumu. In Kisumu they can rely on building more traditional clientelistic relation ships because of the low population. Their prevalent traditional bonds with clients in Kisumu reflects ially based on use of the traditional mode they could be employing brokers to bridge the gap. Studies
189 done before have shown that agents are used as intermediaries w ho recruit clients for agents, establish client networks for them, as well as monitor them, is a common norm in traditional clientelism (Boissevian 1966,24; Eisenstadt and Roniger 1980, 43; Kaufman 1974, 297; Kettering 1988, 425; Powell 1970, 412). The fi ndings above do not show the actual variation in the usage of the mode been visite d for campaign reasons. This comparison has been done by calculating the overall percentages using only the 116 business owners that have been visited by the two politicia ns for campaign reasons which is mainly around election period Table 7 1 shows the results from the computation. Table 7 1 lientelism City N (%) Campaign Visits Kisumu Mombasa Nairobi Total MP 30 (88) 48 (86) 16 (62) 94 (81) Councilor 4 (12) 8 (14) 10 (38) 22 (19) Total 34 (100) 56 (100) 26 (100) 116 (100) The findings presented in Table 7 1 show that in Nairobi 62% of the business owners have been visited by the MP for this purposes compared to the 38% visited by the Councilor. Similarly, in Mombasa and Kisumu 86% and 81% have bee n visited by the MP for the same purpose, compared to 14% and 12%, respectively, by the t. The findings comparing their usage of the traditional mode shows that show that they generally prefer to use the traditional where population density is low Mombasa and Kisumu. The findings shown above also suggest that where socio economic
190 developme nt is higher Councilors find it more profitable to use the modern mode. Its usage could be benefiting them by minimizing time costs and maximizes returns. 7.3.4 MP versus Councilor Usage of Traditional Clientelism (%) Having compared the way MPs and Counc ilors use the modern mode of clientelism, this section compares their usage of the traditional mode. MP Councilor Figu re 7 4 MP versus councilor usage of tradit ional c lientelism (%) Figure 7 4 shows that the variations among the MPs and Councilors have a similar pattern. It is difficult to diff erentiate their usage from F igure 7 4 above because these findings are based on the total sample size (N=360). A focus on only the business owners that they have used the traditional mode on provides a clearer idea of their preferences. Table 7 2 shows the results derived from computing their usage of the traditional mode on the 57 business owners that they used the it on. T able 7 2. lientelism City N (%) Business Visits (Non Campaign) Kisumu Mombasa Nairobi Total MP 9 (28) 5 (26) 1 (17) 15 (26) Councilor 23 (72) 14 (74) 5 (83) 42 (74) Total 32 (100) 19 (100) 6 (1 00) 57 (100)
191 Table 7 2 shows that Councilors use the mode more than MPs. In the three cities combined, 74% of the business owners have mainly received business help from the Councilor during non campaign visits compared to 26% who have had the same from the MPs. In Nairobi, the prevalent usage of the mode is 83% for Councilors versus 17% for MPs versus, 74% versus 26% in Mombasa, and 72% versus 28% in Kisumu. The differences highlight the advantage of proximity to clients that Councilors have since they operate at the local level. This is in line with other studies (Hopkin 2006, 8; Powell 1970, 413; Schmidt 1974, 425) that have shown that it is one of the major factors that facilitates the successful execution of the traditional mode. Contrary to Councilo rs, MPs are not able to frequently interact with clients. During the in depth interviews and focus group discussions, business owners pointed out that they often see Councilors casually walking around in the markets for a wide range of uneventful daily ac tivities, such as shopping, visiting friends, and simply dining at some of the local restaurants. MP visits are very different and typically marked with elaborate pomp and ceremony and one hardly ever see the MP interact with individuals in the casual mann er that the Councilor does. The friendliness demonstrated studies (Auyero 1999, 305; Boissevian 1966, 32; Weingrod 1968, 379), even though, as Scott (1972, 92) has pointe d out based on observations in Southeast Asia that they are kind of instrumental friendship in which the patron cares for the well being and protection of his client, wh ile the client promises to support and aid his patron on
192 more in getting assistance from MP versus the Councilor using a different method. 7.3 .5 Logistic Regression Anal ysis The variation in campaign and non campaign visits shown above enlighten us on how MPs and Councilors employ the two clientelistic strategies but that evidence is not adequate enough. They do not address the role of strong face to face bonds and partic ipation in elections. The findings can be strengthened by examining how the delivery of clientelistic goods is associated with voting behavior and the nature of personal relationships. What are the chances that business owners who vote during elections, on ce or often, will get clientelistic goods from the MP and not the Councilor? What are the chances that business owners who have stronger face to face ties with the MP and Councilor will get business assistance from them? Based on the descriptive findings p resented above, I address one major question: What is the probability that those business owners who have received business assistance from MPs and Councilors are those that they have visited for campaign rather than non campaign reasons? Taking account of the fact that the data is categorical in nature I use the maximum likelihood estimation techniques discussed in the methods section to fit two types of models: binary and ordinal regression models. 7.3 .6 Thr business assistance (mpba) is the dependent variable: The independent variables are: Has never missed voting (nmv), Politician campaign visits (pcv), MP business visits (mpbv), Direct acc ess to the MP (dmp), High level of access to the MP (hmp), Direct access to the Councilor (dac), Two way link with
193 (fpav), Membership in lobby group (mlg), Contrib utes to lobbying often (clo), Contributes to political parties (cpp), Male, Many employees (emp), High capital intensity (hci), Older business owner (obo), Actively seeks government services (asgs), High level of education (hle), Has gotten business educat ion (bizd). The descriptive statistics for these variables are presented in Table C 1, Appendix C. The survey questi ons they were derived from, responses, and how they were coded are provided in Appendix A. Model 1: Pr(mpba = 1) = F ( 0 + nmv nmv + pcv pcv + twlc twlc + fpav fpav + cpp cpp + male male + emp emp + hci hci + obo obo) Model 2: Pr(mpba = 1) = F ( 0 + pcv pcv + mpbv mpbv + hmp hmp + hac hac + clg clo + cpp cpp + male male + emp emp + obo obo + asgs asgs + hle hle) M odel 3: Pr(mpba = 1) = F ( 0 + nmv nmv + pcv pcv + mpbv mpbv + dmp dmp + dac dac + mlg mlg + cpp cpp + male male + hci hci + obo obo + asgs asgs + bizd bizd) The results are presented in T able 7 3 below. The findings for variables that have a statistica ness assistance shown in Table 7 3 above are presented in three related models from several successive tests. 2 The goal was to test a wide range of variables while avoiding problems of collinearity. The multicolli nearity diagnostics for the three models, using variance inflation factors (VIF), show that there is no collinearity in the models. The VIF for all the variables is below 1.4 (see Table C 2, in Appendix C for the complete results). The mean VIF for Model 1 is 2 The effect of other possible differences across the three cities that could be related to the dependent variable and may have not been accounted for was tested using dummy variables. The three models with these dummies are shown in Table C 4 in Appendix C. The dummie s are not statistically significant and the significant effect of the other variables do not change. They do not influence the dependent variable. This means that the variables that have been included in the models account for the differences in getting MP
194 1.09, for Model 2 is 1.14, and Model three is 1.10. The tolerance results also show lack of collinearity. They range between 0.76 and 0.97. The Hosmer L) goodness of fit tests shows that the three models fit the data sufficiently well. T he H L chi squared value and p value for model 1 is 10.13 and 0.2558, for model 2 is 6.06 and 0.6407, and for model 3 is 9.40 and 0.3096. This means that we cannot reject the null hypothesis that the models fit the data. Table 7 3. Logit analyses of gett usines s a ssistance Independent Variables Model 1 Coefficient (SE) Model 2 Coefficient (SE) Model 3 Coefficient (SE) Has never missed voting 1.578 (0.999) 1.846 (1.044)* Politician campaign visits 2.145 (0.863)** 1.524 (0.688)** 1.432 (0.713) ** MP business visits 0.851 (0.797) 0.038 (0.811) Direct access to the MP 2.437 (0.843)*** High level of access to the MP 1.144 (0.354)*** Direct access to the Councilor 1.081 (0.680) Two way link with the councilor 0.266 (0.975) High acces s to the Councilor 0.142 (0.339) 0.625 (0.263)** Membership in lobby group 0.701 (0.781) Contributes to lobbying often 0.687 (0.515) Contributes to political parties 0.743 (1.201) 0.873 (1.217) 0.465 (0.687 ) Male 0.115 (0.579) 0.062 (0.600) 0.098 (0.589) Many employees (large business) 0.439 (0.209)** 0.356 (0.218) High capital intensity (least poor) 0.732 (0.371)** 0.937 (0.398)** Older business owner 0.557 (0.459) 0.905 (0.473)* 0.860 (0.458)* Ac tively seeks govt. services 0.533 (0.697) 0.408 (0.637) High level of education 0.947 (0.197) Has gotten business education 0.791 (0.683) Constant 10.167 (4.108)** 2.118 (1.378)* 10.548 (4.255)** N=360 N=360 N=360 Significant at the p <0. 10, ** p <0.05, *** p <0.01 Since th e models presented in Tables 7 3 are for categorical outcomes that are nonlinear, the coefficients cannot be directly interpreted. However, based on a cursory
195 check of the statistically significant variables, it is not iceable that they support those shown in the previous analysis comparing the usage of the modern mode. Frequent participation in elections and being visited for campaign purposes show up as statistically significant. Campaign visits are consistently signif icant in all the three models. For better interpretation of these findings I use predicted probabilities and odds ratios to reveal the most important characteristics of these relationships. Although there are other methods that can be used, predicted prob analyze the effect of the main variables on the odds that business ow ners got the MPs help. Table 7 4 shows the factor change in the odds of getting MPs help for each of the statistically independent variables in the logit analyses. Table 7 4. ssistance First, the relational aspects associated with the existence of clientelism, level of in the sense that interpersona l sense. Establishing direct access to the MP increases Independent Variables Factor Change Never misses voting in elections 6 .33 Mainly gets campaign visits 4.59 High level of direct access to the MP 8.44 Has a high level of access to the MP 3.14 High level of capital intensity (least poor) 2.56 1.87 Older business owner 2.47 High n umber of employees (large business) 1.55
196 variables constant. To reiterate, what I consider as high level of access in this study is when a business own er says that he or she has mostly contacted the MP through office visits, compared to sending and receiving letters or phone calls. This means that there is a face to face relationship. It is also important for these clients to have strong bonds with the M help are increased by a factor of 3.14, holding all other variables constant. What this means is that a client seeking assistance needs to have interacted more than once with th e MP. This could contradict the descriptive findings by suggesting that the MP uses the traditional mode but further analysis below shows that this is not the case. Lemerchand (1972, 77) has observed from his studies on clientelism that situations where tr aditional clientelism and modern clientelism occur simultaneously are not uncommon (Lemarchand 1972, 77). Hilgers (2011, 578) points out that in these circumstances a patron client relationship can have the characteristics of long term exchanges and face t o face contact with the patron primarily distributing goods to the client during election periods (Hilgers 2011, 578). 7.3 .7 The Effect of Campaign Visits elections and getting camp aign visits show that the MP is most likely to use the modern help are increased by a factor of 4.59, holding all o ther variables constant. Table 7 5 below presents predi cted probabilities showing the effects of campaign visits and level The nonlinear effect of getting a campaign visit is clearly evident. The difference in probability for those who have a hi gh level of access to the MP changes incrementally from one level to the next
197 one, especially for those that have been visited for campaigning purposes. The difference from those who have not had access to the MP to those with low access is .14, from low a ccess to moderate access is .26, and from moderate to high access is .27. Overall, it changes from .07 to .74 from no access to high level of access. Table 7 5. evel of access to MP and campaign visits by p oliticians Politician has visited to campaign No Access to MP Low Access Moderate Access High Access No 0.02 0.05 0.15 0.37 Yes 0.07 0.21 0.47 0.74 Figure 7 5 shows the difference between having direct access to the MP and not getti ng campaign visits versus getting campaign visits. Figure 7 5. aries with regards to age and getting campaign v isits Figure 7 5 shows the nonlinearity of the outcomes with the eff ect of having direct access to the MP increasing with campaign visits and age. The positive effect is shown
198 by the increasing probabilities as the linkage status changes from having no access to the MP to establishing access. For a young business owner tha t has mainly been visited for campaigning, it shows that when the level of access to the MP is low, the probability is .13. For a business that has direct access to the MP, the predicted probability is .64. Although the predicted probability also changes p ositively for a business owner that has not been visited for campaign reasons, it is much lower. For this business owner, when there is no direct access to the MP the probability is .04 and when there is direct access the predicted probability is .30. This finding corroborates the descriptive analyses. Since patrons who employ this mode complete their transactions during elections, as other studies have sown (Hilgers 2011, 577 578; Hopkin 2006, 3; Muno 2010, 8; Wantchekon 2002), a positive relationship betw participation in elections should provide more convincing evidence. 7.3 .8 The Effect of Participation in Elections The results show that participation in elections has a stronger relationship with getting MPs assista nce than campaign visits. Campaign visits are likely to change the probability of getting the MPs assistance by a factor of 4.59 while voting is likely to change it by a factor of 6.33, holding all other variables constant. In other words, those citizens who vote often are likely to be the ones who have received assistance from the MP. This strengthens the claim that MPs use the modern mode by suggesting that for them clientelistic transactions are completed at the polls or once they can confirm that the b usiness owner has voted. As Stokes (2005, 2007) has shown in her research, this is a precautionary measure that politicians who use this mode take to ensure clients fulfill their part of the transaction. The finding does not necessarily negate MPs use of t he trad itional mo de shown in Figure 7 4 Other research (e.g. Hilgers 2011, 578) show
199 that both of them can occur simultaneously. The limitation of this study is that there is no data that can be used to show whether this was at election time but we can im ply losely associated to elections. 3 To clarify, this evidence is not enough to conclude whether the clients actually vote for the MP or not. Usually, it is difficult to know how clientelistic transactions play ou t at the polls. The goal of this study is to show the strategy that is most likely associated with the MP rather than finding out whether they ac tually vote for the MP. Figure 7 6 plots the predicted probabilities. Figure 7 6 Probability of getting M ssistanc e when direct access to the MP varies with regards to age and frequency of participation in elections (v oting) Figure 7 6 considers the effect of having direct access to the MP and the level of participation in elections. The effects of these variables are plotted by holding all other 3 Due to the delicate and tense post election crisis environment I was unable to ask this question directly.
200 variables at their means and allowing these ones to vary. It shows a business owner is that when this business owner has n o direct access to the MP, the probability is .10. When there is direct access to the MP, the predicted probability increases to .56. The predicted probability also increases for a business owner that missed 1 to 3 elections but at a much lower extent. For this business owner, when there is no direct access to the MP the probability is .02 and when there is direct access the predicted probability increases to .17. For a business owner that has missed more than 3 elections or all elections there is a zero pr obability of getting help from the MP, even if t hey have direct access. Figure 7 7 below provides more insight by showing the level of participation in elections where the likelihood of a client gettin Figure 7 7 Probabi ssista nce when the level of voting in e lections varies with regards to capital intensity and direct a ccess to the MP
201 Figure 7 7 takes into consideration the effects of voting frequently in elections and direct access for a less poor business owner. It demonstrates the non linear relationships once again with the probability of getting assistance from the MP increasing with level of participation in elections. When a business owner that has direct access to the MP has not taken part in any election the probability of getting help is zero and when/she has voted frequently, without missing any election, the probability increases to .50. The increase begins at the point where a business owner has voted frequently and only missed 1 to 3 ele ctions (at about level 2.5). At this point the probability is .02. From this point onwards, to those who have not missed any elections The finding shown in Figure 7 7 above indicate s that MPs ensure that there is some level of trust before distributing clientelistic goods and services. It suggests that they first ensure they have a face to face relationship with a client but delay the delivery of goods until the client has participat ed in elections more than once. This centrality of participation in elections is a major feature of timing in the modern mode. This is in line with other studies that have shown this is the case because elections are when the completion of transactions tak es place (Grzymala Busse 2008, 638 639; Hopkin 2006, 3; Wantchekon 2002), which is contrary to the diffuseness of delivery time in the traditional mode as shown in the research that examines this mode (e.g. Scott and Kerkvliet 1977). The fact that other va riables that cannot be directly or uniquely linked to clientelism have a weaker effect on getting assistance highlights the importance of modern clientelism. These variables, however, inform us about the demographical
202 attributes and business characteristic s of those that MPs target. It is the young business variables constant. To examine the effect of poverty on clientelism, I created three categories of poverty based on estimated capital investment levels: least poor, moderately poor, and most poor. Based on this categorization, it is the least poor rather than the most poor that against the argument made in the literature that politicians mostly target the poor with clientelistic offers. This could be because wealthier business owners have the resources for seeking the MPs, who are mostly based in Nairobi. The MPs could also benefit more from targeting the less poor, even if the costs are higher. Such as get ting financial support later. offices are in Nairobi means that they have to travel to the city to meet the MP and this requires resources (see map wi th the three cities in Figure 4 1, Chapter 4). It could be an indication that those with the resources are more likely to interact with the MP than most poor business owners and subsequently get recruited for clientelistic transactions. This finding shows that there are some trade offs bein g made by MPs that need to be researched on further. 7.3 .9 The following binary regression model has been fitted to examine the effect of independent variables on the probability of a business owner ge tting business assistance from the Councilor (cba):
203 Pr(cba = 1) = F ( 0 + pcv pcv + vlo vlo + hac hac + hmp hmp + clo clo + male male + obo obo + asgs asgs + gpa gpa + hle hle + emp emp) The results are presented in the table below: Table 7 6. ssistanc e Independent Variables Logit Coefficient (SE) 0.056 (0.547) Has voted at least once (vlo) 0.865 (0.777) High level of access to Councilor (hac) 0.548 (0.163)*** High level of access to MP (hmp) 0.674 (0.447) Contr ibutes to lobbying often (cl) 0.579 (0.322)* Male 1.184 (0.354) Older business owner (obo) 0.077 (0.203) Actively seeks govt. services (asgs) 0.208 (0.371) Has given party assistance (gpa) 1.624 (1.131) Has a high level of education (hle) 0.031 (0. 110) Many employees (large business emp) 0.030 (0.148) Constant 3.150 (1.058)*** N=360 Significant at the p <0.10, ** p <0.05, *** p <0.01 The model above uses a binary dependent variable based on whether or not the councilor has given a business owner any form of business assistance. 4 The multicollinearity diagnostics for the models show that there is no collinearity in the models. The VIF for all the variables is below 1.3, with the mean VIF for being 1.11. The tolerance results range between .8 1 and .98 (see Table C 3, in Appendix C for the 4 The effect of other possible differences across the three cities that could be related to the d ependent variable and may have not been accounted for was tested using dummy variables in two different models. One model includes Nairobi and Mombasa. The other includes Mombasa and Kisumu. The two models with t hese dummies are shown in Table C 5 and Tabl e C 6 in Appendix C. The dummies for the cities are not statistically significant and the significant effect of the other variables do not change. They do not influence the dependent variable. This means that the variables that have been included in the mo del
204 c omplete results on the tests). The Hosmer L) goodness of fit tests shows that the model fits the data sufficiently well. The H L chi squared value and p value for the model 1 is 10.96 and 0.204 2. This means that we cannot reject the null hypothesis that the models fit the data. Two ordinal regression models are also fitted to ex amine the effect that the independent variables have on the probability of a business ow ner getting business assistanc e from the Councilor frequently (Model 1) and the probability of getting a higher quality of business assistance (Model 2): 5 Model 1: x = 0 + pcv pcv + nmv nmv + cbv cbv + twlc twlc + clo clo + male male + obo obo + emp emp + chs chs + cpp cpp + obl obl Model 2: x = 0 + pcv pcv + nmv nmv + cbv cbv + twlc twlc + clo clo + male male + obo obo + emp emp + chs chs + obl obl Th e results are presented in the T able 7 7 below. Model 1 uses an ordinal dependent v ariable based on how frequently a business owner has received this assistance. There are three levels considered in this variable: None (No business assistance received), Low (business assistance received 1 to 2 times), and High (business assistance receiv ed more than 2 times). Model 2 also uses an ordinal variable based on the quality of the goods that the councilor has given. During the in depth interviews conducted with business owners they mostly mentioned three types of goods: small amounts of money th at they irregularly receive from politicians, state funds for small businesses, and business stalls. Through further probing on the value of these 5 The effect of other possible differences across the three cities that could be related to the dependent variable were also tested using dummy variables in two different models. One model includes Nairobi and Mombasa. The other includes Mombasa and Kisumu. The two models with th ese dummies are shown in Table C 7 and Table C 8 in Appendix C. The dummies for the cities are not statistically significant and t he significant effect of the other variables do not change. They do not influence the dependent variable. This means that the variables that have been included in the models above account for the differences in ss the three cities with regards to quality and frequency.
205 goods I was able to establish that the business stalls have the highest value based on the cost of getting on e as well as the monthly rent payments, followed by the small business loans they can get from the government, and then the small amounts of money they irregularly receive from politicians. Thus the small amounts of money correspond to the low quality of a ssistance measure above, while small business loans and getting a business stall correspond to the high quality of assistance. These goods were then incorporated in a survey question designed to find out the business owners that are likely to get them. Ta ble 7 7. Ordinal analyses of g etting ssistance Independent Variables Model 1 Coefficient (SE) Model 2 Coefficient (SE) 0.049 (0.582) 0.191 (0.617) Has never missed voting (nmv) 0.337 (0.225) 0.136 (0.220) Councilor business visits (cbv) 1.934 (0.596)*** 1.801 (0.611)*** Two way link with the councilor (twlc) 0.456 (0.838) 0.309 (0.812) Contributes to lobbying often (clo) 0.594 (0.315)* 0.798 (0.327)** Male 0.036 (0.346) 0.075 (0.382) Older bu siness owner (obo) 0.270 (0.218) 0.311 (0.231) Many employees (large business emp) 0.006 (0.148) 0.004 (0.161) Has completed high school (chs) 0.636 (0.352)* 0.957(0.390)** Contributes to political parties (cpp) 1.249 (0.885) Has owned the busines s longer (obl) 0.336 (0.168)** 0.263 (0.176) N=360 N=360 Significant at the p <0.10, ** p <0.05, *** p <0.01 The diagnostic test for the two models has been done using the likelihood ratio test. A statistically significant test would show that the pr oportional odds assumption (POS) has been violated. The chi square value for Model 1 is 10.91 (Prob > chi2 = 0.1426). The chi square value for Model 2 is 10.03 (Prob > chi2 = 0.268). Both results are not statistically significant, hence the POS has not bee n violated and the models are valid and appropriate to use. The goodness of fit, using the log likelihood p value show
206 that both models are statistically significant. The log likelihood chi squared for model 1, with 11 degrees of freedom, is 27.77 and the p value is 0.0035, and for model 2 is 25.66 and 0.0042. Based on a cursory check of the statistically significant variables, it is noticeable that the findings shown in the models above support the previous findings on traditional mode. These findings shows that Councilors use the traditional mode more often than the MP. Contrary to the findings show up as related to MPs business ass istance, non campaign (business) visits show variable. Campaign visits and voting in elections are not statistically significant in all the three models (Tables 7 6 and 7 7 ). Table 7 8 presents the factor changes of the statistically significant independent The values shown in Table 7 8 are the factor changes in the odds for a unit change in the ind ependent variables. The ordinal regression models are based on two different ordinal variables. One that looks at how frequently business owners have received business assistance from the councilor and the other one ranks the quality of goods they have rec eived from councilors. First, the findings show above confirm that Councilors use the traditional clientelism approach since business owners that contact them often are the ones that have received business assistance. Having a high level of access to the C ouncilor increases the probability of getting assistance by a factor change of 1.73. The findings show that none of the campaign related and participation in elections variables
207 Table 7 6 and Table 7 7). Instead, it is the non campaign related variables (business visits) that assistance at least once, as well as getting it frequently and at a h igh quality. The study quality assistance but these variables had no statistically significant relationship with any of the independent variables testing traditional or mod ern clientelism. This indicated time transactions that tend to be common in the modern clientelistic mode. Table 7 8. Factors that a ffect ssistance Independent Variable Factor Change Logit M odel Mainly gets business visits 3.20 High level of access to the councilor 1.73 Contributes to lobbying several times 1.78 Ordinal Regression Model (Frequency of Assistance) Mainly gets business visits 6.91 Contributes to lobbying several times 1 .81 Has completed high school 1.89 Has owned business for a longer time 1.40 Ordinal Regression Model (Quality of Assistance) Mainly gets business visits 6.06 Contributes to lobbying several times 2.22 Has completed high school 2.60 The relationsh all the t hree models. As shown in Table 7 8 above, getting business visits increases the
208 probability of getting help from the councilor at least once by a factor change of 3.20, holding all other va riables at their means. Figure 7 8 below plots the predicted consideration the effects of access to the councilor and whether or not politicians have mainl y visited the business owner to offer business help. 7.3.10 The Effect of Getting Business Visits Figure 7 8 below shows an almost linear positive outcome with the effect of politicians visiting to offer business help increasing along with increasing acces s to the councilor. Figure 7 8. ccess to the councilor varies with regards to getting business v isits In Figure 7 8, t here is a positive effect shown by the increasingly positive p robabilities as one moves from low to high level of access. The predicted probability of
209 getting assistance from the Councilor is .21 for a business owner with low level of access to the councilor but when the level of access is at its highest the predicte d probability increases to .58. This indicates that Councilors uses the traditional clientelistic mode. They assist those they have strong relationships with, since having a high level of access means that one has had sever personal contacts with the Counc ilor. The predicted probability also increases for one who has been visited mainly for the purposes of being offered business help but at a much lower level. For this business owner, when the level of access to the councilor is low the probability is .08 a nd when it is high the probability increases to .30. The findings further show that Councilors employ the traditional mode when it comes to the relationship between non campaign visits and getting help frequently and a high quality. The goods with the high est quality that business owners have received are business stalls. The factor change is 6.91 for those who the councilor has visited frequently for purposes of giving business assistance, and the factor change for getting a high quality of assistance is 6 .06, holding all other variables constant. These findings highlight a key characteristic of the traditional clientelistic strategies that confirm long time has been s hown by other studies to be a common aspect of the traditional mode (Boissevian 1966, 22; Hilgers 2011, 568). Councilors have access to business stalls because they are directly connected to the source: local governments. In Kenya, local governments own ma rkets with stalls leased to business owners. The Mayor, who frequent interaction with the Mayor provides them with direct access to stall distribution.
210 These stalls ar group discussions in Kisumu, business owners mentioned that the Councilor often assigns newly opened local government business stalls to their close friends in the market. The com parative changes in predicted probability for each outcome category of the dependent variable highlights the importance of the non campaign visits. They have a stronger effect than the other variables that we would expect to als o h ave a strong effect. Tabl e 7 9 and Table 7 10 compare the predicted probability for these variables at each outcome category. Table 7 9. Discrete change in the probability of g etting requently for the ordered logit m odel Variable Change Level Avg. Change None Lo w High Mainly gets business visits 0 to 1 0.28 0.42 0.10 0.32 Contributes to lobbying frequently Min to Max 0.16 0.23 0.00 0.24 Has completed high school 0 to 1 0.09 0.11 0.03 0.14 Has owned the business for long time Min to Max 0.18 0.27 0.01 0. 26 For a business owner that mainly gets business visits from the councilor, the predicted probability of getting a high frequency of assi stance increases by .32 (Table 7 9 ), and by .28 (Table 7 10 below) for high quality assistance holding all other var iables constant at their means. We would expect that being in the business for a long time would increase the likelihood of getting assistance frequently, but this is not the case. Instead the effect is negative. An extended duration of ownership decreases the predicted probability of getting party assis tance frequently by .26 (Table 7 9 ), holding all other variables constant at their means My interpretation is that these are business
211 owners who have become successful and are less dependent on clientelistic offers. Similarly, lobbying (active group membership) also has a lower effect on both frequent and high qua lit y assistance, of .24 (Table 7 9 ) and .26 (Table 7 10 ) respectively. Table 7 10 Discrete change in the probability of getting high quality of co for the ordered logit m odel Variable Change Level Avg. Change None Low High Mainly gets business visits 0 to 1 0.27 0.40 0.13 0.28 Has completed high school 0 to 1 0.16 0.19 0.04 0.23 Contributed to lobbying frequently Min to Max 0.25 0.37 0.11 0.26 Although the findings confirm that citizens with lower levels of education are targeted by clientelistic politicians, as it has been shown in other studies (Kitschelt 2000, 857; Vicente and Wantchekon 2009, 303), the effect of education in getting help frequently is weaker than that for the others. For a business owner that has completed high school, the predicted probability of getting assistance from a the Councilor frequently decreases by .14, holding all other variables constant at thei r means. It decreases by .23 for getting quality goods. Meaning that educated business owners are less likely to get these forms of assistance. The strong effect of the business visits, ode as a strategy for extending business assistance with minimal consideration of the se demographic factors. Figure 7 9 plots the predicted probabilities for the low and high frequency Figure 7 9 considers a busi ness owner that has owned the business for less than one year and contributes to lobbying activities business visits.
212 Figure 7 business visits v ary with regards to lobbying and duration of o wnershi p probability of getting assistance frequently also increases. When t here is no visit the probability of getting assistance at a high frequency is .07 and when there is a business visit the probability is .33, a difference of .26. The probability of low frequency is an unusual characteristic of ordinal regression models (Lo ng 1997, 132 133). The effect is initially positive and then negative. The probability begins at .25, increases to about .46, and then decreases to .44. This occurs because as the probability of business visits increases more cases move from the low freque ncy category to the high frequency category. Figure 7 10 illustrates the effect that business visits has on the quality of goods.
2 13 business assistance. According to the l iterature, traditional patrons normally distribute durable, high quality goods so as to bind clients to long term debt of obligations that they may never be able to fully reciprocate (Scott 1972, 93). This has been found to ensure that they secure a long t erm following (Scott 1972, 100). Figure 7 10 Probability of getting a high q uali ducation Figure 7 10 takes into consideration a business owner that h as not completed high school and frequently contributes to lobbying activities. It shows that with a high of goods from the councilor is also high. It begins at 20 and ends at .59, a difference of .39. This indicates that Councilors prefer to distribute goods that establish the type of long term patron clients bonds associated with traditional clientelism. A strategy that has been found to be common in other studi es (e.g. Scott 1972, 93). The goal can be to
214 ensure that there are successful outcomes from the strategy. When clientelistic relationships are embedded in long term relationships they can facilitate accountability on the part of the patron and simultaneous ly increase the incentive for the client to comply to the contract than the one shot attempt to buy votes (Hilgers 2011, 578). This means that the durable good can also serve as a strategy for securing votes, only that the patron has to invest more in mai ntaining the relationship through frequently interacting with the client and supplying the goods requested by the client. In contrast, the modern mode associated with the MP did not have any significant relationships with the high quality goods or frequen t distribution of goods. This emphasizes the one time nature of the transactions that take place in the modern mode. 7.4 Overview of Findings Most studies of clientelism examine its effects at the national level and hardly compare how politicians at diffe rent legislative levels use it. An important contribution of this dissertation is the attempt to systematically operationalize the two modes and compare their occurrence and usage. To my knowledge, this is the first one that does so. This dissertation make s a case for within country comparisons that take into account the different positions that politicians occupy and the differences in the contexts within o ne country. The findings in C hapter 7 demonstrate that clientelistic behaviors can intricately vary a t the macrosociological and microsociological levels simultaneously. The findings show that different patterns of clientelism exist at the city and politician level. These findings uphold the model proposed in C hapter 1 (Figure 1 1) by demonstrating the im portant role of variations in urban development. It shows what difference it makes on clientelism when the urban context is taken into account. The
215 analysis demonstrates the implications of the socioe conomic variations reported in C hapter 5. An important feature of the model that these findings also highlight is the importance of examining clientelism at the local government level. As pointed out in Chapter 5, the initiatives of both local government administrations and citizens transform the socioeconomic factors that affect clientelistic strategies. The results in terms of these strategies in different ways, depending on their legislative level and access to resourc es. Politicians at a higher level use the modern mode because they have access to more resources, while those at the lower level use the traditional mode because they have less resources. 6 The latter secure loyalty from clients by building strong interpers onal relationships, while the former purchase it when they need it during elections. They reinforce the importance of distinguishing between the traditional and modern norms of clientelism. In sum, the analysis has shown that citizen politician clientel istic linkages can be understood better if the effects of socioeconomic development are also accounted for at the microeconomic level rather than the commonly used macroeconomic factors. More research is certainly needed to strengthen the findings but they provide a fairly substantial basis for making some theoretical and methodological suggestions for studies on clientelism, including th is one. This is the goal of Chapter 8. 6 A limitation of this study is that it has not accounted for all the factors that could be at play and may also lead to the same findings. This factors will be included and examined as this study evolves.
216 CHAPTER 8 CONCLUSION This dissertation examines clientelism as a common form of i nteraction by citizens and politicians in African countries. It seeks to provide insight on the effect that urban development has on such interactions. The findings show that the effectiveness of clientelistic appeals vary depending on the level of urban d evelopment, the type of clientelistic mode, and the type of politicians. This study attempts to improve our understanding of the intricate dynamics of clientelism in urban areas. Existing studies on clientelism have not examined the effect of urban develop ment on clientelistic strategies employed at the political party level, politicians lumped in one category, and specific types of politicians. C hapter 8 will provide an overview of the main questions and how they have been addressed. It will do so by rec apping the proposed theories, the derived hypotheses, the analysis that was conducted, and the conclusions reached. It then shows the gaps that the study fills in the existing literature on clientelism and makes some recommendations for future studies. Thi s dissertation started by raising a number of questions which any study that tries to examine how urban development affects clientelism needs to consider. The overarching questions are: how does urban development affect the clientelistic strategies employe d by politicians? How does it affect their preferences for one strategy over the other? How do these preferences vary with regards to the type of politician? What are the socioeconomic factors behind the variations in citizen politician relationships in di fferent urban contexts? What is the role of local governments in the relationship between urban development and clientelism? It has addressed these questions using different types of data and methods.
217 8.1 Overview of Data and Methods Utilized The data used in this study was collected through in depth interviews, focus groups, surveys, archival sources, and from a variety of secondary sources. In the three cities compared, 60 in depth interviews were conducted, 3 focus groups, 360 survey questionnaires were administered, and archival data on urban development aspects collected from the local authorities that govern the cities. Three key informant interviews were also conducted. Appendix A provides a detailed description of the procedure employed in gathering the data from interviews, focus groups, and surveys. The data on urban development is based on initiatives that have been undertaken by local authority administrations and citizens. It uses expenditures that the local governments have extended towards inf rastructural projects, the revenues collected from real estate taxes, and revenues collected from small businesses to compare levels of urban development in the three cities. The limitation of this data set is that the it was only available for the years 2 002 to 2009. This is because prior to 2002 the local authorities did not keep complete records. Starting from 2002 the central go vernment, which they were under 1 required all of them to submit financial reports using the same format if they were to get an y financial assistance. This standardization has made the data comparable. It stops in 2009 because from 2010 they were no longer required to submit the reports. The eight year data is, however, sufficient for demonstrating the variation in urban developme nt and rank the cities. To bolster this evidence, the study traces the history of development in the three cities using secondary data from a variety of sources. 1 Kenya passed a new constitution in 2010 that ended the centralized system of government, which placed local authorities under the central government.
218 Different methods have been employed to systematically examine the data. The initial analysi s is mostly based on the use of cross tabulations to analyze the relationship between a set of different categorical variables. They have been primarily used to examine the relationship between urban development and socioeconomic status. This is an importa nt part of the main argument that drives the study. I argue that urban development affects clientelism by improving citizens socioeconomic status. A number of tests have been done to examine the strength of these relationships. First, chi square tests have been done to see whether there is a statistically significant relationship between the categorical variables. The limitation of using chi square is that it is does not say much about the strength of the relationship. To avoid the danger of misinterpreting statistical significance as indicating a strong relationship, three other measure the strength between nominal variables and Gamma to measure those between ordinal variab have few cases in the cells. This test is usually necessary when there are fewer than five cases in each one of the cells. Like the chi squared, it tests whether there is a relationship bet ween the variables. For the next level of analysis maximum likelihood estimation (MLE) techniques have been used since they have appropriate analytical methods for examining survey data. These techniques are appropriate since MLE estimators work well with data collected from small sample sizes like the one used for this study. There is no set rule on the sample size needed, but a minimum of 100 is recommended (Long & Freese 2006, 77). The sample size for this study (N = 360) fulfills this requirement. The
219 advantage of MLE techniques is that they provide consistent ways of estimating parameters, such that they can be used to estimate a variety of models when the type of categorical data used is different. Outcomes from the binary dependent variables are exa mined using binary logit regressions models and ordinal logit regression models used for ordinal variables. The estimates produced by these models are not directly interpretable. MLE techniques are useful for interpretation since they provide a variety of methods for describing the relationship between independent and dependent variables. MLE has various options for interpretation that make it possible to elegantly and concisely summarize complex models. One can be able to try the different interpretation o ptions they provide for examining the effect of variables. The results from the binary regression models have been interpreted using predicted probabilities, discrete change, and odds ratios. The simplicity of predicted probabilities make them suitable for interpretation. This method examines the effect that each independent variable has on the probabilities of observing an outcome. Interpretation is direct because it is possible to examine how change in each independent variable affects the predicted proba bility of an outcome. The disadvantage of using predicted probabilities is that extreme values affect the range by increasing the change in probability from minimum to maximum. Taking this into consideration, the study also examines the discrete change for the outcomes. This method of interpretation has been used for both the binary and ordinary regression interpretation.
220 For binary variables, the analysis examines their effect as they change from 0 to 1, and for those that are not binary, the change in predicted probability has been examined as the independent variables change from their minimum to maximum values. The study examines how the predicted probabilities change when t here is a discrete change in each variable. The benefit of using discrete change is that one can observe the change in predicted probability calculated as an independent variable changes from the starting point to the end point, for instance a change from 0 to 1 in a binary variable. It is also possible to examine the effects of an individual variable of interest by allowing it to vary and holding the others at their means. This effects can be calculated for any amount of change in an individual variable. F or instance, the change for year of education for men and women can be calculated separately. The examination of odds ratios is useful for finding out the chances of an outcome occurring. This study examines the factor changes for both the binary regressio n models and the ordinal regression models. Diagnostic tests for all models have been done so as to ensure that the models are valid and appropriate to use. The most common problem in binary regression models is collinearity of variables. Hence, they have been tested for multicollinearity. This is to prevent the danger of using biased and unreliable coefficient estimates that are outcomes of correlation between the independent variables. The tests conducted check the variance inflation factors (VIF) and t he tolerance values for the variables in each model. These two measures check for inflation of standard errors and biasness of coefficients. The standard rule of thumb is that multicollinearity is present if the VIF values are above 10 and is the tolerance values are close to 0. The most desirable
221 outcome is for both to have a value of 1. The models used in this study meet this standard since the mean VIF for all the BRM is around 1.1 and the tolerance values range at around 0.75 to 0.98. As for the ordinal regression models, a test checking whether they have violate the proportional odds assumption (or parallel regression assumption) has been conducted. Ordinal regression models are based on the assumption that the coefficients at each of the different cut points in the ordered categories are equivalent. In other words, the effect of any independent variable should be the same across the different ordered categories in an ordinal dependent variable. This means that the odds of an outcome are the same at each category. If this is not the case the proportional odds assumption is violated, hence it is inappropriate to use the models. This study finds that this assumption has not been violated in any of the ordinal regression models. 8.2 Summary of Chapters Chap ter 1 points out the gap that exists in the clientelism literature with regards to urban development. It shows that the effects of urban development associated with local government initiatives are not yet understood. This is mostly because existing studi es tend to focus on the effect of national level variables, such as gross domestic product, government expenditures, and national level education. By doing this they have overlooked the effect of socio economic factors within local government jurisdictions This study argues that local governments play a central role in transforming clientelism through the development initiatives they undertake, such as road construction, construction of markets, and development of electricity grids. Citizens within these a lso jurisdictions play a similar role by engaging in property and business development activities. In existing studies on clientelism, these roles are masked by the
222 focus on aggregated development initiatives that central governments undertake within an en tire country. In order to address this gap this study suggests the need to combine theories from clientelism and urban economics. Theories from urban economics provide tools for examining the socioeconomic outcomes of urban development initiatives. This is important because these are the factors that sustain or undermine clientelism. Since clientelism is usually sustained by low socioeconomic status, the argument made is that by understanding how urban development initiatives within local government jurisdi ctions transform these factors we are able to gain insight on the relationship that exists between urban development and clientelism. T he study further points out in C hapter 1 that there is a need to examine the effect of urban development on two types of clientelistic strategies: traditional and modern. Most studies that examine the relationship between development and clientelism have conflated them or lumped them. One reason why this occurs is because the two strategies have not been systematically dist inguished. It suggests the need to distinguish the two types of clientelism by using existing studies from political science, anthropology, and sociology. Chapter 2 justifies the use of a case study method and the case selection. The case that was selecte d for this study is Kenya, which is a typical case with regards to the relationship this study examines. A case study is appropriate for this study because of five main reasons: case studies are useful for building theories on issues that have not been exa mined since they can produce new and valid theories with minimal biasness, they have a high potential for achieving high conceptual validity than large N studies, there is no valid and reliable cross national large N data available on clientelism, a great depth of understanding can be gained on an issue that is barely
223 known, and they allow for triangulation of evidence. The study shows that Kenya is a typical case with regards to three factors related to clientelism: economic development, corruption, and de mocracy. Kenya has the same annual GDP growth rate (4% 5%) as most African countries (40%), it has the same level of corruption as most African countries (48%), and shares the same level of democracy (partly free) with most countries (52%). Kenya also fa lls among the 56% of countries with the average rate of annual urban population growth in Africa (3% to 4%). Its typicality therefore makes it possible to generalize the findings from the study. This is important for both theory building and policy recomme ndations. Chapters 3 and 4 present more specific information on the context in which the study was done. This study was conducted in the informal sector in Kenya. Chapter 3 shows how the sector has rapidly grown and gained economic significance in urban a reas. It is now where most citizens earn their living. The recorded employment in the sector curr ently stands at about 80% In most African countries the ave rage is around 50% to 60% of the population. Given its expansiveness it is an important venue for e xamining clientelism since it is where most of the citizen politician interactions at the local government level take place. It is where most politicians seek votes through both clientelistic an d non clientelistic ways. C hapter 3 also shows the role that t he state and microfinance institutions have played in sustaining the sector. In Kenya, as well as most African countries, it has become a rather permanent feature in urban areas through the support of this two actors. This makes it a suitable place to exam ine the effect that urban development initiatives have on clientelism. These businesses are concentrated Nairobi, Mombasa, Kisumu, which have been exa mined in
224 Chapter 4. It introduces the three cities by showing the history of ur ban development. It attempts to show how they were founded and grown over time. It accomplishes three independence to the current period, provides a history of urban development in the three cities and compares the three cities. It uses secondary data to show why Nairobi, the capital city, has the highest level of urban development, followed by Kisumu, and then Mombasa. Through this examination it introduces the main driving factors of urbanization in Kenya. It essentially lays the foundation fo r the analysis that follows in C hapters 5, 6, and 7 where the level and type of clientelism in the three cities is compared. Since clientelism thrives in areas that are poor, higher levels ca n be expected to be present in cities with a low degree of urban development. Chapter 5 examines this hypothesis. Cities with low development have scarce opportunities for growth since the infrastructural factors that stimulate business growth are lacking. In this context infrastructural capacity is too underdeveloped to increase production levels. This leads to scarcity of basic needs for citizens. Hence, the lack of growth makes clientelistic offers an option for filling the scarcity. In a developed city this may not be the case because citizens have more opportunities for business growth and can depend on their businesses. For them clientelistic offers have to be much more appealing and this is costly for politicians, who end up engaging in such transacti ons at a minimal level. Chapter 5 uses the above logic to examine the effect of urban development on clientelistic transactions that takes place between political parties and citizens. It starts with an extensive examination of the roots of clientelistic l inkages that exist between
225 parties and citizens in Kenya. An analysis using four factors is then done to compare the level of urban development in the three cities. This analysis uses archival data that was retrieved from the Ministry of Local Government i n Kenya, which oversees the operations of local governments. The four factors compared using the data are: the level of infrastructural development, the number of markets owned by local governments, real estate development and business growth. Nairobi ran ks high in all the four dimensions, followed by Mombasa, and then Kisumu. In this case, we should expect to find that clientelistic transactions are the least common in Nairobi, moderate in Mombasa, and high in Kisumu. The examination of clientelistic tran sactions between parties and citizens uses survey data. They are based on a question that asked citizens who have received assistance from parties, whether it was mostly given around election periods or not. If so, it is considered clientelistic. This reas oning borrows from most studies in political science. The findings uphold the expectations laid out above by showing that these transactions are less common in Nairobi and Mombasa but mostly common in Kisumu. But we cannot tell from these findings whether the same transactions can be observed at the politician level since parties can use a wide range of channels to carry out clientelistic transactions, such as agents. This is the goal of C hapters 6 and 7, which investigate the nature of these transactions a t the politician level. Chapter 6 starts off the analysis by conceptualizing and operationalizing clientelism using an approach that has not yet been used in literature. Clientelism is systematically disaggregated into two modes: traditional and modern. T he traditional mode is based on long term clientelistic relations between citizens and politicians, while
226 the modern one is based on one time transactions that take place during elections. Although this distinction is known to exist, most studies in politi cal science conflate the two categories or overlook the traditional mode. This mode has been mostly addressed in anthropology and sociology. I argue that both modes should be considered in studies on clientelism so as to give a more complete understanding of its causes and effects. The effects of clientelism should be compared based on the traditional versus modern distinction. In most cases, the findings in the literature present an image of one general type of clientelism affected by development. But wha t is not possible to decipher from these studies is the comparative effect of development on both modes. This study attempts to fill this gap. This approach can significantly increase our understanding of clientelism. In order to clea rly distinguish the tw o modes, C hapter 5 uses a wide range of studies from anthropology, sociology, and political science. It operationalizes the traditional mode as one undergirded by long term relationships between citizens and politicians, with transactions that do not depen d on election periods but occur throughout the relationship. On the contrary, in the modern mode the relationships are ephemeral since they are established for one time transactions that occur during election periods only. The occurrence of the two modes a cross the three cities is thoroughly examined in Chapter 7 The data used in Chapter 7 is based on a question that asks respondents the main reason why politicians mostly visit them and when the help they have received from politicians was given to them. Fo r the first part they are given choices to pick from that include campaigning and four other that have nothing to do with campaigning but helping them with their businesses. The other question asks them whether most of the
227 help they have received from poli ticians was given during election periods or non election periods. The assistance given during election periods are categorized under modern clientelism while those that are given during non election period are categorized under traditional clientelism. T hey were indirectly rather than directly asked whether they mostly get help through clientelistic transactions that they have with politicians for two related reasons. One, the study was conducted four months after a post election violence that occurred in Kenya, where politicians had been accused of inciting citizens to participate in violent activities. Investigations were then launched to find those who had been involved in the skirmishes. Hence, it would be a highly sensitive question to ask directly. T his would have also decreased the response rate and sample size. Two, it would lead to unreliable answers because the natural inclination at that time, with investigations going on, would have been to lie or skip the question. This was the case when pilot tests were done on the survey questionnaire. Respondents tended to skip or feel uncomfortable answering any sensitive questions related to those elections. The data from the questions that try to find out why and when politicians visit citizens is examine d using cross tabulations and histograms. The occurrence of the two modes is compared in the three cities. The goal is to find out where the traditional mode is most common and where the modern mode is most common. According to studies on the traditional m ode, it is mostly found in rural areas. This means that we should expect to find it most comply used by politicians in the city with low level of urban development (Kisumu). The analysis has shown that clientelistic practices are less prevalent in the cit y that has high urban development (Nairobi), moderately prevalent in the city with
228 moderate urban development (Mombasa) and more prevalent where urban development is low (Kisumu). The literature on clientelism does not distinguish clientelistic practices b ased on the type of politician. This study argues that the use of both modes depends on the type of politician and the level of urban development. Politicians who are at a higher legislative level such as Members of Parliament have more resources to us e for clientelism than those at the lower level such as Councilors. Hence, the mode preferred depends on the context and resources available to a politician. Since the traditional mode is based on establishing long term relationships at the local level, it favors politicians who serve at the local level Councilors. But it only works well where the level of urbanization is low since clients have to be monitored frequently. Again this favors the Councilor because he represents a smaller percentage of citi zens than an MP. On the contrary, MPs have enough resources to buy votes during election periods but since they represent more citizens they may not have the time to establish long term relationships, especially in highly urbanized areas. Given these cons iderations, we should expect the Councilors to use the traditional mode more often and the MPs the modern mode. The findings comparing the occurrence of clientelism show that MPs mostly use the modern mode in the moderately developed city. This suggests th at the costliness of the modern mode in the most developed city prevents MPs from using it often. As for the least developed city, the findings show that they mostly use the traditional mode. Councilors mostly use the modern mode in the most developed city and the traditional mode in the least developed city. This indicates that they are unable to employ the traditional mode in a highly populated city, hence for them the modern mode is preferred in that context.
229 Although it may seem as if both the Councilor and MP use the modern mode in the developed cities in the same way, it is not the case. When their overall usage of this mode is compared, the MP uses it more than the Councilor. To corroborate the findings above, which were examined using the cross tabul ations graphs, two maximum likelihood estimation techniques (binary regression models and ordinal regression models) have been used to test the likelihood of the two politicians using the modern mode versus the traditional mode. For this analysis, the use of the modern mode occurs when citizens who vote often are the ones that are likely to get help from a particular politician and the traditional mode occurs when a citizen is mostly visited for non campaign reasons but to be offered business assistance. Th e data for this examination is based on a question that asks citizens how many elections they have missed. Again, the indirect approach in examining the occurrence mode takes into consideration the sensitive context in which the study was done. In general, the findings show voting often and campaign visits are strongly associated with getting assistance from MPs, while those who have been visited often for non assistance. These fin dings again show that the modern mode is associated with the MP and the traditional mode associated with the Councilor. Taking into consideration the past and recent studies, three broad suggestions can be made for the future. The need to distinguish betwe en the two clientelistic modes examined in this study, especially that of providing more knowledge on the traditional mode, the need to account for the effects of urbanization on clientelism and the need to differentiate how higher and lower level politici ans use the two modes with regards to variations in urban development.
230 8.3 Distinguishing Clientelism and Increasing Focus on the Traditional Mode Although recent studies on young democracies have shed more light on clientelism (e.g. Kitschelt and Kselman 2012, Stokes 2005, Wantchekon 2003), they seem to re examine of the same issues that have been addressed in past studies. Hence, they have not adequately addressed other unexamined aspects of clientelism. What is needed is a new approach that examines the comparative effect of the modern and traditional mode. There is much to be learned in such examinations. Most studies, including recent ones (e.g. Benit Gbaffou 2011, 455; Brender and Drazen 2009) tend to conflate the two modes. There are a few (e.g. Nicht er 2011; rnebring 2012) that acknowledge the existence of these modes but do not go as far as comparing their effects or examine how they are distinctively affected by social, political, and economic factors. Almost all the studies that examine the effect s of clientelism are based on the modern mode (e.g. Keefer 2007; Lynch and Crawford 2011). Since the traditional mode is not often specifically mentioned, there is an underlying assumption that it has similar effects as the modern one. Oversight of the tra ditional mode raises some questions about existing theories because its aspects differ from those of the modern one. To be sure, there have been some studies that have exclusively paid attention to the traditional mode (e.g. Auyero 2001; Banfield and Wilso n 1963; Kitschelt 2000; Levitsky 2003; Powell 1970; Paul 2008). But these studies also tend to provide a one sided understanding of clientelism as they leave out the modern mode. They also fail to provide an intricate understanding of how it is used by dif ferent politicians. What is missing are studies that focus on both at the same time. Most of the earlier studies on clientelism in political science applied a broader meaning of clientelism that included the traditional mode but still did not make an
231 atte mpt to compare the effects of both (e.g. Clapham 1982, 4 5; Eisenstadt and Lemerchand 1981, 271 271; Eisenstadt and Roniger 1984, 43 44; Lande 1977, xii; Scott 1972, 104 105; Weingrod 1968, 377 379, Roniger 1994, 3 5). For instance, studies that treat it a s a social relationship (Omobowale 2008; Taylor 2004, Garcia Guadilla and Perez 2002) strictly focus on the traditional mode but inappropriately make conclusions that apply to both. To be clear, I am not claiming that clientelism has not been distinguished in other ways but highlighting more intricate variations that have not been addressed. There are hardly any studies that try to differentiate it at the different, from operationalization to how it is applied, levels this study has attempted to do. Instead the few that are there address other issues while still maintain the generalized definition. For instance, distinctions are based on the different goods distributed using the strategy: targeted benefits versus club goods (e.g. Robinson and Verdier 2002; Keefer and Valicu 2008), integrated versus non integrated clientelist systems (Ansell and Mitchell 2011), structured versus processual clientelistic exchanges (Bika 2011), clientelism in political versus non political networks (Szwarcberg 2012), and democr atic versus anti democratic forms of clientelism (Benit Gbaffou 2011). Although such evaluations are insightful, those studies continue using it in its conflated form. By doing this, they raise more questions that make their findings tentative but also ope n new avenues for future research. For instance, what is the relationship between these goods and the two modes? Which one is mostly distributed using the modern mode or the traditional one? The fact that both modes are used as strategies for acquiring vo tes in exchange for personal goods has contributed to the oversight of their different features. Since both
232 are used for this similar purpose, the point I seek to make is not that one mode is used as a strategy for acquiring votes while the other one is no t. What am calling for is a consideration of the different procedures involved in both modes. As this study has shown, this differences primarily revolve around the timing of transactions that occur during non election and election periods. Most of the ins ight offered by commonly referenced studies on clientelism is one sidedly based on the latter period (e.g. Lindberg and Morrison 2008; Van de Walle 2007; Vicente and Wantchekon 2009; Young 2009). There have been several recommendations offered for addressi ng the clientelistic transactions that take place during election periods (e.g. Keefer 2009; Robinson and Verdier 2002) but hardly any for addressing those that take place during non election periods. Several questions thus remain unanswered. For example, what is the nature of recurring clientelistic transactions that continue to take place during non election periods? How frequently do politicians who use the traditional mode monitor their clients. Other questions can be raised with regards to different as pects associated with the modes: the politicians that use them, the goods that are distributed through them, the clients that are mostly targeted, and the context that they are most likely to be used in. An appropriate response to these issues requires us to pay more attention to the dynamics of the under examined traditional mode. This shift of attention has both theoretical and practical benefits. Theoretically, it will help us to understand vote buying strategies that are not directly observable at the p olls and develop theories that shed light on them. From a practical standpoint, this will help in enhancing strategies that are designed to mitigate the negative effects of clientelism on electoral outcomes. So far, the perpetual focus on the modern mode o nly provides part of the solution.
233 The modern mode of clientelism has been somewhat over examined, including in recent studies (e.g. Kitschelt and Kselman 2012; Jensen and Justesen 2012; Ichino and Nathan 2012). It, therefore, tends to be inappropriately used as the default definition of clientelism. This is evident from the fact that some scholars (e.g. Green 2010; Lynch and Crawford 2011) do not clearly define the concept in their work but still make conclusions based on it. Most of the time it is simply seen as a way of buying support in exchange for direct benefits (Brender and Drazen 2009, 306). When scholars talk about the pervasive nature of clientelism in African countries (e.g. Lynch and Crawford 2011; Lemarchand 1972; Wantchekon 2003; Hyden 2004) they tend to naturally refer to this view. Studies that examine the negative effect of clientelism on economic outcomes (e.g. Bardhan and Mookerjee 2011, Keefer and Khemani 2004) tend to suggest that it is only the modern mode that leads to such outcomes. Due to the extensive focus on the modern mode, we know it as the only effective one in mobilizing political support in young democracies (e.g. Jensen and Justesen 2012, 4; Hopkin 2006, 3). What we do not know is how that effectiveness compares to that of the traditional mode. This can be said of other factors related to clientelism. For instance, the nature of the links that exists between modern clientelism and demographic attributes are well known than those of the traditional mode. Hence, theories, as w ell as policy recommendations, related to these attributes and the modern mode have been well developed. But the same has not been adequately done for the traditional mode. We know that since educated citizens are less inclined towards modern clientelism ( Fujiwara and Wantchekon 2012; Kitschelt and Wilkinson 2007, 25 26), educating voters can prevent voting based on clientelistic appeals. But we do not
234 satisfactorily know how this finding applies to traditional clientelism. This can be said of other demogra phic aspects like income, age, and gender. Most of the time we inappropriately apply what we know about the modern to the traditional one. The primary focus on the modern mode has sometimes misled us to the conclusion that clientelistic relationships in g eneral are only ephemeral. The image one gets from such a conclusion is that they are only established during elections and then terminated once elections are over. For example, Kitschelt and Kselman (2012, 6) d especially in democracies where the very existence of party organization is in doubt from one electoral cycle to the next, will generally not be able to credibly commit to the provision of continuing benefits over a argument, this study has shown that this is the case for the one time exchanges in the modern mode but not the traditional clientelistic exchanges that are based on durable goods. Scholars that have examined transactions done in the traditional mode have shown that the purpose of the durable goods is usually to create a long term commitment from the voters (e.g. Hopkin 2006, 2; Boissevian 1966, 22; Hilgers 2011, 568). The nature of transactions in the traditional mode requires us to view citizen politicia n relationships as being long term rather than short term (Scott 1972, 100; Eisenstadt and Roniger 1984, 48 49; Mason 1986, 489). It can be a form of demonstrating long term accountability towards citizens for some politicians (Hilgers 2011, 578). The long term nature of the linkages in the traditional mode make it critical to understand it better if we are to suggest ways to mitigate the negative effects of clientelism on the consolidation of democracy. With regards to policies that attempt to
235 undermine th e negative effects of clientelistic practices, it is critical to understand the traditional mode because it is likely to be more persistent than the modern mode. This is because it has strong habitual underpinnings that serve as an entrenched source of the dimension that is related to the practice of exchanging support for votes, it is also a practice based on the existence of an internalized habitus, in other words, p atterns of There are diverse interactions that exist within clientelism (Landini 2012, 9) that have gone unnoticed due to the one sided focus. Clientelism takes many different forms in various contexts and levels of societies (Lande 1983, 440). The existing knowledge based on relationships built around the one time exchange of votes at the polls leaves out voters who take part in them but are also already committed to a politician because they already received durable goods during the non election periods. When both modes are taken into consideration, it seems that there are more than one type of clientelistic voters: those whose transacti on already took place prior to the elections and will continue after elections, those whose transaction will take place close to or during the elections, those who are involved in both transactions. Since most of the insight that has been offered is based on those that engage in one time transactions, it is possible that the prevalence of clientelism in some countries is often underestimated and misunderstood. This is especially the case in countries where clientelism in general is said to be minimal or abs ent without complete evidence that includes the types of
236 clientelistic voters mentioned above. Addressing all those categories of voters in future studies will improve existing theories. Some of the most recent findings show how the findings from this stud y could advance theories on clientelism. A study that examines the effect of primaries on parties with regards to clientelism (Ichino and Nathan 2012) concludes that the effects no and Nathan 2012, 1). Insightful as it is, it falls short of specifying the type of clientelism where such outcomes may not apply. It raises several other questions that can be answered by distinguishing the two modes. Are these effects mostly common whe re the modern mode is prevalent or is it where the traditional mode is prevalent? Or what is the variation in contexts where both modes are present? These questions apply to other studies as well. In another study that re examines an established understand ing of clientelism, Kitschelt and Kselman (2012, 1) challenge the argument that clientelism decreases as economic development increases and consolidation of democracy takes place by showing that it tends to increase as a country develops and consolidates i t democracy. This is an important finding that has several theoretical and policy implications. But when we take into consideration the existence of the modern and traditional mode it seems to raise some certain questions. Does their argument apply to both the modern and traditional mode? Is the argument applicable where only one of the two modes is present or where both are present? It may be the case in one and not the other, especially when we take into consideration that it is persistent in rural areas because it is used as a strategy for accessing resources that states only provide in urban areas (Brinkerhoff and Goldsmith 2002, 2). The insight they offer can be
237 improved by conducting case studies that compare the effect of development and democratizati on on both modes. Theories on clientelism are based on commonly accepted arguments that need to be reevaluated with a different approach towards the concept. For instance, the commonly accepted argument that clientelism undermines democratic electoral outc omes (Devarajan et al. 2012, 3) needs to be further examined since the two clientelistic modes represent different strategies that may have different effects on democratic outcomes. This argument is often based on the modern mode rather than the traditiona l one, in the sense that it refers to one time transactions that take place at the polls. A re examination of the relationship between the two modes and democratic electoral outcomes could potentially lead to different conclusions and new insight. One of t he most problematic aspects in the way clientelism has been treated is the similarity made between the practice and other outcomes of democratization. For instance, some studies have made the case that it can be form of democratic accountability (e.g. Beni t Gbaffou 2011, 455; Beck 2008). This proposition tends to be based on a generalized form of clientelism although it could play out differently in when clientelism is distinguished. Hence, it is questionable whether both modes are forms of accountability. This is not clear from the existing literature. Answers to these questions have important policy making implications. Knowing the exact relationships can help in the development of appropriate policies for preventing the negative outcomes of clientelism th at usually affect the distribution of public services. There is a widely upheld view that clientelism continues to dominate African politics (Lynch and Crawford 2011, 287). This has been accepted from a generalized
238 perspective that does not show how the m odern and traditional mode may vary around the continent (e.g. Ichino and Nathan 2012, 9). The dominant image presented by existing literature is one of a uniform type of modern clientelism occurring across the continent. This image has led to the conclusi on that the modern mode is more dominant in Africa than the traditional one (Resnick 2012,4). Yet, there are countries where a good measure of democratic accountability exists along with clientelism, such as Ghana (Lindberg 2012, 1). Such findings contradi ct those that argue that clientelism diminishes when democratic consolidation occurs (Keefer 2007). These contradictions suggest a need to reconsider how the norm has been conceptualized and operationalized. It is possible for both clientelism and procedur al democracy to exist where the mode that is tied to such a democratic state is the modern one. This is because the existence of a well functioning procedural democracy assumes that elections are not adulterated by negative aspects of vote buying at the po lls. A change in methodologies may be required in order to address these contradictions. What this case study of Kenya has shown is that theories that have been built using large N studies need to be tested by conducting within case studies that examine t he occurrence of both modes. Due to methodological advancements, scholars now have better tools that can be used to improve our understanding of clientelism. To get more insight on the effect of long term traditional clientelistic relationships there is a need for panel studies that use experiments. But these methods alone will not make a difference if the way we conceptualize and operationalize the norm is not modified. For instance, the innovative experiments that have been recently used to examine client elism (Vicente and Wantchekon 2009; Aker et al 2010; Collier and Vicente 2010)
239 still have the same problem of conflating the two modes. They essentially address the clientelism associated with vote buying during elections but leave the one that is not di rectly associated with elections unaddressed. Although we know that education reduces voter engagement in clientelistic practices from one such study (Fujiwara and Wantchekon 2012), we are still unable to say what the comparative effect of education on bot h modes looks like. Extensive focus on the modern mode has also led to the failure to distinguish the use of the two modes in both the national and local government level. Most of the attention in the vast literature tends to be given to national level is sues (e.g. Diaz Cayeros et al. 2001; Gibson 1997; Johnston 1979; Keefer 2007; Kurer 1993; Medina and Stokes 2002; Robinson and Verdier 2002; Tullock 1980). In most on this literature it is also not clear which politicians are being referred to (e.g. Ansell and Mitchell 2011; Benit Gbaffou 2011, Hellmann 2012; Sanchez and Lozano 2012; Szwarcberg 2012; Weitz Shapiro 2012; Grzymala Busse 2008; Keefer and Khemani 2004; Kurer 1993). d Manson 2011; Whitehead 2012; 1087; Ichino and Nathan 2012; Calvo and Murillo 2004; Eisenstadt and Roniger 1980; Fatton 1986; Fox 1994). The literature therefore presents one commonly upheld image that needs improvement: politicians or patrons serving at the national government level as being clientelistic in the modern sense. This dissertation has presented a more complex image showing that MPs who deal with national level issues are more likely to use the modern mode and least likely to use the trad itional one, while Councilors who deal with local government issues are more likely to use the traditional mode and least likely to use the modern one. By
240 doing this, it has shown that the both modes exist in the two levels of government. Due to lack o f studies that show such variations, we do not fully understand how the use of both mode affects democratic governance. This is particularly the case at the local government level. Oversight of this level has important policy implications related to democ ratic distribution on public goods and services. The fact that citizens get most of their public services at the local government level (Evans 1989; Lewis 1998, 142; Barasa and Eising 2010, 1) means that understanding how both modes are used at this level matter significantly for policy recommendations. It suggests a need for a more complex image. The fact that this has not been adequately addressed in the literature means that some existing theories are rather weak and policy recommendations offered only t arget part of the problems associated with lack of democratic distribution. From the past and recent studies that this study has reviewed, there is a recent one that somewhat addresses the gaps it has highlighted and echoes some of the suggestions this s tudy has made. This study (Nichter 2011) demonstrates the additional knowledge that can be gained by distinguishing the two modes in other contexts. It was done in the healthcare sector in Brazil. Nichter (2011) compares the delivery of healthcare benefits in two types of clientelism: electoral clientelism and relational clientelism delivers all pay offs to citizens before voting, and involves the threat of opportunis tic defection by citizens. By contrast, relational clientelism continues to deliver benefits to citizens after voting, and involves the threat of opportunistic defection by conflating the two categories and reducing them to the modern mode (Nichter 2011, 3). He also
241 points out that most of the literature pays very little attention to clientelism at the local level (Nichter 2011, 7). He finds that relational and electoral co exist in the healthcare sector in Brazil but what varies are the quality and quantity of goods distributed using the two strategies. What he does not show, however, is how the use of both modes vary around the country. He (2011, 32) aptly points out that s ince the modes are not separated scholars talk past each other. The following extract from his study (Nichter 2011, 32) appropriately illustrates this problem. Failing to distinguish between electoral and relational clientelism may result in scholarly conf need not conflict. Although neither Kitschelt nor Stokes makes this distinction e xplicit, they focus on entirely different strategies of clientelism. On the one hand, Stokes focuses on a specific form of electoral clientelism vote buying that rewards citizens for voting against their preferences during a given election. This focus leads Stokes to argue that clientelism holds citizens, rather than politicians, accountable for their actions: votes, reward them for their support and punish them for defection i s what patterns of relational clientelism, which he argues involve elite demands for selective incenti ves will be held accountable by them and no Kitschelt 2000: 852; see also Kitschelt et al 2010, 292). It is not uncommon to find such contradicting views in the literature on clientelism. Their existence s uggests a need to reconsider how we conceptualization the norm by comparing the occurrence of both modes and their use by politicians serving at different legislative levels. In another recent study that seems to have taken note of the one sided focus on t he modern mode, Ornebring (2012, 498) challenges scholars to move beyond the focus on the modern mode. He argues that clientelism
242 does not have to be related to securing votes and so expanding its definition beyond practices associated with buying votes du ring elections sheds light on the role of other actors, such as the media (rnebring 2012, 499). All the above suggestions ultimately culminate in how they affect practical outcomes that are at the center of policy oriented studies on clientelism. The prim ary goal of most policy driven studies (e.g. Keefer 2007; Keefer and Khemani 2004; Medina and Stokes 2002; Robinson and Verdier 2002; Ansell and Mitchell 2011) is to find ways of reducing the negative effects of clientelism on competitive political process es associated with a well functioning democracy. For this goal to be fully achieved both modes have to be examined thoroughly so that the solutions suggested are appropriately customized. Policies that are designed to undermine clientelism can be effective if we know better which clientelistic mode is being targeted. Based on the findings presented in this study, it is possible that the undesired effects from politicians who use the traditional versus those who use the modern mode will be different. Similar ly, the strategies suggested for undermining their clientelistic practices will also have to be different. Policies that seek to end the occurrence of clientelism in future can be more helpful if they take this into consideration. This is because those tha t address the one time transactions that occur during election periods may not appropriately apply to transactions based on long term relationships. It is possible that some of those that are in place today, mostly focusing on the modern mode, are not effe ctive since they have left out practices under the traditional mode. This is an issue that needs to be taken into consideration in future.
243 8.4 Beyond Kenya The findings from this study have shown that that there is a negative relationship between urbanizat ion and clientelism in Kenya. The question is whether they can be generalized to other African countries, since urbanization rates are similar to that of Kenya. I make a case for examining this relationship in other African countries based on an explorator y cross national study. This analysis focuses on 36 Sub Saharan Africa countries. Although the relationship between urbanization and other political and economic variables has already been established, that between clientelism and urbanization has not. Thr ough this study we can start establishing a basis for the recommendation that future studies on clientelism should examine it at an in depth level. In this regard, it is also important to find out whether the relationship is similar to the one shown in th is study. The analysis includes a wide range of variables that have already been linked to urbanization and covered in this study. The goal is not to elaborate on them again but to simply acknowledge them and then use them as control variables. I do not in tend to compare their effects with that of clientelism since this is not the purpose of the analysis. In most studies on urbanization in developing countries, the relationship between urban growth trends and economic factors rather than political factors has been established (e.g. Bertinelli and Strobl 2007; Black and Henderson 1999; Cohen 2004; Storper and Scott 2009; Waddell 2007). This study has tried to link all the three factors by combining lessons from these studies. Similar to what has been shown i n the present study, most existing studies show that urbanization and economic factors have a positive relationship (e.g. Black and Henderson 1999, 253 Todaro and Smith 2009, 321). For instance, urbanization and GDP per capita rat es are
244 often positively correlated (Davis and Henderson 2003, 99), although in some African cities the correlation is negative (Asabere and Owusu Banahene 1983, 377). The relationship between economic growth and urbanization in some African countries has d iffered from others, such that urbanization has occurred without much economic growth (Fay and Opal 2005; Poelhekke 2009, 8). Another popular factor that has been associated with urbanization since the 1990s, due to the effects of globalization, is opennes cities, which are the contact points and the business focus of international trading opennes s markets to underdeveloped areas, increasing their importance and hence reducing the importance of capital cities (Krugman and Elizondo 1996, 146). Ades and Glaeser (1995, 196) argue that low levels of international trade increase the degree of concentrat ion of resources in few urban areas. Unlike the national focus that such arguments are based on, the examination in this study has shown how they play out in small businesses at the local level. Other factors that affect these levels of concentration are t he business environment and regulatory policies (Ades and Glaeser 1995, 199). Both of these factors have also been incorporated in this study. Political factors have not been widely examined but they have been acknowledged as critical stimulants and restri ctors of urban population growth, especially in places where urban concentration has been directly affected by policies and politics (Davis and Henderson 2003, 98). In a study on the political economy of the urbanization process, Davis and Henderson (2003, 98) found that infrastructural investments facilitate urbanization. The present study has highlighted how this finding
245 plays out at the local government level in Kenya. In some cases, the influences of political factors on urban population growth is not n ecessarily a direct one, but indirect. sectors to the other usually affect th e labor composition of the two sectors. The findings from this study show that the relationship that exists between urbanization and clientelism is indirect. This relationship affects democratic outcomes as well. The effect of political factors is usually related to the way in which democratization shapes government operations (Henderson and Wang 2007). In essence, the type of democratic policies that a government adopts and implements largely affect the rate of urban population growth through distribution of resources (Poelhekke 2009). What we still do not know is how the relationship between clientelism and urbanization affects such outcomes. This is the gap that scholars of clientelism should consider addressing in future research. Focusing on urbanizat ion has the benefit of highlighting how clientelism varies within one country. This insight lacking in existing studies. The examination of clientelism in the three cities in Kenya has shown that generalizations that present a uniform image of all voters i n one country as clientelistic can be misleading. For ue, it can vary in different urban contexts within one underdeveloped country. The socioeconomic factors that feed clientelism are constantly
246 and Opal 2000, 14) and levels of income (Rakodi 1991, 542; Annez and Buckley 2008, 1; Njoh 2003, 173; Davis and Henderson 2003, 99) are rapidly changing in African countries (references). The changes depend on the level of urbanization. The conclusion that is arrived at in most studie s (e.g. Fatton 1986; Chappell 1989; Calvo and Muliro 2004; Keefer 2007; Van de Walle 2007; Young 2009; Kitschelt and Kselman 2012, 25) present countries as a whole as dominated by clientelistic transactions while disregarding that there could be urban and rural areas where clientelistic practices are not be prevalent. Given the increasing rate of urbanization in Africa (Kessides 2005; UN Habitat 2010), scholars of clientelism can improve our understanding of the norm by taking into consideration the variati ons that exist across urban areas. This can be based on both small N and large N case studies. The exploratory regression analysis that follows examines whether there is a possible relationship between clientelism and urbanization that could be explored fu rther. The analysis is based on a snapshot of data for 2009 from 36 Sub Saharan Africa countries. I have used the 2009 data since it is the latest year that has complete data for most countries on the variables included. Urbanization, which is the dependen t variable, has been operationalized as urban population growth (see Table C 4 in Appendix C for a description of all the independent variables and the descriptive statistics). Due to lack of a uniform and comparable data set on clientelism from all Africa n countries, I have used corruption as a proxy. This is an appropriate proxy because corruption is closely related to clientelism (Muno 2010, 14; Persson et al. 2003; Singer 2009, 9) in the sense that narrowly self seeking public officials tend to use reso urces accessible to them for clientelistic purposes (Harriss White 1996; Philip 1997;
247 Robinson 1998, 4). It is sometimes associated with vote buying (Kawata 2006, xii), which is a central aspect of clientelism (Wantchekon 2003, 400). Corruption has also be en found to be common where clientelistic interactions are prevalent (Kawata 2006, xii; Manzetti and Wilson 2007, 949). I also use this proxy to find out the direction of the relationship between clientelism and urbanization. The question is whether it is a negative one as the one shown in this study. This analysis does not attempt to develop a model that can be used to explain the political and economic factors that affect urbanization but to merely find out whether there is a possibility that clientelism has a relationship with urbanization, while cont rolling for other factors. T able 8 1 below presents the findings. Table 8 1. The relationship between clientelism and u rbanization in Africa Dependent Variable: Urbanization (urban population growth) Mode l 1 (N=36) Model 2 (N=36) Independent Variables Coefficient ( p value) Coefficient ( p value) Corruption (proxy for clientelism) 0.029 (0.098) Government effectiveness 0.238 (0.639) 1.393 (0.028) Political stability 0.120 (0.640) Regulatory qualit y (policies) 1.783 (0.029) 2.282 (0.008) Rule of law 0.043 (0.945) Failed state 0.099 (0.018) Business Freedom Openness to trade 0.001 (0.863) 0.020 (0.093) FDI 0.014 (0.228) 0.004 (0.684) GDP growth (annual %) 0.057 (0.043) GDP per capita g rowth (annual %) 0.048 (0.077) Adjusted R 2 0.341 (0.013) 0.516 (0.001) N=36 N=36 Th e findings presented in Table 8 1 show a similar relationship between clientelism and urbanization that has been observed in the case study of Kenya. Similar to the f indings from Kenya, corruption (as a proxy for clientelism) has a negative relationship with urbanization. This is an indication that in countries that have high levels
248 of urbanization, clientelism may be less prevalent and where the level of urbanization is low it may be widespread. But since corruption is just a proxy in this case, this finding should be taken as tentative and as a starting point. It would have been helpful to go beyond the simple analysis if there was data on clientelism from these count ries. The hope is that this finding will provoke considerations for future studies that specifically use data on modern and traditional clientelism. The challenge that lies ahead is in coming up with innovative methods for collecting cross national data on the two modes. Small N studies that make it possible to do more in depth examinations will also be needed. They will make it possible to better examine the effect of both modes. Within this cases it is important to examine clientelistic interactions at b oth the national and local level, and compare how different politicians use them. The question is whether other within case studies in Africa will reveal similar or different variations to those that have been observed in Kenya. Whatever the revelations wi ll be, it is certain that they will improve our understanding of clientelism and enhance policies made to prevent the negative effects of clientelistic transactions.
249 APPENDIX A DATA COLLECTION METHODOLOGIES The data used in this study was collected using five different methods. 120 individual level questionnaires, 20 semi structured interviews, and one focus group was administered in each of the three cities in Kenya (Nairobi, Mombasa, and Kisumu), yielding a total sample size of 360 for the survey, 60 fo r the interviews, and three focus groups. In addition, key informant interviews from three people who have first hand knowledge of the relationship between politicians and the informal sector. The data on urban development in the three cities was collecte d from the Ministry of Local Government archives. Three markets were selected in each city based on variation in constituencies. This made it possible to examine distinct citizen politician linkages. This study began with an attempt to examine the nature o f politics in general in the informal sector. Although there was no particular issue driving it at the beginning, there were several themes. The main themes were the role of government policy in business nature of citizen politician linkages. The exploratory nature of the study at the beginning hence led to the use of three methods. Since this research was conducted in three cities that are distant from each other, it was ne cessary to use research assistants (RAs) to ensure that the data collection was being done within the same period. To ensure that the assistants employed uniform data collection techniques I first spent one week acquainting them on the study and training t hem on the three methods. During this week I held day long training sessions. This week was also used to map the markets that had been selected for the study. The goal was to ensure that we understood the general nature of
250 businesses in the markets, the st ructure of the market, and the number of shops in the market. This information was important for the purpose of systematically recruiting a fair number of participants from each market and selecting markets that had a variety of businesses. To map the mark ets we would walk around them while counting the shops and sketching the general design. Using these maps as a guide in the recruiting process prevented oversampling one cluster of similar business owners and also made it possible to start each day using d ifferent entrances so as to increase the variation of responses. By using different entrances, I ensured that business owners did not share the questions prior to our arrival or even know we were collecting data in the market. T o prevent predictability of responses in this market during one day we would collect data in different rows and ensure that we do not re visit the same row two days in a row. I began the study with the focus group discussions, which were based on open ended questions. The goal was t o use the responses to these questions to find out common themes that could be explored further. Two business owners were selected from each of the three markets selected in each city. Those that were selected had to be politically active in the market. Th e goal was to find business owners that would be able to provide the most insight on the issues that were to be discussed. But the selection was also to be random. In order to accomplish these two goals I established a criteria that would be used for recru iting. The criteria was business owners who had been in business for more than 15 years, were members of a group in the market, and knew the political representatives in the constituency. I set 15 years as a benchmark so as to ensure that they were busines s owners that had participated in at least one election. With the help of three research assistants, one in each city, we visited the
251 markets and used systematic random sampling technique to visit business owners. In each market that we visited we selected several businesses and when we encountered a business owner that met the criteria we gave them a flyer that provided them with the details of the focus group and a phone number to call the next day if they are interested in participating. The following d ay, the first two business owners to call me from each market were selected for the focus group discussion. Those that were finally selected had to consent to being recorded using a tape recorder. Although this is not a perfect method since I may have sel ected those that have access to a phone, it was the most appropriate one. It made it possible to give all the business owners some fair chance to be recruited for the discussion, rather than selecting the first two in each market using only the systematic random sampling method. To create the questions that were used during the focus groups, I reviewed the literature on citizen politician relationships and read newspapers on the same issue in Kenya. Each focus group discussion was recorded using a tape reco rder. I then transcribed these interviews and examined them thoroughly to select the common themes. The themes that stood out were the close relationship that business owners have with microfinance institutions, the lack of government services in the marke ts, rampant insecurity, harassment by local government authority officials, and weak citizen politician linkages. I used these themes to develop questions for next stage, semi structured interviews. Semi structured interviews were more appropriate, as oppo sed to in depth interviews, since I was getting assistance from three RAs in different locations and it was important that the information from the interviews be comparable. Moreover, although semi structured
252 interviews are different from unstructured inte rviews they still have much of the freewheeling quality of unstructured interviewing and require the same skills (Bernard 2006, 212). In order to accommodate the language differences across the three cities, the questions were translated into the commonly spoken languages. English and Swahili in Nairobi, Luo in Kisumu, and Swahili in Mombasa. Prior to the interviews, I held another week long training session with the RAs. The goal was to ensure that the questions would be administered in the same way in the three cities. The prompts were to be done in the same way, the time allocated for answering each question was to be the same, and clarification of questions was to be done in the same way. The goal was to increase the comparability of the responses. To ma ke this exercise successful pilot test were conducted with under my supervision during the one week training session. This helped in identifying areas that would potentially lead to problems and address them. The participants for the interviews were also r ecruited using systematic random sampling within each market. Using this technique we recruited business owners only by first visiting their shops and seeking appointments with them. This is because some of the business owners have employees and are not al ways available. For those who were absent we left the employee(s) with short memos requesting for their participation and also informing them of when we would be stopping by to conduct interviews. This was to ensure that the business owners would be availa ble on the set date. This approach was quite successful as they were always available on the day of the interview and also gave maximum participation. In each city 20 business owners were recruited and interviewed, leading to 60 interviews. To ensure that the RAs were conducting the interviews as expected I used
253 two strategies. One, I conducted the first five interviews in each city as they watched and learned. They then conducted the next five interviews under my supervision. This ensured that there was co nsistency in how the questions were being administered in the three cities. Two, All these interviews were tape recorded and notes taken as well. The purpose of recording the interviews was to ensure that the RAs followed the same format in my absence. On ce all the interviews had been completed I transcribed all the recorded tapes while looking for common themes. This is when my research interest in citizen politician linkages emerged. The finding that stood out, and also led to the application of the next two methods (surveys and key informant interviews) was the nature of relationships between the business owners in the sector and politicians. The findings from these interviews showed that there were only a few business owners that had managed to get subs tantive forms of assistance from politicians. Hence, I was curious about the nature of the linkages that these business owners had with the politicians. I wanted to find out what was unique about these few business owners that receive assistance from busi ness owners. Based on the information gathered on the linkages, I decided to develop a survey that scrutinizes the citizen politician linkages further. Prior to the administering of the surveys, I held another one week training session with the RAs since i t had aspects that all of them needed to be familiar with, such as when to skip questions and how to use the vignettes. I also wanted to ensure that they were able to effectively administer the survey. Most importantly, I used this week to do pilot tests o n the questionnaire and make modifications where necessary. With the assistance of the RAs, we were able to conduct 360 surveys (120 in each city) from nine markets. The
254 surveys were translated into the appropriate languages widely spoken in each city. The y were administered in English and these languages. The data from the surveys were entered on a daily basis so as to get a data set that I could examine while in the field and see if there are any patterns that can be investigated further. For the most par t, the preliminary data that I examined showed the same results as those from the interview. In the last stage, with one week left to depart from Kenya, I interviewed three key informants so as to indirectly corroborate the findings from the politicians s tand point. Although I tried to directly contact politicians I was unable to get an appointment with a one week notice. Most of them were willing to offer me an interview between two weeks and a month. Hence, I decided to interview the key informants inste ad. These key informants were: 1) A personal assistant for the one of the current Deputy Prime Ministers in Kenya; 2) A scholar and politician who lost in his bid for a parliamentary seat at the nomination stage and later worked as a strategist for other M .P.s, and 3) a Party ODM). I found these informants using the snow ball technique. I would have found more key informants but I was short of time. I conducted one hour in depth interviews with them based on the same themes that I used for the semi structured interviews. They provided a perspective of the citizen politician linkages from the For the analysis on urban development in the three cities this study uses archival financial data from the local governments of the three cities Nairobi, Mombasa, and Kisumu. At the time of this study, Kenya was still under a centralized system of government. Hence, the local governments were under the Ministry of Loc al
255 Governments (MOLG). Under this system the state wielded a substantial degree of power over the local governments in a number of important ways, such as, determining the financial base of the local governments, and their annual budget. All their financia l plans had to be approved by the Minister for Local Governments. For them to get approval and financial assistance from the state they had to submit annual financial plans. These conditions made it possible to collect data on urban development initiatives from the MOLG. I, therefore, visited the MOLG archives during my fieldwork and collected this data. I retrieved the data from financial reports submitted starting from 2002 to 2009. The data collected prior to 2002 had problems of authenticity and incompl eteness. They had not been audited and some months had no data. On the other hand, the reports that were compiled from 2002 by the city councils are comparable because they have been edited and also use a similar format required by the MOLG for recording t he data. They strictly followed the set guidelines because the MOLG made it a prerequisite for getting any financial assistance for supplementing their annual budgets. The guidelines generally required local governments to submit annual financial accounti ng records, provide statements of debtors and creditors, and Revenue Enhancement Plans, which provide a clear description of the urban development initiatives within their jurisdictions. The authenticity of the data from these reports is high because they are thoroughly audited before funds are issued. The last report I use is the 2009 one because it is the latest one that was available. There were no reports beyond 2009 because the central government had started a process of devolution.
256 APPENDIX B QUE STIONS AND VARIABLES This appendix provides the actual questions for the variables used in this dissertation, as well as descriptions for those for the constructed variables. Information on the coding of variables is also provided. Dependent Variables: 1. Have you ever received any form of assistance(s) to your business from the M.P./Councilor? Coding: 1 if yes; 0 if no 2. How often have you received this assistance from the Councilor? Codin g: 0 if None; 1 if 1 to 2 times (Low), 2 if more than 2 times (high) 3. I first asked the following question: What kind of assistance have you mostly received from the councilor? Financial assistance to start this business Finan cial assistance to boost my business With the registration of the business With getting this stall The responses are then lumped as low and high based on two factors: 1) the average amount that business owners get as financial assistance and the actual am ount it would take a business owner to register a business or register a business. The average amount is based on responses from business owners when asked about the amount they have received. The actual amount of registration or getting a stall was retri eved from the local government department responsible for these two issues. 2) the other factor that determined now the responses lumped are the number of responses in each category. The number of business owners that received financial assistance to start their business and those who received a stall were too few. The two categories on financial assistance were, therefore, lumped and the other two also lumped. This led to two recoded categories: Low quality coded as 1: financial assistance to start a busi ness or to boost a business. High quality coded as 2: registration of the business or getting a stall.
257 Independent Variables 1. Has voted at least once Have you voted in any national elections? (we clarified that these are elections that include Councilors MPs, and Presidential candidates in the ballot, which is usually the case in Kenya) Coding: 1 if voted; 0 if not 2. Elections missed voting How many times have you missed to vote in elections? Missed all elections More than 3 times 1 to 3 times None Coding : 1 if has missed voting in all national elections; 2 if missed more than 3 national elections; 3 if missed 1 to 3 national elections; 4 if not missed any national election 3. MP/Councilor business visits Has the Councilor/MP ever visited your busine ss when in the market? Yes No What is the main reason why the MP/Councilor visited your business? To campaign To provide financial assistance To provide me with a new business stall To issue me a business permit To solve other business problems Based o n the responses from the question above, I generated two new variables for each of the two politicians based on the following questions: Has the MP/Councilor mainly visited the business owner to campaign? The campaign responses here are recoded as 1 (Yes ) and all the other responses as 0 (No) Has the MP/Councilor mainly visited the business owner to offer business assistance?
258 The four responses associated with offering business responses are lumped and recoded as 1 (yes) and campaign is recoded as 0 (No ) 4. Politician campaign visits: this is a new variable that was created by combining MP and Councilor responses from variable 6 and generates a new variable based on the following question Have politicians mainly visited the business owner to campaign or o ffer business assistance? Coding: campaign responses are coded as 1 and the rest as 0. 5. Direct access to the MP/Councilor How have you mainly contact the M.P./Councilor? Personally written letters Telephone calls Visiting him/her at the office Coding : 1 if contact is through letters (low level); 2 if contact is through phone (moderate level); 3 if contact is through office visits (high level) I then transformed the variable into a binary one based on two categories: direct (telephone calls and office visits) and indirect (letters), coded as 0 for indirect and 1 for direct. 6. High level of access to the MP/Councilor How many times have you contacted the M.P./Councilor? None 1 to 2 times 3 to 4 times More than 4 times Coding : 0 if none; 1 if 1 to 2 (l ow); 2 if 3 to4 (moderate); 3 if more than 4 times (high) 7. Two way link with the councilor This is a new variable by adding the responses in the following questions: a sum of two indicates the presence of a two way link Has the Councilor ever visited y our business when in the market? 1 if yes and 0 if not
259 Have you ever contacted the Councilor? (we clarify that this means that they successfully reached/found him/her) 1 if yes and 0 if not 8. P and Councilor) This is a variable that was created by combining the responses from the following question that was asked separately for both the MP and Councilor. Have you personally been visited by anyone who works for the MP/Councilor here in the mar ket for any particular reason since you started your business? up question: ? The original respo nse categories to this question are: None 1 to 2 times 3 to 4 times 5 to 6 times More than 6 times The responses for both MP and Councilor are then summed up in the new variable. Using the totals I create four new categories, based on four ranges of val ues, in the variable that are coded as: 0 for none, 1 for low, 2 for moderate, and 3 for high. 9. Membership in lobby group better service to these types of markets? Coded as: 1 if he/she is a member; 0 if not 10. Contributes to lobbying often How often do you contribute funds in a year? Coded as: 0 if None; 1 if 1 to 2 times; 2 if More than 2 times 11. Contributed to political parties Have you ever contributed money to political parties, e.g. ODM, PNU, KANU etc.? Coding: 1 if he/she has ever contributed; 0 if not 12. Male
260 Coding: 1 if business owner is male; 0 if female 13. Many employees (large business) How many employees do you have? Coded as: 0 if no employees; 1 if one employee; 2 if two employees; 3 if three employees; 4 if more than 3 employees 14. High capital intensity ( ) What is the type of business? ( only asked if it is nt obvious ): there are 30 types of businesses entered. I then asked business owners for an estimate of the money they personally spent in buying starting their business, including purchasing equipment, and then used those figures to place businesses in three categories: low moderate, high. I asked for estimates because almost all the business owners who took part in the study do not keep records, and for those who do they did not have them readily available. Coded as: 1 if low capital intensity; 2 if moderate capital inten sity; 3 if high capital Intensity (see Table 5 1 in Chapter 5) 15. Older business owner In which age range do you lie? Responses coded as: 1 if 20 years old and below; 2 if 21 to 30 years old; 3 if 31 to 40 years old; 4 if 41 to 50 years old; 5 if more th an 50 years old. 16. Actively seeks government services As a business person in this market and a stakeholder of government services do you take practical actions to demand from the government for better services to your business? Responses coded as: 1 if business owner has achieved business education; 0 if not 17. High level of education What is the highest level of formal education you have completed?
261 Coding: 0 if no formal education; 1if completed primary school (low); 2 if completed high school (moderat e); 3 is has some post high school education (high) 18. Has completed high school Created from the above responses, with those who have completed high school, have some post high school education, or completed post high school education categories considered to have completed high school. The rest have not. Coding: 1 if business owner has completed high school; 0 if not 19. Has achieved business education Do you have any formal business education? Coding: 1 if business owner has achieved business education; 0 if not 20. Has owned the business longer How long have you been in this business? Responses coded as: 1 if one to four years; 2 if five to eight years; 3 if nine to twelve years; 4 if thirteen to twenty years; 5 if more than 20 years 21. Business Challenges Which of the following business challenges do you mostly face? Responses coded as: 1 for difficulty in paying rent; 2 for difficulty in purchasing stock often; 3 for difficulty in acquiring business loans, and 4 for other
262 APPENDIX C SUPPL EMENTARY TABLES AND FIGURES Table C 1. Descriptive s tatistics Variables Mean Std. Deviation Minimum Maximum Dependent Variables 0.04 0.206 0 1 0.13 0.337 0 1 0.16 0.426 0 2 Quality of Councilors help 0.15 0.466 0 2 Independent Variables Has voted at least once 0.91 0.280 0 1 Elections missed voting 3.49 0.902 1 4 MP business visits 0.04 0.200 0 1 Councilor business visits 0.06 0.240 0 1 Politician campa ign visits 0.11 0.318 0 1 Direct access to the MP 0.05 0.235 0 1 Direct access to the Councilor 0.13 0.334 0 1 High level of access to the MP 0.14 0.527 0 3 High level of access to the Councilor 0.46 0.935 0 3 Two way link with the councilor 0.05 0.22 4 0 1 0.31 0.763 0 3 Membership in lobby group 0.13 0.128 0 1 Contributes to lobbying often 0.09 0.396 0 2 Contributed to political parties 0.06 0.245 0 1 Male 0.44 0.497 0 1 Total employees (proxy for business size ) 0.73 1.194 0 4 High capital intensity (proxy for poverty) 1.73 0.779 1 3 Older business owner 2.66 0.850 1 5 Actively seeks govt. services 0.40 0.403 0 1 Highest level of education 1.98 0.797 0 3 Has completed high school 0.58 0.495 0 1 Has achieve d business education 0.27 0.443 0 1 Has owned the business longer 2.18 1.241 1 5
263 Table C 2. Variance inflation factors (VIF) and tolerance values for binary r e gression ariable Variables Model 1 VIF (Tolerance) Model 2 VIF (Tolerance) Model 3 VIF (Tolerance) Ever get MP help (dep. variable) 1.11 (0.90) 1.13 (0.88) 1.10 (0.91) Elections missed voting 1.03 (0.97) 1.04 (0.97) Politician campaign visits 1.25 (0.80) 1.07 (0.94) 1.09 (0.92) MP busine ss visits 1.22 (0.82) 1.18 (0.85) Direct access to the MP 1.17 (0.85) High level of access to the MP 1.31 (0.76) Direct access to the Councilor 1.14 (0.88) Two way link with the councilor 1.16 (0.86) High level of access to Councilor 1.23 (0 .81) 1.10 (0.91) Membership in lobby group 1.12 (0.89) Contributes to lobbying often 1.05 (0.95) Contributed to political parties 1.05 (0.95) 1.12 (0.89) 1.15 (0.87) Male 1.05 (0.96) 1.06 (0.95) 1.09 (0.92) M any employees (large business) 1.07 (0.94) 1.11 (0.90) High capital intensity (least poor) 1.05 (0.96) 1.06 (0.94) Older business owner 1.03 (0.97) 1.03 (0.98) 1.04 (0.97) Actively seeks govt. services 1.20 (0.83) 1.12 (0.89) Highest level of educat ion 1.14 (0.88) Has gotten business education 1.06 (0.95) Mean VIF 1.09 1.14 1.10
264 Table C 3. Variance inflation factors (VIF) and t olerance values for binary regression ari able Independent Variables VIF (Tolerance) variable) 1.10 (0.91) 1.04 (0.96) Has voted at least once 1.04 (0.96) High level of access to Councilor 1.24 (0.81) High level of access to MP 1.2 3 (0.81) Contributes to lobbying often 1.04 (0.96) Male 1.05 (0.95) Older business owner 1.02 (0.98) Actively seeks govt. services 1.21 (0.83) Contributed to political parties 1.07 (0.93) Has a high level of education 1.13 (0.88) Many employees (lar ge business) 1.10 (0.91) Mean VIF 1.11
265 Table C 4. ssistance (with dummy v ariables for Nairobi and Mombasa) 1 Independent Variables Model 1 Coefficient (SE) Model 2 Coefficient (SE) M odel 3 Coefficient (SE) Has never missed voting 1.511 (0.977) 1.825 (1.040)* Politician campaign visits 1.587 (0.848)* 2.057 (0.787)** 1.471 (0.757)* MP business visits 0.715 (1.255) 0.235 (1.241) Direct access to the MP 2.589 (0.882)*** Hig h level of access to the MP 1.284 (0.385)*** Direct access to the Councilor 1.108 (0.687) Two way link with the councilor 0.160 (1.012) High access to the Councilor 0.131 (0.332) visits 0.525 (0.252)** Membership in lobby group 0.652 (0.822) Contributes to lobbying often 0.599 (0.539) Contributes to political parties 0.687 (1.190) 0.810 (1.248) 1.446 (1.333) Male 0.066 (0.581) 0.144 (0.599) 0.135 (0.599) Many employees (large business) 0.436 (0.207)** 0 .336 (0.219) High capital intensity (least poor) 0.722 (0.409)* 0.918 (0.449)** Older business owner 0.666 (0.459) 0.779 (0.463)* 0.789 (0.456)* Actively seeks govt. services 0.850 (0.745) 0.560 (0.683) High level of education 0.061 (0.198) Has gotten business education 0.046 (0.693) Nairobi 0.101 (0.713) 0.361 (0.755) 0.385 (0.773) Mombasa 0.001 (0.874) 0.906 (0.887) 0.055 (0.916) Constant 9.588 (4.026)** 2.254 (1.464) 10.719 (4.367)** N=360 N=360 N=360 Significant at the p <0. 10, ** p <0.05, *** p <0.01 1 Kisumu has been left out of the model so as to prevent perfect multicollinearity. However, i t was included in other models together with Mombasa, while maintaining the same variables, and the output had the same values as those shown.
266 Table C 5. ssistance (with dummy v ariables for Nairobi and Mombasa) Independent Variables Logit Coefficient (SE) 0.173 (0.559) Has voted at l east once (vlo) 1.038 (0.786) High level of access to Councilor (hac) 0.693 (0.161)*** High level of access to MP (hmp) 0.679 (0.447) Contributes to lobbying often (cl) 0.711 (0.303)** Male 0.249 (0.352) Older business owner (obo) 0.131 (0.208) Actively seeks govt. services (asgs) 0.150 (0.379) Has given party assistance (gpa) 1.115 (1.070) Has a high level of education (hle) 0.043 (0.112) Many employees (large business emp) 0.055 (0.150) Nairobi 0.529 (0.397) Mombasa 0.175 (0 .177) Constant 3.611 (1.058)*** N=360 Significant at the p <0.10, ** p <0.05, *** p <0.01
267 Table C 6. Logit analysis of getting c ou ssistance (with dummy v ariables for Mombasa and Kisumu) Independent Var iables Logit Coefficient (SE) 0.224 (0.553) Has voted at least once (vlo) 0.931 (0.795) High level of access to Councilor (hac) 0.638 (0.157)*** High level of access to MP (hmp) 0.628 (0.449) Contributes to lobbying often (cl) 0.683 (0.302)** Male 0.253 (0.351) Older business owner (obo) 0.100 (0.208) Actively seeks govt. services (asgs) 0.259 (0.369) Has given party assistance (gpa) 1.208 (1.070) Has a high level of education (hle) 0.012 (0.110 ) Many employees (large business emp) 0.029 (0.147) Mombasa 0.149 (0.176) Kisumu 0.060 (0.190) Constant 3.419 (1.114)*** N=360 Significant at the p <0.10, ** p <0.05, *** p <0.01
268 Table C 7. Ordinal analyses of gett ssistance (with dummy v ariables for Nairobi and Mombasa) Independent Variables Model 1 Coefficient (SE) Model 2 Coefficient (SE) 0.108 (0.584) 0.114 (0.622) Has never missed voting (nmv) 0.356 (0.225) 0.140 (0.222) Councilor business visits (cbv) 1.988 (0.592)*** 1.892 (0.609)*** Two way link with councilor (twlc) 0.211 (0.819) 0.253 (0.799) Contributes to lobbying often (clo) 0.553 (0.318)* 0.776 (0.331)* Male 0.008 (0.346) 0 .086 (0.384) Older business owner (obo) 0.328 (0.222) 0.380 (0.238) Many employees (large business emp) 0.032 (0.151) 0.009 (0.160) Has completed high school (chs) 0.683 (0.357)* 0.960 (0.396)** Contributes to political parties (cpp) 1.305 (1.080 ) Has owned the business longer (obl) 0.303 (0.166)** 0.268 (0.177) Nairobi 0.254 (0.374) 0.022 (0.434) Mombasa 0.134 (0.166) 0.198 (0.152) N=360 N=360 Significant at the p <0.10, ** p <0.05, *** p <0.01
269 Table C 8. Ord ssistance (with dummy v ariables for Mombasa and Kisumu) Independent Variables Model 1 Coefficient (SE) Model 2 Coefficient (SE) 0.139 (0.584) 0.212 (0.626) Has never misse d voting (nmv) 0.345 (0.228) 0.131 (0.223) Councilor business visits (cbv) 1.976 (0.590)*** 1.887 (0.607)*** Two way link with the councilor (twlc) 0.275 (0.819) 0.275 (0.797) Contributes to lobbying often (clo) 0.533 (0.320)* 0.734 (0 .336)** Male 0.017 (0.346) 0.084 (0.385) Older business owner (obo) 0.323 (0.224) 0.389 (0.238) Many employees (large business emp) 0.001 (0.148) 0.029 (0.160) Has completed high school (chs) 0.651 (0.352)** 0.988 (0.394)** Contributes to polit ical parties (cpp) Has owned the business longer (obl) 0.327 (0.166) 0.280 (0.177) Mombasa 0.127 (0.161) 0.211 (0.148) Kisumu 0.150 (0.169) 0.194 (0.175) N=360 N=360 Significant at the p <0.10, ** p <0.05, *** p <0.01
270 Table C 9 Dependent and independent v ariables Dependent Variable Description Source Urban population growth (annual %) Urban population is the midyear population of areas defined as urban in each country and reported to the United Nations. United Nations, World Urbanization Prospects. Independent Variables Description Source Freedom from Corruption (proxy for clientelism ) This factor relies on Transparency (CPI), which measures the level of corruption in 152 countries, to determine the freedom from corruption scores of countries that are also listed in the Index of Economic Fre edom The CPI is based on a 10 point scale in which a score of 10 indicates very little corruption and a score of 0 indicates a very corrupt government. Quality of Government Government Effectiveness single grou ping responses on the quality of public service provision, the quality of the bureaucracy, the competence of civil servants, the independence of the civil service from political pressures, and the credibility of the for the government to be able to produce and implement good policies and deliver public goods. World Bank Governance Indicators. Political Stability whi ch measure perceptions of the likelihood that the government in power will be destabilized or overthrown by possibly unconstitutional and/or violent means, including domestic violence and terrorism. World Bank Governance Indicators. Regulatory Quality incidence of market unfriendly policies such as price controls or inadequate bank supervision, as well as perceptions of the burdens imposed by excessive regulation in areas such as foreign trade and business de velopment. World Bank Governance Indicators.
271 Table C 9. Continued Independent Variables Description Source Rule of Law measure the extent to which agents have confidence in and abide by the rules of soci ety. These include perceptions of the incidence of crime, the effectiveness and predictability of the judiciary, and the enforceability of contracts. Together, these indicators measure the success of a society in developing an environment in which fair and predictable rules form the basis for economic and social interactions and the extent to which property rights are protected. World Bank Governance Indicators. Failed States Index The Failed States Index includes an examination of the pressures on stat es, their vulnerability to internal conflict and societal deterioration. Fund for Peace Business Freedom The business freedom score encompasses 3 components, all weighted equally; starting a business, closing a business, obtaining a license. Each of thes e raw components is converted into a scale graded from 0 to 100, where 100 represents the maximum degree of business freedom. Doing Business study. Openness to Trade, Current Prices (%) Exports plus imports as a percentage of GDP. Measured at current prices. United Nations Statistics Divisions National Accounts Foreign direct investment, net inflows (% of GDP) Foreign direct investment are the net inflows of investment to acquire a lasting management interest (10% or more of voting stoc k) in an enterprise operating in an economy other than that of the investor. It is the sum of equity capital, reinvestment of earnings, other long term capital, and short term capital as shown in the balance of payments. This series shows net inflows in th e reporting economy and is divided by GDP. IMF International Financial Statistics and Balance of Payments databases, World Bank, Global Development Finance, and World Bank and OECD GDP estimates.
272 Table C 9. Continued Independent Variables Description So urce GDP growth (annual %) Annual percentage growth rate of GDP at market prices based on constant local currency. Aggregates are based on constant 2000 U.S. dollars. GDP is the sum of gross value added by all resident producers in the economy plus any pr oduct taxes and minus any subsidies not included in the value of the products. It is calculated without making deductions for depreciation of fabricated assets or for depletion and degradation of natural resources. World Bank national accounts data, and O ECD National Accounts data files. GDP per capita growth (annual %) Annual percentage growth rate of GDP per capita based on constant local currency. GDP per capita is gross domestic product divided by midyear population. GDP at purchaser's prices is the sum of gross value added by all resident producers in the economy plus any product taxes and minus any subsidies not included in the value of the products. It is calculated without making deductions for depreciation of fabricated assets or for depletion an d degradation of natural resources. World Bank national accounts data, and OECD National Accounts data files.
273 REFERENCES Acock, Allan C. 2006. A Gentle Introduction to Stata Texas: Stata Press. South Asia: Getting Beyond the Structure Political Studies 52: 1 18. Savings and Development 1: 5 22. Ades, Alberto F. The Quarterly Journal of Economics 110 (1): 195 227. Africa Research Bulletin. 1966. Volume 3. Cambridge, MA: Blackwell. ences in the Incidence of Journal of Development Studies 35(6): 36 58. and Tim Sparkman. 2007. Attracting Investment to Kis umu: Opportunities and Challenges The Earth Institute, Millennium Cities Initiative, Columbia University. ment of Economics, Oxford University. http://www.iig.ox.ac.uk/research/48 cellphones election mozambique.htm Akivaga, S. K., W. Kulundu Bitonye., and M.W. Opi. 1985. Local Authorities in Kenya Nairobi: Heinmann Educational Books. Akoten, John E., Yasuyuki Credit Access and Its Impacts on Micro and Small Enterprises: The Case of Economic Development and Cultural Change 54 (4): 927 944. African Economic Development in a Cambridge Journal of Economics 25(3): 265 288. Management Science 53(5): 760 776. Journal of Theoretical Politics 13(3): 249 70.
274 African Perspe ctives of Development eds. Himmelstrand Ulf, Kinyanjui Kabiru and Mburugu Edward. New York: James Currey 107 117 Review of African Political Economy 22 (65): 301 320. Allu The Faces of the Italia n Christ ian Democratic Party Politics 3: 23 52. Urban and Regional Imbalances in Economic Development 17(1): 1 14. Cities 6: 202 208. banization and Growth, eds. Michael Spence, Patricia C. Annez, and Robert M. Buckley. Washington, D.C.: World Bank 1 46 Amenya, Gibson N. 20 Finance Foru m, University of Nairobi, Kenya 96. Anangwe, Amukow From the District Focus Strategy for Rural Development" In Perspectives on Local Government Management and Development in Africa. Durban Westville: Department of Public Administration, Univ ersity of Durban Westville. Azania: Archaeological Research in Africa 36 37 (1): 138 154. ntelism and Policy Change: The Bulletin of Latin American Research 30 (3): 298 312. Asabare, Paul K. and K. Owusu Ghanaian (African) Citie Journal of Urban Economics 14 (3): 370 379. to Credit By Small Research Paper: African Econ omic Research Consortium. www.aercafrica.org/documents/rp111.pdf
275 University World Institute for Develop ment Economics Research, Helsinki, Finland. Theory and Society 28: 97 334. Auyero, Javier. 2001. ival Networks and The Legacy of Evita Durham: Duke University Press. Environment & Urbanization 20 (1): 231 242. Bairoch, P. 1988. Cities and Economic Development: From Dawn of History to the Present Chicago: University of Chicago Press. Banerjee, Nirmala. 1985. Women Workers in the Unorganized Sector: The Calcutta Experience Hyderabad: Sangain Books. Banfield, Edward C. and J ames Q. Wilson. 1963. City Politics Cambridge: Harvard University Press. Institutional Fragility: The Political System in the Dominican Republic (1978 Latin America n Perspectives 39 (6): 198 211. World Institute for Development Economics Research, United Nations U niversity, Helsinki, Finland. http://www.wider.unu.edu/publications/working papers/2012/en_GB/wp2012 097/ Reforming Local Authorities for Better Service Delivery in Developing Countrie s: Lessons from RPRLGSP in Kenya eds. Tiberius Barasa and Wim Eising. Nairobi, Kenya: Institute of Policy Analysis and Research 45 60 Barasa, Tiberius, and Wim Eising, eds. 2010. Reforming Local Authorities for Better Service Delivery in Developing Coun tries: Lessons from RPRLGSP in Kenya Nairobi, Kenya: Institute of Policy Analysis and Research. Reforming Local Authorities for Better Service Delivery in Developing Coun tries: Lessons from RPRLGSP in Kenya eds. Tiberius Barasa and Wim Eising. Nairobi, Kenya: Institute of Policy Analysis and Research 203 212
276 Barbara, Grosh. 1987. Performance of Financial Parastatals in Kenya. Nairobi: Institute of Development Studies. B arkan, Joel D Elections in Independent Africa ed. Fred Hayward. Boul der and London: Westview Press, 213 237 In Governance and Politics in Africa eds. Goran Hyden and Michael Bratton. Boulder, CO: Lynne Rienner, 167 192. Barkan, Joel the Politics of Reallocation in Kenya." The Journal of Mod ern African Studies 27 (3): 431 453. Barkan, Joel, and R. Henderson. 1997. Toward credible and legitimate elections in Washington, DC: International Foundation for Election Systems. Local Level Politics: Social and Cultural Perspective eds. Marc J. Swartz. Chicago: Aldine 107 130 Experience and Lessons from Selected African Cou 04/174, International Monetary Fund, Washington, DC. www.imf.org/external/pubs/cat/longres.aspx?sk=17710.0 Journal of Southern African Studies 8(2): 187 213. Sweden International Development Cooperation Agency www.sida.se/publications. Beck, Linda. 2008. Brokering Democracy in Africa: The Rise of Clientelist D emocracy in Senegal New York: Palgrave Macmillan. Journal of African Economies 13(1): 1 43. Benit How does Local Democracy Help the Poor Access the State? Stories of Accountability and Clientelism in Journal of Asian and African Studies 46 (5): 453 464. Berg Comparative Politics 14 (4): 397 415.
277 Canadian Journal of African Studies 43 (3): 462 506. Bernard, H. Russe ll. 2006. Research Methods in Anthropology: Qualitative and Quantitative Approaches 4th ed. New York: Altamira Press. Berry, Sara. 1993. No Condition is Permanent Madison: The University of Wisconsin Press. Journal of Urban Economics 56(1): 80 96. Urban Studies 44(13): 2499 2510. Bigelow, R., Mbajan J. Cotter, E Mbajah, and P. Ondeng. 1987. Mid Term Evaluation of the Rur al Enterprise Program in Kenya USAID: Nairobi. The Sociological Review 59 (2): 348 370. Bird, K. and Ryan P. 1998. An Evaluation of DFID Support to the Kenya Enterprise Pr Evaluation Report EV: 605. Department for International Development, London. outh Africa A Case of the Economic Society of Sout h Africa, Durban, South Africa Journal of Political Economy 107(2): 252 284. Blau, P. 1964. Exchange and Power in Social Life New York: Wiley. Sociologishe Gids 16: 365 78. Blumer, Herbert. 1969. Symbolic Interactionism Englewood Cliffs, NJ: Prentice Hall. Blun ch, Niels The World Bank, Washington, DC. Royal Anthropological Institute of Great Britain and Ireland 1(1): 18 33.
278 Comparative Political Studies 36(4): 355 380. The Eastern Africa Social Science Research Review 19 (2): 1 21. American S ociological Review 52 (2): 224 239. Understanding the bureaucratic executive state, 1952 Review of African Political Economy 33 (107): 11 31. tizen Perceptions of Local Government Responsiveness in Sub World Development 40(3): 516 527. 09, Department of Economi cs, University of Connecticut, Storrs, Mansfield, CT. http://digitalcommons.uconn.edu/cgi/viewcontent.cgi?article=1165&context=econ _wpapers Bratton, Michael and Nicolas van de Walle. 1997. Democratic Experiments in Africa: Regime Transitions in Comparativ e Perspective Cambridge, England: Cambridge University Press. The American Economic Review 99 (2): 304 Brinkerhoff, Derick W., and Arthur A. Goldsmi th. 2002. Clientelism, Patrimonialism and Democratic Governance: An Overview and Framework for Assessment and Programming Bethesda, MD: Abt Associates. Agglomeration on Multifacto Journal of Economic Geography 8: 181 209. on Political Parti es and Legislative Organizations in Parliamentary Democracies, New Haven, CT, December 2002. Africa Today 46 (1): 95 117.
279 Microfinance in Africa: Is it Either the Problem or the World Development 25 (7): 1081 1093. The American Economic Re view 62(5): 842 853. Cain, Ferejohn, John Ferejohn and Morris Fiorina. 1987. The Personal Vote: Constituency Service and Electoral Independence Cambridge, MA: Harvard University Press. n the Argentine American Journal of Political Science 48 (4): 742 757. Camic, Charles, Philip Gorski, and David Trubek. 2005. Society: A Critical Companion (1st Edition) Palo Alto, CA: Stanford University Press Campbell, John K. 1964. Honour, Family and Patronage, A Study of Institutions and Moral Values in a Greek Mountain Community Oxford: Clarendon Press. the Democratic R Democratization 9 (3): 53 71. International Studies Quarterly 36(3): 245 270. turational European Journal of International Relations 6(4): 435 63. Finance and Development 22 (2): 24 27. Charmes, J. 2000. "The Contribution of Informal Sector to GDP in Developing Countries: Assessment, Estimates, Methods, Orientations for the Future", Paper presented at the Fourth Meeting of the Expert Group on Informal Sector Statistics (Delhi Group), Geneva, August 2000. Chao ive Analysis of Member Based Africa. New York: UN Development Program. Retrieved from http://www.undp.org/sum/microsave. ism in Ivorian The International Journal of African Historical Studies 22(4): 671 696.
280 World Development 17(3): 351 355. Comparative Politics 14 (2): 169 189. Cheema, G. S., ed. 1993. Urban Management: Policies and Innovations in Developing Countries Praeger, Westport. Review of African Political Economy 20: 74 88 SAIS Review 21(1): 1 10. Jour nal of Eastern African Studies 2 (2): 166 184. Journal of International Development 8(2): 179 193. Third World Quarterly 14: 423 438. World Development 32(1): 23 51. Cohen, Barney. 2004. World Development 32(1): 23 51. Projections, and Key Challeng Technology in Society 28: 63 80. and Industry Location Decisions: The Impacts of Spatial and Industrial Regional Science and Urban Econom ics 35(3): 215 237. Cohen, Ronald. 1966. "The Dynamics of Feudalism in Bornu." Boston University Papers on Africa, Vol. II Boston: Boston University Press. Political Science: The State of the Discipline II ed. Ada W. Finifter. Washington, D.C: American Political Science Association 105 19
281 Oxfor d. http://www.csae.ox.ac.uk/workingpapers/pdfs/2008 16text.pdf Political Parties and Democracy eds. Larry Diamond and Richard Gunther, 173 205. Baltimore: Johns Hopkins Universit y Press. Coquery African Studies Review 34(1): 1 98. Frien ds, Followers, and Factions: A Reader in Political Clientelism ed. Steffen W. Schmidt. Berkeley: University of California Press 337 353 Craig, Kim and Goodwin Development: MicroSave Africa and the Study No. 8. Consultative Group to Assist the Poor: The World Bank, Washington, DC. Of Local Public Admini stration and Development 23: 77 88. Italian Politics and Society 60: 2 5. Da Cruz, Fernando, Kerstin Sommer, and Ombretta Tempra. 2006. Nairobi Urban Sector Profile. Nairobi: U nited Nations Human Settlements Programme (UN HABITAT). 98, Development Studies Institute, Lon don School of Economics and Political Science, Houghton Street London. www.isn.ethz.ch/isn/Digital Library/Publications/Detail. Journal of Urb an Economics 53 (1): 98 125. Economic Development and Cultural Change 3(1): 6 26. and National World Development 27(1): 55 65.
282 rch Paper No. 26. Kenya Rural Enterprise Programme, Nairobi. Financial Development and Economic Growth: Explaining the Links ed. C. Goodhart Basingstoke: Macmillan/Palgrave. http ://discovery.ucl.ac.uk/15036/ Dege Consult. 2007. Local Level Service Delivery, Decentralization And Governance: A Comparative Study Of Uganda, Kenya And Tanzania Education, Health And Agriculture Sectors Denmark: Nordic Consulting Group. Devarajan, Sha 036, John F. Kennedy School of Government, Harvard University. web.hks.harvard.edu/publications/getFile.aspx?Id=723 Devas, Making: Citizen Participation and Local Accountability. Examples of Good and Bad Practice in Department, United Kingdom.www. bham.ac.uk/IDD. Making Citizen Public Administration and Development 23: 307 316. Devereux, Michael P., Rach Journal of Public Economics 91(3): 413 435. Diaz Public Good s as Electoral Investments: A Portfolio Diversification Model of Policy Politician Linkages in Democratic Politics, Duke University, March 30 April 1 2001. rminants of Success of Special Journal of Politics 58(4): 1132 55. Douglas, Jack. 1970. Understanding Everyday Life Chicago: Aldine. to Generate Better Theory: A Rejoinder to Eisenhardt Academy of Management Review 16(3): 613 619. Earth Institute. 2007. Invest in Kenya: Focus Kisumu New York: The Earth Institute at Columbia University.
283 Eckstein, H. 1975. Case Study and Theory in Political Science. In Handbook of political science Vol. 7 eds. F. J. Greenstein and N. W. Polsby. Reading, MA: Addison Wesley 79 137 The Academy of Management Review 14(4): 532 550. Eisenhardt, K. The Academy of Management Review 16(3): 620 627. Eisenstadt, Shmuel, and Ren Lemerchand, eds. 1981. Political Clientelism, Patronage and D evelopment London: Sage. Client Relations and a Model o Comparative Studies in Society and History 22(1): 42 77. Eisenstadt, Shmuel, and Luis Roniger. 1984. Patrons, Clients and Friends: Interpersonal Relations and the Structure of Trust in Society Cambridge: Cambridge University Press. Reforming Local Authorities for Better Service Delivery in Developing Countries: Lessons from RPRLGSP in Kenya eds. Tiberius Barasa and Wim Eising. Nairobi, Kenya: Institute of Policy Analysis and Research 63 100 litions of Convenience and Commitment: Political Global and Area Studies, Germany. http://repec.giga hamburg.de/pdf/giga_08_wp68_elischer.pdf. Estevez, Federico, Beatriz Magal oni and Alberto Diaz Diversification Model of Policy Choice. Paper presented at the Conference on Stanford University, May 2002. onal Development and Rural Urban Policy: Past Experience Urban Studies 26: 253 266. African Studies Review 29(4): 61 78. Fay, Marianne and Charlotte Opal. 20 Development and Transportation Division, The World bank, Washington D.C.
284 Theory The Info rmal Sector Revisited eds. D. Turnham, B. Salome, and A. Schwartz, 49 69. Paris: OECD. Qualitative Inquiry 12(2): 219 245. llenges to Party Development and Democratic Consolidation: Journal of Asian and African Studies 46 (6): 578 592. Fowler, Alan. 1989. New Scramble for Africa: Non Governmental Development Organiz ations and New Donors in Kenya Nairobi: The Ford Foundation. Fowler, Alan and Kinyanjui Kabiru. 2004. Indigenizing Foreign Seed on African Soil, the story of K REP Nairobi: Kenya Rural Enterprise Programme. n from Clientelism to Citizenship: Lessons World Politics 46(2): 151 184. The Journal of Modern African Studies 34 (4): 597 607. Francis, E Multiple livelihoods: Cases from Eastern and The Journal of Development Studies 38(5): 167 190. Oxford Economic Papers 20(2): 250 274. Freedom House. 2012. Free dom in the World Comparative and Historical Data. www.freedomhouse.org. Accessed May 24, 2012. Regional Science and Urban Economics 12 : 161 96. No. 56. Development Research Institute, New York University, New York. http://nyudri.f iles.wordpress.com/2011/10/driwp56.pdf. The Journal of African History 14 (2): 275 290. Garcia Latin American Perspectives 29 (5): 90 109.
285 International Journal of Urban and Regional Research 14 (4): 648 666. American Studi es Association, Chicago Citizenship: Exchanges, Gifts & Rights in Politician Linkages in Democratic Politics, Duke University, March 30 April 1 2001. Patrons an d Clients in Mediterranean Societies eds. Ernest Gellner and John Waterbury London: Duckworth 1 7 Gellner, Ernest and John Waterbury, eds. 1977. Patrons and Clients in Mediterranean Societies London: Duckworth. George, Alexander L. and Andrew Bennet t. 2005. Case Studies and Theory Development in the Social Sciences Cambridge, MA: MIT Press. American Political Science Review 98(2): 341 354. Gerring, John. 2007. Case study research Principles and practices New York, NY: Cambridge University Press. Gertzel, Cherry J. 1970. The Politics of Independent Kenya, 1963 8 Nairobi, Kenya: East African Publishing House. Gertzel, Cherry, M. Goldschmidt and D. Rothchild, eds.1969. Governmen t and Politics in Ke nya: A Nation Building Text Nairobi: East African Publishing House. Public Choice 120: 267 300. Gibson, Clark C., and Jame Electoral Studies xxx: 1 6. World Politics 49(April):339 70. Giddens, Anthony. 1976. New Rules of Sociological Method: A Positive Critique of Interpretive Sociologies London: Hutchinson. Giddens, Anthony. 1979. Central Problems in Social Theory London: Macmillan.
286 Giddens, Anthony.1984. The Constitution of Societ y: Outline of the Theory of Structuration Berkeley and Los Angeles: University of California Press. Gilbert, Alan. and Josef Gugler. 1992. Cities, Poverty and Development: Urbanization in the Third World, 2nd ed Oxford: Oxford University Press. Glaeser Edward L., Hedi D. Kallal, Jose A. Scheinkman, and Andre Shleifer. 1991. Journal of Political Economy 100 (December): 1126 52. Goffman, Erving. 1974. Frame Analysis New York: Harper and Row. American Sociological Review 48: 1 17. Paper presented at the ANZSYS07 Conference. Auckland, New Zealand, December 2007. Scale Small Business Advancement National Center, University of Central Arkansas. http: //sbaer.uca.edu/research/usasbe/1996/pdf/23.pdf Accessed 06/12/12. Scale Journal of Small Business Management 35(1): 66 72. European Journal of Political Research 4: 149 74. Studies in Comparative International Development 45: 83 103. Greenstone, Da Comparative Studies in Society and History 8 (2): 199 210. Grzymala F Comparative Political Studies 41: 638 673. Gulyani, S., Talukdar, D. and Potter, C. 2006. Inside Informality: Poverty, Jobs, Housing and Services in Nairobi Slums. Report No. 36347 KE, Africa Water and Urban Unit 1. Washington, D.C: World Bank Gunther, Richard, P. Nikiforos Diamandouros, and Hans Journal of Democracy 7: 151 159.
287 Hagopian, Frances. 1996. Traditional Politics and Regime Change in Brazil Cambridge and New York: Cambridge Un iversity Press. Sized Cities on Population Regional Development Dialogue 11(Spring): 60 76. Operation Habitat International 21 (1): 65 77. Harriss IDS Bulletin 27(2): 31 19. Hart, Kei Journal of Modern African Studies 11(1): 61 89. Political Studies XLVI: 951 957. Hays, Shar Sociological Theory 12(10: 57 72. The Review of Economics and Statis tics 86(4): 959 972. Heiberger, Richard. M., and Holland, Burt. 2004. Statistical Analysis and Data Display: An Intermediate Course with Examples in S Plus, R, and SAS Springer. http://springeronline.com/0 387 40270 5. ffects of Urban Concentration on Economic Cambridge, MA. http://www.nber.org/papers/w7503. The World Bank Research Obse rver 17(1): 89 112. So Journal of Economic Growth 8(1): 47 71. The Role of Ins Regional Science and Urban Economics 37: 283 313. wide Governance Networks in Nairobi: Towards Contributions to Political Rights, Influence and Service Delivery for Poor and Middle Habitat Internat ional 34: 59 77.
288 Herbert, H. Werlin., 1974. Governing an African City: A Study of Nairobi New York: African Publishing Company. Studies in Ethnicity an d Nationalism 11 (3): 527 537. Hildebrand, Mark, and Brian Holst. 2010. A Rapid Urban Diagnostic and Proposed Intervention Strategy for DIG in Mombasa, Kenya Bethesda, MD: Development Innovations Group. tretching: Differentiating Among Theory and Society 40(5): 567 588. World Development 22(3): 469 482. Himbara, David. 1994b. Keny an Capitalists, the State, and Development Nairobi: East African Educational Publishers. Credit Markets World Bank Economic Revi ew 4(3): 235 250. World Affairs 150(1): 21 30. Africa 54 (3): 72 91. Holmquist, F. W., Frederick S. W., and Michael D African Studies Review 37(1): 69 105. Journal of Asian and African Studies 33(4): 345 358. eptualizing Political Clientelism: Political Exchange and Political Science Association. Philadelphia, PA, August 31 September 3, 2006. Economic Development in the Third Cities February: 41 57. African Journal of Economic and Sustainable Development 1 (1): 4 26.
289 tion and The Journal of Developing Areas 21(4): 383 402. Latin American Review 32 (3): 209 223. icro Finance Programmes in Kenya as Supported through the Dutch Co Financing finalreport.pdf. Accessed December 2009. E conomic Development and Cultural Change I32(2): 277 302. House History Compass 7(6): 1444 1458. Te Urban Forum 20: 271 292. Hulme, D., John Kashangaki and Harry Mugwanga 1999. Dropouts Amongst Kenyan Microfinance Institutions Nairobi: MicroSave. Huntington, Samuel P. 1968. Political order in changing socie ties New Haven, CT: Yale University Press. Huriot, Jean Marie and Jacques Francois Thisse. 2000. Economics of Cities: Theoretical Perspectives Cambridge MA : Cambridge University Press. gle Party Systems: British Journal of Political Science 2 (4): 389 420. Representation and Accountability. NORD SUD aktuell.www.ethno. unizh.ch/csfconference/files/abstracts/Hyden_Abstract.pdf. Performance? Clientelism and Intra American Journal of Political Science (Forthcoming). Ikiara Scale Enterprise Sector: The of the Network of the African Scientific Organization, Nairobi, October 1989.
290 Ikiara, Gerrisho on Small and Micro Enterprise Promotion in Changing Policy Environment, The Hague, Netherlands, September 30 October 2, 1991. Polity 32(1): 45 66. Infotrack East Africa. 2008. A Synopsis of the Role of Members of Parliament in National Leadership, Governance and Development in Kenya. Nairobi Kenya: Infotrack East Africa Ltd. www.infotrackea.co.ke/articles/index.php=article&id=77 Nairobi, Kenya: Institute of Economic Affairs. ww.ieakenya.or.ke/document s/The%20Point%2054.pdf (April 12, 2012). International Labor Organization. 2007. Informal Sector in Africa Geneva: International Labor Office. International Labor Organization. 1972. Employment, Incomes, and Equality: A Strategy for increasing Productiv e Employment in Kenya Geneva: International Labor Office. International Trade in East Asia, NBER East Asia Seminar on Economics Vol. 14 eds. Tak atoshi Ito and Andrew K. Rose. Chicago: University of Chicago Press 7 52 Comparative Politics 16 (4): 421 442. to Collective Action: Looking for Agency in Social An International Journal 9(1): 1 16. European Political Science Association, Berlin, Germany, June 2012. The American Political Science Review 73(2): 385 398. Jones, Charles. I. 2001., Introduction to Economic Growth 2nd ed New York: Norton. International Studies Quarterly 55: 151 175.
291 Habitat International 31: 87 99 Enterprises in Technovation 19: 117 126. UNHABITAT, Nairobi, Kenya, November 2006. nformal Settlements in Kisumu, Kenya, Implemented by t Environment and Urbanization 22(1): 217 239. Kenya Institute for Public Policy Research and Anal ysis, Nairobi, Kenya. East African Alternatives 3: 48 50. Client Concept and Macro Politics: Prospects Comparative Studies in Society and History 16(1): 284 308 Kauzya, John Department of Economic and Social Affairs. Kawata, Junichi, ed. 2006. Comparing Political Corr uption and Clientelism Burlington, VT: Ashgate. Nairobi, Kenya, September 2010. American Journal of Political Science 51 (4): 804 821. Keefer, Philip. 2009. The Ethnicity Distraction: Political Credibility Clientelism and Partisan Preferences in Africa Washington, DC.: World Bank. Economic and Political Weekly 39(9): 935 43. Journal of Law, Economics, and Organization 24 (2): 371 406.
292 Evaluation Quarterly 3(4): 661 678. http://www.knbs.or.ke/ accessed March 7, 2008. www.knbs.or.ke/Census%20Results/KNBS%20Brochure.pdf. http://www.knbs.or.ke/ (accessed March 7, 2010). http://www.knbs.or.ke/ accessed April 18, 2011. Kenya National Bureau of Statistics. 2010 Economic Survey Re ports, 1975 2010 Nairobi, Kenya. Kenya Open Data. 2012. Census Volume 1 Question 1 Population, Households and Density by Sublocations 2009 Nairobi, Kenya. https://opendata.go.ke/Population/Census Volume 1 Question 1 Population Households a/wd27 eki2 Saharan Africa: Implications for Economi Working Paper No. 97. Urban Development Unit, The World Bank, Washington D.C. Kessinger, Tom G. 1974. Vilyatpur: 1848 1968. B erkeley: University of California Press. The Journal of Interdisciplinary History 18 (3): 419 447. Kimenyi, Mwangi. 2004. Achieving Universal Primary School Education in Keny a. Nairobi: Kenya Institute of Public Policy Research and Analysis. Research and Analysis, Socia l Sector Division, Nairobi, Kenya. Journal of International Educational Development 16 (4): 417 426.
293 SAIS Review XXI 1(Winter Spring): 97 108. King, K. and Abuodha, C. 1991. The Building of an Industrial Society: Change and Development in Kenya's Informal (Jua Kali) Sector, 1972 to 1991 Nairobi, Kenya: University of Nairobi. Development, Geneva, Switzerland Kitchin, Garvin. 1980. Class and Economic Change in Kenya: The Making of an African Petite Bourgeoisie. London: Yale University Press. Level Planning Process to Address the Impacts of Climat Cities 27: 466 475. Comparative Political Studies 33: 845 879. Kitschelt, Herbert, Kirk A. Hawkins, Juan Pablo Luna, Guillermo Rosas and Elizabeth J. Zechmeister. 2010. Latin American Party Systems New York: Cambridge University Press. Comparative Political Studies XX (X): 1 32. Kitschelt, Herbert, Zdenka Mansfeldova, Radoslaw Markowski and Gabor Toka. 1999. Post Communist Party Systems: Competition, Representation, and Inter Party Cooperation. Cambridge: Cambridge University Press Kitschelt, Herbert and Steven Wilkinson. 2007. Patrons, Clients, and Policies: Patterns of Democratic Accountability and Political Competition Cambridge, UK: Cambridge University Press. Patronage And Resistance In The Multi http://www.columbia.edu/~jk2002/publications/klopp_phdthesis.pdf Structuralism and Comparing Political Corruption and Clientelism eds. Kawata, Junichi. Burlington, VT: Ashgate 1 22
294 ISOCARP Congress, September 2010. Finance Institutions in Kenya: Client Profiles and University. tructure and Agency: Outline of Mobilization: An International Journal 10(1): 19 33. Local Government Power Relations in Mediating Space in Ath West Virginia University. https://eidr.wvu.edu/eidr/documentdata.eIDR?documentid=1319. KPMG. 2008. Kisumu, Kenya Potential Opportunities for Investors United Kingdom: KPMG. Krugman, Paul, and Raul L. Elizondo. 1996. Journal of Development Economics 49: 137 150. Public Choice 77: 259 273. Lamb, G. 1974. Peasant Politics: Conflict and Developm Lewes: Julian Friedmann. International Political Science Review 4 (4): 435 454. Land, C. ollowers and Factions: A Reader in Political Clientelism, edited by Steffen Schmidt, James C. Scott, Car Lande, and Laura Guasti, xiii xxxvii. Berkeley: University of California Press. Landini, F a Clientelist Rural Journal of Community and Applied Social Psychology (Forthcoming). Journal of Modern African Studies 18: 657 673. Legovini, A. 2002. Kenya: Macro Economic Evolution since Independence. Nairobi, Kenya: United Nations Development Programme. http://www.ke.undp.org/kenya%20macro%20economic%20evolution%20since% 20independence.doc. Accessed April 16, 2012.
295 sm and Ethnicity in Tropical Africa: Competing Solidarities in Nation American Political Science Review 66 (1): 68 90. Comparative Politics 4 (2): 149 178. Labor Based Party Adaptation in Argentina (1983 Department of Political Science, University of California, Berkeley. Levitsky, Steve Unionization of Argentine Peronism and the Transformation of Working Class Politician Linkages in Democratic Politics, Duke University, March 30 April 1 2001. Levitsky, Steven. 2003. Transforming Labor Based Parties in Latin America: Argentine Peronism in Comparative Perspective New York: Cambridge University Press. l Transition in Africa: The Quest World Politics 49: 92 129. RURDS 10 (2): 142 156. Lewis, Paul Politics 22 (1): 17 23. The American Political Science Review 65 (3): 682 693. Lindberg, S. I. Neopatrimonialism Rather than Counter Democratization 14 (2): 121 140. Clientelistic?: Survey Ev Political Science Quarterly 123 (1): 95 122. Journal of Modern African Studies 48 (1): 117 142. Lindberg, Party Politics 1 17. http://ppq.sagepub.com doi:10.1177/1354068811436030
296 World Development 19 ( 6): 651 670. Local Authorities Transfer Fund. 2010. Annual Report and Review of Local Authority Financial Performance, 2002 2009 Nairobi, Kenya. Long, J. Scott. 1997. Regression Models for Categorical and Limited Dependent Variables Thousand Oaks, Cali fornia: SAGE Publications. Long, J. Scott, and Jeremy Freese. 2006. Regression Models for Categorical Dependent Variables Using Stata College Station, Texas: Stata Press. Lubell, Harold. 1991. The Informal Sector in the 1980s and 1990s Paris: OECD. Lu September 2004. 2010 : Democratization 18 (2): 275 310. Urban Studies 42 (7): 1077 1097. Urban Edge 1 (3): 3 3. Journal of Modern African Studies 30 (2): 221 236. Working paper No. 25. French Institute for Research in Africa, Nairobi, Kenya. Macharia, Kinuthia. 1997. Social and Political Dynamics of the Informal Economy in African Cities New York: University Press of America. Manzetti, Luigi and Caro Comparative Political Studies 40(8): 949 970. Studies in Comparative Inter national Development 34 (2): 3 30. Mainwaring, Scott. 1999. Rethinking Party Systems in the Third Wave of Democratization: The Case of Brazil Stanford CA : Stanford University Press. Cahiers d'ttudes Afric aines II (6): 315 26.
297 Conference on Citizen Politician Linkages in Democ ratic Politics, Duke University Comparative Political Studies 18: 487 518. Cities 15 (6): 463 471. The Political Economy of Kenya e d. M. Schatzberg. New York: Praeger. Wasunna Owino, 183 218 Nairobi: Institute of Policy Analysis and Research. World Development 25 (7): 1095 1110. loyment Growth World Development 22,(12): 1881 1894. World Development 26,(1): 61 74. risis, Informalization and the Urban Informal Sector in Sub Development and Change 26: 259 284. Paper presented at the conference Informal Institutions in Latin America, Universit y of Notre Dame, IN Conflict Talk ed. Allen D. Grimshaw, 160 170. Cambridge, UK: Cambridge University Press. Menon, Balakrishna, James Mutero, and Simon Machar 32, International Studies Program, Andrew Young School of Policy Studies at Georgia State University, Atlanta, Georgia. http://aysps.gsu.edu/isp/files/ispwp0832.pdf Michie, Barr Client Relations: 40. Microfinance Information Exchange. 2012. www.mixmarket.org/profiles reports.
29 8 y: The Case of Sarawak and Sabah in Asian Survey 13(10): 891 907. Miro, Carmen A. and Joseph E. Potter. 1980. Population Policy: Research Priorities in the Developing World in Kenya: The Contribution of Research in Symposium on Urban Development for Economic Growth and Poverty Reduction, Washington D.C., December 2003. y Governance for Poverty Alleviation in Local Regional Development Dialogue 25 (1): 88 105. Mitullah, Winne V., Hamisi O. Mboga, Rahab Mundara, Edward Cherono, Augustine Odipo, Elijah Agevi, and J. P. Mbugua. Democracy at the local level in East and Southern Africa: Profiles in Governance Stockholm, Sweden: International Institute for Democracy and Electoral Assistance. dustrial Growth in Coastal China: Economic World Bank Economic Review 11 (2): 293 325. he Role of Women in Africa, Canada, October 2005. Science 319: 761 764. Agglomeration Economies? Regional Science & Urban Economics 13: 525 545. Moomaw, R. Journal of Urban Economics 40: 13 37. Comparative Studies in Society and History IV(4): 473 94. Bank in Kenya, 1980 African Affairs 85(338): 107 119. Moyi, Eliud and Peter Njir Policy Research and Analysis, Productive Sector Division, Nairobi, Kenya.
299 Moyi, Eliud and Eric Ronge. 2006. Taxation a nd Tax Modernization in Kenya: A Diagnosis of Performance and Options for Further Reform Nairobi: Institute of Economic Affairs. Journal of Modern African Studies 22 (3): 399 427. Journal of Contemporary African Studies 29 (1): 99 117. Muia, Daniel M. 2006. Local Government Reforms in Kenya: A Study on Civic Engagement, Inc lusivity and Gender Mainstreaming in the LASDAP Process in Selected Local Authorities Nairobi Kenya: Action Aid Decentralization and Devolution in Kenya: Ne w Approaches eds. Thomas N. Kibua and Germano M. Mwabu Nairobi: University of Nairobi Press 77 102 Reforming Local Authorities for Better Service Delivery in De veloping Countries: Lessons from RPRLGSP in Kenya eds. Tiberius Barasa and Wim Eising. Nairobi, Kenya: Institute of Policy Analysis and Research 11 28 Institutional, Legal a Reforming Local Authorities for Better Service Delivery in Developing Countries: Lessons from RPRLGSP in Kenya eds. Tiberius Barasa and Wim Eising. Nairobi, Kenya: Institute of Policy Analysis and Research 29 44 Muia, D. Reforming Local Authorities for Better Service Delivery in Developing Countries: Lessons from RPRLGSP in Kenya eds. Tiberius Barasa and Wim Eisi ng. Nairobi, Kenya: Institute of Policy Analysis and Research 175 200 Labour Office. Munc k, Geraldo. L., and Jay Verkuilen. 2005. Research designs. Encyclopedia of Social Measurement 3: 385 395. Land Degradation & Development 17: 97 108.
300 workshop on Neopatrimonialism in Various World Regions, GIGA German Institute of Global and Area Studies. Hamburg, Germa ny Municipal Council of Mombasa. 2009. Strat egic Management Plan: July 2009 June 2012 Mombasa, Kenya: Municipal Council of Mombasa. scale Enterprises in Nairobi, Habitat International 30(1): 127 143. Murumba, C. K., 1987. Sharing Re sponsibility and Resources for Effective Central Local Relationships in Kenya. Planning and Administration 14(1): 100 109. Beyond State Failure and Collapse eds. George Klay Kieh, Jr New York: Rowman and Littlefield 115 150 Destruction: The Journal of Contemporary African Studies 24 (1): 1 28. Musingi, John K., Shadr Growth Of Mombasa Coastal Town and Its Implication for Surface and Impacts of Urban Growth on Surface Water and Groundwater Quality ed. Bryan E. Wallingford, UK: IAHS Press. Mutua, Research Paper No. 2010 007, Buffalo Legal Studies Research Paper Series, University of Buffalo Law School, New York. http://ssrn.com/abstract=1533530. Journal of Business and Public Affairs 30 (1): 50 56. The Jour nal of Modern African Studies 46 (2): 267 285. Activities: The K Hotel, November 1996. Mwenda, Albert K. 201 Research Paper No. 24, Institute of Economic Affairs, Nairobi, Kenya. Africa n Affairs 104/416: 449 467.
301 Review of African Political Economy 30 (96): 343 350. ion American Polit ical Science Review 91 (3): 599 616. Cities in the Developing World ed. Gugler, J. Oxford: Oxford University Press. cro and Small Enterprises (MSEs) in Sub Contemporary Microenterprise: Concepts and Cases ed. Mark S. Munoz. Northampton, MA: Edward Elgar Publishing 87 101 elational Clientelism? Healthcare and Political Science Association in Seattle, Washington, September 2011. nt Finance Institute for Public Policy Research and Analysis, Private Sector Development Division. Massachusetts: Edward Elgar Publishing Limited. Njeru, E. H. N., and Njoka, J Negotiating Social Space: East African Micro Enterprises eds. P. O. Alila, and P. O. Pedersen. Trenton, NJ: Africa World Press 141 174 Saharan Africa. Cities 20 (3): 167 174. Rep Scaling Up Poverty Reduction: Case Studies in Microfinance Proceeding s of Conference on Global Learning Process for Scaling up Poverty Reduction, sponsored by the Con sultative Group to Assist the Poor, World Bank, Shanghai, China, May 2004. Journal of Democracy 7(April): 34 51. Urban Economics New York: McGraw Hill. Obudho R.A.19 The Urban Challenge in Africa: Growth and Management of its Large Cities ed. Carole Rakodi. Tokyo: United Nations University Press 292 334
302 in Spatial Planning in International Conference on Urban Growth and Spatial Planning of Nairobi, Kenya, December 1988. Social Economic http://ssrn.com/abstract=2064178 or http://dx.doi.org/10.2139/ssrn.2064178. Kenya 1940 93, edited b y Bethwell A. Ogot & William R. Ochieng, xi 1. Oxford: James Currey. International Studies Perspectives 2: 161 176. Okoth, Rose, and A. Odipo. 2005. Local Governance And Povert y Alleviation In Kenya: Lessons of experience from Local Government Reforms Nairobi, Kenya: Kenya Local Government Reform Programme. the World Bank Urban Research Symposium, Washington D.C., December 2003. Clientelism in East Party Politics 1 18: doi: 10.1177/1354068811436039 Omobowale, Ayo Afrika Spectrum 43 (2): 203 224. African Studies 61 (2): 209 22 1. Emerging Metropolitan Region: Development 46th ISOCARP Congress, Nairobi, Kenya. Some New Policy Awareness for University Administrators and Trainers, Thika, Kenya, February 1992. rn The International Journal of Press/Politics 17 (4): 497 515.
303 GeoJournal 62: 117 128. Ouma, S.J. 1990. The Transformation of the Informal Sect or: The Co operative Perspective Nairobi: Shirikon Publishers. GeoJournal 71: 143 157. eting at the Grassroots Decentralization and Devolution in Kenya: New Approaches eds. Kibua, Thomas N. and Germano Mwabu. Nairobi: University of Nairobi Press, 199 234. Oyugi, Walter O. Local Government in the Third World: The Experience of Tropical Africa ed. Mawhood, P. New York: John Wiley and Sons 107 140 Politics and Administration in East Africa ed. Walter Oyugi. Nairobi: Konrad Adenauer Foundation 153 191 Parker, Joan C., and Tanya R. Torres. 1994. Micro and Small Scale Enterprises in Kenya: Results of the 1993 National Basel ine Survey Bethesda, MD: Development Alternatives. Research Group on International Agreements a nd Development, Netherlands. http://www.hiva.be/docs/paper/P23_lirgiad wp7.pdf International Review of Economics 55: 209 227. an Micro Enterprises: The Negotiation of the Their Labour, Gender, and Rural Development Economics and Social Justice: Essays in Honour of Ian Livingstone eds. Michael Tribe, John Thoburn, and Richard Palmer Jones. Burlington, VT: Ash gate Publishing Company, 208 219 Journal of Modern African Studies 37 (1): 109 135. tic Parties: A Experiments, University of Cape Town, South Africa, December 2011.
304 Penfold Latin American Politics and Society 49(4): 63 84. Polity 44: 625 642. of Korean and Japanese Asian Journal of Political Science 16(2): 111 129. International Review of Economics 55: 209 227. Persson Torsten, Guido Tab ellini, Francesco Trebbi. 2003. Electoral rules and corruption. Journal of the European Economic Association 1(4): 958 89. Political Studies 45(2): 436 462. Piattoni, Simona, ed. 2001a. Clientelism, Inter ests, and Democratic Representation Cambridge: Cambridge University Press. Clientelism, Interests, and Democratic Representation ed. Simona Piattoni. Cambridge: Cambrid ge University Press 1 30 Pitt Rivers, J. 1954. The People of the Sierra London, Weidenfeld and Nicholson. Journal of Development Economics 96 (2): 461 475. Development Express 4: 4 8. Portes, Alejandro and Saskia Sasen American Journal of Soci ology 93 (1): 30 61. Saharan African Urban 18: London School of Economics, London, United Kingdom. The Urban Informal Sector in Sub Saharan Africa: From Ban to Development Southern Africa 25(2): 151 167. ociety and Clientelist Politics. American Political Science Review 64: 411 425.
305 Pyke, F., and Sengenberger, W. 1992. Industrial Districts and Local economic Regeneration Geneva: International Institute for Labor Studies, International Labour Organization. Rakodi, Carole, Rose Gatabaki Con flict Environment and Urbanization 12 (1): 153 170. Renaud, B. 1981. National Urbanization Policy in Developing Countries Oxford: Oxford University Press. Republic of Kenya. 1986. Sessional Paper No. 1 of 1986 on Economic Management for Renewed Growth Nairobi: Government Printer. Republic of Kenya. 1989a. Development Plan 1989 1993 Nairobi, Kenya: Government Printer. Republic of Kenya. 1989b. A Strategy for Small Enterprise Development in Kenya: Towards the Year 2000 Nairobi, Kenya: Governme nt Printer. Republic of Kenya. 1991. Statistical Abstract Nairobi, Kenya: Government Printer. Republic of Kenya. 1992. Sessional Paper No. 2 of 1992 on Small Enterprises and Jua Kali Development Nairobi, Kenya: Ministry of Planning and National Develo pment. Republic of Kenya. 2002. National Development Plan 2002 2008 Nairobi, Kenya: Government Printer. Republic of Kenya. 2003a. Economic Recovery Strategy for Wealth and Employment Creation 2003 2007 Nairobi, Kenya: Ministry of Economic Planning and National Development. Republic of Kenya. 2003b. Report by the Task Force on Micro and Small Scale Enterprises Nairobi, Kenya: Ministry of Labour and Human Resource Development. Republic of Kenya. 2009. Directorate of Rural Planning Overview Nairobi, Ke nya: Ministry of State for Planning. Comparative Political Studies XX (X): 1 28. Rhyne, Elisabeth and Robert Peck Christen. 1999. Microfinance Enters t he Marketpla ce Washington, DC : USAID.
306 Saharan Africa: Its Contribution to Forest Management, Livelihoods, and Environmental Conservation 37(1): 35 44. Richardson, Ha The Journal of Developing Areas 15(1): 97 118. Working Paper 3205, Public Policy. Center for Economic Policy Researc h, Washington, D Corruption and Development ed. Mark Robinson. New York: Taylor and Francis 1 15 T he Experience of Post Habitat International 23(4): 511 534. Habitat International 25: 115 142. Ronge, Eric, L Discussion Paper No. 20/2002, The Kenya Institute for Public Policy Research and Analysis, Nairobi, Kenya. Roniger, L uis. 1990. Hierarchy and Trust in Modern Mexico and Brazil New York: Prager. International Journal of Comparative Sociology 35 (3/4): 207 220. Clientelism and the Changing Nature of Ayata, 1 18. Boulder, CO: Lynne Rienner. Comparative Politics 36(3): 353 375. Journal of African Media Studies 1(1): 155 171. mong the Secondary Nairobi, Kenya.
307 Journal of Social Development in Africa 17(1) : 81 98. Development in Kenya: K Harvard Institute for Interna tional Development, Harvard University. Rosenthal, Stuart S., and William C. S The Review of Economics and Statistics 85(2): 377 393. ht tp://www.ruralpovertyportal.org/web/guest/country/home/tags/kenya. Accessed June 3, 2012. Canadian Journal of Political Science 5 (1): 104 119. Sch Elections for Sale: The Causes and Consequences of Vote Buying ed I. Schaffer,1 16. Boulder: Lynne Rienner. Comparati ve Politics 6 (3): 425 450. Client Politics and Political Change in Southea st Asia. American Political Science Review 65(1): 91 114. Scott, Leg Friends, Followers, and Factions: A Reader in Political Clientelism ed. Steffen W. Schmidt. Berkeley: University of California Press 439 458 Selection Techniques in Case Study Political Research Quarterly 61(2): 294 308. International Labour Review 116(3): 343 352. Sewell, Journal of Sociology 98(1): 1 29. Urban Affairs Quarterly 21: 47 53.
308 hange Perceived in Relation Climatic Change 85:271 284. disciplinary Politics 19(3): 139 144. Silverman, Sydel Ethnology 4(2): 172 90. Theory and Society 18(2):181 211. ormalizing the Informal Sector in a Changing South Africa: Small World Developmen t 20 (7): 1092 1045. Clientelism and Corruption. Association Meeting, Toronto, ON, Canada. Smith, Michael G. 1960. Government in Zazzau, 1800 1950 London: Oxford University Press. ries: World Development 21 (6): 901 903. Smoke, Paul., 1994. Local Government Finance in Developing countries: The case of Kenya Nairobi, Oxford University Press. L ocal Governance in Africa: The Challenges of Democratic Decentralization eds. Dele Olowu and James S. Wunch. Colorado: Lynn Reinner 211 234 Multi Africa n Affairs 95: 501 527. 1895 African Studies Review 29 (4): 17 42 Steen, J., Catelijne C. and J ennifer W Network Theory and Str Strategic Organization 4(3): 303 312. The Journal of Commonwealth & Comparative Politics 35 (3): 2 7 52.
309 Stokes, Susan C. 1995. Cultures in Conflict: Social Movements and the State in Peru Berkeley: University of California Press. Stokes, S. American Political Science Review 99 (3): 315 325. The Oxford Handbook of Comparative Politics eds. Susan Stokes and Carles Boix. New York: Oxford University Press 604 627 Stone, Clarence N. 1989. Regime Po litics: Governing Atlanta, 1946 1988 Lawrence: University Press of Kansas. Canadian Journal of African Studies 1 (Winter): 33 56. Stren, Richard. E., 1978. Housing the Urban Poor: Policy, Politics, and Bureaucracy in Mombasa Institute of International Studies, University of California, Research Series No. 34, Berkeley. Stren, Richard. and Gombay, C. 1994. The Alleviation of Urban Poverty at the Municipal Level: A Rev iew of the Literature Nairobi, Kenya: UN HABITAT. Stren, Richard E., and Rodney R. White, eds. 1989. African Cities in crises: Managing rapid Urban growth London: West Views Press. reativity and Journal of Economic Geography 9: 147 167. Syagga, P., Mitulla W. and Karirah Gitau S. 2001. Nairobi Situation Analysis Consultative Report Nairobi: Collaborative Nairobi Slum Upgrading Initiative, Government of Kenya and Unit ed Nations Centre for Human Settlements (UN HABITAT). Review of African Political Economy 27 (85): 427 441. Case Social Networks 34: 230 240. African Affairs 77 (308): 297 320.
310 Tarrow, Sidney. 1967. Peasant Communism in Southern Italy New Haven: Yale University Press. Bulletin of Latin American Research 23 (2): 213 227. Journal of the International African Institute 63 (3): 371 396. Throup, David, and Charles Hornsby. 1998. Multi Party Politics in Kenya Oxford: James Currey. To daro, M. P. and Stephen C. S 2009. Economic Development New York: Pearson. World Development 17(7): 1067 76. Transparency International. 2012. Corruption Perceptio ns Index www.transparency.org. Accessed May 23, 2012. Toward a Theory of the Rent Seeking Society eds. J.M. Buchanan, R.D. Tollison and G. Tullock, 39 50. College Station: A&M T exas University Press. UNCHS (United Nations Center for Human Settlements) and ILO (International Labor Organization). 1995. Shelter Provision and Employment Generation Geneva: International Labour Office. UNESA (United Nations Department of Economic an d Social Affairs/Population Division). 2008. World Urbanization Prospects: The 2007 Revision New York: United Nations. http://www.unicef.org/kenya/overview_4616.html. Accessed June 3, 20 12. United Nations. 1990. World Urbanization Prospects New York: United Nations. United Nations. 1996. Informal Sector Development in Africa New York: United Nations. UN HABITAT. 2005. Kenya Urban Sector Profile Nairobi, Kenya: United Nations Human P rogramme. UN HABITAT. 2005. Situation Analysis of Informal Settlements in Kisumu Nairobi: United Nations Human Settlements Programme.
311 UN HABITAT. 2006. Kisumu Urban Sector Profile Nairobi: United Nations Human Settlements Programme. UN HABITAT. 2009. Kisumu City Development Strategies (2004 2009) Nairobi: United Nations Human Settlements Programme. UN State of the World Cities 2010/2011: Bridging the Urban Divide Nairobi: United Nations Human Settlements Programme. Democratization 18 (2): 443 461. Systems in Cambridge Review of International Affairs 13 (1): 14 28. 07. Mario Einaudi Center for Interna tional Studies, Cornell University, Ithaca, NY. Patrons or Policies? Patterns of Democratic Accountability and Political Competiti on eds. Herbert Kitschelt and Steven Wilkins, 50 67. New York: Cambridge University Press. The Journal of Modern African Studies 40(2): 297 321. Van Loo, in the Russian Duma: Alternative Politician the Conference on Citizen Politician Linkages in Democ ratic Politics, Duke University Vicente, Oxford Review of Economic Policy 25(2): 292 305. Journal of the American Planning Association 68(3): 297 314. Journal of Geography and Regional Pla nning 3(4): 114 125.
312 UN Habitat, Local Democracy and Decentralization in East and Southern Africa Nairobi: UN Habitat. What is a Case? Exploring the Foundations of Social Inquiry eds. Charles C. Ragin and Howard S. Becker, 121 138. Cambridge: Cambridge University Press. Wanjohi, Nick. 1997. Political Parties in Kenya Nairobi: Lengo Press. Wantchekon, Leonard. 2002 Theoretical and Comparative Per ting Behavior: Evidence from a a Field World Politics 55 (April): 399 422. Party Politics 3(4): 533 548. Patrons and Clients in Mediterranean Societies eds. Ernest Gellner and John Waterbury, 41 52. London, UK: Duckworth & Co. Weng, Yen Landscape and Urban Plannin g 81(4): 341 353. Habitat International 25(2): 209 227. cialist Gender and Society 5(3): 373 389. Journal of Contemporary African Studies 16(1): 39 64. Weitz tical Competition, Poverty and Incentives for Good Political Science Association, Chicago, IL, April 2007. Weitz t of American Journal of Political Science 56 (3): 568 583.
313 Comparative Studies in Society and History 10: 377 400. and the Capacity to Fight: Democratization 19 (6): 1086 1116. Sector: Cultural Despair or a Politics of Multiple Alle Comparative Political Studies 24(1): 31 55. Widner, Jennifer. 1992. The Rise of a Party State in Kenya Berkeley: University of California Press. Ivory Coast Comparative Politics 26 (2): 127 147. Wilson, James Q. 1973. Political organizations New York: Basic Books. Jour nal of Mixed Methods Research 3(1): 7 25. World Bank. 1992. Kenya Local Government Finance Study: Report No. 8997 KE World Bank, Washington D.C. World Bank. 2002. Kenya: An Assessment of Local Service Delivery and Local Government in Kenya, Nairobi Wor ld Bank, Washington D.C. World Bank. 2006. Kenya Inside Informality: Poverty, Jobs, Housing and Services In Washington, DC: World Bank. World Bank. 2012. World Development Indicators and Global Development Finance Washington, D.C.: The World Bank. Africa. Public Administration and Development 21: 277 288. Employment in Developing Countries: Th Economic Development and Cultural Change 44 (2): 289 31. http://www.afrobaromet er.org/papers/AfropaperNo106.pdf
314 Zeigler, Donald.J., Stanley D. Brunn, and Maureen Hays Cities of the World: World Regional Urban Development 3rd Edition eds. Brunn S.D., Williams J.F. and Zeigler D.J. Lanham MD: Rowman and Littlefield Publishers 3 48 Zulu, Eliya M., Donatien Beguy, Alex C. Ezeh, Philippe Bocquier, Nyovani J. Madise, Health Dynamics in Nairobi City's Slum Settleme Journal of Urban Health: B ulletin of the New York Academy of Medicine 88 (2): S185 S199.
315 BIOGRAPHICAL SKETCH Levy Odera was born and grew up in Kenya, where he completed his primary and secondary education. He later attended college at the Kenya Sc hool of Professional Studies (currently Inoorero Univeristy), where he received a Diploma in information t echnology. From 2002 to 2006 he attended college at Valdosta State University, Georgia, U.S ., where he attained a B.A. degree in political s cience, a Bachelor of Bus iness Administration degree in e conomics, and a Master of Public Administration. At the end of 2006 he returned to Kenya to consult for a business project. In 2007 he entered the PhD program in political science at the University of F lorida where he received a degree in political science in the spring of 2010 and his Ph.D. in the spring of 2013.