<%BANNER%>

Agriculture, GDP and Inequality in Sub-Saharan Africa

Permanent Link: http://ufdc.ufl.edu/UFE0021138/00001

Material Information

Title: Agriculture, GDP and Inequality in Sub-Saharan Africa Cross-Country Analysis of the Impact of Agricultural Production and Exports on Income Inequality
Physical Description: 1 online resource (240 p.)
Language: english
Creator: Ledermann, Samuel T
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: africa, agricultural, agriculture, development, doha, income, inequality, sub, trade, wto
Geography -- Dissertations, Academic -- UF
Genre: Geography thesis, M.A.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Over the past decades, global measures of income inequality have become a focal point in analyses of development. While a vast amount of global studies have supported the popular belief of a rise in both within- and between-nations income inequality over the past century, fewer studies have dealt with an in-depth analysis of income inequality in sub-Saharan Africa. A plethora of studies on the other hand has investigated patterns of growth, finding that increases in trade resulted in increases in growth. My research bridges these two bodies of work by focusing on agricultural (export) production and its impact on inequality and development in sub-Saharan Africa. My study provides an overview of the applicable theories and literature on development and income-inequality, agricultural export strategies and trade liberalization in the sub-Saharan African realm. Using the most recent UN-WIDER database on World Income Inequality, I develop an empirical cross-national regression model of agriculture?s relationship with development and inequality measures at the national scale, as well as a qualitative comparative case study analysis of select African countries. Finally, I discuss policy implications of the findings for agricultural-exporting nations, especially in the context of the continuing World Trade Organization's (WTO) Doha Round negotiations.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Samuel T Ledermann.
Thesis: Thesis (M.A.)--University of Florida, 2007.
Local: Adviser: Goldman, Abraham C.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2009-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021138:00001

Permanent Link: http://ufdc.ufl.edu/UFE0021138/00001

Material Information

Title: Agriculture, GDP and Inequality in Sub-Saharan Africa Cross-Country Analysis of the Impact of Agricultural Production and Exports on Income Inequality
Physical Description: 1 online resource (240 p.)
Language: english
Creator: Ledermann, Samuel T
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: africa, agricultural, agriculture, development, doha, income, inequality, sub, trade, wto
Geography -- Dissertations, Academic -- UF
Genre: Geography thesis, M.A.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Over the past decades, global measures of income inequality have become a focal point in analyses of development. While a vast amount of global studies have supported the popular belief of a rise in both within- and between-nations income inequality over the past century, fewer studies have dealt with an in-depth analysis of income inequality in sub-Saharan Africa. A plethora of studies on the other hand has investigated patterns of growth, finding that increases in trade resulted in increases in growth. My research bridges these two bodies of work by focusing on agricultural (export) production and its impact on inequality and development in sub-Saharan Africa. My study provides an overview of the applicable theories and literature on development and income-inequality, agricultural export strategies and trade liberalization in the sub-Saharan African realm. Using the most recent UN-WIDER database on World Income Inequality, I develop an empirical cross-national regression model of agriculture?s relationship with development and inequality measures at the national scale, as well as a qualitative comparative case study analysis of select African countries. Finally, I discuss policy implications of the findings for agricultural-exporting nations, especially in the context of the continuing World Trade Organization's (WTO) Doha Round negotiations.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Samuel T Ledermann.
Thesis: Thesis (M.A.)--University of Florida, 2007.
Local: Adviser: Goldman, Abraham C.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2009-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021138:00001


This item has the following downloads:


Full Text

PAGE 1

1 AGRICULTURE, GDP AND INEQUALITY IN SUB-SAHARAN AFRICA: CROSS-COUNTRY ANALYSIS OF THE IM PACT OF AGRICULTURAL PRODUCTION AND EXPORTS ON INCOME INEQUALITY By SAMUEL THOMAS LEDERMANN A MASTERS THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS UNIVERSITY OF FLORIDA 2007

PAGE 2

2 2007 Samuel Thomas Ledermann

PAGE 3

3 To all whove supported and inspired me throughout the years and let me travel high above the skies. May the ladder of development never get kicked away.

PAGE 4

4 ACKNOWLEDGMENTS I thank all for the great support Ive received from my thesis committee, with a special thank you to Dr. Abe Goldman, for his great a nd challenging advice, Dr. Julie Silva for her inspiration and believing in the significance of my work, and last but not least, Dr. Barbara McDade for her constant support and kindness.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........8 LIST OF FIGURES................................................................................................................ .......13 ABSTRACT....................................................................................................................... ............15 CHAPTER 1 INTRODUCTION................................................................................................................. .16 2 REVIEW OF LITERATURE.................................................................................................20 Agricultural Development and Inequality..............................................................................20 Impact of Commodities...................................................................................................21 Impact of Policies............................................................................................................24 Impact of Dependency on Cash Crops............................................................................26 Impact of Climate............................................................................................................29 Impact of Decline in Terms of Trade..............................................................................29 Impact of International Trade Policy...............................................................................31 Impact of Input Use and Modernization..........................................................................32 Impact of Land................................................................................................................34 Impact of Urbanization....................................................................................................36 Impact of Manufacturing.................................................................................................38 Impact of Infant Mortality...............................................................................................41 Outlook........................................................................................................................ ...........43 3 HYPOTHESES, DATA AND METHODOLOGY................................................................46 Hypotheses..................................................................................................................... .........46 Data........................................................................................................................... ..............47 Dependent Variable.........................................................................................................47 Gini coefficient (gini)...............................................................................................47 Independent Variables.....................................................................................................50 Agricultural production (agvalue)............................................................................50 Food exports (tradfood), raw agricult ural exports (tra draw) and total agricultural exports (totagexp) as per centage of merchandise exports and as percentage of GDP................................................................................................50 Arable land (ha) per capita (arabland)......................................................................51 Urbanization (urbaniza) and population density (popdensi)....................................51 Before WTO (prewto)..............................................................................................51 Gross Domestic Product per capita (gdppppca) and LN GDP per capita (loggdpppp)...........................................................................................................51

PAGE 6

6 Infant mortality (infantmo).......................................................................................52 Methodology.................................................................................................................... .......52 4 DEVELOPMENT AND AGRICULTURAL GR OWTH IN SUB-SAHARAN AFRICA....59 Case Study: Burkina Faso.......................................................................................................59 Case Study: Cte dIvoire...................................................................................................... .63 Case Study: Ghana.............................................................................................................. ....70 Case Study: Senegal............................................................................................................ ...74 Case Study: Ethiopia........................................................................................................... ....77 Case Study: Kenya.............................................................................................................. ....82 Case Study: Uganda............................................................................................................. ...85 Case Study: Malawi............................................................................................................. ...88 Case Study: Mozambique.......................................................................................................91 Case Study: South Africa....................................................................................................... .95 Assessment of Case Studies....................................................................................................99 Comparing David to Goliath...........................................................................................99 Conflict and Its Aftermath.............................................................................................100 Poverty, Infant Mortality and Inequality.......................................................................101 Diversification or Lack There Of..................................................................................102 Ability to Modernize.....................................................................................................104 Droughts and Other Environmental Impacts.................................................................105 5 COMPARATIVE ANALYSIS ON AGRICUL TURE, GROWTH AND INEQUALITY..123 Main Hypothesis................................................................................................................ ...123 Results........................................................................................................................ ...123 Test of Assumptions......................................................................................................124 Secondary Hypotheses..........................................................................................................131 First Hypothesis: Role of Agriculture on GDP.............................................................131 Second Hypothesis: Amount of Arable Land................................................................133 Third Hypothesis: Dependency on Few Agriculture Exports.......................................134 Fourth Hypothesis: Agricultur al Policies and Inequality..............................................134 Fifth Hypothesis: Urban Areas and Inequality..............................................................135 6 DISCUSSION................................................................................................................... ....160 Empirical C ontributions........................................................................................................160 Policy Implications............................................................................................................ ...166 Strengthening the WTO negotiations............................................................................166 Pro-Poor Development Strategies focu sed on Agricultural Diversification.................167 7 CONCLUSION AND FURTHER AVENUES....................................................................169 APPENDIX A NORMALITY.................................................................................................................... ..171

PAGE 7

7 B OUTLIERS..................................................................................................................... ......178 C NORMALLY DISTRIBUTED ERROR..............................................................................185 D HOMOSCEDASICITY........................................................................................................190 E NONLINEARITY................................................................................................................191 F MULTICOLLINEARITY....................................................................................................192 G INDEPENDENCE................................................................................................................200 H COMPLETE DATASET......................................................................................................201 I REGRESSION RESULTS WI THOUT MISSING VARIABLES.......................................205 J GDP REGRESSION.............................................................................................................208 K GDP REGRESSION WITHOUT MISSING VARIABLES................................................212 L DEFINITIONS AND DATA RANGES...............................................................................215 M ROBUSTNESS OF MODELS.............................................................................................224 LIST OF REFERENCES.............................................................................................................229 BIOGRAPHICAL SKETCH.......................................................................................................240

PAGE 8

8 LIST OF TABLES Table page 2-1 Development outcomes......................................................................................................45 3 List of secondary hypotheses.............................................................................................54 3-2 Definition of inequality measured.....................................................................................54 3-3 Sources................................................................................................................... ............54 3-4 Countries................................................................................................................. ...........55 4-1 Summary Burkina Faso......................................................................................................59 4-3 Burkina Faso, food produc tion per capita index, 1965=100..............................................62 4-4 Summary Cte dIvoire..................................................................................................... .64 4-6 Cte dIvoire, food produc tion per capita index, 1965=100..............................................66 4-7 Production in Cte dIvoire 1999-2005 of commodities facing declines since 2001........70 4-8 Summary Ghana............................................................................................................. ....71 4-11 Summary Senegal.......................................................................................................... ...74 4-14 Summary Ethiopia......................................................................................................... ...78 4-15 Ethiopia, food producti on per capita index, 1993=100.....................................................80 4-17 Summary Kenya............................................................................................................ ....83 4-20 Summary Uganda........................................................................................................... ...86 4-23 Summary Malawi........................................................................................................... ...89 4-26 Summary Mozambique.....................................................................................................92 4-30 Summary South Africa.....................................................................................................96 4-2 Overview of agriculture a nd development in Burkina Faso............................................107 4-5 Overview of agriculture a nd development in Cte dIvoire............................................109 4-9 Overview of agriculture and development in Ghana.......................................................110 4-10 Ghanas exports, 1980 2000 (value as percentage of exports)....................................111

PAGE 9

9 4-12 Overview of agriculture and development in Senegal....................................................112 4-13 Senegal, food production per capita index, 1965=100...................................................112 4-16 Overview of agriculture and development in Ethiopia...................................................113 4-18 Overview of agriculture and development in Kenya......................................................115 4-19 Kenya, food production per capita index, 1965=100......................................................115 4-21 Overview of agriculture and development in Uganda....................................................116 4-22 Uganda, food production per capita index, 1965=100....................................................116 4-24 Overview of agriculture and development in Malawi....................................................117 4-25 Malawi, food production per capita index, 1965=100....................................................117 4-27 Overview of agriculture and development in Mozambique...........................................119 4-28 Diverse poverty and health figures in Mozambique.......................................................120 4-29 Mozambique, food producti on per capita index, 1965=100...........................................120 4-31 Overview of agriculture and development in South Africa............................................121 4-32 South Africa, food producti on per capita index, 1965=100............................................121 4-33 Infant mortality, GDP&Gini...........................................................................................122 5-1 Model 1 regression results...............................................................................................137 5-2 Model 2 regression results...............................................................................................138 5-3 Model 3 regression results...............................................................................................139 5-4 Regression results Gini coefficient..................................................................................140 5-5 Regression results GDP ppp per capita............................................................................140 5-6 Regression results GDP ppp per Capita...........................................................................141 5-7 Correlations of agricult ural exports and inequality.........................................................142 5-8 Correlations of agricultur al exports and development.....................................................147 5-9 Correlations of arable land (ha per person)......................................................................149 5-10 Correlations of value share of 3 of Top 20 agricultural exports.....................................153

PAGE 10

10 5-11 Correlations of cash crops dummy variable.....................................................................154 5-12 Correlations of agricultural policies, inequality and growth..........................................155 5-13 Correlations of urbaniza tion, inequality and growth......................................................158 5-14 Correlations of la ndlocked dummy variable...................................................................159 A-1 Model 1:Test of normality...............................................................................................171 A-2 Model 2:Test of normality...............................................................................................171 A-3 Model 3:Test of normality...............................................................................................171 A-4 Arable land test of normality...........................................................................................172 A-5 Population density test of normality................................................................................172 A-6 Value of Top 3 Exports Tests of Normality.....................................................................172 A-7 GDP ppp per capita test of normality..............................................................................174 A-8 Trade agricultural raw materials tests of normaliy..........................................................176 A-9 Post-Transformation Tests of Normality.........................................................................177 B-1 Model 1 residual statistics............................................................................................... .178 B-2 Model 1 outliers.......................................................................................................... .....179 B-3 Model 2 residual statistics............................................................................................... .180 B-4 Model 2 outliers.......................................................................................................... .....181 B-5 Model 3 residuals statistics..............................................................................................182 B-6 Model 3 outliers.......................................................................................................... .....183 C-1 Normality test Model 1....................................................................................................186 C-2 Test of normality Model 2...............................................................................................187 E-1 Model 1 nonlinearity...................................................................................................... ..191 E-2 Model 2 nonlinearity...................................................................................................... ..191 F-1 Correlations Model 1...................................................................................................... .192 F-2 Coefficients Model 1...................................................................................................... ..192

PAGE 11

11 F-3 Correlations Model 2...................................................................................................... .193 F-4 Coefficients Model 2...................................................................................................... ..193 F-5 Correlations Model 3...................................................................................................... .194 F-6 Correlations Model 3...................................................................................................... .195 F-7 Alternate stepwise regression model...............................................................................195 F-8 Alternate model summary................................................................................................196 F-9 Alternate model ANOVA................................................................................................196 F-10 Alternate model coefficients............................................................................................197 F-11. Alternate model excluded variable..................................................................................198 F-12 Collinearity diagnostics.................................................................................................. .199 G-1 Model 1 summary........................................................................................................... .200 G-2 Model 2 summary........................................................................................................... .200 G-3 Model 3 summary........................................................................................................... .200 H-1 Complete dataset.......................................................................................................... ....201 I-1 Model summary............................................................................................................. ..205 I-2 Coefficients.............................................................................................................. ........205 I-3 Collinearity diagonstics.................................................................................................. .205 I-4 Residuals statistics...................................................................................................... .....206 J-1 Descriptive statistics.................................................................................................... ....208 J-2 Model summary............................................................................................................. ..208 J-3 ANOVA..................................................................................................................... ......208 J-4 Coefficients.............................................................................................................. ........209 J-5 Collinearity diagnostics.................................................................................................. .209 J-6 Residuals statistics...................................................................................................... .....210 K-1 Variables entered......................................................................................................... ....212

PAGE 12

12 K-2 Model summary............................................................................................................. ..212 K-3 ANOVA..................................................................................................................... ......212 K-4 Coefficients.............................................................................................................. ........213 K-5 Collinearity diagonstics.................................................................................................. .213 K-6 Residuals statistics...................................................................................................... .....213 L-1 Arable land (ha) per capita...............................................................................................215 L-2 Value share of top 3 agricultural expo rts (as % of top 20 agricultural exports)..............216 L-3 Urbanization (% of urban population) and population density........................................217 L-4 Agriculture, value added (as % of GDP).........................................................................218 L-5 Food exports (as % of total merchandise exports)...........................................................219 L-6 Agricultural raw exports (as % of total merchandise exports)........................................220 L-7 Total agricultural expo rts (Food and Raw Ag) as % of total merchandise exports.........221 L-8 GDP ppp per capita & log GDP ppp per capita...............................................................222 L-9 Agricultural exports (% of GDP).....................................................................................223 M-1 Comparison of regre ssion models robustness................................................................225

PAGE 13

13 LIST OF FIGURES Figure page 3-4 Regression model.......................................................................................................... .....52 3-1 Frequency of years of inequality measured.......................................................................56 3-2 Quality of inequality data................................................................................................ ...57 3-3 Histogram of Gini......................................................................................................... .....58 4-1 Average cereal yiel ds in select regions............................................................................108 4-2 Producer price (local currency/tonne)..............................................................................114 4-3 Export quantity of coffee (green) and sesame seeds........................................................114 4-4 Producer price of tobacco (unm anufactured) in Malawi in 1991-2003...........................118 5-1 Regression equation Model 1...........................................................................................123 5-2 Regression equation Model 2...........................................................................................123 5-3 Regression equation Model 3...........................................................................................123 5-4 Scatterplot agriculture value added..................................................................................143 5-5 Scatterplot food exports as pe rcentage of merchandise export........................................144 5-6 Scatterplot agricultural raw materials e xports as percentage of merchandise exports....145 5-7 Scatterplot square-root ra w agricultural exports as percen tage of merchandise exports.146 5-8 Scatterplot of l og of GDP ppp per capita.........................................................................148 5-9 Scatterplot arable land................................................................................................... ...150 5-10 Scatterplot value share of three of the top 20 agricultural exports versus GDP ppp per capita......................................................................................................................... .......151 5-11 Scatterplot value share of three of th e top 20 agricultural exports versus Gini coefficient.................................................................................................................... ....152 5-12 Scatterplot of population density.....................................................................................156 5-13 Scatterplot of urbanization...............................................................................................157

PAGE 14

14 5-14 Scatterplot of non-linear (cubic) relationship between urbanization and log of GDP ppp per capita................................................................................................................. ..157 A-1 Before WTO................................................................................................................ .....173 A-2 GDP per capita transformations.......................................................................................174 A-3 Histogram of log GDP ppp per capita transformation.....................................................175 A-4 Histogram of raw trade agricultural goods......................................................................175 A-5 Histogram of raw ag riculture transformation..................................................................176 A-6 Histogram trade food exports...........................................................................................177 C-1 Histogram GINI Model 1.................................................................................................185 C-2 Residuals plot Model 1....................................................................................................185 C-3 Histogram standardi zed residuals Model 2......................................................................186 C-4 Regression plot residuals Model 2...................................................................................187 C-5 Histogram standardi zed residuals Model 3......................................................................188 C-6 Regression plot residuals Model 3...................................................................................188 D-1 Scatterplot Model 1....................................................................................................... ...190 D-2 Scatterplot Model 3....................................................................................................... ...190 I-1 Residuals scatterplot..................................................................................................... ...207 J-1 Histogram regression residuals........................................................................................211 K-1 Histogram regression residuals........................................................................................214

PAGE 15

15 Abstract of Masters Thesis Pr esented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Arts AGRICULTURE, GDP AND INEQUALITY IN SUB-SAHARAN AFRICA: CROSS-COUNTRY ANALYSIS OF THE IMPACT OF AGRICULTURAL PRODUCTION AND EXPORTS ON INCOME INEQUALITY By Samuel Thomas Ledermann August 2007 Chair: Abraham Goldman Major: Geography Over the past decades, global measures of income inequality have become a focal point in analyses of development. While a vast amount of global studies have supported the popular belief of a rise in both withinand between-na tions income inequality over the past century, fewer studies have dealt with an in-depth analysis of income ine quality in sub-Saharan Africa. A plethora of studies on the other hand has investigated pa tterns of growth, finding that increases in trade resulted in increases in growth. My research bridges these two bodies of work by focusing on agricultural (export) production a nd its impact on inequality and development in sub-Saharan Africa. My study provides an overview of the appli cable theories and literature on development and income-inequality, agricultural export strategi es and trade liberalization in the sub-Saharan African realm. Using the most recent UN-WIDE R database on World Income Inequality, I develop an empirical cross-national regressi on model of agricultures relationship with development and inequality measures at the national scale, as well as a qualitative comparative case study analysis of select African countries. Finally, I disc uss policy implications of the findings for agricultural-exporting nations, especially in the context of the continuing World Trade Organizations (WTO ) Doha Round negotiations.

PAGE 16

16 CHAPTER 1 INTRODUCTION Karl Polanyi brilliantly remarked on capitalism half a century ago in The Great Transformation that laissez-faire was planned, pla nning was not (Polanyi 1957, 141). Through this succinct, yet powerful exclamation, he high lighted to several generations of scholars the importance of actors and their actions for the current state of our economies and dominant ideologies. Entering the twenty-first century, ou r so-called age of globalization, a heated debate has emerged concerning the current dominant systems multidimensional impact on developing nations and the respective winners and losers (O'Brien and Leichenko 2003). Just as during Polanyis era, most current studies analyze globalizations forces th rough the traditional measurements of growth and, for the more socially inclined, poverty. It has become increasingly clear that analogous to laissezfaire, globalization itself is a social construct, shaped and supported by its own movement of actors, such as characterized by Be neria as the gendered Davos Man (Beneria 1999; O'Brien and Leichenko 2003). Traditional studies of agricult ural development in Africa ha ve focused most prominently on the relationship between measures of agricu ltural production (i.e. yield per hectare) and development (i.e. growth in Gross Domestic Product (GDP) per capita ppp). More recently, these studies have added trade as a key force impacting development in sub-Saharan Africa and beyond. With the latest (stalled and currently revived) World Trade Organization (WTO) Doha development rounds foci on agricultural liber alization, the significan ce of crop exports on African nations is most likely to increase accordingly, especially in countries with agricultural export potential to the ma rkets that liberalize most (that is East Asia and Europe) (Hertel and Winters 2006, 26). At a global scal e, the long term, conservative estimate of the impact of the Doha Development Round for developing countries are projected to be a reduction of the poor

PAGE 17

17 living under US$ 1 per day by 9.7 million in 2015 (Hertel and Winters 2006, 27).1 Consequently, the rationale for my focus on agri cultural exports rests on their (projected and) continued importance in relationship with develo pment (i.e. growth) in sub-Saharan Africa, as the largest sector in terms of employment for most African countries. Granted the focus on agricultural exports, a ke y additional indicator has gradually risen to the forefront of these debates concerning the so -termed winners and lose rs of globalization: inequality. With increases in accessible data, su ch as the World Income Inequality Distribution database from the World Institute for Devel opment Economics Research (WIDER) by the United Nations (UN), a plethora of work has contribut ed to this growing body of interdisciplinary scholarship (Beer and Boswe ll 2001; Kayizzi-Mugerwa 2001; Pa luzie 2001; Dollar and Kray 2002; Wade 2002). The scholarship largely build s upon and critiques the theoretical foundation laid by Simon Kuznets infamous inverted ushaped curve of inequality and development (Kuznets 1955).2 These complex studies differ all chronologically and chorologically, investigating both multiple time periods and scales of analyses. The inequality debate has also entered the mainstream realm of policymakers, reflected both in the publishing of the 2006 World Development Report by the World Bank entitled Equity and Development (World Bank 2005), as well as the links create d towards global security in th e post 9/11 arena (Sachs 2001). In terms of scale, they range from household micro-level surveys (Adams 1995), regional studies (Silva 2007) to global assessments (Krugman and Venables 1995). Inequality as a result from 1 Coupled with gains in productivity in both the manufacturing and agricultural sector from increasing trade liberalization, the number of people living under US$ 2 per day by 2015 would be reduced under a Doha scenario over the long term by an estimated 29.6 million people, and under full liberalization scenario by 193.2 million people, or 4.9% (Hertel and Winters 2006). 2 Kuznets argues that inequality follows an inverted u-shap e that increases as countries shift from agriculture to manufacturing, and decreases once they enter the more developed stages of economic development. Given the problematic data availability, he remarks that this 1955 landmark paper is perhaps 5 per cent empirical information and 95 per cent speculation, some of it possibly tainted by wishful thinking (Kuznets 1995, p.26).

PAGE 18

18 trade also plays a key role in the concept of the new econom ic geography, as put forth by Krugman (1991). Out of all these works, w ith a few exceptions (Collier and Gunning 1999; Kayizzi-Mugerwa 2001), an abysmal geographical gap appears upon closer review: Sub-Saharan Africa. Not only has Sub-Saharan Africa, as a region, received the least attention, the attention that it has received has usually only b een superficial. Focusing on nu merous explanatory variables, ranging from geography (landlocke d), institutions (ethnicities, education) to policy (trade openness, rule of law) to economic s (manufacturing), scholars have struggled to disaggregate the diversity and complexities found amongst inequality measurements in this most impoverished region of the world (Bloom and Sachs 1998; Collier and Gunning 1999; Dollar and Kray 2002; Easterly 2006). Subsequently, while they broa dly agree that Sub-Saharan Africa is a special case, as reflected in the stat istically significant regional dummy variables found in Dollar and Kray (2002, 207) and Easterly and Levine (1997, 1210), the search for the missing link for explaining the variations of inequality still appear s to be ongoing for the African continent. This thesis consequently is an attempt at providing a new perspective on an old variable, agriculture, focusing specifically on exports, to highlight the importance (or lack thereof) of that sector on development and inequality. My study consists of six main sections. It provides an overview of the literature on agricultural development and inequality in subSaharan Africa and beyond. The hypotheses are articulated and methodology is presented, as well as the data sources. Of the seventeen African nations included in the dataset, ten representative country-level case studies were examined more closely regarding their agricultural developm ent post-1990. These, mostly qualitative, case studies are followed by the findings of the quantitative analysis, involving the development of a

PAGE 19

19 regression model explaining patte rns of inequality in sub-Sa haran Africa. By focusing ultimately on agricultural export policies, my study fits well in the vast body of work that has shown that policies are key in determining gr owth and inequality (Persson and Tabellini 1991; Cramer 1999). My studys ultima te goal is to take a forwardlooking stance, trying to provide additional quantitative insights to policy makers on the importance (or lack thereof) of the agricultural sector and its respective policies fo r achieving the goal of su stainable and equitable development. My study concludes with the policy implications of this research and highlights future avenues of study.

PAGE 20

20 CHAPTER 2 REVIEW OF LITERATURE Agricultural Development and Inequality Agriculture plays the key role across multiple scales in sub-Saharan Africa: local, national and international. At the local scale, it remains the most important source of formal (traded) or informal (subsistence/consumption) income for the livelihoods of a vast majority of people in sub-Saharan Africa. Agricultural export commodities account for a large share of most farmers incomes, given its prime role as the main source of foreign exchange in sub-Saharan Africa. For an analysis of exports impact on inequality, it is important to identify them by type, as one would expect that exporting primary commodities may increase or decrease income inequality, depending on the pattern of ownership and the resources concerned (Wood 1994, 14; italics added). At the national scale, from a governmental pe rspective, agriculture provides the main source of foreign exchange and revenue for a majority of non-oil and mineral exporting governments, even in the current ag e of globalization. Gains from ag riculture thus still rest to a great extent at the heart of their national deve lopment strategy (Wood and Mayer 2001, 392). At the international scale, given the fact that 30 African nations have become net-food importers instead of net-food expor ters over the last th irty years (Peacock 2005), changes in both global supply and demand are of highest importa nce for two main reasons They influence the volatile commodity markets, especially in sectors where commodities are heavily dominated by one importer or producer and a great degree of dependency exists (e.g. cashew nuts in India, cotton in China) (Cramer 1999; FAO 2004). Esp ecially during years of widespread severe droughts, an increase in global pr ice for crops, such as maize, can have devastating consequences as the governments are unable to pay for the much-needed food imports (Wines 2005).

PAGE 21

21 Overall, experts and scholars alike agree that agriculture is an extremely complex sector, shaped by myriad economic, political, social and environmental forces. Consequently, the literature review has been subdivi ded into various interlinked s ectors dealing w ith agricultural export production. Impact of Commodities One of the most important distinctions in agricultural export production is between processed versus unprocessed agricultural goods, with the former leading to greater inequality as their industries have historica lly sought cheap inputs in orde r to make up for their overall inefficiencies and mismanagements. As argued by Bates, the larger the fraction of output consumed by domestic processing firms, the gr eater the pressures on governments to keep down the prices offered to farmers (2005, 124). Consid ering growth, it is howe ver unclear whether or not a focus on processed goods would lead to its increase or decrease (Mozambican case study) (Cramer 1999). This is most likely to depend upon th e degree of inefficiency within the industry and the potential imperfections of the intern ational market, as the processed goods are theoretically gaining a larger share of the goods commodity-c hains value, and subsequently are able to garner higher profit margins (FAO 2004). This efficiency is likely to be correlated with the skill level available within the country, as countries w ith higher levels of skill per worker tend to export more of their primary products in processed form (Wood and Mayer 2001, 375). If the expected negative impact of processed good s on inequality holds true, it would appear to support Kuznets hypothesis on the inverted ushaped relationship between inequality and development. Consequently it would appear to go against the emerging tenor, which has argued that the Kuznets hypothesis [] no longer ho lds up to empirical a nd theoretical tests (Ackerman et al. 2000, 299).

PAGE 22

22 The importance of focusing on processed goods ha s increased over the past years, with the gap between processed and unprocessed agricultural goods widening during the era of globalization, as exports of pro cessed agricultural products grew 6% per year during the period 1981-2000, compared with 3.3% for primary pr oducts (FAO 2004, 26). Disaggregated along the levels of development, it becomes however cl ear that developed countries have harvested the largest share of this increase in growth, as the global share of developi ng countries of processed agricultural exports dec reased from 27% in 1981-1990 to 25 % in 1991-2000 (ibid., 26). This figure is even more abysmal for Least-Develope d Countries (LDCs), whose share fell from a negligible 0.7% to 0.3% over the same period (ibid., 26). The in ability to maintain or actual decrease in share of processed goods by these lesser developed c ountries is an indication of a vast problem of inequality that is penetrating agricultural commodity chains, as they have become increasingly dominated by a few transn ational enterprises and distribution companies with significant market power (ibid., 27). Unfortunately, as a common thread throughout this discussion, there is a clear shortage of studies investigating this rela tionship. Conventional wisdom is however that a shift from subsistence farming to the market-oriented, cashcrop, production of goods results in an average increase in incomes, with an analogous increase in inequality, as the wealthy are best able to reap the benefits of this transition (at the initial stages) through thei r improved capital availability and bargaining power. The opposing view, mos tly argued for by proponents of the Green Revolution, however continues to exist, stating that cash crops can (and do) have a favorable impact on rural income distribution by providi ng the poor with new employment and incomeearning opportunities (Adams 1995, 467). Overall, as discussed by Moradi and Baten, due to

PAGE 23

23 missing data, this hypothesis was never tested co mprehensively in the previous literature (2005, 1235). The general importance of the type of agricult ural exports however is further highlighted by Easterly, who suggests in hi s recent finding by focusing on sugarcane and wheat, that agricultural endowments specifica lly the relative abundance of land suitable for wheat to that suitable for sugarcane predict st ructural inequality and that st ructural inequality predicts development outcomes (Easterly 2006, 32).3 These findings that cash crops, in his case sugar cane, lead to rural inequality is further discussed by Adams, who undertakes a very solid indepth study of inequality in rura l Pakistan (1995). Coincidentl y, the leading cash-crop of his region, sugarcane, has a large and negative impact on income distribution (ibid., 468). In addition, while too small to be significant, co tton has also been observed to have a negative impact. Contrastingly, stre ngthening the emphasis on the type of crop exported, he found that income from the main food (wheat, rice) and li vestock crops (fodder, barley) has an equalizing effect on income dist ribution (ibid., 468). Finally, in addition to the previous evidence that export crops are more likely to increase inequality than food crops, inst itutions also seem to have treated crops differently and subsequently either enhanced or lessen thei r purely economic distribu tional impact. Cooksey, focusing on the case of maize in Tanzania, finds that the price of maize, th e main staple of the urban poor, was too important politi cally to be left to the merc ies of marketing board and cooperative lobbyists (2003, 76). Export crops, on the other ha nd, given their reduced national 3 Structural inequality is different, acco rding to Easterly, than market inequa lities, with the former reflecting such historical events as conquest, colonizatio n, slavery, and land distribution by the state or colonial power; it creates an elite by means of these non-market mechanisms. (Easterly 2006, p.2)

PAGE 24

24 political importance, could be more easily accommodated through national-local patronage politics and systematic rent-seeking (ibid., 76). In summary, while case studies exist s upporting respective points of view, the overwhelming consensus appears to state that the type of expor t crops clearly matters when analyzing income inequality as well as growth. The types of export commodities produced were historically greatly influenced by colonial policies, and continue to be impacted by policies well into the 21st century. Impact of Policies Sidestepping path dependency, much of the structure of agricultura l production, such as inputs, subsides, tariffs and taxation policies provided, is dependent on institutional policies. Two contrasting institutional features found in sub-Saharan African countries are marketing boards, which generally regulated export crops through a quasi monopoly, and structural adjustment policies (SAP) and liber alization strategies undertaken in the 80s and 90s, with their ultimate goal of trimming governments ability to regulate the agricultural market. A potential shift in a country from one mar ket ideology to the other has a significant impact on whether or not investments are made by the government in agriculture, ultimately affecting inequality and growth. It was under policies of import substitution, that the belief reigned that a peasant path based on intensified, agrarian based rural live lihoods was still a possible and conceivable development option (Bebbington 1999, 2024). Ho wever, since the early and mid-1990s, an observable shift has occurred towards that of ne oliberal economic reforms [where] significant part of the peasant economy is in ma ny instances not viable (ibid., 2024). Overall, both SAP and marketing boards thus ar e clear examples of (inter)national policies and ideologies governing agriculture and impacting local farmers. Structural adjustment policies

PAGE 25

25 have increasingly become unpopular since their imple mentations in the early 1980s, as they have failed to achieve their ultimate goals of incr easing growth through a wi de-array of toolsets: budget cuts, privatization programs, deregulation of markets, trade li beralization, easing of controls on foreign investment, shifts from impor t substitution to export promotion development models (Beneria 1999, 67ff). While th eir intentions were to also in crease food security, as in the case of southern Africa, the results are more ambiguous. Leichenko and OBrien find that especially national-level price reforms that el iminate price controls on agricultural commodities have allowed some farmers to earn higher profit s, [while leaving] many farmers vulnerable to both price instability and drought (2002, 11-12). Increasingly, a vocal body of cr itics has argued that another main goal of SAPs, increases in trade openness, has not lead to increases in growth and standard of living. As a matter of fact, focusing on globalization, new estimates that incorporate measuremen t of inequality both between and within countries suggest that inequal ity has grown even more rapidly in recent years than previously thought (O'Brien and Leichenko 2003, 95). It has been argued, from a politicaleconomic perspective, that libe ralization and privatiz ation promote specific interest groups economic interests, [... which] were favored befo re such schemes were deployed (Ackerman et al. 2000, 322-323). Hence, opposite to conventio nal neoliberal wisdom, countries that intervened in the aftermath of the SAP experien ced more economic growth, not less. According to World Bank studies, countries that strongly intervened exper ienced both higher rates of growth in domestic manufacturi ng and lowered rates of manuf acturing imports. Intervention, therefore, is not universally b ad (Grant and Agnew 1996, 734). The shift from marketing boards to SAPs conse quently is not a clear linear one, as some countries have resorted back to interventionist policies. As highlighted in the case of Tanzania,

PAGE 26

26 both the market for agricultural inputs and output s were gradually liberal ized in their initial stages, resulting in a collapse of the interna l maize market, large falls in the production of traditional export crops, and a consequent increase in subsistence agriculture and rural poverty (Cooksey 2003, 70). These reforms however were reversed by a later generation, as export crop liberalization in particular, have been hotly resisted by significant players in the system who consider there has been enough externally-driven liber alization [] (ibid., 70). This trend of non-linearity of the process of liberalization has also been observed by Oyejide et al. (1999) who found it to have reversed in seven out of ten African countries, in many cases (again including Kenya) more than once (cited in Gunning 2000, 9). On the other hand, the picture is not entirely negative, as SAP reforms appear to have worked in Uganda since th eir implementation in 1987, rais[ing] incomes in urban and especially rural areas at the same time. Civil service wages were increased from a low level. Reforms provided incentives for production, especially in the coffee sector, through greater use of th e market, removal of export taxes and related exchange rate reforms and resulted in hi gher coffee and food production in the southern parts of the country, raising incomes. At the same time, world coffee prices improved (Kayizzi-Mugerwa 2001, 24). These findings are furthermore supporte d by Deininger and Okidi. (2003). Impact of Dependency on Cash Crops As highlighted by the case of Uganda, the su ccess or failure of a policy is sometimes closely related to the development of the world s commodity markets. Scholars tend to agree, analogous to the famous resource curse, that an increase of dependenc y on a single cash-crop comes with an expected increase in overall vulne rability. Given high pric e volatility and relative decline over the past decades of many agricultural commodities, a reduction of growth can be expected over the long-run.

PAGE 27

27 A recent FAO report highlights in dramatic di splay that a dependency on a few cash crops results in countries becoming h ighly vulnerable to unfavorable market or climatic conditions (FAO 2004, 20). This picture is espe cially grave in sub-Saharan Af rica, as 37 out of 42 countries classified as Heavily Indebted Poor Countri es (HIPCs) by the IMF and World Bank, rely on primary commodities for more than half of thei r merchandise export earnings (ibid., 20). The picture looks exceptionally grim in the agricultura l sector, as overall pric e volatility appears to have declined over the last 20 years, yet pri ces of many agricultural commodities remain highly volatile (ibid., 21). Furthermore, there is very low demand elasticity for several agricultural goods, with some cash crops actually experiencing an overall decline in demand, such as cotton under the pressure from synthetic fibres. Last ly, differentiating between types of commodities, the instability tends to be higher for agricultura l raw materials and tropical beverages than for temperate-zone products (ibid., 21). Overall, it thus appears that dependency on a few primary commodities is a widespread economic disease in Africa. As found by Grant and Agnew, the overall structure of Africas trade has remained remarkably stable over tim e, and Africa remains heavily dependent on the export of primary commodities (1996, 737). This focus on primary commodities had a negative impact on Gross Domestic Product (GDP) growth in most African countries, as the GDP of all commodity exporters grew by only 1.4% [b etween 1980 1992], whereas the GDP of manufacturing exporters gr ew by 6.8% (ibid., 737). A common thread in the literature is the emphasis on the negative economic and environmental impacts of monoculture cash croppi ng on growth and inequality vis--vis a more diversified agricultural scheme. Looking at 28 countries in sub-Sahara n Africa over six 5-year periods from 1950 to 1980 with anthropometric es timates, Morati and Baten find that their

PAGE 28

28 regression results indicate clearly that the existence of a (singl e) cash-crop industry reduces a regions average height by almost one centimete r and increases intra-regional inequality (2005, 1252). Furthermore, they find the opposite effect for a more diversifie d cash-cropping strategy, as heights increase nearly by three millimeters and the CV decreases by 0.10 for each additional cash-crop industry, while inequality is consistent ly lower (ibid., 1252). Consequently, in an attempt to speak to policymakers about the furt her impact of globalizat ion, they advocate in favor of a specialization in cash crops and simultaneous support fo r the process of globalization, arguing that it could have overwhelmingly pos itive effects on a regions development, decreasing inequality even furt her if a strategy of diversifi cation is pursued simultaneously (ibid., 1252). These findings are furthermore supported by Beer and Boswell, who also find a negative relationship between commodity c oncentration and physical quality of life. Consequently, they also argue in favor of diversific ation, as it increases th e ability to better we ather downturn in the global commodity market (Beer and Boswell 2001). However, it is important to keep in mind that there is no direct link between diversificatio n and inequality, as dive rsification per se does not necessarily level rural incomes (Tiffen 2003, 1359). According to Tiffen, both rich and poor are heavily dependent upon nonfarm income s for their income composition, with rich farmers usually able to invest outside of the pr imary sector, agriculture This would suggest however that even during periods of lower prices for the most im portant cash crop, rich farmers are the ones able to display the greatest resilien cy, consequently potenti ally increasing the gap between rich and poor. It is within this context that inequality might come into existence even without the overwhelming economic pressure s faced by farmers in the South overall.

PAGE 29

29 Impact of Climate It is evident that climate play s a major role at the local scal e on inequality, due to varying degrees of resilience and adaptability in the populations. With the occurrence of global warming, climate issues have become an even mo re pressing concern for the future. Its impact on farmers, coupled with liberalization, appear s to be however mixed, as both outcomes exist where farmers who formerly had difficulty adap ting to climatic variability may become less vulnerable to drought-related food shortages [ whereas] adaptation strategies may be constrained or limited for other farmers [] (Leichenko and O'Brien 2002, 14). Furthermore, while climate change might have a negative impact on a region, its overall economic outcome might still be positive if accompanied by rising wo rld prices for that regions crop (ibid., 3). Focusing solely on droughts, it is clear that they have a great impact on overall cereal production, even though some crops are more drought -resilient than others (ibid., 7). Given the likelihood of wealthier farmers having multiple ad aptation and livelihood st rategies, as well as access to more drought-resistant crop varieties, severe droughts or other environmental crises, e.g. pest outbreaks, are likely to in crease inequality a nd decrease growth. Impact of Decline in Terms of Trade As has been discussed previousl y, a historical decline in pric es received for agricultural goods had detrimental impacts on both growth and inequality in sub-Saharan Africa over the past. At the globally aggregated scale, terms of trade within th e agricultural sector worldwide neither rose nor fell significantly between 1961 and 2002 (FAO 2004, 12). However, once again, the picture looks especially dire when focusing on the terms of trade for the LDC. Agricultural terms of trade feel by half from a peak in 1986 to a low in 2001 (ibid., 12). As these countries heavily depend on their commodity export incomes to provide for food imports, overall food security has b een significantly undermined over these past decades.

PAGE 30

30 Furthermore, decreasing terms of trade have al so increased the countries respective debt burden. This phenomenon has been especially dram atically captured in the relationship of the terms of trade between agricultural commodities and manufactured imports, which were even more persistent and more damaging, increasing the debt burden of developing countries (ibid., 12). As a matter of fact, from 1961 to 2001, average prices of agricultural commodities sold by LDCs fell by almost 70% relative to the price of manufactured goods purchased from developed countries (ibid., 12). This drop ha s resulted in a severe shortfa ll of revenues for countries in sub-Saharan Africa, as it has been declared the r egion that has suffered the most from declining terms of trade [ costing them] the equivalent of 119% of their combined annual gross domestic product (GDP) in lost revenue (ibid., 13). Grant and Agnew however dispute the above pi cture, arguing that orthodox development economists showed that the term s of trade in sub-Saharan Af rica has actually declined less compared to their Latin American and As ian counterparts in th e 1980s, and yet Africa experienced much the worst export performa nce (Grant and Agnew 1996, 730). While their time period might differ, they nevertheless appear to argue in favor of lib eralization as actually reducing structural inefficiencies removing heavy taxation of crops for example that are holding a comparative advantage. Overall, it however appears to be clear that a link exists between worse terms of trade indicators and worse prices for main export crops in sub-Saharan Africa. These changes in terms of trade had an impact on inequality both prio r to reform efforts unde rtaken during the 1980s. Such was the case in Cote dIvoire, where inequa lity has increased due to interaction[s] of external conditions coffee and cocoa prices wi th internal inequalitie s in land ownership and wages for civil servants [] (Kayizzi-M ugerwa 2001, 22). Even during liberalization, it

PAGE 31

31 continued to play a role, according to Kayizzi, as the government had to cut civil service wages and salaries and export prices put pl antations under pressure (2001, 22). Impact of International Trade Policy One of the main driving forces for the above-d iscussed worsening of the terms of trade are attributed to unfair and illegal agricultural policy put in place by the most developed nations. Agricultural trade was largely excluded from pro cesses of international trade liberalization prior to the 1980s due to a multitude of political, soci al and economic reasons. Politically, during the Cold War era countries made agricultural self-su fficiency a security pr iority. Consequently, while manufactured goods covered under the GATT were freely traded, agricultural goods were still heavily protected an d internally supported. It was within this context that the GATT trade ministers, under the le adership of the United States, launched the Uruguay Round (the first major trade round). Its main achievements were average cuts in tariffs and export subsidies of 36%, and an ambitious reduction in domestic subsidies. This trade round was proceeded by the current Doha Development Round, which holds the potential of providing th e biggest liberalization of the agricultural sector. While the round has come to a halt (and been revived ag ain after the World Economic Forum in Davos, Switzerland of 2007), the previous cuts and harm onization efforts are in place, and membership in the WTO is currently associated with an increasingly out-ward oriented policy which emphasizes exports, subsequently increasing grow th. Onafowara and Owowye, focusing on their trade policy dummies, find signi ficantly positive effects of trade orientations on economic growth for five countries Cameroon, Cote d Ivoire, Kenya, Sudan, and Zambia at the 5% level and better (Onafowora and Owoye 1998, 503). As they additionally find a positive impact of orientation from in-ward to ou t-ward, they argue that it is po ssible to stimulate real economic growth in sub-Saharan African countries through an outward-looking strategy of export

PAGE 32

32 expansion (ibid., 504). In conclu sion, they advocate in favor of continued trade liberalization, as their results suggest the need for continued trade liberalization in order to enhance economic growth [of sub-Saharan Africa] in th e current world economy (ibid., 505). A unique way through which GATT/WTO membership would result in greater growth is through the endogeneity of trade pr otection. Membership creates mo re uniform tariff rates, as tariff declines in each indus try are proportional to that indu strys prereform tariff levels, alleviating concerns about endogeneity, at leas t in the economic sense (Goldberg and Pavcnik 2004, 227). Consequently, the most protected sect ors are the ones experien cing the greatest cuts, potentially transferring additional resources from heavily protected manufacturing towards the agricultural sector. It however would be wrong to argue that me mbership in the WTO is a win-win, when measured in absolute terms, as agreements tend to primarily benefit a few nations, typically the initial signatories (O'Brien and Leichenko 2003, 95-96). Other nations sign on in order to minimize their absolute losses (ibid., 95-96). In relationship to the impact on growth, there however appears to be few studies that link WTO membership in particular, and trade openness in general, to inequality. For example, Dollar and Kray find little evidence that these policies and institutions [openness to international trade, macroeconomic stabilit y, moderate size of government, financial development, and strong property ri ghts and rule of law] have sy stematic effects on the share of income accruing to the poorest quintile (Dollar and Kray 2002, 196). Impact of Input Use and Modernization A defining feature of both monoculture and cash-cropping since the mid-century in the developed, and increasingly since the later part of the 20th century in the developing nations, is the gradual (attempted) modernization and mechan ization of agriculture. A lack of access to

PAGE 33

33 inputs has been identified as one of the most common obstacles in peasant agriculture and for achieving growth (Wood 1994, 46). As most states heavily taxed agriculture through lower prices they tended to subsidize inputs in order to appease powerful political fo rces (Bates 2005, 6). In need of political support in the countryside, their rural confederates include tenants and managers on state production schemes [as well as] elite-level farmer s, [ and] the more widespread group of progressive farmers who have become dependent on state-sponsored programs of subsidized inputs (ibid., 120-121). While th ey are placed at a disadvantage through the artificially lowered agricultural prices, they still support the government exactly because of these subsidized inputs. In addition, once producers have obtained a strong relative advantage (as re flected potentially in higher yields) in their production, they actually entertain weaker political bargaining power. This is based upon the observation that the stronger their relative advantage, the longer they will persist in growing a crop under conditions of falling prices the more thoroughly they can be squeezed, in short, by adverse pricing policies by the government (Bates 2005, 127). Unfortunately, poorer farmers were largely i gnored in policy schemes on both ends, as they did not receive fac ilitated access to the inputs and were stressed by the lower prices. There are clear signs however that rural citizens, as observed in the Tanzanian PRSP, are eager to be granted better access to credit and perceive high costs of farm inputs as hindering their own development (Ellis and Mdoe 2003, 1381). Given the periods of structural adjustment and subsequent at tempts at reducing the ability of states to subsidize such schemes favoring we althy constituencies, it is important to observe more closely the effect of agricultural inpu t price fluctuations. As argued by Wood, price fluctuations and actual prices fo r inputs could be much more si gnificant than capabilities for

PAGE 34

34 local peasants and casual, agricultural wage laborers in determining real income entitlements (2003, 457). But even in an entirely efficient and liberal market scheme for fertilizer usage, its usage might not be beneficial to small farmer s for certain crops. As found by Cooksey in Tanzania in the case of maize, be nefits of using such fertilizer s might be rather small, and on purely economic terms the mass of small farm ers would not have profited from systematic inorganic fertilizer use (2003, 71). As a matter of fact, gi ven the unequal di stribution of fertilizer usage prior to liberalization efforts, it comes as no surprise that the wealthiest farmers might be the main loser from liberalization in a post-structural adjust ment environment (ibid., 72). These impacts of liberalization however might not be as strongly felt in a state with higher oil-revenues, such as Nigeria for exampl e, who can afford to subsidize inputs. In summary, a high fertilizer and tractor usage per capita should indica te the existence of a high number of large farmers, as the distinguish ing characteristic of la rge sellers is their good management quality, which has enabled them to accu mulate resources (from livestock sales, crop production or non-farm incomes) for investme nt in intensification (Tiffen 2003, 1354). Subsequently, no matter the benefi ts of using such improved inputs and equipment, the two main investments local farmers in Kenya would like to undertake when resources became available were equipment (67%) a nd inputs (17%) (ibid., 1357). Impact of Land Besides intensification, extens ification or the expansion of agricultural production is another key aspect of agricult ural production. Of the trinit y of labor, land and capital, land consequently is of vast importance within an Af rican agricultural context. Dollar and Kray find that countries with increased arable land per wo rker have lower income share of the poorest quintile, which is consistent with Leamer et al. (1999) who find that cropland per capita is

PAGE 35

35 significantly associated with hi gher inequality in a cross-sect ion of 49 countries (Dollar and Kray 2002, 215). This emphasis on land is one of the most discu ssed agricultural variables in the literature, as reflected by the instrumental works of De ininger and Squire (1998), who find a negative impact of land inequality on inequality and gr owth. This is furthermore supported by Alesina and Rodrik, who have articulated that their estimated coefficients imply that an increase in, say, the land Gini coefficient by one standard deviatio n [] would lead to a reduction in growth of 0.8 percentage points pe r year (1994, 481). As far as African case studies, there appear s to be a less clear-cut picture of the relationship between inequality and land inequa lity. South Africa, for example, has been identified as having a very unequal land distri bution, which has a large impact on inequality when coupled with a policy mix that taxed the ag ricultural sector directly and indirectly through industrial protection and overvalued exchange rates (Binswanger and Deininger 1997, 196263). Ghana, Nigeria, Tanzania and Uganda on the other hand have agrarian structures dominated by family farmers, where land inequa lity has played a lesser role (ibid., 1962). Inequalities are rather the resu lt from excessive taxation and little support that was provided which went primarily to relatively inefficient, but politically powerfu l large producers (ibid., 1962). There are however some qualifications and cauti ons that need to be taken into account. While Adams also finds in the case of Pakistan that agricultural income is highly correlated with land owned, his research also shows that livestock income is not linked with land ownership (Adams 1995, 472-474). Furthermore, a change in land law can not be assumed to have a predictable effect on farmers, as the chan ge needs to be coupled with changes in other

PAGE 36

36 institutions affecting agricult ure: market networks and crop pricing policies or marriage and gender relations, for instances (L each, Mearns, and Scoones 1999, 241). Nevertheless, it appears valid that the greater th e availability of agricultural land per capita, the more agricultural practices are viable. If farmers are inheriting smaller farms in a land scarce environment, for example, they are often una ble to find the necessary capital to intensify production from a small land holding [and] are incr easingly forced to look to other occupations for a living []4 (Tiffen 2003, 1358). These opportunities ar e often believed to be found in urban areas and the manufacturing sector. Impact of Urbanization A vast literature exists on the relationship of urbanization to in equality and growth simultaneously, both at a global scale, as well as for sub-Saharan Africa. One of the best known relationship stems from Lipton ( 1977). The argument goes that a greater urban population acts as an indicator for an increased urban bias consequently resulting in heavier taxation on agriculture, ultimately increasing inequality whil e decreasing growth. The problem with this proposition however is that the crit ical link is placed on taxation. If that taxation occurs mostly on cash crops, which are generally of greater importance to wealthier farmers, it actually might decrease inequality as smaller farmers are less dependent on th em for their income source. However, given that the focus most often rests on subsidizing urban food sources, the effect is most heavily expected on food crops. This nega tive effect could be offset through increasing domestic demand for food crops by the growing urbanized, non-ag ricultural population. Given the likelihood of changes in nutritional preferen ces for the urban population, this latter effect might be reduced in reality. 4 There certainly is the possibility to obt ain access to land, yet be unable to in tensify production if labor constrains exist and capital is not available to off-set the constraint.

PAGE 37

37 The literature nevertheless appears to support an increase in the likeli hood of an urban bias with a higher urban population. As observed by Ba tes, where urban in terests came to power, they have adopted policies more hostile to the interest of farmers (Bates 2005, 122). On the contrary, governments dominated by rural prod ucers have intervened less forcibly in the markets for outputs and have also provided more favorable subsidies in the markets for inputs (ibid., 122). In terms of geographical distan ce from the urban areas, as refl ected in the measurement to the capital city, studies show that the inequality is lower in distant regions. On the other hand, [] relative to the distance of the remotest region, distance has a positive effect on inequality (Moradi and Baten 2005, 1253). Consequently, Mora di and Baten conclude that the size of the country matters, as remote regions of large coun tries such as Tanzania, Mali or the Ivory Coast display smaller inequality, whereas distant regi ons of small countries such as Rwanda and Uganda have higher inequality (ibid., 1253). Tiffen is one of the main proponents of the argument that urban areas are playing an increasingly larger role for domestic farmers, an d acts as a main generator of change in farming systems, particularly, but not only, in the semi-arid areas (Tiffen 2003, 1344). She actually finds that most African countri es are not exporting more than 20% of agricultural production value, so 80% is locally consum ed. Crops internally consumed are now much more important than crops exported [] (ibid., 1351). While th is would indicate for the urban areas to be a clear driver of growth in the rural areas, it is li kely that inequality actually will increase at the national level, as the urban market is attrac ting not only their products but also their labor (ibid., 1344).

PAGE 38

38 This inequality between the ur ban and rural areas is well-doc umented in the literature and has been described as a major aspect of inequal ity in Africa [] stem[ing] from colonial rule rather than globalisation (Kayizzi-Mugerwa 2001, 10). This gap, however, appears to have lessened from an inequality standpoint through trade liberalization during the 1980s and 1990s. According to Kayizzi, by boosting export crops, tr ade liberalization actually reversed rural decline in some countries. Wh ere they were grown on small peasant holdings, the impact on rural welfare was clea rly positive (2001, 10). While this observation might be closely rela ted to the actual crises faced during these decades, it is likely that during periods of positiv e growth, the urban-rural bias is likely to increase. One of the most prominent eviden ce for such a phenomenon is provided by Krugman, who argues that increasing retu rns are in fact a pervasiv e influence on the economy [] determining the geography of real economies (Krugman 1991, 10ff). Impact of Manufacturing Other occupations outside the agricultural sect or are often found in the formal or informal sector of manufacturing. Agricult ural development strategies have increasingly taken a dualistic form, as the sector has been poised against manuf acturing. As argued by Beer and Boswell, it is not development per se, but dualism and diffusi on processes that are the keys to explaining income inequality (2001, 6). Consequently, a fo cus on key indicators of the agricultural exports has the potential for unearthing a great degree of these dualistic and diffusional processes. According to Bates, if count ries are able to tax other sectors for revenue besides agriculture, they not only use subsidy programs, but are more lenient on reducing taxation, as governments with access to nonagricultural sour ces of funds often impose lighter levels of taxation on export crops (2005, 123). Consequently, we could expect that from an agricultural

PAGE 39

39 perspective, countries with larger non-agric ultural exports might experience a lowered inequality, as revenues are more evenly distributed This political analysis is furthermore suppor ted by Ellis and Mdoe, who find that Poverty Reduction Strategy Papers (PRSP) programs by the World Bank and IMF in Tanzania, with their goal at subsequent decentraliz ation and improving rural liveli hoods, resulted in increases in taxation and corruption. While th is problem has been realized by the PRSP, it fails to make specific proposals [] and [draws] no strategi c connections between rural taxation and the demands for increased rural revenue generation that results from new distri ct councils seeking to create budgets over which they can exercise their own control (Ellis and Mdoe 2003, 1381). Granted these findings, they have been qualifie d in terms of size and maturity of the nonagricultural sector, as ty pically initial non-ag ricultural sectors are still taxing heavily agriculture. These first nonagricultural sector s are usually administration a nd defense, which often cull any small agricultural surplus that exists without contributing much to [agriculture] (Tiffen 2003, 1347). Focusing on the size of agricu ltural exports, several studies provide further support for focusing on the ratio between agricultural and manufacturing exports. Bourginion et al., for example, argue that countries dependent on few cash crops are subject to an important risk, which affects the level of macro economic activity as well as the households income distribution. Such a systemic risk is thus a potential source of inequality among households in a poor exportdependent country (Bourguignon, Lambert, and Suwa-Eisenmann 2004, 369). While their study is less applicable, as they focus on a mode l showing the various im pacts of liberalization policies based around Ivory Coast, they nevert heless highlight the importance of focusing on agriculture exports, as export pr ice risk is seldom taken into account when evaluating the

PAGE 40

40 overall degree of inequality within a country (ibid., 369). While dependency on a single cashcrop is discussed in an earlier section, such reasoning has received support from dependency theorists, who argue that large export sectors are positively correlated with income inequality, as export sectors are largely foreign owned, which monopolizes internal ca pital and repatriates profits, stagnating the domestic sector (Beer and Boswell, 3). The positive relationship between the non-agri cultural sector on in equality is also supported indirectly by Bebbingtons analysis of livelihoods, as he finds that the more viable rural livelihoods share the common characteristics that th ey are able to sustain or increase their access to resources beyond the traditional agri cultural sector (1999, 2028). The ability to diversify the income through other opportuni ties has a positive impact on growth (and inequality) is furthermore supported from a gendered perspective by a study of womens livelihoods in Benin. Mandel found that mobility and ability to diversify is key, as women with substantial mobility tend to have greater autono my and independence in their economic activities than those with limited mobility (2004, 278). This depicted positive relationship between the ratio of agricultural to manufacturing exports and inequality however appears to r un counter the previously mentioned Kuznets hypothesis, which would predict an increase in inequality as a re sult from development of the manufacturing sector (Kuznets 1955). Combined with the impact of globalization, sectors that have imported heavily foreign machinery and ha ve modernized are likely to experience an increase in demand for skilled workers (Gol dberg and Pavcnik 2004, 239). If coupled with increases in productivity that a re shared with workers in th e form of higher wages, the increased wage premium clearly would increase national inequality measures (ibid., 244). Goldberg and Pavcnik (2003) is even more applicable to our case of sub-Saharan Africa, as they

PAGE 41

41 created a model, which displayed an increase in the informal sector in the aftermath of trade liberalization. Given this sectors associati on with lower wages and worse job quality, trade liberalization could in principl e contribute to increased inequa lity (Goldberg and Pavcnik 2004, 247). Using empirical data, if we accept the assu mption that richer countries have lower agriculture to manufacturing expor ts ratios, these findings are di sputed by studies that find no support for different relationships between rich and poor countries and inequality, contrary to the Kuznets hypothesis that inequality increases with income at low levels of development (Dollar and Kray 2002, 208). As a matter of fact, indicators, such as th e percentage of labor force in agriculture, have been found to be associat ed with income inequali ty at lower levels of development, with higher percentages indicatin g lower income inequality (Beer and Boswell, 6).5 From a policy perspective, it is important to note that overall, manu facturing exports in Africa have actually declined over the years, and both policies of competitive advantage and import substitutions have equally not been a de velopment panacea. Consequently, recent tenors articulate that a healthy rela tionship between agriculture and manufacturing exports might be most beneficial, as African economies might be be tter served by a degree of diversification than by the specialization mandated by the comparat ive-advantage thesis (Grant and Agnew 1996, 734). Impact of Infant Mortality Infant mortality has in the past decade rece ived considerably more attention, both as a proxy for overall human welfare, as well as due to its observed negative impact on economic 5 This picture however is different in the case of th e United Sates, where a small number of workers in the agricultural sector exist, yet inequality is relatively high.

PAGE 42

42 growth (Deininger and Okidi 2003, 486). One of its most famous proponents is Amartya Sen, who has advocated in favor of a greater focu s beyond basic economic indicator in order to receive a fuller picture of a persons quality of lif e. As such, he argues that mortality information not only highlights the capabiliti es of a person, but furthermore can throw light also on the nature of social inequality [ as b]iases in economic arrangements are often most clearly seen through differential mortality information (S en 1998, 23). Placing the impact of mortality on economic income into the African context, Bloom a nd Sachs find that high ra tes of fertility with falling levels of infant and child mortality in Africa result in a substantial drag on African economies by reducing their productive capacity pe r capita (1998, 4). Furthermore, due to the heavy population skew towards the younger levels of the population pyramids, this youth-heavy age distributions also tend to be associated with lower rates of savings and investment [] and therefore slower econom ic growth (ibid., 4) While a clear relationship appears to exist be tween economic growth and infant mortality, it is unclear whether or not a di stributional relationshi p exists as well. As shown in Table 2-1, adopted from Wood (2002, 49), inequality follo ws a unique path in relationship the other measures of development. While sub-Sahara n Africa has by far the lowest GDP ppp per capita, highest poverty rates and highest infant mortality, Latin Americas inequality is even higher, thus showcasing the complex relationship between these two variables. Furthermore, while focusing on the impact infa nt mortality can have on inequality, the relationship can also be the reverse. As f ound in a study of the US, inequality increases economic segregation, which increases infant deat h, with infants living in states with higher level of inequality holding a gr eater chance of dying than those in poorer states (Mayer and Sarin 2003, 19). Given these findings, yet reported, even if complex, relationship between income

PAGE 43

43 inequality, development and infant mortality, close at tention will be paid to this variable in both the qualitative and quantitative an alyses of Chapters 4 and 5. Outlook Overall, the importance of the agricultural se ctor in Sub-Saharan Africa is unlikely to diminish soon. While scholars disagree on whether or not agriculture will hold the key for development (Thirtle, Lin, and Piesse 2003), it is clea r that it will continue to play a vast role in the future, as the share of agriculture in Af rican primary output and exports seems likely to increase in the future (Wood 2002, 14). This is furthermore stressed by the multitude of untapped inputs and proclaimed scientific advan ces that have so far largely sidestepped the African continent, such as the Green Revolution, irrigation, inorganic fert ilizer and geneticallymodified crops. Some preliminary caution however is in order, as agriculture, while accounting for a large portion of income in sub-Saharan Africa, might no t significantly explain respective inequalities. As highlighted by Congeu et al., no matter how interesting the inequa lities within the agricultural sector may be per se, they argue that in their study of five countries they did not find it as the major source of substantial ine quality deviations between their case studies (2006, 14). While their findings could be disc oncerting, given the authors primary goal of simply focusing on the agricultural sector for hi ghlighting its importance (or lack thereof) on inequality (and growth), their wa rnings are well taken. However, other studies have highlighted the vast role agriculture plays. Adams found that in rural Paki stan, agricultural income makes the largest contribution to overall inequality. Depending on the year, the results suggest that agricultural income accounts for 35.5% 45.6% of overall inequality (Adams 1995, 472). Concerning measurements of development, such as the Gross Domestic Product per capita (purchasing power parity), the literature finds them useful to a certain degree. As argued by

PAGE 44

44 Bates, due to the greater share of food in pe oples (especially urban populations) basket of income, nations with lower per-capita incomes ar e thus more likely to adopt policies in support of low-priced food (Bates 2005, 124). Furthermore, it has been noted that no correlation exists between the share of agricultural exports in total expor ts [] with GDP per capita (Wobst 2003, 74). As a note of caution, Wade points out that GDP per capita might just be an alternate measurement of inequality, as the average in come of each country weighted by population is interesting only [] an approximation to what we are really interested in [, inequality] (Wade 2002, 14). An additional concern is that common wis dom of Africa might indicate that not enough variations exist between countries when looking at agriculture a nd development indicators, given their clustered status of largely underdeveloped countries. This fear ha s been dispelled by Wood and Mayers main findings, which showed that [s ub-Saharan] Africa is internally diverse [ and its] export structure and resource combin ations vary widely (Wood and Mayer 2001, 389). This diversity of export structur e and resources is exactly what well examine in the ten case studies of sub-African nations, arranged by regions, starting w ith Burkina Faso in Western Africa.

PAGE 45

45 Table 2-1. Development outcomes Sub-Saharan Africa Latin America High-Income OECD Per capita GDP ($000, PPP) 1995 1.56.4 26.9 Poverty (% of population <$1/day) 1999 46.715.1 Income inequality (Gini, most recent year) 46.154.7 38.4 Infant mortality (per 1000 live births) 1995 94.033.6 7.1 Sources: Adapted from Wood (2002). Weighted by population.

PAGE 46

46 CHAPTER 3 HYPOTHESES, DATA AND METHODOLOGY Hypotheses Based upon the literature review the thesis will test the fo llowing two main hypotheses: (1) Countries with higher shar e of agriculture value adde d (as% of GDP) have lower inequality, (2) Countries with higher share of agricultural exports (as% of total merchandise exports) have lower inequality. These hypotheses are derived from the main assumption that the agricultural sector significantly explains differences among the inequali ty levels of sub-Saharan African nations, as measured by the Gini coefficient. The resulti ng inequality model will be tested by regressing agricultural with nonagricultural variable s that have historically s eemed to be significantly related to inequalit y, including human capital6 and urbanization (Deininger and Squire 1998; Castello and Domenech 2002). Building from this main hypothesis, five othe r hypotheses will be tested (Table 3-1): As a note of explanation, GDP per capita ppp has been chosen in order to observe the impact of agricultural production and exports vis-vis inequality. Given this studys focus on inequality, it has not been elevated to the sa me importance as our main hypotheses, yet however is an integral part of the study, especially in connection with policymaker s historical main foci on income, development, growth and poverty. The third hypothesis tests th at countries with a higher dependency on a few (cash -) export crops have greater in equality; in other words, countries with a greater diversity in export crops have a lower ine quality. In addi tion, the fourth hypothesis tests that countries afte r 1994, signifying the aftermath of most structural adjustment 6 Traditionally, human capital is measured by educational levels (years of schooling). Due to lack of data availability, infant mortality has been chosen as a proxy, as it has been found to hold a very significant and quantitatively not unimportant negative effect on attainment of schooling in the population [] (Deininger and Squire, 1998, p.273).

PAGE 47

47 policies, the founding of the World Trade Organization (WTO) from its predecessor (GATT) and the implementation of the Uruguay Round have a higher inequality. Th e fifth hypothesis tests whether a higher percentage of the urban populati on of a country is correlated with greater inequality. In addition, severa l secondary hypotheses have been considered. However, due to lack of available (or complete) data, significance or other constraints, these have not been included here.7 Data Data used for this thesis comes from a variet y of sources. The data source is the most recent UN-WIDER database on World Income Ine quality, Version 2.0a (World Institute for Development Economics Research 2005), which in cludes the large World Bank dataset created by Deininger and Squire (1996). This represents the main dependent va riable of inequality examined in this study and the main hypotheses. The data for the independent variables come from a variety of official UN or FAO datasets most notably the up-to-date FAOSTAT database (FAO 2006), the African Development Indicator s (World Bank 2005) and the World Bank Development Indicators (UN 2006). Given the colla boration and interdepe ndence between these international organizations, the data is fairly compatible as one of the above organizations themselves reference the others variables. A detailed definition a nd data ranges of each independent variable is listed in Appendix L. Dependent Variable Gini coefficient (gini) This data set is the largest and most commonl y used secondary source for inequality data, listing 4664 Gini coefficients, ranging from before 1960 up to 2003, and includes 40 countries 7 They include tests on the value of agricultural export s per capita (FAO database had missing variables or not reliable), terms of trade (lack of available data), dr oughts (issue of time and susceptibility), Foreign Direct Investments (FDI) (lac k of relevance to agriculture), fertilizer and tractor data (lack of reliability).

PAGE 48

48 from sub-Saharan Africa. A gini coefficient is calculated based upon the respective share of national income held by each percentile of the popul ation, resulting in a Gini coefficient of 0 for perfect equality, and a Gini coefficient of 1 for pe rfect inequality. For ev ery countrys value, the database lists the source and quali ty of the data, a reported (and calculated) Gini coefficient, quintile and decile shares, survey means and medians along with the income shares of the richest 5% and the poorest 5% (World Institute for Development Economics Research 2005). A conscious decision has been made to rely on consumption data rather than income data. While the former has been observed to be not clos ely tied to short-term fluctuations in income [and is] smoother and less variable than income, gathering reliable income data is very difficult in a (rural) developing country setting (UNU-WIDER 2005, 4). While some people disagree with the use of consumption data given the que stion of quantifying consumption patterns, it is broadly agreed that where the rural agricultural sector is larg e [] consumption data should be used8 (UNU-WIDER 2005, 4). These studies were all undertaken at the household level, and ideally measured both food and non-food consum ption, the use-value of durable goods and housing. The dataset also reports two different Gini coefficients, and the study uses the first Gini coefficient calculated by WIDER using methods developed by Tony Shorrocks and Guang Hua Wan to estimate [it] from decile data almost as accurately as if unit record data were used (UNU-WIDER 2005, 9). Finally, the database al so uses a quality ranking, which indicates whether or not the income concept or the survey quality could be judged. Based upon this metadata, fifty da ta points have been chosen in sub-Saharan Africa. The three main criteria applied were that they were no t allowed to be of the lo west quality rating (4), should come from a uniform, reliable source, a nd should include at least two data points at 8 In sub-Saharan Africa, income data is historically rarely collected in the first place, with th e major exception for South Africa.

PAGE 49

49 different times for each country. 46 of the 50 data points are based on consumption data (Table 3-2), while only one data poi nt from South Africa in 1997 is based on gross income9. These datasets cover a total of 23 y ears (Figure 3-1), with the earli est data point starting in 1980 (Nigeria) and the latest in 2003 (Mozambique). The vast majority of the data points were inevitably of lesser quality (quality 3 out of 4, with 1 being the best), given the sub-Saharan African environment, accounting for 78% of the sample10 (Figure 3-2). The main source of the data points from the UN-WIDER dataset come s from Deininger and Squire 2004 World Bank dataset, which accounts for 68% of the data points. The remain ing 32% come from other World Bank sources, with the exception of the 2003 Mozam bican case, which was reported in a journal article by James, Arndt, and Simler (2005) (Tab le 3-3: Sources). Finally, while seventeen countries were initially include d, Senegals data point s have been removed because of various data problems, which are li kely based on reporting errors11. Consequently, the current data set covers sixteen sub-Saharan Afri can nations, a representative sa mple of the entire continent (Table 3-4: Countries).12 The mean Gini of the 50 cases is 49.6 with a standard deviation of 7.39 (Figure 3-3: Histogr am of Gini). 9 Comparing income to expenditure-based data, authors have usually adjusted the former to match the latter. The adjustment, as undertaken by Frazer (2 006) was 4.3 points, wher eas Deininger and Squire (1996) adjusted by 6.6 points on average. No adjustment in our case was undertaken, as these previous cases related to expenditure data, not consumption. 10 The one missing value for which no quality has been assessed is the 2003 Mozambican Gini coefficient published in 2006, which has been included in the dataset. 11 Senegals Gini, according to the survey dropped by more than 30 points fr om 1994 to 1991 to a low of 29.3. A different survey in 1994, however, has found that its Gini was 41.0 for that year. While there might be other such cases of error, I have attempted to eliminate any data that seems unreasonable and consequently no other data point has been eliminated. 12 The main inequality regression models use only a minimum of two and a maximum of three data points per country.

PAGE 50

50 Independent Variables Agricultural production (agvalue) To measure the value of agri cultural production, and its import ance to the economy, the sotermed agricultural GDP is measured, as reported by the UN as Agriculture Value Added (as% of GDP) (UN 2006). Agriculture incl udes forestry, hunting, and fishi ng, as well as cultivation of crops and livestock production according to the ISIC divisions 1-5 (UN 2006). It is calculated by measuring the net output of a sector after adding up all output s and subtracting intermediate inputs (UN 2006). Food exports (tradfood), raw agricultural expo rts (tradraw) and total agricultural exports (totagexp) as percentage of merchandise exports and as percentage of GDP Sporadic absolute values of agricultural export s are available in some countries. However, the most reliable value available for each year needed for the datase t is the variable of food exports as percentage of merchandise exports and to a lesser degree, the raw agricultural exports as percentage of merchandise exports with the latter measuring primarily cotton that is not accounted for by the former (UN 2006). Thes e two separate variables have also been combined into an aggregate total agricultural exports as pe rcentage of merchandise exports Furthermore, this independent aggregate variable has also been measured in relationship to the percentage of GDP In addition, the degree the agri cultural export sector is dependent on a few commodities is measured by calculating the value of the top three export commodities as a percentage share of the value of the top 20 export commodities in any given year (FAO 2006). Stated positively, this variable measures the degree export diversification has taken place.

PAGE 51

51 Arable land (ha) per capita (arabland) The estimated amount of arable land per capita. Data is available ev ery year and has been calculated. Arable land is defined as is land under temporary crops (double-cropped areas are counted only once), temporary meadows for mowing or pasture, land under market and kitchen gardens, and land temporarily fa llow (less than five years) (FAO 2006). Urbanization (urbaniza) and population density (popdensi) Urbanization is measured as the urban popul ation as a% of total population. Since definitions can vary of what comprises urba n areas, some caution should be used (UN 2006). Furthermore, population density has also been used as an alternate measure of populations potential role on inequality. Before WTO (prewto) A time dummy has been created to indicate, whet her or not the Gini coefficient was before or after the end of the Uruguay r ound and the creation of the WTO, with a 1 indicating before or in 1994. Beyond attempting to assess the impact of the WTO, it also is aimed at including the aftermath of the structural adjustment programs, wh ich were mostly in their final stages or ended by 1994. Gross Domestic Product per capita (gdppppca) and LN GDP per capita (loggdpppp) In order to measure overall development, GDP per capita, purchasing power parity is measured (in international $) (UN 2006). This va riable has also been l ogged in an attempt to reduce the outlier of South Africa.13 Furthermore, the variable is also used as a dependent variable (Secondary Hypotheses) in order to test for the relationship be tween agriculture and development. 13 Using purchasing power parity, we are able to be tter compare the relative ability for same amounts of money to purchase a given set of goods across countries.

PAGE 52

52 Infant mortality (infantmo) Childrens infant mortality rate (deaths per 1000 live births) is used as an approximation to indicate primarily human welfare, development, and to a lesser degree human capital. Since the variables are only available in 1980, 1990, 1995, 2000, 2004, the value closest to the respective year was chosen (UN 2006). Methodology Two different methods will be employed in this thesis. First, a primarily qualitative case study analysis is undertaken to de velop a narrative of the agricultural struct ure and development history of individual countries, which are not told by the numbers alone. The ten case studies have been chosen in order to represent a dive rse, representative sample across the African continent. The chapter will concl ude with a comparative analysis. The quantitative study is undertaken in Chapter 5. It involves the cr eation of a regression model (enter/stepwise method) aimed at testing the above main hypotheses (Table 3-1). Given that some data points are missing, a separate mo del has been created re placing them with the mean. Thorough empirical analysis will be unde rtaken to study whether or not collinearity exists. Normality will be tested, and leverage an d outliers will be taken into considerations. The following main model will be tested: ( GIN I ) 0 1( PREWT O ) 2( A GVALU E ) 3( TRADFOOD ) 4( TRADRAW ) 5( LOGGDPPP ) 6( INFANTMO ) 7( URBANIZA ) u whereby1( PREWTO ) 2( AGVALUE ) 3( TRADFOOD ) 4( TRADRAW ) Agriculture Figure 3-4. Regression model14 14 PREWTO: Before WTO; AGVALUE: Agriculture, value added (as% of GDP); TRADFOOD: Food exports (% of merchandised exports); TRADRAW: Agricultural Raw expo rts (% of merchandised exports); LOGGDPPP: LOG of

PAGE 53

53 As such, my study combines both quantitative and qualitative sections in order to better understand the role played by agricu lture on inequality in Africa. GDP ppp per capita (international $) ; INFANTMO: Infant mortality (deaths per 1000 live births); URBANIZA: urban population (% of total population).

PAGE 54

54 Table 3. List of secondary hypotheses Secondary Hypotheses 1 Countries with higher share of ag riculture value added (as% of GDP) have lower GDP per capita ppp; 1 2 Countries with higher share of agricultural exports (as% of total merchandise exports) have higher GDP per capita ppp; 2 Countries with more arable land per capita have gr eater inequality; 3 Countries with greater dependency on a few export crops have greater inequality; 4 Inequality in African countries incr eased after the WTO agreement in 1994; 5 Countries with higher percentage of urban population have greater inequality. Table 3-2. Definition of inequality measured Frequency* Percent Consumption 4692.0 Consumption / Expenditure 24.0 Expenditure 12.0 Income, Gross 12.0 Total 50100.0 *Refers to the number of data points Table 3-3. Sources Frequency Percent Deininger & Squire, World Bank 2004 34 68.0 James, Arndt, and Simler (2005) 1 2.0 World Bank Poverty Monitoring Database 2002 7 14.0 World Bank, Africa Department 1 2.0 World Bank, World Development Indicators 2004 7 14.0 Total 50 100.0

PAGE 55

55 Table 3-4. Countries Frequency Percent Western Africa (50%) Burkina Faso 24.0 Cameroon 24.0 Cote d`Ivoire 612.0 Ghana 48.0 Gambia 48.0 Guinea 24.0 Nigeria 510.0 Eastern Africa (26%) Ethiopia 36.0 Kenya 36.0 Malawi 24.0 Tanzania 24.0 Uganda 36.0 Southern Africa (24%) Mozambique 24.0 South Africa 24.0 Madagascar 48.0 Zambia 48.0 Total 50100.0

PAGE 56

56 YEAR2003 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1989 1988 1987 1986 1985 1980 8 6 4 2 0 Figure 3-1. Frequency of years of inequality measured

PAGE 57

57 Missing 3 2 1 Figure 3-2. Quality of inequality data Range from 1 -4, with 1 indicating the highest quality, 4 worst. Missi ng case is from Mozambique.

PAGE 58

58 GINI62.5 60.0 57.5 55.0 52.5 50.0 47.5 45.0 42.5 40.0 37.5 35.0 32.5 30.0 12 10 8 6 4 2 0 Std. Dev = 7.39 Mean = 49.6 N = 50.00 Figure 3-3. Histogram of Gini

PAGE 59

59 CHAPTER 4 DEVELOPMENT AND AGRICULTURAL GR OWTH IN SUB-SAHARAN AFRICA Case Study: Burkina Faso Burkina Faso is a landlocked West African na tion located within th e cotton belt and just south of the Sahel region (Bassett 2001, 17) It has experienced cons iderable growth in GDP since 1990, just above 4% per year, with GDP pe r capita, adjusted for pu rchasing power parity (PPP), increasing from constant 2000 US $ 889 in 1990 to 1093 in 2005 (FAO 2006; UN 2006) (Table 4-1 and 4-2 for all data ). Similar to its neighboring Sahelian countries, it remains a largely rural agricultural societ y dependent upon a few major cash crops for export revenues. In 1990, an estimated 92.4% of the total la bor force were agricultural workers15 (Earthtrends 2003) Agricultures contribution to the Gross Domes tic Product (GDP), in terms of value added, increased from 28% in 1990 to 31% in 2005, with the mean averaging at 31.7% from 1990 to 2005 (Table 4-1). Table 4-1. Summary Burkina Faso Mean GDP Growth (%) 4.13 Mean Agricultural Exports (% of GDP) 7.41 Mean Agriculture, Value Added (% of GDP) 31.69 Source: World Bank (2005), UN Developm ent Indicators (2006) and FAO (2006) Similarly, agricultural exports, measured in US$ million, have increased from US$ million 92.10 in 1990 to 313.70 in 2004, a growth of over 340%. The role of agricultural exports, even when adjusted for overall growth in the GDP has increased, accounting for 5.3% in 1990 to 9.9% in 2004, with a mean for the four years of data available of 7.4%. Given the heavy reliance of most West African agricultural exporters on a few commodities, mostly cotton, considerable fluctuations are in order when ta king into context the volatility of the global commodity markets 15 It is important to note that an estimated three million Burkinabis are migrant laborers who work in Cte dIvoire and send home remittances, contributing significantly to the economy (new-agri.co.uk 2006).

PAGE 60

60 (Norton 2004, 56) Burkina Fasos top four export commodities in 1989-1991 accounted for 87%, decreasing to 83.8% in 1999-2001, and re covering to 86.75 and 91.35% in 2003 and 2004 (Table 4-2). As such, the share of the top four commodities seems to follow a similar path as the share of agricultural exports previously analyzed. What follows is that Burkina Faso, similar to its neighboring states, such as Mali, has failed to diversify its agricultural export commodities and holds a large dependence on cash crops similar to most lesser developed nations (LDCs) in sub-Saharan Africa. In 2004, cotton lint remained the king of these cash crops Its value totaled over 264 million US$, followed by a distant second, sesame seeds, with 10 million US$. This difference between the overall top two export commodities was vastly smaller in the years 1990 and 2000 when the share of agricultural exports as percenta ge of the GDP was small as well. As reflected in both 1990 and 2000, cottons value accumulated to 86 and 75 million US$ respectively, with the second highest export commodity, cattle, av eraging more than 7 and 8 million US$ (FAO 2006). An estimated three million, mainly subsistence farmers, depend on cotton production for their income whilst a further three million are i nvolved in the industry (new-agri.co.uk 2006). The average income of such cotton farmers is hovering around $250, which is strongly influenced through the artificially low global agricultural prices depressed by Western agricultural subsidies cu rrently disputed at the WTO (Baf fes 2004, 2005; Stiglitz 2006, 85f). Cotton overall is marketed by the parastatal, SO FITEX, with the cotton company still hold[ing] a monopsony on cotton purchases, but producers acqui red 30% of the companys shares in 1999 (Baffes 2004, 13). One of the main reason for th e phenomenal growth of cotton, averaging 12% per year from 1995/96 to 2004/05, is the provision of a guaranteed minimum price by the ginners

PAGE 61

61 one year before selling cotton fiber, which pr ovides stability to the farmer while allowing profitability for the parastatal (Goreux 2005). Furthermore, in October 2003, the Cotlook A Index, which is the world price for high quality co tton used in the West African case, reached a six-year high of US$ 1.55 per kg, significan tly boosting production and export revenue (FAO 2004). The value of cotton exports has thus increased dramaticall y, raising from US$ 103 million in 2002 to US$ 224 million in 2003 and US$ 261 million in 2004 (FAO 2006). However, since 2004 the world prices have declined again, dropping by approximately 25%, showcasing the price volatility in a globalized market distorted by Western export subsidies and Chinas vast influence (Cotlook 2007). Livestock used to be another mainstay of Burkina Fasos export economy. However, the number of livestock per person appears to have decreased drastically. While Burkina Faso averaged the most sheep and goats per person, w ith 5.74 head, in sub-Saharan Africa in the early 1980s, the number has dropped now to around on e per person by 2004. Cattle, on the other hand, has not experienced the same dramatic dr op, as per capita cattle stocks was 0.41 in 2004, compared with 0.40 in 1961 and 0.42 in 1980 (FAO 2006). Overall, export tr ade in livestock has decreased from 25.2 million US$ in 1980 to 19.7 million US$ by 2002, while total agricultural exports have increased from 79.6 million US$ in 1980 to 271.9 million US$ in 2002 (FAO 2005). Other major cash crops produced that c ontributed to this growth are groundnuts and sesame. Subsistence crops are sorghum, millet, maize, and rice (new-agri.co.uk 2006). While this dominance of cotton certainly appe ars problematic, theoretically increases in cotton production through heavier fe rtilizer usage can also results in higher yields for other crops due to fertilizer residues. As opposed to much of sub-Saharan Africa, which has failed to increase cereal yields in re lationship to per capita growth with production from 1979-81 to 1999-

PAGE 62

62 2001 dropping by 11%, Burkina Faso has increased per capita cereal production by 34% over the same time period. Furthermore, while still re sting below the sub-Saharan African average crop yield of 1221 kg per ha, it has al so increased its yiel ds by 53% to 880 kg per ha from 1979-81 to 1999-2001 (Earthtrends 2003). Comp ared to both South Americ a and Asian countries, these yields figures however still are extremely low, with South America averaging 3016 kg per ha and Asia (excluding the Middle East) with 3294 kg pe r ha (Figure 4-1). The average yields of cereals in 1999-2001 of Burkina Faso pale in comparison to the aver age yields of a select Asian agricultural exporters, ra nging from a low of 2661 kg per ha in Thailand to a high of 4084 kg per ha in Vietnam (Figure 4-1) (FAO 2006). Nevert heless, comparatively more impressively is Burkina Fasos achievement of increasing its per capita food production by around 25% since 1965, which is a significant achie vement given the population gr owth to almost 14 million people by end of 2004 (Table 4-3). Table 4-3. Burkina Faso, food pr oduction per capita index, 1965=100 1965 1970 1975 1980 1985 1990 1991 1992 1993 1994 100.0 97.2 89.9 80.0101.493.1108.7110.6 121.3108.2 1995 1996 1997 1998 1999 2000 2001 2002 2003 112.2 119.8 104.2 121.3121.499.5129.9124.3 127.0 Source: FAOSTAT (2006) The reason for that increase in production ou tput, according to Gray and Kevane (2001) rests in a strategy of intensif ication. They argue that the dominant population-degradation narrative does not hold true for Burkina Faso, as farmers, faced with an increasing pressure of smaller plots of land given the increases in population, have responded through corresponding increases in intensification, la Boserup. Consequently, they find that farmers in the southwestern cotton zone in Burkina Faso, living in the more densely populated village has more intensive production practices, [ and] farm ers are making changes in their management strategies that result in improved soil quality (Gray and Kevane 2001, 574). The village with

PAGE 63

63 dense population showed that fertilizer use was in general very high; it was applied to 65% of all fields [ while m]anure use was lower [with ] only 20% of fields were manured (Gray and Kevane 2001, 578). As a matter of fact, fertilizer usage intensity, measured in kg per hectare of cropland, was above the sub-Saharan African average in 1999 of 12 kg/ha, reaching a high of 15 kg/ha (Earthtrends 2003). While farmers overall a ppear to have responded, there however also diverging response paths, with wealthier farmers being able to better se cure land tenure rights than poorer farmers, widening the income gap to be analyzed in the late r chapters (Reardon and Tayler 1996; Gray and Kevane 2001). It however is important to keep in mind that Gray and Kevanes study focused on three villages in the hear t of Burkina Fasos southwestern cotton zone during the 1995-96 agricultural season (Gray a nd Kevane 2001). While providing us with a narrative of inequality in that area, it is impor tant to point out that the following inequality analysis, by taking place at the national level, wi ll be insensitive to any regional differentiation, and thus their case study can only provide a glimps e into the complexities measured by the Gini coefficient. Case Study: Cte dIvoire Cte dIvoire, a densely populated West Af rican nation of over 18 million people, is named after the thriving tr ade of ivory in the 17th century which attracted European traders and ultimately its French colonizers (Britannica 2006). Consequently, it enjoys a situational advantage compared to its north ern neighbors by having access to th e Atlantic Ocean, as well as a less arid climate. The southern region of Cte dIvoire is dominated by cocoa beans, with total production accounting for an estimated 37% in 2004 and 40% in 2003 of total global export value of cocoa beans (FAO 2006) While overall exports were relatively low, with 332,000

PAGE 64

64 metric tons in 1980 compared to the second la rgest exporter of cocoa beans, Ghana, it has increased to more than 1.2 milli on metric tons by 2003 (World Bank 2005) Its northern, more arid region, however, similarly to the southern re gions of Burkina Faso, is part of the West African cotton belt, with the largest political regions (d epartments) being Korhogo, Mankono, Boundiali, and Ferkessedougou. As of 1995-96, these regions themselves accounted for approximately 60 to 70% of overall cotton production in the country (Bassett 2001, 16) While cotton has also been considered an agricultural success story in that country, the country has experienced disappointing economic growth over the past decade compared to Burkina Faso and currently is embroiled in a low level civil wa r. The major exception to this narrative of economic decline is for the years following the 50% currency devaluation in 1994 by the West African franc (FCFA), as pressured for by the In ternational Monetary Fund (IMF) and the former colonial power, France, temporarily revivi ng economic growth (Bassett 2001, 148). GDP has increased by only 1.4% per year on average sinc e 1990. Of the 16 years measured, only seven years reported positive growth figures (19941999, 2004), with the other nine experiencing negative or zero growth. Conse quently, its GDP per capita, as adjusted for purchasing power parity (PPP), decreased from constant 2000 US$ 1807 in 1990 to US$ 1401 in 2005 (Table 4-4 and 4-5 for the data). Table 4-4. Summary Cte dIvoire Mean GDP Growth (%) 1.44 Mean Agricultural Exports (% of GDP) 25.60 Mean Agriculture, Value Added (% of GDP) 25.69 Source: World Bank (2005), UN Development Indicators (2006) and FAO (2006) While the agricultural sector is largely dependent upon a few major cash crops for export revenues, with cotton leading the way, its division of labor appears to be diversified compared to

PAGE 65

65 its neighbors, as only an estimated 59.9% of the total labor for ce were agricultural workers in 1990 (Earthtrends 2003) In comparison to Burkina Faso, these sixty% interestingly account for a similar share of total value added of agricultu re, 26%. Furthermore, the mean contribution of the agricultural exports to the GDP are significantly higher, 25.6%, compared to 7.4% in Burkina Faso, reflecting agricultures importance as a ma in source of foreign cu rrency. Especially in recent years, when there have been incentives for smaller farmers to grow cotton in Cte dIvoire, these export earnings were growing rapi dly. Agricultural expor ts, measured in US$ million, have increased by 87% from $1.65 billion in 1990 to $3.09 billion in 2004. The role of agricultural exports, even wh en adjusted for overall changes in the GDP, has grown substantially, from 19.9% in 1990 to over 30% by 2003 and 2004, with a mean increase of 26% over the four years measured. However, as opposed to Burkina Fasos high dependence upon four export commodities, averaging around 90%, C te dIvoires top four agricultural export commodities averaged around 70% of all agricult ural exports, with a high of 76% from 19891991 to a low of 69% in 200416 (Table 4-5). From an overall agricultural production sta ndpoint, Cte dIvoire has been able to increase its per capita production of cereals (as measured in t ons per person) since 1979-81 to 1999-2001 by 15%. This is a signific ant increase in contrast to th e overall decline in per capita cereals production sub-Saharan Africa of 11% to 135 tons per person in 1999-2001. Compared to Burkina Faso, its actual average cereal yiel ds are above those of the sub-Saharan African 16 These contrasts are sometimes not as stark when focusing on the top 3 of the top 20 export commodities for a given year. In 1998, for example, cotton lint, cattle and gr een beans accounted for 89% of total value of the top 20 exports in Burkina Faso. In Cte dIvoire, on the ot her hand, cocoa beans, green coffee and cocoa paste also accounted for about 86 % in 1998. However, only three years ear lier, cocoa beans, green coffee and cotton lint accounted for only 57% of the total share of exports of the top 20 commodities. While both cocoa or cotton dominate the value share of the top 3 export commoditie s over the past decade, there appears considerable fluctuations in their share depending heavily upon their value (market price), environmental impacts (droughts), and overall planting decisions made by the mostly small-scale farmers of Cote dIvoire. These factors clearly reach across scales, from the international to the local.

PAGE 66

66 average of 1221 kg per ha, with 1307 kg per ha, an increase by 51% in its yields since 1979-1981 (Earthtrends 2003) Correspondingly, while average fertiliz er use intensity, measured in kg/ha, declined from 11.87 kg/ha in 1979-81 to 10.23 kg/ha, it has reached an estimated new historic heights with 16.4 kg/ha in 1997 and 15.8 kg/ha in 2002 (FAO 2006) According to the FAO Food Production Index per capita estimates, Cte d Ivoire has increased its levels by around 10% since 1965, with as high as 121.7 in 2000. Intere stingly, as opposed to Burkina Faso, Cte dIvoire never experienced a drop below 100, which a ppears to indicate a re latively stable supply for food production over the past 40 years (Table 4-6). Table 4-6. Cte dIvoire, food production per capita index, 1965=100 1965 1970 1975 1980 1985 1990 1991 1992 1993 1994 100.0 102.9 124.7 113.2109.5108.2106.6105.4 103.8102.7 1995 1996 1997 1998 1999 2000 2001 2002 2003 113.7 119.5 116.9 116.0119.5121.7116.8114.1 109.5 Source: FAOSTAT (2006) Overall, Cte dIvoire is cl early part of the cotton belt, but has a more diversified agricultural export base. Beside s cotton in the northern region17, there has been a dramatic increase in coffee and cocoa production [ ] between 1950 and 1990 through agricultural extensification (Bassett 2001, 9) According to Bassett, these two cash crops, however, have had significantly lower yields than the in ternational average, around 300-450 kg/ha, and the subsequent expansion in total output of th ese two mainstays of the Ivorian economy has historically been the result of pioneer agriculturalists establis hing new plantations in sparsely settled areas of the tropical rainforest (ibid., 9) 17 Interestingly, the reason for farmers in the north grow ing cotton despite bad financial conditions often result from the lack of a profitable alte rnative cash crop, especially since no othe r crops have received as much support and attention.

PAGE 67

67 This path of extensification production runs counter to the intensification approach taken with cotton production. According to Bassett, obse rving the astonishing gr owths of yields by an average of more than 4% per year from 1965 84, he confidently concludes that there is ample evidence to label this achievement a cotton revolution and one of the few agricultural success stories in sub-Saharan Africa (ibid., 9) In the past decade, there have been many dynamic changes in the cotton sector, incl uding privatization of the cotton sector, a devaluation of the currency, the FCFA, introduction of new, unsatisfactory cotton varieties (GL7), and a recent recovery of producer prices. While some pr ominent agricultural economists and political scientists perceive poor and il literate farmers as incapable of having their voices heard and influencing their fates, farmers in the Ivory Coas t have used diverse copi ng strategies. Not only did they use both diversification and extensification strategies in the face of decreasing terms of trade and profit margins, they also exerted pr essure through their individual farm management decisions as well as through thei r organizations (Bassett 2001, 173) While the fate of the producers under the newly privatized system in the new millennium is unclear18, it seems evident that the prior monopolistic Compagnie francaise pour le deve lopment des fibres textiles (CFDT) system that existed since 1949 and coopera ted with the local agricu ltural service of Cte dIvoire (IRCT) and the established Compagnie ivorien pour le dveloppement des fibres textiles (CIDT) was very successful in providing the Ivorian cotton farmers with (subsidized) input packages and guaranteed purchasin g prices, while garnering all the profits as the sole buyer from controlling the ginning and sa le of cotton (Bassett 2001, 176) Nevertheless, they did not 18 The CFDT system has undergone restructuring throughout the past decades, was renamed DAGRIS in 2001 and has currently undergone privatization, as of 2006, as the French government sold its majo rity shares in the company (UNCTAD 2006). Furthermore, the parastatal CIDT was taken over by three other companies in 1998: Ivoire Coton (northwest block), owned by IPS of the AgaKhan group; Compagnie Cotonniere (northeast block), owned by the Aiglon, a Swiss company; and Nouvelle CIDT (USDA 2002).

PAGE 68

68 succeed in improving these rural livelihoods, as inc omes in real terms steadily declined over the period 1970-94 and that cotton grow ers received a relatively small sh are of the market value of their crop (Bassett 2001, 180) By 1998, several private companie s have built cotton gins, and due to better seed prices, yields and rainfall, record growth se asons have been achieved, and a new cotton boom seemed to have taken place (Bassett 2001, 181) However, in the last few years, this apparent boom appears to have cooled down according to the latest FAO statistics, which show that the quantity of cotton seed produced dropped from a high of 220,000 tons in 2000 to a low of 161,000 tons by 2004, with a simultaneous drop in the area under cultivation from 330,000 ha to 206,360 ha in 2004 (FAO 2006). Unfortunately, the main export commodity, cocoa, has experienced a similar bust, being affected by the current state of poli tical turmoil in Cte dIvoire, as well as one of its underlining causes (Woods 2003; Kurti 2005) As argued by Woods (2003), Cte dIvoires property rights regime has been a major cause of the recent ethnic conflict and civil war there, as the tutorat has been put in place, which is an institution that decides over access to land by immigrants (Chauveau 2003; Richards 2004, 15) According to Woods, it en couraged a rent-seeking process that shifted social and political interactions fr om a relatively peaceful dynamic to a more conflictual one, as the rent derived from the e xploitation of tropical fore st land declined (Woods 2003, 642) With decreasing availability of forest land to be cleared, the so-called forest rent, benefits accrued from planti ng cocoa trees, diminished and became increasingly difficult to obtain. Consequently, by 1998, the 1967 decree da ting back to the rule of Felix HouphouetBoigny stating that land belongs to the pers on who cultivates it with its aim at spurring cultivation, was replaced with a law enforced by the current Gbagbo government giving priority of access to traditional, ancestral land rights (K urti 2005). Ultimately, as not to lose political

PAGE 69

69 support, the current administration failed to ad dress the property rights issues, rather blaming foreigners and Muslim Northerners for the weak economy and lack of availa ble land, resulting in the north rebelling and the onset of a civi l war (Woods 2003, 14). The Gbagbo government has consequently seeded the venomous flow ers of Ivoirite, according to Lemarchand, a murderous ideology which calls into question the citizenship rights of perhaps as many as three million long-time Ivoirian residents of forei gn extraction (Lemarchand 2003, 21). While the conflict, involving the rebel par ties in the North and the governme nt in the South is certainly more complex than only resolving around cocoa land, it seems clear that agri culture plays a large role. Richards goes so far as proposing that the reason for the popularity of the rebel forces found in West Africa rests on their attempt at inst ituting an agrarian reform against the village elders and political elites, which have exploite d the poor, landless and uneducated for their labor and services for centu ries (Richards 2004, 15f) He cites the case of a young RUF commander, who created a rural farm development programme after demobilization for 60 former comrades and over 600 villagers with the ultimate ai m of creating a mass movement for agrarian change (Richards 2004, 16) While the ultimate outcome of the civil war is yet to be determined, an analysis of changes in total production, as measured in 1 tonnes, from 2001 to 2002 and 2001 to 2003 indicate that out of the 60 total commodities reported by the FAO, only 14 in 2002 and 21 in 2003 have declined in production compared to 2001. Since the civil unrest has been ongoing over the past ten years below the surface, these fi gures might not correctly capture the full effect the civil war had on agriculture (FAO 2006) As a matter of fact, both cotton and cocoa

PAGE 70

70 production seems to have recovered in the last years again, and out of the main cash crops, only coffee and to a lesser case pineapples, have experienced a continuous decline19 (Table 4-7). Table 4-7. Production in Cte dIvoire 1999-2005 of commodities facing declines since 2001 Production (1'000 mts) 1999 2000 2001 2002 2003 2004 2005 2002 <2001 2003 <2001 Avocados 2.1 2 1.51.211.051 11 Cassava (fresh and dried) 1,681 2,100 2,0872,0742,0602,1282,198 11 Coffee, green 307 336 20918214015496 11 Cucumbers and gherkins 29.36 27.22 25.6424.9917.3517.9218.52 11 Large Pelagic fish 28.93 33.68 33.0123.121.321.321.3 11 Millet 65 75.8 7370456060 11 Molluscs (excl. cephalopods) 0.08 0.05 0.080000 11 Nuts, nec 80.2 80.22 85.7283.0766.9474.5283.09 11 Pineapples 257 268.2 284.82271.92242.23248.99194.5 11 Sheep and goat meat 10.51 9.49 9.619.439.039.439.28 11 Source: FAOSTAT (2006) Case Study: Ghana Half a decade after having gained independence as the first African nation, Ghana continues to present itself as one of the economic and political success stories in Africa and optimism for its future prevails. Fifty years afte r the revered first leader of independent Ghana, Kwame Nkrumah, called for pan-Africanism, Ghana has emerged as one of the leaders on the African continent, hosting the 2007 meeting of the African Union (AU) dedicated to one sole agenda item the AUs Study on an AU Govern ment: Towards a United States of Africa (Wanyeki 2007). 19 Nevertheless, a clear drop in th e FPI can be identified post-2002.

PAGE 71

71 Table 4-8. Summary Ghana Mean GDP Growth (%) 4.44 Mean Agricultural Exports (% of GDP) 15.27 Mean Agriculture, Value Added (% of GDP) 38.69 Source: World Bank (2005), UN Development Indicators (2006) and FAO (2006) Compared to its neighboring c ountries of Burkina Faso in the North, Togo to the East and Cote dIvoire to the West, Ghana has achieved as tonishingly consistent high rates of growth over the last two decades, with a mean GDP grow th of 4.44% from 1990 to 2005, and relatively low population growth rates of 2.05% in 2005 (Table 4-8 and 4-9). Agriculture continues to be the largest contributor towards the GDP, as measur ed in value added, contributing 39% by 2005 (UN 2006). For 22 million people of Ghana, agriculture continues to be the main source of income, as an estimated 56% of the labor force are part aking in agriculture in 2004, a figure that is similar to Cote dIvoire, but very low compared to their northern neighbor of Burkina Faso (FAO 2006). While gold has risen to the top of the count rys source of foreign exchange, aptly named Gold Coast, agricultural exports account for ove r 15%, with slightly higher figures identified for 2003 and 2004 with 18.19% and 20.09% respectively (Table 4-8). Ghana consequently represents an African nation th at has achieved a sustainable ec onomic transformation, including development of the manufacturing sector without neglecting the ag ricultural sector. According to Stiglitz, this success was aided by new poli tical leadership in the 1990s that was more receptive to pursuing good economic policies, including investment promotions and the creation of one of the worlds most profitabl e stock market (Stiglitz 2006, 41; Moss 2007, 231). This dynamic, mostly liberal, strategy has resulte d in one of the highest GDP per capita, PPP, of sub-Saharan Africa of a non-oil exporting nation, growing from 1589.10 US$ in 1990 to 2149.00 US$ in 2005, an increase of over 35% over these 15 years (UN 2006).

PAGE 72

72 One of the main reasons for the economic succe ss is Ghanas ability to slowly reduce the trap of dependence on global price volatilities experienced in othe r African nations. Historically, cocoa beans, butter and paste have been the three main agricultural export commodities, accounting for example in 1987 for 97% of the top 20 export commodities value (FAO 2006). Cocoa, one of the worlds most important tree crop, was introduced in Ghana during the late 19th century and experienced a rapid growth (Edwin and Masters 2005, 491). Prior to independence in 1957, cocoa accounted for more than 70% of the countrys export earnings from 1900 to 1950 and became the worlds leading exporter by 1910 holding 30 40% of the global market share (Konandu-Agyemang and Adanu 2003, 516; Ed win and Masters 2005, 491). Observing the top four agricultura l exports in 1989-1991, the dependence within the ag ricultural sector is equally dramatic with 96.09%. However, starti ng in 2000 a considerable can be observed, with the share of the top four agri cultural exports slow ly decreasing from 86.60% in 2000 to 85.77% in 2003 and 83.30% in 2004 (Table 4-9). As a ma tter of fact, by 1998, pineapples has risen into the top three commodities exported, gaining a valu e of 11.7 million US$. This reflects Ghanas leading efforts in nontraditional exports under the ongoing structural adjustment programs, reduc[ing] the vulnerability of the economy by diversifying the sources of export earnings, but also engendering participation in the export trad e by regions that have been traditionally left behind (Konandu-Agyemang and Adanu 2003, 513). By 2000, cocoa itself accounted only for 22.5% of Ghanas overall exports, with minerals, mostly gold and to a lesser extent diamonds, accounting for 40%, and timber, the third main e xport, ranging at 9.0%. Nontraditional exports (NTE), on the other hand, have grown to 13.6% of total exports (Konandu-Agyemang and Adanu 2003, 517). It is important to note that cocoa production however has not experienced a bust (Table 4-10). While it has declined in the 1970s and earl y 80s, in 1984, numerous

PAGE 73

73 economic reforms, including regula rly increasing farm gate pr ices, spurred a recovery and production level, thanks to improved genetic vari eties and continued fertilizer usage, have recovered (Edwin and Masters 2005, 491). The range of NTEs is wide and large, mostly horticultural products, including pineapples, flowers, and bananas to wood products such as carvings, coffins, and wodden poles (KonanduAgyemang and Adanu 2003, 519). Beyond these more traditional exports, they also include the increased trade in items that previously were assumed to hold no value, such as cocoa waste, orange peels, and cornhusks, as well as reptiles (Konandu-Agyemang and Adanu 2003, 519). The latter appears to be by far the most exotic commodity traded, yet possibly also one of the fastest growing. By 1999, exports of around 7 000 adult and 30 baby pythons amounted to annual export earnings of 1.4 US$ million (Konandu-Agyemang and Adanu 2003, 519-520). In summary, Ghana appears to be well on its way to transition from its trinity exports of cocoa, minerals and timber, towards a more diversified and dynamic economy. While poverty rates are as high as 90% in parts of the northern and central area of Ghana, the country has secured continued internati onal support, including a US$ 547 million antipoverty Compact from the US Millennium Cha llenge Corporation (MCC) in 2006 (MCC 2006). Divided into three sector s of agriculture, transportation and rural development, the key is to alleviate the poverty of over 230,000 Ghanaians a nd to enhance the livelihood and welfare of one million Ghanaians in total (MCC 2006). Si nce clear spillovers exist for each section, farming receives a total of US$ 241 million, with an emphasis to provide training for further commercialization, irrigation improvements, incr eased processing and access to credit (MCC 2006). Consequently, Ghana appears well on the road towards continued strong and hopefully equal growth.

PAGE 74

74 Case Study: Senegal Located at the Western tip of sub-Saharan Africa facing the Atla ntic Ocean, Senegal, reaching independence from France in 1960, is well known as one of the most stable democracies in Africa. Since 2000, the forty year majority rule of the socialist government has ended and Senegals opposition l eader, President Abdoulaye Wade was elected, ending a four term rule from 1981 to 2000 by former Presid ent Abdou Diouf (CIA 2007). As of March 2007, the eight-plus year old Wade was re-elected as president in what was termed a (relatively) free and fair election (IRIN 2007). Table 4-11. Summary Senegal Mean GDP Growth (%) 3.38 Mean Agricultural Exports (% of GDP) 4.36 Mean Agriculture, Value Added (% of GDP) 52.88 Source: World Bank (2005), UN Development Indicators (2006) and FAO (2006) Overall, agricultural performan ce has been declared to be in a state of crisis by the government over the 1970s and 80s, as [agricult ural] growth has not exceeded the population growth rate (Government of Senegal 1994). Comparing agricultural cereal per capita production has continued to dr op, as they decreased by 27% from 1979-81 to 1999-2001 to 113 tons per person (Earthtrends 2003). Senegal overall continue s to experience very high population growth, with the average growth fi gure ranging from a high of 2.89% in 1989-1991 to a low of 2.36% in 2005, decreasing every year sli ghtly. However, overall population increase is very large, with a growth of almost 50% over the past 15 year s, from around 8 million in 1990 to more than 11.5 millions in 2005. Starting with a seri es of liberalization and structural adjustment efforts in 1980 and throughout the next two decad es, Senegal has enjoyed an increase in GDP per capita (PPP) over the last 15 years fr om around 1,400 US$ in 1989-1991 to 1615 US$ in

PAGE 75

75 2005 (Table 4-11 and 4-12 for all data). While some years were lower, such as 2002, from 2000 to 2005, five out of the six years experienced GDP growth that was greater than population growth. The role of the agricultural sector in comp arison to the GDP appears to have decreased slightly, from 19.45 in 1989-1991 to around 18% in 2005 as measured in terms of value added. Correspondingly, agricultural exports have decrea sed significantly from 5.67% in 1989-1991 to 3.47% in 2004 in relationship to the GDP. Most significantly and telling ha s been the decline in the share of the top four commodities of all agricultural exports, as measured in value. While they averaged 81.80% in 1989-1991, the sh are has decreased to only 56.5% by 2004, representing a major overall shift in agricultural production in Senegal. The colonial government in Senegal was very successful in rapidly pushing for a transformation of a subsistence economy into an export-oriented, cash cr op economy (Kelly et al. 1996, 12). The main cash crop that was prom oted was peanuts in combination with a subsistence crop of millet or sorghum, with the governmental aid of provision of inputs and credit, import of rice from Asia, and guaranteed purchasing of cash crop s (ibid., 12). This policy of agricultural support continued into th e post-colonial era, w ith the government taking gradually over the former French trading houses with a government pean ut and input marketing parastatal (ibid., 13). This indirect subsidy pr ogram by the government worked for periods of strong international price for peanuts and good yields, howev er faltered as farmers experienced increasing droughts and were unable to pay back their credits. As the fiscal deficit accumulated by the government through its control of the agricultural market became increasingly unsustainable, it began to institute structural ad justment programs for the agricultural sector in 1980 under the pressure of major international d onors and international organizations, notably

PAGE 76

76 the IMF, WB, US and France. The ultimate ai m of these policies were to curtail direct government involvement in the agricultural sect or and to eliminate government subsidies and taxes to the greatest exte nt possible (ibid., 18). While some private actors have entered the market, it has overall enjoyed a reduction in overall fertilizer usage, as credit and subsidy programs were no longer in place, and stagnating yields from 1960 until 1990. However, since 1979-81, average crop yields appear to have increased again by 24% to 854 kg per ha, still around 1/3 lower than the sub-Saharan average (Earthtrends 2003). While the Food Production Index per capita (Table 4-13) appe ars to have some serious data i ssues (as seen in the base year of 1965 and 1975), it has remained stagnant and not improved over the past forty years, but rather decreased, as seen in the drop in 2002 (FAO 2006). Upon closer investigation, the 2002 agricultural season experienced a prolonged dry spell, a 7% lowe r cereal output compared to 2001, with a particularly sharp fa ll of millet by 21.7%, and subse quent dramatic increases in price for that crop20 (FAO 2002). Similarly, farmers had increasing difficult y obtaining access to fertilizer, as a restriction has been put in place that only farmers with a 100% repayment record are eligible to further credit for inputs (seed and ferti lizer) (FAO 2002). Overall, the government of Senegal is still reluctant to fully liberalize its markets, as it continues to set producer prices and applies high tariffs on pr ocessed groundnuts [20%] to encourage domestic processing or oil production (D iop, Beghin, and Sewadeh 2005, 227). In terms of export commodities, groundnuts (oil and/or cake) still ac count usually for the largest share of all exports, howev er no longer hold the top share as measured in value in recent years. In 1995, the top 3 commodities, oil of groundnuts, cotton lint and Arabic gum, accounted for approximately 80% of the top 20 export va lues. Prior to 2000, an estimated one million 20 Yield drops were even more dramatic for other food cr ops, such as rice (54%), maize (45%) and sorghum (34%), with the first two known to be the most sensitive to irregular rainfall (FAO 2002).

PAGE 77

77 people, about ten% of the total population, was believed to be engage d in groundnut processing and farming, contributing about 2% of [GDP] and 9% of exports in Senegal (Beghin et al. 2003, 4). By 2000, oil of groundnuts, oil of palm and cake of groundnuts were the three export commodities that accounted for only 67% of the top 20 export values. Finally, by 2004, cotton lint was the most valuable export commodity, netting 29 millions US$, with oil of groundnuts and prepared food accounting for second and thir d respectively. Overall, by 2004, these top 3 export commodities accounted only for 50% of th e top 30 export values share (FAO 2006). In summary, while maintaining around the same shar e of agricultural out put, it appears that Senegalese farmers have undertaken the important step of diversifying their production away from groundnuts, one of the rare cash commoditie s experiencing lower vol atility yet long-term, recently rapid price declines, and Senegals prim e export of groundnut oil, which is increasingly being replaced by cheaper vegetable oils (B eghin et al. 2003; Diop, Beghin, and Sewadeh 2005). Case Study: Ethiopia Unfortunately, when depicting Ethiopia, one qui ckly is tempted to denounce it as a country of over 70 million people ravaged by constant hunger, omnipresent food insecurity and overwhelming famines. Marked by 15 major fa mines over the hundred years before the epic famine in the mid-1980s that caused the deaths of over half a million people, Ethiopia has become synonymous with crisis (Block and Webb 2001, 335). Given this dominant, dramatic narrative of famine in Ethiopia, it should be ve ry interesting to take a closer look at its agricultural production and developmental figures since the 1990s, especially in comparison to other sub-Saharan African nations. Located in the horn of Africa, Ethiopia ha s been left landlocked since 1993 in the aftermath of its violent conflict with Eritrea. As a matter of fact, th is ongoing struggle only intensified the problems encountered of starvati on, highlighting the politi cal dimensions of food

PAGE 78

78 insecurity, as it has cost as ma ny as 50,000 lives and resulted in deteriorating conditions of food security in the border region (White 2005). While the recent conflict with Somali Islamist forces appears (or at least appeared) to be under control, as the Islamist fighters have fled the capital city of Mogadishu and the Somalias interim govern ment took over, it nevertheless indicates the constant political insecurity f aced within the region, worsening peoples lives and the countrys overall development (BBC 2007). Table 4-14. Summary Ethiopia Mean GDP Growth (%) 3.38 Mean Agricultural Exports (% of GDP) 4.44 Mean Agriculture, Value Added (% of GDP) 52.88 Source: World Bank (2005), UN Development Indicators (2006) and FAO (2006) This development of the country, as meas ured by the average GDP growth from 1990 2005, has been better than one could expect, with an average growth of 3.38%. However, hidden in that figure is the fact that the economy has grown at tremendous swings, as the GDP for example dropped by 11% in 1992, only to rise from the ashes with a growth of 13% the next year in 1993, likely reflecting the granting of i ndependence to Eritrea, ending several years of strive. In a similar fashion, 2002 and 2003 witnesse d zero or negative GDP growth, with a vast rebound to 12% by 2004. This is likely related to the complex 2002-2003 famine, which slowed down the economy. After a bumper harvest in 2001-2002, prices dropped rapidly and several farmers reported a net loss that did not cover th eir expenses. Consequently, the following year, in order to mitigate losses and costs of labor, se veral farmers reduced seed, fertilizer, fuel and planting acreages. However, as a drought h it Ethiopia that next year, yields dropped dramatically and a dramatic f ood shortage become evident (Thur ow 2003). As reflected in the agriculture value added figures for these years, it actually increased from the 2002 to the 2003

PAGE 79

79 season by 2%, with agricultural exports reach ing a high of 450 million US$ in 2003, indicating that plenty of food was exported. According to Thurow and Kilman, what took place was the perverse situation where US ai d policy, opposing the purchasing of local food during famines in favor of flooding markets with US grain, made the situation worse as American food traveled on roads that ran right past local warehouses fi lled with the 2002 Ethiopian harvest (Thurow and Kilman 2005, 2). It is important to note that even with 2001-2002 being a year with a bumper harvest, it is counted as a year where there was significant shortage of rai n, indicating an incidence of drought, according to the World Ba nk (Table 4-16). An analysis of the variable however indicates that it is, as ack nowledged by the World Bank, the only indicator [] based on subjective considerations (World Bank 2005, 242). A D consequently was assigned to a country if a significant shortage of rain unfavorably affected its agricultural production (ibid., 242). While this variable attempts to take into considerations the regiona l differences, it is clear that given the size of Ethiopia, th is subjectivity might be stressed greater than at other countries. Nevertheless, out of all mainland sub-Saharan Af rican countries, Ethiopia experienced by far the most droughts from 1980 2004, with 15 years, followed by Zambia with 11 years of droughts that negatively affected its ag ricultural production. Thus, these figures appear to confirm the dominant narrative surrounding Ethi opias crises (ibid., 240). Ho wever, as seen in the Food Production per capita index for Et hiopia (Table 4-15), the patterns are far more complex, as significant deviations in terms of food security appear to exist between the droughts. As a matter of fact, the index appears to have climbed up to a record high in 2002, while dropping slightly in 2003, both being drought years. Given the political di mensions of Ethiopia and the fact that data was not available prior to 1993, these figures should be evaluated with great caution.

PAGE 80

80 Table 4-15. Ethiopia, food produc tion per capita index, 1993=100 1993 1994 1995 1996 1997 1998 100.0 97.1 104.9118.2115.7 106.7 1999 2000 2001 2002 2003 110.5 113.2 121.5122.0114.6 Source: FAOSTAT (2006) The main solution proposed by international acto rs to such a crisis is often through an increase in food supply, with soil depletion named a major reason for poor yields. As a matter of fact, per capita production (tons per person) of cereals has dropped by 14% since 1979-1981 to 1999-2001. Furthermore, average crop yields (kg per ha) have declined by 2% over the same time period to 1164 kg per ha, which is not only be low the sub-Saharan average, but also 11% worse than the overall performance in the region, and 43% worse than the world (Earthtrends 2003)! Thus, an increase in fertilizer usage has generally been advocated as a popular approach towards amending these insecurities (Kristof 2005). While the most recent FAO statistics only show fertilizer usage until 2002, they have increased from a low of 7.2 kg per ha in 1993 to 14.1 kg per ha in 2002, with an estimated high of 16.9 kg per ha in 1996. The main applications of fertilizer in Ethiopia are mai ze, teff and coffee (Camara and Heinemann 2006, 10). Taking into consideration that 33 African nations fertili zer intensity usage rests below 10 kg per ha, Ethiopias current usage is actually relatively high. Taking into account that agricultural exports a ccount for approximately 90% of all exports in Ethiopia, the sectors overall export performance of agricult ure as percentage of GDP is extremely low ranging from 3.70% from 1999-2001 and increasingly slightly to 5.48% in 2003 and 4.12% in 2004, with each of those years affected by droughts (INTRACEN 2001). The overall agricultural exports, measured in millions US$, follow a similar pattern, as agricultural exports ranged from 290 million US$ in 1999-2001 to 450 million US$ in 2003. Aggregate share of the top four ag ricultural exports of total exports sim ilarly have decreased lately from a

PAGE 81

81 high of 84.14% in 1989-91 to a low of 71.39% in 2004, indicating a possible tr end in agricultural production. However, upon closer ob servation it becomes evident that Ethiopia continues to be heavily dominated by one major cash crop commodity: coffee. An estimated 1.2 million coffee farmers, mostly small-holders working on less than half an acre of land, and over 15 million households are either indirectly or directly dependent on coff ee for their livelihoods (Oxfam 2002). Measured in export value, coffee accounted for 70.8% of the top 20 agricultural exports in 1995, 77.7% in 1997 and 76.4% in 2000 (FAO 2006). Since 2000, Ethiopia has ranked amongst the top 15 exporters of green coffee, with the crop accounti ng around 55% of all agricultural exports in 2003 and 2004 (FAO 2006). The second largest cash crop, sesame seeds, app ears to have risen relatively quickly in comparison to coffee exports, as indicated in Figure 4-3 below.21 Whats astonishing is that even though sesame seeds exports have increas ed rapidly, coffee expor ts actually have not dropped, but rather increased over the past years. The lower export values are consequently attributed not to a decline in coffee production, bu t rather a drastic drop in producer prices over the past years in coffee (Figure 4-2). While bot h commodities have experienced severe declines, coffees producer prices have declined by approxi mately 40% over the span of four years from 2000 until 2003. Even though sesame seeds have also experienced a decrease in producer price, its decline is approximately 31% from its hi gh of 3 to its low of 1 in 2002. Coffee however has dropped just in one single year fr om a high of 7 to a low of 4, a drop of 44%, seriously jeopardizing the liv elihoods of local coffee farmer s, especially when coupled with droughts that reduce coffee production by as much as 20 to 30% (Oxfam 2003; FAO 2006). Furthermore, at the national level, the decrea se in world prices due to oversupply and low 21 Please note that the 2001 value has been omitte d, as it was reported as 28100 tonnes.

PAGE 82

82 demand elasticity has critically re duced foreign revenues for the gove rnment, as well as its taxes received from coffee exports22 (INTRACEN 2001). This overall crash in producer prices dates as far back as in 1989, when the international coffee market was liberalized, and new countries entered the already saturated market, such as Vietnam, which has become the worlds second largest coffee producer (Bloomer 2004). Given the inefficient and complex commodity ch ains of coffee production in Ethiopia, it is unlikely that an increase in worl d price would result in a signifi cant trickle-down effect for the farmer. As a matter of fact, Ethiopian coffee farmers on average only receive an estimated 27% of the export price, compared to 70% in Uganda and 80% in Kenya23 (Oxfam 2002, 8). Ultimately, neither coffee or sesame beans mi ght be the main agricultural income for a majority of these smallholders in the future. Numerous farmers have switched their production away from coffee towards chat, a mild amphetamine-like stimula nt illegal in the US that garners a significantly higher market price as global demand has increased (Bloomer 2004). Case Study: Kenya As one of the most prominent African nations, Kenya was ruled by one of the infamous big men of Africa, Jomo Kenyatta. After his death in 1978, Kenyatta left behind a legacy of tension between his Kikuyu and other ethnic groups as well as repressive state machinery in place for his successor, Daniel arap Moi who ruled until 2002 (Moss 2007, 43). As of today, the country is ruled by a fragile coalition govern ment, which appears to be devoted towards economic reforms, yet riddled by rivalries and jockeying for power (USAID 2005, 4). 22 The government has actually cut its export tax during the crisis in recent years. 23 These percentage figures are significa ntly higher for farmers in a co-operativ e, as they eliminate numerous chains of the coffee trade, including the aucti oning house. While not dealing with a parastatal, as in the Ivorian case, it appears that these farmers have found alternate ways in orde r to increase their share of the purchasing price (Oxfam 2002).

PAGE 83

83 Located on the Eastern border of Africa, Kenya is home to both one of the most productive agricultural regions, the Kenyan highlands, and a major hub for commerce and trade in Africa in Nairobi and port in Mombassa (CIA 2007). Riddled by corruption, Kenya has achieved disappointing growth over the past decade, av eraging 2.19% from 1990 to 2005 as measured in GDP (Table 4-17). Kenya has re ceived disappointingly li ttle foreign direct investment (FDI), below the sub-Saharan average, reflecting the general distrust by investors after the corruption scandals. It nevertheless appear s that with the passing of the landmark anti-corruption legislation and several other measures intended to restore its internationa l reputation, Kenya might be well positioned to enjoy considerable outside investments and economic growth in the near future (USAID 2005). Table 4-17. Summary Kenya Mean GDP Growth (%) 2.19 Mean Agricultural Exports (% of GDP) 8.24 Mean Agriculture, Value Added (% of GDP) 30.00 Source: World Bank (2005), UN Developm ent Indicators (2006) and FAO (2006) Even though the service and industry sector s dominate the economy, including smallerscale manufacturing and tourism, agriculture co ntinues to be a key component of Kenyas economy with a mean of 30% of its GDP ste mming from value added agriculture, dropping slightly to around 27% by 2005 (Table 4-18). As of 1990, an estimated 79.5% of agricultural workers existed as a percentage of the total labor force (Earthtrends 2003). The average GDP per capita (PPP), as measured in constant 2000 in ternational dollars, actually decreased from 1132.42 $ in 1990 to 1042 $ in 2005, illustrative of the poverty faced by approximately half of the 34.3 million residents of Kenya, especially in the rural countryside (CIA 2007). From an agricultural perspective, f ood production, as estimated through the Food Production per capita Index (Table 4-19), has actually not impr oved over the past 25 years,

PAGE 84

84 achieving only a value above 100 three (1990,1991,1992) out of the last 15 years. This actual decrease in per capita production is reflected also in the drop by 33% of per capita production, as measured in tons per person, of cereals fr om 1979-81 to 1999-2001. At the turn of the millennium, the 94 tons per person annual cereal yields were even 30% lower than the subSaharan average of 135 tons per person (Earthtr ends 2003). On a positive note, fertilizer use intensity, measured in kilograms per hectare, however is well above the sub-Saharan African average, reaching 27.7 kg per ha in 2002, compared to 12.4 kg per ha in sub-Saharan Africa. This is a substantial increase from only ten years ago, when it was as low as 21.6 kg per ha in 1992 and 14.7 kg per ha in 1982 (FAO 2006). Given th e prevalence of large scale farming in Kenya, however, it is likely that the overall fertilizer consumpti on is very unevenly distributed (and applied), heavily favoring th e exporting horticulture sector. The 1.3 US$ billion of agricultural exports in 2004 accounted for approximately 9% of total GDP, a figure that has increased substantia lly from its lower value of 6.31% in 1989-91 and 8.13% in 1999-2001. The undisputed top export continues to be tea, which accounts for approximately half of the top 20 agricultura l exports in 2004 with US$ 463 millions. While coffee (and to a lesser degree suga r) was the other tradit ional main export commodity, they have been replaced, similar to the case of Ghana, by non-traditional exports, such as pineapples and green beans. These two are part of one of the main success stories found in sub-Saharan agriculture surrounding horticu ltural produce production. Encompassing the production of vegetables, frui ts and fresh cut flowers, the horticulture sector is estimated to have generated jobs that directly support a half a million workers, small farmers and their families in Kenya (English, Jaffee, and Okello 2004, 1). Horticulture has grown dramatically, overtaking in recent years co ffee as the second largest merchandise export,

PAGE 85

85 and accounting for two-thirds of all growth in agricultural expor ts (English, Jaffee, and Okello 2004, 1). Fresh fruit and vegetable exports are estimated to have grown by over 500% to 164 $ million from 1991 to 2002, while the export of cut flowers has enjoyed si milarly vast growth rates of approximately 450% to 175 million $ from 1991 to 2002 (English, Jaffee, and Okello 2004). Most of these export commodities are produced through contract farming, a practice which clearly favors larger scale producers w ho have the resources of land, technology, labor, and credit to undertake la rge-scale, flexible pr oduction under contract for European importers (Barrett et al. 1999; Stock 2004, 219). Given the st ringent food standards and quality demands put in place by the European Union and the cont racting large-scale buyers, small-scale farming production of horticulture has become increasingl y problematic in a market heavily dominated by a few major players. Overall export produc tion is dominated by two-dozen large-scale farms, which account for 75% of the industry (English, Jaffee, and Okello 2004, 4). These problems not only relate to the cut flower sector, but also to the e xporting of vegetables as smallscale farmers are facing pressures to obtain food safety certificati on and to comply with ethical and environmental principles (English, Jaffee, a nd Okello 2004). The sector nevertheless is believed to be a successful tool towards pove rty reduction, if not th rough the increasing of farming income, but rather of providing increased opportunities for sources of off-farm income for the (urban) poor. Case Study: Uganda Uganda is one of the few African countries that has not only been able to achieve high growth rates in the 1990s, averaging 6.31% from 1990 to 2005 (Table 4-20), but also significantly reduce poverty, thus achieving pro-poor growth (Moss 2007, 173). GDP per capita purchasing power parity has increased by 55% from a low around 878 US$ in 1989-1991 to a high of 1363 US$ by 2005 (UN 2006). While a higher income does not have to reflect a

PAGE 86

86 reduction in poverty, Ugandas economic growth appears to have been translated into substantial increases in household consumption and a reduction in income poverty (IMF and IDA 2001, 2). Based on three household surveys undertaken during the 1 990s, the% of people living below the poverty line, also called the pov erty head count index, declined from 56% in 1992 to 44% in 1997, and then to 35% in 2000, an overall decrease by 38% from 1992 to 2000 (IMF and IDA 2001, 2). Table 4-20. Summary Uganda Mean GDP Growth (%) 6.31 Mean Agricultural Exports (% of GDP) 4.47 Mean Agriculture, Value Added (% of GDP) 42.56 Source: World Bank (2005), UN Developm ent Indicators (2006) and FAO (2006) Given this major success, it however is importa nt to note that ineq uality over the same period, measured from 1992 to 2000 has increas ed from 0.39 to 0.47 on the Gini scale, an increase of 20% (World Institute for Devel opment Economics Research 2005). Large regional inequalities continue to exist, especially with the ongoing civ il war in the North, where poverty has increased since 1997, while it has declined in all other regions. Furthermore, rural to urban dualisms exist, as % of the poor [are] livi ng in rural areas in 2000 (IMF and IDA 2001, 2). In addition, consumption growth rates vary highly between classes, as the richest decile has experienced an increase by 20% in consump tion since 1997, while the poorest quintiles consumption levels have only grown by 8% (ibid.) Given this urban and upper-class bias of the growths of consumption levels, the policy makers of the IMF and the In ternational Development Association (IDA) proclaim that t he need to raise rural income to reverse the adverse trends in inequality is recognized [] (IMF and IDA 2001, 3). This rural economy in Uganda continues to be dominated by agricu lture. Estimated in 1990, 84.5% of the total labor force were agricultu ral workers, with value added agriculture

PAGE 87

87 contributing an average of 42.56% of GDP from 1990 to 2005 (Table 4-21). However, as dramatically reflected, agricultu ral exports only contribute to 4.47% of the countrys GDP, not necessarily reflecting the increasing export pr oduction boosted by intern ational donors. On a side note, Uganda is unique in comparison to a majority of sub-Sa haran African countries, in that its economy does not heavily depend upon other mi neral and oil exports. As of 2004, gold exports (to Congo) amounted to only 73.8 US$ million. Total mineral production was as low as $41 million in 2000, projected to rise to betw een $128 and $203 million in 2008 (Yager 2004). Furthermore, value added of agriculture has de clined from 55.49% of GDP in 1989-1991 to 34% of GDP in 2005. Analogous, the share of the top four agricultural export commodities has dropped from a high of 92.19% in 1989-91 to a low of 66.01% in 2004, as farmers have increasingly struggled w ith the dominant export commodity, coff ee, and its great price volatility. As of 2004, coffee, tea, tobacco and cotton were the four main agricultural export commodities ranked in order of ex port value. Coffee alone accounted for more than 40% of the total exports of the top 20 commodities with $124 millions, followed by tobacco leaves with $41 millions, tea with $37 millions and cotton with $32 millions (FAO 2006). In 2004, Uganda was the 14th largest exporter of green coffee behind Ethiopia (11th), yet ahead of major coffee producers of Cote dIvoire (17th) and Kenya (20th). This leading position appears to be held relatively constant, with the major excepti on of 2001 and 2003, where coffee exports have dropped to a low of $51 and $37 millions respectiv ely (FAO 2006). Similar to Ethiopias fate, Uganda has experienced dramatical ly the problems associated with a more than fifty% drop in world coffee prices, a doubling of the petroleu m price, combined with poor weather and the continued advance of coffee wilt disease (Wetzel 2003). Over all, the share of coffee, constituting 48.7% of total export earnings in 199 7/98 declined dramatically to as low as 19.2%

PAGE 88

88 by 2001/02 (UNECA 2003, 66). Even through the peri ods of such vast declines in terms of trades, the Ugandan economy has been able to ach ieve surprisingly solid growth figures, mostly through an increase in nontraditional exports, such as fish, fl owers, and hides and skin. These NTEs, for example, increased by over 80% ove r the from 1998/99 to 2001/02 growing from a total of US$ 165 million to US$ 300 million (Wetzel 2003). In summary, Uganda has been able to increase its export diversity to the extent of reducing the devastating impact on its economy of the rapid decline in terms of trad e of its largest export commodity, coffee. It consequently has entere d similar markets to Kenya and Ghana, while avoiding the heavy dependency faced by Ethiopia. Ev en in the face of these successes, Uganda continues to struggle to keep up overall agricultura l production with annu al population growth over three%, as its per capita production yields of cereals remained sta gnant over the past two decades, even though still above the sub-Sahara n average (Earthtrends 2003). Food production per capita (Table 4-22) reflects a similar picture, showing a boom in per capita production in 1970 and 1975, yet stagnating around 95 in the 21st century (FAO 2006). Case Study: Malawi Malawi holds the infamous label as one of th e worlds poorest and l east developed places on earth. According to the most recent Huma n Development Report, Malawi is ranked 166th out of 177 countries, recording the 6th lowest life expectancy at birth with 39.8 years and the second lowest GDP per capita, PPP, of 646 US$ in 2004, or 596.75 US$ if expressed in constant 2000 US$ (UNDP 2006). Given its low development stat us, it comes as no surprise that Malawis economy is heavily dependent on agriculture. As of 1990, an estimated 86.6% of its work force were agricultural workers, with agriculture accounting on average 38.29% of its GDP from 1989 to 2005 (Earthtrends 2003). Haunted by seven incidences of drought from 1991 to 2004, Malawi is popularly depicted as a similar basket case in need of US and in ternational food aid as

PAGE 89

89 Ethiopia (Wines 2005). Both count ries display similar patterns of food insecurity. Malawis variation in domestic cereal production from 1992 to 2001 averag ed 26.7% from the mean, while Ethiopias averaged 18.5%, both we ll above the sub-Saharan Africa n mean of 6.5% (Earthtrends 2003, 2003). Food aid from 1998 to 2000 accounted fo r 29.4% of Malawi imports, a figure that is well above the sub-Saharan average of 19.9% (E arthtrends 2003). However, while of course differing in terms of shortages of rain, a simp le comparison shows that Ethiopias situation during the same period was significa ntly different, as an astonish ing 321.3% of its total imports were food aid (Earthtrends 2003) While droughts, such as the one in 1993 and 2005 clearly have a devastating impact on local food production, if they are coupled with global increases in maize prices due to a regional shortfall of rai n, such as in 2005, Malawi is unable to afford purchases from its neighbors. Consequently, while people have been able to adapt to occasional droughts and the resulting hunger cr ises, increasing corn prices are at the r oot of the 2005 famine, displaying the interconnect edness of political-economic a nd ecological factors, as well as Malawis huge dependence on the agricultural sector (Wines 2005). Table 4-23. Summary Malawi Mean GDP Growth (%) 3.04 Mean Agricultural Exports (% of GDP) 25.09 Mean Agriculture, Value Added (% of GDP) 38.29 Source: World Bank (2005), UN Devel opment Indicators (2006) and FAO (2006) As a matter of fact, Malawis agricultural sector accounts on averag e 38.29% of its GDP, with the agricultural exports accounting 25.09%, a staggeringly high number (Table 4-23). Agriculture is consequently placed at the core of the governments strategy for earning foreign currency. These averages from 1989 to 2005 have declined slightly over the years, from 45.47% of GDP in 1989-1991 to 34.70% in 2005. However, th is last figure, similar to the low 30.05% in

PAGE 90

90 1995, has clearly been affected by incidences of droughts (and lower tobacco prices)24. Throughout the years, with the exception of 2004, a drought year, agricult ural exports as a percentage of GDP has not declined signi ficantly, measuring 28.83% in 1989-91, 25.36% in 2000, 25.96% in 2003 and a lower 20.22% in 2004. What however has declined slightly is its share of the top four agricultu ral export commodities, which dropped by approximately three% from 1989-91 to 2004 to a very hi gh level of 89.96% (Table 4-24). In 2004, its top four export commodities were classical cash crops, including tobacco leaves (258 million US$), centrifugal sugar (42 million US$), tea (39 million US$), and cotton lint (13 million US$). Out of the top 20 agricu ltural exports, totaling 388.9 million US$, tobacco accounted for 66% in 2004, and an estimated 76% in 2000 (due to better prices; Figure 4-4). Of all total exports in 2004, estimated at 483.1 milli on US$, tobacco leaves produced by both smalland large-scale farmers accounted for around 53 % of them, indicating the great lack of diversification present in Malawi s export structure. A model cr eated by the International Food and Policy Research Institute (IFPRI) showed that a potential decrease of the world tobacco price would be exceptionally harmful to the Malawian economy; a 40% decline in the tobacco prices may cause Malawis tobacco production to fall mo re than 50%, and reduces its export revenues by 66% (Diao et al. 2002, 6-7). While local tobacc o prices in Malawi have recovered from a low in 1995, any drop in price would have devasta ting impacts on the economy, as the government heavily relies on its revenue through the regula ted auctioning of tobacco exports, corporate taxes paid by tobacco producers and traders, as well as some minor export tax revenue from tobacco trade (Diao et al. 2002, 12). Furthermor e, an increase in tobacco production does not 24 Food Production per capita Index has dropped to an all time low in 1992 of 57.8 points.

PAGE 91

91 necessarily result in increased f ood security compared to farmers who do not produce tobacco, as (gender) inequalities tend to be more persiste nt in such households (Peters and Herrera 1994). Given this rather bleak assessment, there ar e several positive signs surrounding Malawis agricultural sector. First, pe r capita cereal production in Mala wi has actually increased by 8% from 1979-81 to 1999-2001 to 234 tons per person, which is 99 tons per person more than the sub-Saharan African average (Earthtrends 2003). In addition, even more promising is the increase in average crop yield by 41% over the same time span to 1,634 kg per ha, as well as the almost double (98%) increase in average produc tion to 2.65 million metric tons in 1999-2001 (Earthtrends 2003). While these increases certainly are substantial, they have only been able to halt a decline experienced sin ce 1965, which was especially pronounced during the 1980s and 90s, as measured in the food production per capita index (Table 4-25). While Malawi continues to struggle with droughts, produc tion figures appear to have r ecovered, possibly creating space for the creation of policy refo rms towards diversification. Case Study: Mozambique Mozambique has experienced a tumultuous past which included a harsh, destructive and unusually long period of Portuguese colonial rule of almost fi ve centuries. Having finally achieved independence in 1975, what followed we re years of civil war and failed Marxist policies, large-scale emigration by whites, severe droughts and economic dependence on its mighty southern (neo-colonial) neighbor, Sout h Africa (CIA 2007). As of 2004, the signs are considerably more optimistic as the countrys leader for 18 years has been succeeded by a democratically elected president, Arma ndo Emilio Guebuza, who appears to embrace wholeheartedly liberal free mark et economics called for by the main international donors and organizations (CIA 2007). It thus appears that the country has been able to alter its path towards deeper impoverishment through the establishmen t of rapid growth rates since the mid 1990s.

PAGE 92

92 Average economic growth was 6.44% from 1989 to 2005, with a high of 13.10% annual growth in 2001 alone. This growth has resulted in a significant increase of its GDP per capita (PPP) from 686.11 US$ to 1219.93 US$ in 2005, starting its transformation from one of the poorest to one of the fastest growing na tions (Table 4-26 and 4-27). Table 4-26. Summary Mozambique Mean GDP Growth (%) 6.44 Mean Agricultural Exports (% of GDP) 1.91 Mean Agriculture, Value Added (% of GDP) 32.89 Source: World Bank (2005), UN Devel opment Indicators (2006) and FAO (2006) Alongside this process up on the ladder of deve lopment, Mozambiques primary sector has decreased in importance, falling from a high of 42% in 1989-91 to a low 23% in 2005 in terms of value added to the GDP. However, indicating th e increasing liberalization and removal of trade barriers, actual agricultural exports have increased from 43.7 million US$ in 1989-1991 to 123.6 million US$ in 2004. Besides this absolute increase, they have also increas ed in relationship to the rapidly growing GDP figures in the most recent years. Agricultural exports initially fell from 1.97% of the GDP in 1989-1991 to 1.32% in 1999-2001, yet grew to 2.08% in 2003 and 2.31% of the GDP in 2004. Throughout these developments the share of the t op four agricultural exports has remained relatively high and stab le between 81 and 84% of total exports. While the narrative told by these growth figures appear to reflect a success story of an African country that has finally achieved a turn around out of the traps of underdevelopment so many others are caught in, severa l points of cautions need to be raised. First, Mozambique continues to experience very high levels of poverty (Table 4-28). 38% of its population continues to live below US$ 1 per day and an o fficial poverty headcount in 1996 estimates that 69% of the total population lives in poverty (World Bank 2005). According to a recent study, this figure has dropped significantly to 54% by 2002 over a period where the economy grew by

PAGE 93

93 62% cumulatively (James, Arndt, and Simler 2005, 1). However, analyzing the case further they find that this relative reduction in povert y has occurred in conn ection with an analogous increase in inequality, especially in the urban areas of Maputo City, the countrys capital. On a countrywide level, the Gini coefficient increas ed only marginally from 0.40 in 1997 to 0.42 in 2002. However, in urban areas closer to South Africa, particularly Maputo City [] it rose from 0.44 in 1996-97 to 0.52 in 2002-03 (James, Arndt, and Simler 2005, 1f). These urban areas are not only experiencing great population growth they al so foster the continued high prevalence of HIV/AIDS with 12.2% of the a dult population living with HIV/AIDS in 2003. According to latest UN figures, HIV prevalence ha s dramatically worsened over the last years from 8.2% in 1998 to 16.2% in 2004 of the adu lt population. Most st aggering are the HIV positive rates in Maputo City, wher e prevalence rates increased from less than 1% in 1988 to 18% in 2002, a figure that was well above the country-wide aver age of 12.2% in 2003 (UNAIDS 2007). Not surprisingl y, given the HIV/AIDS epidemic population growth rates have continuously fallen (with the exception of 198991) from 3.32% in 1994-96 to 1.88% in 2005 to a total of approximately almost twenty million people by 2005. Agriculture continues to play by far the larges t role for a majority of the almost twenty million people. While in 1979-1981 agricultural wo rkers represented 84% of the total work force, that figure has dropped by 1% to 83% by 1989-91 and has since been estimated to have never dropped below 80%, with the latest esti mate in 2004 showing that 80% of the population worked in the agricultural sector. Given the population growth rates, it comes as no surprise that the number of workers in the agricultural sect or has increased from an absolute low of 5.6 million in 1979-81 to over 8 million by 2004 (FAOSTAT 2005).

PAGE 94

94 In 2004, the top four main export commoditie s were tobacco leaves (32 million US$), cashew nuts (28 million US$), cotton lint (23 milli on US$) and raw centrifugal sugar (18 million US$). Sesame seeds was the fifth most valuable export commodity (9 mill ion US$), with maize, the distant sixth valuable one (2 million US$) (FAO 2006). From 1994 2004, the top three export commodities have always been dominated by three of those four export commodities. One of the most important and infamous cash cr ops with a long history in the Mozambique case are cashew nuts. Starting under colonization in the 1930s, cashew production was expanded rapidly on mainly small and medium, and typi cally African, farms (Cramer 1999, 1252). This expansion was so dramatic that Mozambique became by the early 1970s [] the largest producer of cashew nuts in the world and a domes tic processing industry had begun to emerge (ibid., 1252). This processing industry, even with the aid of export tariffs on unprocessed cashew nuts, however faltered in a post-independence setting, leading to two main waves of policy reform that washed over the cashew sect or during the 1990s, pr ivatization and later market liberalization (ibid., 1253). According the Cramer, these aggressive liberalization policies, indicative of the commitment towards the free market by the government and placed at the top of the list by the World Bank, put in je opardy the whole processi ng industry, an industry that is severely constrained inte rnally, most importantly by low and variable level of supply of raw cashew nuts, and its variable quality (ibid., 1257). Consequently, by 1994-1996, Mozambiques share of the world raw cashew nut production market has dropped from 42.7% in 1969-71 to 4.1% (ibid., 1257). Recent production figur es and export figures indicate that cashew production has recovered again, growing fr om 33,423 Mtons in 1995 to 57,894 Mtons in 2000, while processed (shelled) cashews exports remain small (FAO 2006). Given this recent recovery of at least cashew supply yet lack for gaining ma rket share in the processed segment, Cramers

PAGE 95

95 concluding critique concerning th e international markets still ri ngs true: International markets for the production and trade of pr ocessed commodities are rich in imperfections and if a country is to promote such exports then its task is to pr omote effective market imperfections that work to its own advantage, not to abandon competitive hopes [] (Cramer 1999, 1262). Consequently, rather than getting the price right, it is a matter of getting the politics right in order to foster development in an attempt to gain a gr eater share of the commodity chain through agroprocessing. Finally, the current state of Mozambiques economy appears mixed. Observing the estimated changes in food production index per capita, the magnitude of the dire state and challenges ahead for Mozambique appear evid ent. Food production levels have dropped dramatically since independence, reaching new lows during the periods of liberalization and privatization, only recovering slightly in the late 1990s, and currently are indexed around 70, 30 points below its 1965 base value (Table 4-29). However, agricultural cereal production figures are a lot more promising. While per capita production (tons per pe rson) is still well below the sub-Saharan African average of 135 tons per pe rson with 91 tons per person, it has increased by 65% since 1979-81 to 1999-2001. Similarly, the aver age crop yield (kg per ha) has increased by 54% over the same period to 929 kg per ha, while resting below the sub-Saharan African average of 1,221 kg per ha (Earthtrends 2003). In summary, Mozambique, while enjoying very high growth rates, appears to con tinue to struggle into the 21st century with a large deficit of production and development it acquired in the 1980s and 1990s in the aftermath of bad policies, droughts, internal civil strive a nd intense international competition. Case Study: South Africa South Africa, as has been well established in the literature, represen ts both a geographical and economic outlier in sub-Saharan Africa. Located at the southern tip of the continent, the

PAGE 96

96 country has enjoyed a unique hi story under apartheid rule by the right-wing white National Party that rose to power in 1948 and ruled until 1994, when Nelson Mandela was elected as president in the first free elections (Moss 2007, 25). It was during this s econd half of the 20th century that South Africa became the economic powerhouse of s ub-Saharan Africa, emerging out of its state of underdevelopment and becoming the most indus trialized nation in sub-Saharan Africa. While its mean growth figures average below those of the faster growing nations with 2.31% from 1990 to 2005, its GDP per capita (PPP) is by far the largest with US$ (constant 2000) 11,044 in 2005 (Table 4-30 and 4-31 for all data). Growth figures a ppear to be stronger post-independence, as they averaged 0.35% in 1989-91 and 3.55% by 1994-1996. It was during the late 1980s that South Africas economy was es pecially struggling unde r the internationally imposed financial sanctions, which undermin ed business confidence [] and sharpened divisions among the white elite (Crawford 1999, 19). On the ot her hand, the percentage of value added of agriculture appears to have decl ined from a high of 4.87% in 1989-91 to a low of 3.00% in 2005. Agriculture nevertheless continues to play an important role, as exports have increased from 1.864 billion US$ in 1989-91 to 3.421 billion US$ in 2004. This increase of agricultural exports is also refl ected in the increasing percentage of exports, as they averaged around 1.65% both in 1989-91 and 1999-2001, but gr ew to 2.01% in 2003 and 2.25% in 2004. Table 4-30. Summ ary South Africa Mean GDP Growth (%) 2.31 Mean Agricultural Exports (% of GDP) 1.89 Mean Agriculture, Value Added (% of GDP) 4.00 Source: World Bank (2005), UN Developm ent Indicators (2006) and FAO (2006) Overall, South Africas economy is well-diver sified, including its agricultural sector. While increasing over the past five years, the sh are of the top four agri cultural export goods only measured 37.05%. Given the countrys Mediterra nean climate (besides six others, including

PAGE 97

97 sub-tropical and semi-deserts), its agricultural exports are uni que to Africa. In 2004, its top export was wine, measured at 533 million US$, followe d by various fruits in the next four spots: grapes (283 million US$), oranges (271 million US $), apples (181 million US$), and prepared fruit (174 million US$) (FAO 2006). Of all the agricultural exports in January 2007, 66% went to the European Union (EU), with less than 10% going to Northeast Asia and 7% to China. Reflecting both the type of goods exported, as well as the lack of regional trade, only 0.5% went to West Africa and a mere 0.1% to Central Afri ca of the exports in January 2007 (DTI 2007). While these export commodities are of large impor tance to the overall economy, they are mostly exported by large-scale, capital intensive white farmers. Given the countrys long history of apartheid and a slow proc ess of land redistribution, the primary sector is still marked, to use the narrat ive put forth by the national tourism alliance and a recent personal visit, by a visible dual economy with both well-developed commercial farming and more subsistence-based production in the de ep rural areas (Southafrica.info 2007). An estimated 1.271 million people were employed in the agricultural sector as of March 2004, slightly more than 10% of the total employed workforce of 11.95 millions (Statistics South Africa 2004). This figure is only slightly lo wer than the previous estimate of 13.4% of agricultural workers as a percen tage of the total labor force in 1990 (Earthtrends 2003). Focusing on agricultura l inputs, South Africa takes on an utter dominance in the subSaharan African realm. For example, of the tota l thousand metric tons of fertilizer used in 1999, South Africa used 804,000 metric to ns out of a cumulative total of 2.12 million metric tons in sub-Saharan Africa. Furthermore, in terms of using mechanized machinery, South Africa holds an estimated 100,000 tractors in 1997 out of a to tal of 261,984 tractors in sub-Saharan Africa. Taking into account the lack of capital and resources in many African countries to undertake

PAGE 98

98 repairs, this number might even be more dom inating (Earthtrends 2003) Finally, 8.6% of the cropland in South Africa was estimat ed in 1999 to be irrigated, which is more than twice as large as the sub-Saharan average of 3.8% (Earthtrends 2003). These dramatic differences in agricultu ral modernization result in correspondingly significantly higher yields. S outh Africas average crop yields measured in kg per ha, are almost twice as great as the sub-Saharan Af rican average, averaging 2,334 kg per ha in 19992001 compared to 1,221 kg per ha (Earthtrends 2003). While per capita production has decreased over the past two decades following the growth of the other sectors, average yields have increased by 11% for cereals from 197981 to 1999-2001 (Earthtrends 2003). The population growth, with a high of 2.24% 2000 and a current low of -0.70% in 2005, appears not to play a major factor given South Africa is mu ch further down the path of the demographic transition than any other neighbors25. Consequently, even though the most recent estimates on the Food Production Index show that it has dr opped as low as to 75.2 points in 1995, ranking 88.7 points in 2003, South Africas agricultural s ector appears healthy (Table 4-32). A major problem of course needs to addresse d more in detail, which is the extensive inequality and expansive number of people living in poverty in this wealthiest of nations in subSaharan Africa, resulting from huge inequalities in access to productive [land assets], basic infrastructure and capital as well as to educat ion and skills (DTI 2007). In 1997, inequality, as measured by the Gini coefficient, has increa sed by 0.6 points from 1993 to 60.1, which is the 4th highest value of the sample countries under study (World Institute for Development Economics Research 2005). More equal land access provided through the willing seller willing buyer policy is by many believed to be the key determinant towards decreasing the inequality (World 25 Unfortunately HIV/AIDS plays a large reason as well for the lower levels of population growth.

PAGE 99

99 Bank 2005). In terms of poverty, an estimated 86% of the rural population is living below 2/3 of the national mean per capita income, a number that is considerably lower in the urban areas with only 40% of the population living below those 2/3 (World Bank 2005, 309). Reflecting the inadequacies of the absolute poverty measuremen t of percentage of population living under US$ 1 a day, only 11% lived under that measuremen t, calculated between 1994 and 2002 (World Bank 2005, 309). Referring to the Human Developmen t Index (HDI) of the UN, South Africa is ranked 121st out of 177 countries, ranking especially low on the life expectancy at birth measurement, which averaged only 47 years, or 156th out of the 177 countries (UNDP 2006). Assessment of Case Studies Comparing David to Goliath As becomes evident, the various countrie s studied cover a broad range of Africas development levels, with diverging results. The strata of the case studies ranges from a country heavily dependent upon its agricultural sector, ye t struggling to gain a foothold on the ladder of development, such as Malawi, compared to the ca se of South Africa, which is at the verge of entering the club of industrialized developed nations. While the fo rmer is marked by a lack of diversification, heavy dependence on cash crops and great vulnerab ility to climate, the latter displays a modern, highly divers ified agricultural sector that am ounts to a decreasing share to the overall GDP, yet overall increases in total va lue of agricultural exports. The economic disparities, as measured in GDP ppp per cap ita in 2005, are vast: $596.75 in Malawi versus $11,044 in South Africa. While these countries differ in about every aspect of economic comparison, they have one statistic in common: income ine quality. The dramatic inter-country difference in per capita income is followed by similar levels of inequality at the national level in 1993 as measured as a Gini index: 62.00 in Malawi versus 59.50 in South Africa. Inte restingly, income inequality is

PAGE 100

100 likely one of the only variables where Malawi su rpasses South Africa, as it lowered it by 1997 to 49.30, while South Africas variable increased slightly to 60.10. Without diverting into an ela borated discussion on the respectiv e reasons for this finding, it becomes evident: First, that there are several co mplex development paths that can lead to the same level of inequality. Sec ond, inequality is a key aspect fo r development policy in Africa, not only because of its growth retarding impact (Fosu 2005), but also because of its ability to provide us with an improved narrative of a countrys distribution of income and wealth, especially in connection with an average value, such as GDP ppp pe r capita. It is within this context that a comparison of the case studies yields the following findings. Conflict and Its Aftermath First, our sample covers a dive rse range of countries strugglin g or recovering from internal strive, civil war or external c onflicts. One of the most promin ent and pressing cases is Cote dIvoire, which has been involved in an increasing civil war that is linked to increasing pressures of land availability and inequalit y, and the subsequent struggle over that resource. This pressure on unequal land holdings is to a degree similar to the non-violent struggles faced in South Africa over land redistribution before th e end of Apartheid, negatively impacting income inequality. Internal warfare has also resulted in vast in equalities in Uganda, as the Northern region is plagued with civil conflict in a country that ha s otherwise made great strides towards increasing development and reducing poverty. While not likel y to have a direct impact on inequality, Ethiopias ongoing international conf lict with its neighboring countries is likely not only to be a drain upon its already limited resources, but also result in lowered inte rnational investors confidence for funding private enterprises. Last, but not least, Mozambique appears to represent a country that has recovered into the 21st century from its internal unrest, achieving similar to Uganda astonishing growth figures.

PAGE 101

101 Poverty, Infant Mortality and Inequality Secondly, the cases of both Mozambique and Ug anda are especially noteworthy, as they highlight the unique relationship of inequality in the process of growth and development. Ghana, while not having experienced civil strive over the past ten years of its fifty, at times bumpy, years since independence, can be added to those countries, since all these three countries have displayed astonishing growth rates over the past years after an implemen tation of a pro-export, liberal development strategies with the aid of international development agencies and donors26. For example, Mozambique has achieved astonish ing growth rates since 2001, with the GDP growing by 13.10% in 2001 and never dropping be low 7% annually. Similarly, GDP ppp per capita has increased each year, growing from $992.80 to $1219.93. Uganda and Ghana have grown correspondingly, averaging 5% GDP grow th or more since 2001, with GDP ppp per capita increasing each consecutive year. While poc kets of great poverty exist in each country, such as in the north of Ghana and Uganda, or in their urban areas, the countries have overall had impressive results in decreasing poverty at a na tional scale as discusse d in the above country studies. Given their impressive growth figures and improvement in poverty, they all displayed a unique pattern of change in inequality: the Gini coefficient increased whil e agricultures share of value added to the GDP decreased. Mozambique s Gini coefficient grew minimally from 40.0 to 42.0 from 1997 to 2003, Ugandas grew from 39.0 in 1992 to 46.90 in 2000, and Ghanas most recent inequality indicator from 1998, 50.70, is significantly higher than the 1992 value, 39.70. Consequently, there appears to be a poten tial relationship between higher growth rates that result in a reduction of pove rty, but corresponding increase in inequality, with agricultures 26 Ghanas astonishing growth started earlier before the other two countries.

PAGE 102

102 importance decreasing, as measured in terms of value added as percentage of the GDP27. This relationship between poverty and inequality however is clearly complex. Similarly, the relationship between infant mortality, a proxy for development, and inequality is neither clear. As displayed in Table 4-33, no clea r relationship appears apparent between infant mortality and Gini coefficient, as highlighted by the case of South Africa, which experiences the largest GDP ppp pe r capita, the lowest infant mo rtality, yet one of the highest Gini coefficients. The country with the wors t infant mortality, Malawi, having 133 deaths per 1000 live births, on the other hand holds also th e lowest GDP ppp per capita, yet one of the higher Gini coefficients (49.30) For the two countries just above and below 100, Cameroon with 95 and Zambia with 102 deaths per 1000 live births, they hold va stly different Gini coefficients (44.20 and 57.40 respec tively) and also different levels of average income, further complicating any initial interpretation. In sum, while an overall trend appears appa rent of a relationship between GDP ppp per capita and infant mortality, this relationship does not appear to exist, at least initially, with inequality in our case studies. Diversification or Lack There Of Having discussed the countries respective performance in te rms of achieving high growth rates, four countries in the sample stand out as having been exceptionally successful at diversifying their agricultural production. These include the afore-mentioned Uganda and Ghana, as well as South Africa and Kenya. Countri es that have failed to diversify, for example, are Malawi and Mozambique, whose share of the top four agricu ltural export good stayed stable around 89 and 82% respectively from 1990 until 2004. 27 As pointed out by Julie Silva, Mozambiques growth rate s are largely a result for exam ple of the construction of a dominating aluminum smelting plant, which is majority-owned by South African investors.

PAGE 103

103 The main road leading towards diversifica tion involved the increa sing privatization and liberalization of domestic market s, with the sector entering into the expo rt of non-traditional commodities (NTE). As a whole, with the major exception of a few designated cash crops, such as cotton in West Africa, most African countri es have successfully (at least attempted) to liberalize and privatize their mark eting boards by the midand late -1990s, as well as reduce their export tariffs on crops in the afte rmath of the WTOs Uruguay Round. As mentioned previously, esp ecially Kenya and to a lesser extent Ghana have been exceptionally successful at that, with both countries dropping thei r percentage share of the top four agricultural e xports by more than ten% from 1990 2004. Uganda appears to have achieved even higher diversifica tion, as their share dropped from a high of 92.19% to a low of 66.01% in 2004. The case of Uganda however cl early not only an incr easing diversification strategy, but also the dramatic impa ct a drop in international prices for one of the four cash crops exported can have on the overall va lue exported, as was the case with coffee, Ugandas main export crop, in 2001 and 2003. Consequently, while NTEs have grown, such as fish, the overall growth is not as substantial as in the previous two cases, and a lo wer share of the top four export commodities does not indicate a higher value of overall agricultural exports necessarily. The lowest share of the top four agricultura l exports in the case study has South Africa, hovering around 35%, as would be expected given its place on the ladder of development. Its main agricultural exports are not made up of any traditional ex port crop, while overall agricultural exports have increased successively from 1.8 billion US$ in 1989-91 to 3.4 billion US$ in 2004. A lower share of the top four ag ricultural commodities does consequently not necessarily indicate lowe r levels of importance of the agri cultural sector to the economy.

PAGE 104

104 Senegal is an opposite case in point, where the countrys share has decreased from a high of 81.80% in 1989-91 to a low of 56.5% in 2004. Th e overall value of agricultural exports has actually decreased from a high of 186 million US$ in 1989-91 to 182 million US$ in 2004. The change in Senegal, for example, is subsequently related to the governments fiscal inability to continue support for its exports, su ch as peanuts, and any subseque nt diversification is likely due to the lack of (input) subsidies, such as seeds an d fertilizer, creating a di sincentive, as opposed to the successful entering of farmers into new marketing opportunities. Overall though, a lack of diversif ication is likely to always re sult in reduced growth in the long-run due to increased vulnerability to fluctuat ions in producer prices, weakening terms of trade and droughts. This impact is highlight ed by Burkina Fasos focus on cotton, Ghanas history with cocoa, Ethiopias dependency and Ugandas struggle with coffee, and Malawis dependency on tobacco. Returning to Mala wi, given the high de pendency on tobacco, a projected decrease in worl d prices is to have a devastating impact as discussed previously. Similarly, however, it can be argued that gi ven the large amounts of livelihoods depending upon tobacco for their income, an increase in produc er prices certainly has a positive impact on growth. Furthermore, it appears that the increas e in producer price for tobacco (Figure 4-4) in Malawi from 1993 to 1997 correlates with a drop of the Gini coefficient from 62.00 in 1993 to 49.30 in 1997. In summary, it consequently appears that an improvement in producer prices for a commodity produced mainly by poorer, smaller-s cale farmers, could result not only in increased growth, reduced poverty, but also a decrease in inequality. Ability to Modernize Faced with a lack of subsidized inputs, poor er farmers are usually less able to find the financing necessary to purchase fe rtilizer, if available. In seve ral (poor) countries in our sample, we however have observed increases in fertilizer usage, such as Burkina Faso, Cote dIvoire,

PAGE 105

105 Kenya, as well as South Africa. These countries however differ widely in their structure of farms. In West Africa, including Ghana, mostly small-scale farmers make up the landscape and the national scale increased fertiliz er usage is likely to be connect ed to them. On the other hand, both Kenya and South Africas agricultural export landscape consists of (dominant) large-scale farms, which clearly are the main purchaser of the increased fertilizer (and machinery). Consequently, while the increases in fertilizer usage are positive signs from the perspective of agricultural export produc tion, the different make-up of the sector highlights the diverging impacts they are likely to have on inequality. Droughts and Other Environmental Impacts A final impact that appears to weigh heav ily on agricultural deve lopment are droughts. Given that most African nations lack the capacity for irrigati on (with the exception of South Africa), their farmers are vulnerabl e to increasing variability in rainfall. The two most prominent examples are Ethiopia and Malawi, with both ex periencing numerous drought years during the study period. While the origins and results of droughts are complex, rangi ng from the politicaleconomic to environmental, it becomes clear that the more severe ones have a ruinous impact on the economy, as reflected by the lack of or nega tive growth in 2002 and 2003 in Ethiopia, as well as the negative impact on food production per capita as seen in the drop in 2002 in Malawi. As drought occurrences are likely to in crease in the near future, especi ally in the countries in the Sahel, Ethiopia, Malawi and Mozam bique, it appears that the abil ity for farmers to cope with such shocks will come increasingly under pressure It is also important to note, that beyond droughts, pests are for example another major issue pl aguing farmers, such as seen with coffee in Uganda or observed with cotton in the Sahe l, only magnified by accounts of increased desertification.

PAGE 106

106 The impact of droughts however is complex on inequality, as based upon our case studies, one could not infer whether or not more drought -prone countries experience higher or lower inequality levels. The case in poi nt is Ethiopia, which enjoys the lowest Gini coefficients of our sample, with a low of 29.70 in 2000. This coeffici ent is dramatically lower than Malawis, a similarly drought-prone country, wh ere the latest Gini coeffici ent in 1997 equals 49.30. While this figure is actually only sli ghtly higher than Ethiopias 199 7 coefficient of 48.20, Ethiopias Gini coefficient measured 32.70 in 1995, wher eas Malawis previous 1993 value was 62.00, showing the (averaged) lower value of inequality28. Having highlighted some of the similarities a nd differences between th e case studies, the next chapter will undertake a quant itative investigation of the impact and interactions of those agricultural variables on in equality and development.29 28 For Ethiopia, 1997 and 2000 were years that experience d incidences of droughts. For Malawi, 1997 experienced an incidence of drought (World Bank 2005). 29 While a comparison between the African countries and some more-developed Asian counterparts has preliminarily been investigated, it has not been explicitly d ealt with here. The general understanding however is that these different continents varied widely in their availability of land, labor and capital, resulting in diverging development outcomes. For further information, see (Thirtle, Lin, and Piesse 2003) and (World Bank 1993).

PAGE 107

107Table 4-2. Overview of agriculture and development in Burkina Faso Series 1989-19911994-19961999-2001 2001 2002 2003 2004 2005 Population, total (million) 8.539.8311.30 11.6412.0212.4212.8213.23 Population growth (annual%) 2.932.762.94 3.083.193.243.203.12 Agriculture, value added (% of GDP) 28.1732.2534.11 33.0031.0031.0031.0031.00 GDP per capita, PPP (constant 2000)^ 920.22937.92998.00 1026.001035.001067.001076.001093.00 GDP growth (annual%) 2.704.234.73 6.004.007.004.005.00 GDP (constant 2000 US$ million) 1809.542127.162638.20 2754.142875.323062.223181.653334.37 Agricultural Exports (constant US$ million)92.10130.23 291.44313.70 Agricultural Exports (% of GDP) 5.265.01 9.529.86 Share of top 4 agricultura l exports (%) 87.1783.77 86.7591.35 Legend: ^: International $ Source: UN World Development Indicat ors (2006), UN (2006) and FAO (2006)

PAGE 108

108 050010001500200025003000350040004500 Cereal Yield (kg/ha) Asia (excl. Middle East) South America Burkina Faso Indonesia Malaysia Thailand VietnamGraph 1: Avera g e Cereal Yield (kg/ha) in Select Regions 1999-2001 Figure 4-1. Average cereal yields in select regions

PAGE 109

109Table 4-5. Overview of agriculture and development in Cte dIvoire Series 19891991 19941996 19992001 2001 2002 2003 2004 2005 Population, total (million) 12.6614.7616.72 17.0517.3417.6017.8718.15 Population growth (annual%) 3.432.872.12 1.871.661.531.511.57 Agriculture, value added (% of GDP) 32.8424.8622.84 24.0025.0024.0022.0022.00 GDP per capita, PPP (constant 2000)^ 180215611576 15361485142814251401 GDP growth (annual%) 0.635.22-0.55 0.00-2.00-2.002.000.00 GDP (constant 2000 US$ million) 8306893210549 1043610266100951026110230 Agricultural Exports (constant US$ million) 16502130 32163093 Agricultural Exports (% of GDP) 19.9420.43 31.8630.14 Share of top 4 agricultu ral exports (%) 76.7169.73 72.1169.64 ^: International $ Source: UN World Development Indicat ors (2006), UN (2006) and FAO (2006)

PAGE 110

110 Table 4-9. Overview of agricu lture and development in Ghana Series 1989-1991 1994-19961999-2001 2001 2002 2003 2004 2005 Population, total (million) 15.4817.7219.87 20.3120.7621.2121.6622.11 Population growth (annual%) 2.792.542.21 2.202.192.162.112.05 Agriculture, value added (% of GDP) 46.4438.5136.13 36.0036.0036.0038.0039.00 GDP per capita, PPP (constant 2000)^ 158917051893 19121955201420882149 GDP growth (annual%) 4.574.004.10 4.004.005.006.006.00 GDP (constant 2000 US$ million) 328940374988 51875420570260336385 Agricultural Exports (constant US$ million) 403.00521.00 1037.001212.00 Agricultural Exports (% of GDP) 12.3410.47 18.1920.09 Share of top 4 agricultu ral exports (%) 96.0986.60 85.7783.30 ^: International $ Source: UN World Development Indicat ors (2006), UN (2006) and FAO (2006)

PAGE 111

111 Table 4-10. Ghanas exports, 1980 2000 (value as percentage of exports) Year Cocoa Minerals Timber NTE Others 1980 65.0 31.03.0n/a 1.0 1990 40.2 17.222.5n/a n/a 1994 25.2 46.3n/a9.4 n/a 1996 35 40.99.39.1 5.7 1997 31.3 41.111.510.6 5.5 1998 29.7 34.38.210.9 19.0 1999 27.6 37.28.612.4 23.2 2000 22.5 40.09.013.6 24.9 Adapted from Konandu-Agyemang and Adanu (2003, 517) Sources: (Ghana Export Promotion Counc il; Government of Ghana 1997, 1998; Ghana Statistical Services 2000)

PAGE 112

112Table 4-12. Overview of agricultu re and development in Senegal Series 1989-1991 1994-1996 1999-2001 2001 2002 2003 2004 2005 Population, total (million) 7.989.1210.34 10.6010.8611.1211.3911.66 Population growth (annual%) 2.892.592.46 2.442.412.392.372.36 Agriculture, value added (% of GDP) 19.4519.8819.88 2216181718 GDP per capita, PPP (constant 2000)^ 1398.121323.901435 14601450150415601615 GDP growth (annual%) 0.704.394.62 51766 GDP (constant 2000 US$ million) 323635374411 45914642494652515577 Agricultural Exports (constant US$ million)186135 175182 Agricultural Exports (% of GDP) 5.673.10 3.543.47 Share of top 4 agricultu ral exports (%) 81.8068.12 59.3256.5 ^: International $ Source: UN World Development Indicat ors (2006), UN (2006) and FAO (2006) Table 4-13. Senegal, food produc tion per capita index, 1965=100 1965 1970 1975 1980 1985 1990 1991 1992 1993 1994 100.0 56.4 92.2 40.250.749.149.944.348.147.8 1995 1996 1997 1998 1999 2000 2001 2002 2003 52.1 47.0 43.1 47.856.454.149.630.945.9 Source: FAOSTAT (2006)

PAGE 113

113Table 4-16. Overview of agricultu re and development in Ethiopia Series 1989-1991 1994-1996 1999-2001 2001 2002 2003 2004 2005 Population, total (million) 51.1656.5564.2965.7867.2268.6169.9671.26 Population growth (annual%) 3.522.952.372.282.172.061.941.83 Agriculture, value added (% of GDP) 54.7957.5747.0146.0042.0044.0046.0048.00 GDP per capita, PPP (constant 2000)^ 754.32725.15781.00824.00805.00767.00843.00896.00 GDP growth (annual%) -2.526.876.628.000.00-3.0012.009.00 GDP (constant 2000 US$ million) 602564107919846784668204921310018 Agricultural Exports (constant US$ million) 290.00450.00380.00 Agricultural Exports (% of GDP) 3.705.484.12 Share of top 4 agricultu ral exports (%) 84.1483.8379.4271.39 Incidence of drought* DDDDD ^: International $ *: Significant shortage of rain (excluding 1991 and 2005) Source: UN World Development Indicat ors (2006), UN (2006) and FAO (2006)

PAGE 114

114 2000 2001 2002 2003 Sesame seed Coffee, green 0 1000 2000 3000 4000 5000 6000 7000 8000 YearsGraph 2: Producer Price (Local Currency/tonne) of Sesame Seeds and Green Coffee in Ethiopia (2000 2003) Source: FAOSTAT (2006) Sesame seed Coffee, green Figure 4-2. Producer price (local currency/tonne) Graph 3: Export Quantity of Coffee (green) and Sesame Seeds Ethiopia from 2000 20050 20 40 60 80 100 120 140 160 180 200 200020012002200320042005 Years Coffee, green Sesame seed Figure 4-3. Export quantity of co ffee (green) and sesame seeds

PAGE 115

115Table 4-18. Overview of agricu lture and development in Kenya Series 1989-19911994-19961999-2001 2001 2002 2003 2004 2005 Population, total (million) 23.427.230.7 31.432.032.733.534.3 Population growth (annual%) 3.422.742.23 2.172.132.142.222.33 Agriculture, value added (% of GDP) 29.2832.0331.80 31.0028.0028.0027.0027.00 GDP per capita, PPP (constant 2000)^ 113210581018 10421022102910471042 GDP growth (annual%) 3.443.732.42 4.000.003.004.003.00 GDP (constant 2000 US$ million) 104661142112866 1326213314136831427614676 Agricultural Exports (constant US$ million)6661033 12911296 Agricultural Exports (% of GDP) 6.318.13 9.449.08 Share of top 4 agricultu ral exports (%) 81.2275.67 68.1970.66 ^: International $ Source: UN World Development Indicat ors (2006), UN (2006) and FAO (2006) Table 4-19. Kenya, food produc tion per capita index, 1965=100 1965 1970 1975 1980 1985 1990 1991 1992 1993 1994 100.0 99.1 96.992.294.2103.3103.6100.294.696.3 1995 1996 1997 1998 1999 2000 2001 2002 2003 95.0 88 89.089.09691.598.396.989.7 Source: FAOSTAT (2006)

PAGE 116

116Table 4-21. Overview of agricultu re and development in Uganda Series 1989-19911994-19961999-2001 2001 2002 2003 2004 2005 Population, total (million) 17.7620.8924.33 25.1125.9626.8727.8228.82 Population growth (annual%) 3.573.093.14 3.253.343.423.483.52 Agriculture, value added (% of GDP) 55.3948.1537.37 36.0031.0032.0032.0034.00 GDP per capita, PPP (constant 2000)^ 87810381249 12701301131113391363 GDP growth (annual%) 6.139.006.21 5.006.004.006.006.00 GDP (constant 2000 US$ million) 307142995919 62196622691273007706 Agricultural Exports (constant US$ million)206.00272.00 115.00359.00 Agricultural Exports (% of GDP) 6.704.59 1.664.92 Share of top 4 agricultu ral exports (%) 92.1981.40 67.7666.01 ^: International $ Source: UN World Development Indicat ors (2006), UN (2006) and FAO (2006) Table 4-22. Uganda, food produc tion per capita index, 1965=100 1965 1970 1975 1980 1985 1990 1991 1992 1993 1994 100.0 160 161.995.497.5101.1 99.896.598.894.6 1995 1996 1997 1998 1999 2000 2001 2002 2003 96.2 88.2 88.393.194.394.7 96.596.994.3 Source: FAOSTAT (2006)

PAGE 117

117 Table 4-24. Overview of agricultu re and development in Malawi Series 19891991 19941996 19992001 2001 2002 2003 2004 2005 Population, total (million) 9.42 10.13 11.5111.8012.0712.34 12.6112.88 Population growth (annual%) 3.80 1.42 2.642.432.302.20 2.162.16 Agriculture, value added (% of GDP) 45.47 30.05 38.7238.7839.0139.77 38.8834.70 GDP per capita, PPP (constant 2000)^ 518.20 540.62 574.41542.66547.70568.99 595.70596.75 GDP growth (annual%) 5.26 4.60 -0.12-4.972.866.07 7.122.55 GDP (constant 2000 US$ million) 1256.88 1405.76 1705.581656.771704.141807.59 1936.321985.75 Agricultural Exports (constant US$ million) 362.4 432.6469.2 391.6 Agricultural Exports (% of GDP) 28.83 25.3625.96 20.22 Share of top 4 agricultural exports (%) 93.36 93.4788.95 89.96 Incidence of droughts D D DD ^: International $ Source: UN World Development Indicat ors (2006), UN (2006) and FAO (2006) Table 4-25. Malawi, food produc tion per capita index, 1965=100 1965 1970 1975 1980 1985 1990 1991 1992 1993 1994 100.0 96.3106.5 103.291.972.777.857.8 80.664.0 1995 1996 1997 1998 1999 2000 2001 2002 2003 75.4 86.680.3 97.6108.1121.9126.490.14 99.3 Source: FAOSTAT (2006)

PAGE 118

118 Graph 4: Producer price of Tobacco (unmanufactured) in Malawi in 1991 -20030 200 400 600 800 1000 1200 1991199219931994199519961997199819992000200120022003 Years Malawi Figure 4-4. Producer price of tobacco (unmanufactured) in Malawi in 1991-2003 Source: FAOSTAT (2006)

PAGE 119

119 Table 4-27. Overview of agricultu re and development in Mozambique Series 1989-19911994-19961999-2001 2001 2002 2003 2004 2005 Population, total (million) 13.4815.8417.91 18.3018.6819.0519.4219.79 Population growth (annual%) 1.363.322.17 2.132.061.991.931.88 Agriculture, value added (% of GDP) 41.8336.5027.06 24.0024.0024.0023.0023.00 GDP per capita, PPP (constant 2000)^ 686.11692.14913.52 992.801008.021065.571142.861219.93 GDP growth (annual%) 4.136.307.53 13.108.167.907.497.70 GDP (constant 2000 US$ million) 2217.502597.413918.68 4272.634621.294986.375360.085772.80 Agricultural Exports (constant US$ million)43.751.8 103.6123.6 Agricultural Exports (% of GDP) 1.971.32 2.082.31 Share of top 4 agricultu ral exports (%) 81.8383.75 83.6982.05 ^: International $ Source: UN World Development I ndicators (2006) and FAO (2006)

PAGE 120

120 Table 4-28. Diverse poverty and health figures in Mozambique % of population living under US$ 1 a day (between 1994 and 2002) 38.00 National poverty headcount (% of total population in 1996) 69.00 National poverty headcount (% of total population in 2002) 54.00 Gini coefficient (1997) 0.40 Gini coefficient (2002) 0.42 Adult rate (15-49 years) of % population with HIV/AIDS (2003) 12.20 Adult rate (15-49 years) of % population with HIV/AIDS (2004) 16.20 Source: World Bank (2005) Table 4-29. Mozambique, food produc tion per capita index, 1965=100 1965 1970 1975 1980 1985 1990 1991 1992 1993 1994 100.0 108.2 98.3 81.968.476.865.654.4 59.655.8 1995 1996 1997 1998 1999 2000 2001 2002 2003 65.1 71.8 74.7 78.277.468.370.070 70.8 Source: FAOSTAT (2006)

PAGE 121

121Table 4-31. Overview of agriculture and development in South Africa Series 1989-19911994-19961999-2001 2001 2002 2003 2004 2005 Population, total (million) 35.2139.1343.91 44.8145.3545.8345.5145.19 Population growth (annual%) 2.112.172.24 1.831.181.06-0.70-0.70 Agriculture, value added (% of GDP) 4.874.223.44 4.004.004.003.003.00 GDP per capita, PPP (constant 2000)^ 997792939488 96459819100311047211044 GDP growth (annual%) 0.353.553.08 3.004.003.004.005.00 GDP (constant 2000 US$ million) 11068611630713232 136512141549145761152276159738 Agricultural Exports (constant US$ million)1863.002151.00 2937.003421.00 Agricultural Exports (% of GDP) 1.681.62 2.012.25 Share of top 4 agricultu ral exports (%) 36.8131.35 33.4337.05 ^: International $ Source: UN World Development I ndicators (2006) and FAO (2006) Table 4-32. South Africa, food pr oduction per capita index, 1965=100 1965 1970 1975 1980 1985 1990 1991 1992 1993 1994 100.0 95.7 104.8105.491.792.9 92.476.486.490.1 1995 1996 1997 1998 1999 2000 2001 2002 2003 75.1 88.5 88.481.185.093.3 87.792.488.7 Source: FAOSTAT (2006)

PAGE 122

122 Table 4-33. Infant mortality, GDP&Gini Country Year Gini GDP ppp per capita Infant Mortality South Africa 1997 60.108777.9745.00 Ghana 1998 50.701754.6162.00 Kenya 1997 44.50982.4073.00 Madagascar 2001 47.40862.6484.00 Uganda 2000 46.901248.8885.00 Gambia 1998 50.201587.6392.00 Cameroon 2001 44.202018.2895.00 Zambia 1998 57.40749.95102.00 Tanzania 1993 47.70441.42103.00 Mozambique 2003 42.001130.03104.00 Burkina Faso 1998 62.50943.88107.00 Cote d`Ivoire 1998 44.401603.59115.00 Ethiopia 2000 29.70639.06116.00 Nigeria 1997 50.20804.32120.00 Guinea 1994 55.101587.63129.00 Malawi 1997 49.30555.57133.00 Source: WDI (2006) and UN-WIDER (2005) Note: Latest years available used in regressi on models. Infant mortal ity measures deaths per 1000 live births. Since the variab les are only available in 1980, 1990, 1995, 2000, 2004, the value closest to the respective year was chosen.

PAGE 123

123 CHAPTER 5 COMPARATIVE ANALYSIS ON AGRICUL TURE, GROWTH AND INEQUALITY Main Hypothesis Results Based upon the previously discus sed model, the following results have been computed for Model 1, 2 and 3 (Table 3-5): Gini 58.96 0.198 x BeforeWTO 0.407 x AgricultureValueAdded 0.195 x TotalAgricultureExport .238 xLogAgriculturelExportsGDP 0.195LogGDPppp .005ChildrensInfantMortality .550Urbanization Figure 5-1. Regressi on equation Model 1 Gini 86.25 0.329 x BeforeWTO 0.632 x AgricultureValueAdded 0.538 x FoodExports .217 xSquareRootAgriculturalRawMaterialsExport 0.110LogGDPppp .116ChildrensInfantMortality .396Urbanization Figure 5-2. Regressi on equation Model 2 Gini 75.21 0.248 x BeforeWTO 0.580 x AgricultureValueAdded 0.431 x FoodExports .880 xSquareRootAgriculturalRawMaterialsExport 0.252LogGDPppp .168ChildrensInfantMortality 1.938Urbanization Figure 5-3. Regressi on equation Model 3 Of the seven independent variables, four agricultural variables are highly significant (Before WTO dummy, Agricultur e Value Added, Food Exports, Total Agricultural Exports), while one variable, urbanization, found by other studi es is significant at the .05 level in Model 1 and 2, and the .1 level in Model 3 (Deininger and Squire 1998, 272). The regressions (Table 5-1, 5-2, 5-3) overall yields adjusted R2 value of 0.382 (Model 1), 0.516 (Model 2) and 0.403 (Model 3).

PAGE 124

124 Test of Assumptions Given the fact that the regre ssion includes real-world data fr om Africa, we would expect certain analyses of assumptions to be rather mess y. Every feasible effort has been undertaken in order to not violate the basic assumptions of multiple regressions and corrections and manipulations, if necessary, were undertaken.30 First, concerning normality (Appendix I for data), five of the variables31 used in the model (as well as arable land, population density and value of the top thr ee agricultural e xports) did not pass the Shapiro-Wilkens test initially of .05 significance used for smaller sample sizes.32 These variables, GDP ppp per capita, Food and Agricu ltural Raw Exports, as well as the others underwent a closer analysis, which is detailed sp ecifically in the Appendices, mostly for Model 3. While certain limitations existed, given the nature of the dataset, for th e independent variables not included in the model the following transforma tions (generally square -root or logarithmic) improved their normality, log of Arable Land and log of Population Density, with Value of Top 3 Agricultural Exports not improving. Concerning the values of interest in our model, GDP pe r capita ppp was heavily influenced by the two data points of South Africa, which were significant and non-err oneous outliers. Since these outliers are not a re sult of data error, a conscious deci sion has been made to include them in this model, and possible transformations inve stigated. The logged variable clearly yielded superior results to the other tw o transformations, improving the Sh apiro-Wilkens statistics from 30 The sky seems to be the limited when undertaking empiri cal analyses. A case in point is the residual analysis, which was undertaken to the extent best possible, yet w ithout for example mapping them etc. This appears in line with current research, which rarely publishes on residuals, or undertakes similar anal yses (Dollar and Kray 2002; Grimm and Gunther 2004). 31 The dummy variable, Before WTO, is excluded from this discussion. 32 Beyond the Shapiro-Wilkens test, special attention was paid to skewness and kurtosis.

PAGE 125

125 .436 to .856 for Model 3, and .857 for Models 1 and 2. While it still is no t significantly normal, having made an attempt to reduce their impact increasingly justifie s their initial inclusion in the model. This interpretation becomes especially evident when comparing the two dramatically contrasting histograms fo r Model 3 (Appendix A). For the other variable, Exports of Agricultu ral Raw Materials, a similar problematic situation arises, as Burkina Fasos amount of cott on and hides exports results in the creation of a dramatic outlier, with its agri cultural raw materials exports amount to 68% in 1998, well above the mean of 10.5%. Consequently, as the outlier has been verifie d, a square root transformation yielded vastly superior results (histograms in Appendix A). While the final transformed variable is not passing the normality test, it is vastly supe rior, with .850, to the previous .584 value of the Shapiro-Wilkens test in Model 3, and a superior .892 for Model 2. Last but not least, the normality of the Food Exports variable is closely evaluated. As becomes apparent with an observation of the hist ogram, the variable is very heavily influenced by the replaced missing variables, and ultimatel y no transformation was undertaken, as it yielded already a Shapiro-Wilkens value of above .9 for Model 3 and .925 for Model 2. Concerning the unique variables used in Model 1, total agricultural trade exports as share of total merchandise exports [TOTAGEXP] and as share of the GDP [LNAGXGDP], the latter improved due to a log-transformation, with th e former not improving. Ultimately, both variables yielded a normality value above .85 (.887 and .864), which appears satisfactory given the limitations. It is within this context that the post-tra nsformation normality tests using Shapiro-Wilkens provided sensible and lucid improvements, as presented in the post-transformation table in Appendix A.

PAGE 126

126 As discussed above, considerable outliers exist in the datase t that might need remedial measures. In order to further evaluate their importance and impact, an analysis of Cooks D, leverage and Mahalanobis distance was undertaken on the whole dataset for the above regression (Appendix B: Outliers). The mean Cooks D of Model 1 was .031, with a high of a low.164. Furthermore, the average leverage (centered) value was .184 with a maximum of .544, while the Mahalanobis distance averaged 6.816, with a maximum of a low 20.130. For Model 2, the mean Cooks D was .028, with a maximum of .127. For the centered leverage value, the average was .184, with a maximum of .707. Finally, for the Ma halanobis distance, th e average measured 6.816, with a maximum of 26.171. The mean Cook s D of Model 3 was .027, with a high of .416. Similarly, the average leverage (centere d) value was .140 with a maximum of .613. Mahalanobis distance recorded a mean of 6.86, with a maximum value of 30.054. Concerning the Cooks D, a simple analysis shows that none of the values listed in Appendix B are above 1 in all three of the m odels, which is a commonly used cut-off value (Garson 2007). An alternate test used for the cut-offs is 4/[n-k-1], where n is the number of cases 49 (or 37 respectively), and k is the number of independent variables (7). Based upon this measure, our cut-off value would be significantl y lower, at 0.095 (or 0.1 33 respectively). While these cut-offs are very low, the only value above it in Model 3 is the 1998 Burkina Faso in, with .416, reflecting its Raw Agricultural Exports figure In Model 2, none of the values are above the 0.133 cut-off. In Model 1, only Nigeria (1997) is slightly above this cut-off at 0.164. Focusing on the leverage measur e, using the .5 cutoff, only Bu rkina Faso in Model 2 and 3 again is above it, with .707 and .613 respectively. If applying an altern ate measure of 2p/n for Model 3, where p is the number of parameters including the intercept, we receive a cut-off

PAGE 127

127 around .33. Based upon this cut-off value, Sout h Africas 1993 and 1997 variables are on the brink in Model 3, relating to their GDP outlier.33 Finally, using a third alternate measurement, the Mahalanobis distan ce, Burkina once again is by far the greatest in both Model 2 and 3, w ith a value of 26.17 and 30.05, and South Africas 1993 and 1997 value next with 16.7 and 15.7 in Mode l 3, yet below 15 in Model 2 respectively. Consequently, it appears clear that these three potential outliers exist, yet they are all affirmed to not stem from data error and thus have been left in the model while clearly taken into considerations for the further analyses. For Mode l 1, only Nigeria in 1997 is above 15, yet with 20.13 clearly below the cut-off of 30. The next assumption tested relates to the normally distributed errors of the model (Appendix C: Normally Distribute d Error). As can be clearly seen in the histograms for each model of the standardized residua ls, they appear to fit well the normal curve, as well as the P-P plot. A formal test of Shapiro-Wilkens shows great normality for all models, with values ranging from of .979 to .984, at significance of .690 or greater, well above the .05 significance level. Plotting the residuals, homoscedas ticity can be identified in the models. One of the most common scatter plots of the standa rdized predicted values and th e standardized residuals have been plotted in Appendix D for Model 1 and 3, and it becomes clear that no pattern can be identified in terms of their structure. Furthermore, a simple test of nonlinearity ha s been undertaken (Appendix E). As stated by Garson, as a rule of thumb, an indicator of possi ble nonlinearity is when the standard deviation of the residuals exceeds the sta ndard deviation of the dependent (Garson 2007). In both Model 33 If considering the overall relationship of their GDP ppp per capita value to the others, it appears that South Africas value fits well into the model.

PAGE 128

128 1 and 2, the standard deviation of the Gini (7.918) is greater than the respective standard deviation of the unstandardized residuals (5.61 and 4.96). In Mode l 3, the standard deviation of the unstandardized residuals of 5.28 does not exceed the Ginis value stan dard deviation of 7.39, showing that nonlinearity is unlikely to be an issue. One issue specific to multiple regression models is multicollinearity (Appendix F: Multicollinearity). Correlation it self is clearly not a problem, as the Pearson correlation values never exceed .7 in all models, with the larg est correlations existi ng in Model 3 between urbanization and log GDP ppp per capita, with .619, and Model 1 and 2 log GDP ppp per capita and Agriculture value added, with -.689. Be yond that, there are thre e types of tests for multicollinearity undertaken. First, investigating tolerance, none of the va lues in Model 2 and 3 are below the cut-off value of .20, with log GDP being closest in both Model 2 and Model 3. Only Model 1 appears to have a low tolerance value for Total Agricultur al Exports, with .184. A second similar test for multicollinearity is the variance inflation factor (V IF), which has a general cut-off value of larger than 4.0. As becomes evident, in the Model 3, we also find no independent variables that exceed that point, with log GDP ppp per capita the clos est with 2.799. For both Model 1 and 2, log GDP ppp per capita appears slightly ov er, while total agricultural expor ts (as share of merchandise exports) again are the highest, w ith a VIF of 5.4. Finally, when looking at the condition indices for Model 3, we find that the eighth dimension of even that model displays a very high condition over 80, indicating that there is possibly a proble m of multicollinearity. C onsequently, a closer analysis of the variance proportions shows that the insignificant variables in our model, LogGDP and SqrootRawTrad caused t hose high influences.

PAGE 129

129 The problem of multicollinearity, if taken seriously, appears straightforward to address, as it would involve the altering of our regression m odel from an enter regression to a stepwise method (Appendix F for the stepwise results). Th is endeavor has been undertaken for Model 3, but also confirmed for the other m odels to resolve all the problems of multicollinearity. After this change, we find that none of the values are less than .20 in terms of tolerance or greater than 4.0 in terms of VIF. Furthermore, the resulting condition index is clearly superior, with only 17.29 as opposed to the previous value of 80. Urbaniza tion now seems to have the greatest influence, with .81, followed by the value added by agricultur e with .49. While this value certainly is above the more stringent cut-off of 15 (as opposed to 30 traditionally used), the fact that not two values are greater than .5 is broadly interprete d as a sign that multicollinearity is not a defining issue (Garson 2007). We also find that Cook s distance and leverage values are very significantly lower. A final test was undertaken to assure th at the assumption of independence was met (Appendix G: Independence). This assumption is usually tested using the Durbin-Watson test, with a value ranging from 1.5 to 2.5. Using the en ter regression model, th e Durbin-Watson test clearly is within that range for all models Model 1 has a value of 1.807, Model 2 has 2.062 and Model 3 has a value of 2.131, indicating independence of the observations. In summary, the main issue that needs to be taken into account ar e outliers (mainly in Model 3) and multicollinearity (mostly Model 1 and 2) that exist. However, the latter can easily be addressed and become negligible when undert aking a stepwise regression. A summary of the dataset used for Model 3 has b een included in Appendix H. To further highlight this case in point, a fourth regression wa s undertaken, without replacing the missing values with the mean, as well as eliminating Burkina Faso. While this

PAGE 130

130 dramatically reduces the degrees of freedom, the situation has been amended by using a stepwise procedure on the same independent variables ch osen in the previous regression model. The resulting adjusted R2 is considerably higher, with .509 th an the previous regression, with the Pre-WTO and Urbanization no longer significant. Overall, the model appears to confirm however the previous findings that agriculture is related to inequality, with higher shares in agricultural food exports and higher agricultural production va lue added decreasing inequality. Given that we had some serious problems with outliers and multic ollinearity in the previous regression, it is importa nt to note that this regressi on is significantly improved. Multicollinearity sta tistics show that outliers are not an issue, as both the tolerance values (.804) and the VIF values (1.244) are significantly be yond their cut-off values. Furthermore, the condition index for the respective final model is well below the traditional cutoff value (30) at 6.807, displaying no multicollinearity. In addition, an analysis of outliers (e.g. Cooks D) shows that the maximum value of .417 is considerably lower than the previous models outliers. Independence has also been observed for the model, with a Durbin-Watson value of 1.901. Finally, as a caveat for future research, it however appears that a quadratic equation might be better fitting than the linear regression model, as displayed in the scatterplots of the standardized predicted values against the sta ndardized residuals (Appendix I). Overall 15 variations of the regression models has been undertaken and both r-squared and the independent variables are c onsistent, indicating that the re gression models, especially the agricultural independent va riables, are tremendously robust (A ppendix M). More specifically, in each of the 15 models, agriculture value added (a s% of GDP) is highly significant. Food exports (as% of merchandise exports), when used in eight models is significant also in each of them. Total agricultural exports, on the other hand, as a percentage of merchandise exports, included in

PAGE 131

131 the other seven models, is significant in two cases. As a percentage of overall GDP, total agricultural exports however are in none of the seven cases significant. Similarly, GDP ppp per capita and infant mortality were not significant, whereas urbanization wa s significant in five models. The Before WTO dummy variable was significant in six cases, especially in models with higher degrees of freedom. Fi nally, the adjusted R-squared va lue of the regression models ranged from a minimum of .32 to a maximum of .614. Secondary Hypotheses First Hypothesis: Role of Agriculture on GDP Given our interest in both growth and inequality, a similar regression model was constructed surrounding the GDP ppp per capita variable (Appe ndix K). As opposed to the previous inequality model, this model was run with a stepwise regression, which only included agricultural variables and did not replace any mi ssing variables, as its ultimate goal is to highlight any differences between inequality and GDP from an agricultural perspective34. Similar to the inequality models, agriculture value added (as% of GDP) proved to have a significant negative coefficient (.368), which is a bit weaker (a nd less significant) than the variables relationship with inequality. Furt hermore, the previously significant WTO dummy variable no longer proved significan t, yet the created agricultural diversity variable was highly significant at .016 with a standardiz ed coefficient of -.449. In sum, these two variables yielded a total adjusted R2 of 0.513, which is similar to the amount of variation in the Gini variable explained by the independent agricultural variable s. What the resulting regression equation thus shows is that increased shares of agricultura l production are related with lower GDP ppp per capita, confirming that agricultural activities ar e associated with lower levels of economic 34 The complete list of variables used is listed at the bottom of the regression table.

PAGE 132

132 development. However, more inte restingly is the finding that the value share of th e top 3 of the top 20 agricultural export crops is correlated with a lower per capita GDP ppp, suggesting the negative impact of dependency on a few cashcrops in the long-run for the economies under study, or the benefits of an ove rall increased divers ity in export produc tion, both within and between the agricultural sector and the other sectors35. Running a simple linear regression with the missing variables replaced with the mean based upon the inequality model 1 (Table 5-1), we increase our degrees of freedom slightly, but gain a similar regression result. The same two i ndependent variables, agri culture value added (as percentage of GDP) and the valu e share of the top 3 of the to p 20 exports, remain significant, with an overall similar impact and strength. Most tellingly, however, is that the model fails to pass the test of independence, with a low Durb in-Watson of 1.133, holds similar properties of multicollinearity, with a higher condition index va lue (due to enter-model) of 20.977 and a lower maximum value of Cooks distance with .275 (Appendix J). As shown in the models, agriculture value a dded (as percentage of GDP), food exports and total agricultural exports as percentage of tota l merchandise exports had a negative relationship with inequality (and the former with GDP). This relationship has been depicted in the scatterplots below. While there appears no correlation between the Gini coefficient and agricultural raw exports share of total exports, by excluding Burkina Faso and using the transformed root-squared variable a clear linear relationship become s apparent (see scatterplot). A summary of the correlations has b een listed below, showing that all three agricultural variables (production, food and agricultural raw material exports) are signifi cantly correlated with the Gini 35 A brief analysis of the assumptions (Appendix K) shows that the regression passes independence (Durbin-Watson: 2.410), multicollinearity (Tolerance: .567; VIF: 1.765; Cond ition Index: 13.622), has no outliers (largest Cooks D: 1.256), and a normal distribution of its standardized residuals (see Histogram).

PAGE 133

133 variable, causing us to reject the null-hypothesis that no statisti cally significant relationship exists between agricultural exports and inequality. In relationship to GDP ppp per capita, we ha vent found in the regr ession models that countries with higher shares of agricultural exports have higher GDP per capita. As a matter of fact, neither the amount of raw agricultural nor food exports share of total exports was significantly correlated (Table 56 ). The strongest potential relationship appears to exist between agricultural food exports and the log GDP ppp per capita, yet this is only significant at the .139 level. Consequently, we reject the hypoth esis that a statistically significant relationship exists between agricultural expor ts and GDP per capita (PPP). Furthermore, on a final note, as becomes appare nt, there is also no st atistically significant relationship between the Gini coefficient and our chosen measures of economic development, GDP per capita. Second Hypothesis: Amount of Arable Land As seen in the above tests of inequality and growth, the amount of arable land was not significant in the model. Consequently, this section briefly interroga tes (through correlation analysis and scatterplots), whethe r or not an inverse re lationship exists, in which countries with more arable land per capita have greater in equality, as proposed by the literature. As seen in the scatterplot, there appears a significant relations hip between the Gini coefficient and the amount of arable land per capita, with larger am ounts of arable land associated with greater inequality. This correl ation is statistically si gnificant at .015, with a correlation of .342. Testing the same variable against both GDP ppp per capita and Log GDP ppp per capita, we interestingly fi nd that they are not significantly correlated. Consequently, our second hypothesis is confirmed that there exis ts a significant positive relationship between an increase in arable land per capita and an increase in income inequality.

PAGE 134

134 Third Hypothesis: Dependency on Few Agriculture Exports As already has been shown in the previous regression models, the third hypothesis has been confirmed in relationship to GDP ppp per capit a (Figure 5-8, with South Africa as outliers). However, it is unclear whether or not it holds a st atistically significant relationship on inequality as well. As depicted in the scatterplot below, a very weak negative relationship appears to exist, hinting at a potential decrease in inequality as the sh are of the top three agricultural export goods increases. As depicted in th e table below, this correlation is not significant at .145, thus we reject our hypothesis that countri es with greater dependency on a few exports crop have greater inequality, while affirming the latter in relationship to GDP ppp per capita (and log GDP ppp per capita). On a side note, to further investigate the relationship, a cash crop dummy was created in order to preliminarily evaluate th eir potential relationship to inequa lity. Interestingly, countries, whose largest agricultural export for a given year was either cotton, coffee, tea or cocoa, had a statistically significant lower Gini coefficient (Tab le 5-9). This finding seems to contradict what Moradi and Baten (2005) have found in their st udy, where cash crops were negatively correlated with increased intra-regional inequality. However, as the sample appears heavily biased towards years and Gini values from countri es that exported cash crops (37) compared to those years that cash crops were not the top export (13), a great degree of caution is in order. Furthermore, given their different scale of their study, our nationa l level analysis might not observe the same phenomena. Fourth Hypothesis: Agricultural Policies and Inequality As a proxy for agricultural policy, the time dummy has been created, which has already been proven significant in the previous regression models on inequality. A correlation analysis of the dummy between both the Gini and GDP ppp per capita shows that this statistically

PAGE 135

135 significant relationship only holds tr ue for inequality, yet is not si gnificant for our measures of growth. Consequently, this an alysis confirms our previous findings that the time-proxy for agricultural policy is positively correlated with th e Gini coefficient at the 0.05 level, indicating that the Gini values were higher pre-1994, and have dropped since the formation of the WTO and the implementation of the Uruguay round results In other words, inequality in African countries appears to have actually decreased after the WTO agreement in 1994, disproving our previous hypothesis. Fifth Hypothesis: Urban Areas and Inequality As weve found already in several regression m odels, the rate of ur banization was the one non-agricultural independent variab le explaining inequality. Th is hypothesis will nevertheless be tested in order to observe the strength of the correlation, bo th in relationship to income inequality and development. Interestingly, there does not appear to be a strong linear association between Gini and urbanization, highlighting the uniqueness of the variable in combination with the agricultural variables. A statistical significance however does exist between the GDP ppp per capita, its logged version and urbanization. This relationship is positive, indicating that higher rates of urbanization is related with greater levels of economic income, as measured through GDP ppp per capita, at a high significance below the 0.001 le vel. Thus, it appears our fifth hypothesis is also placed in question that countries with hi gher percentage of urba n population have greater inequality. Population density has also been used as an alternate, approxi mate measurement of urbanization. The only significant correlati on found, however, is a very weak one with urbanization, indicating that th e greater urbanization value, th e lower the population density, a finding that actually is contradict ory to what would be expected. Given its lack of correlation

PAGE 136

136 and the superiority of the urbani zation variable, this issue is on ly of minor importance. The weak correlation between the Gini coefficient an d population density can al so be observed in the scatterplot below. In summary, the rate of urbanization does not ap pear to be significantly correlated with the Gini coefficient, but has been found to play a si gnificant role in the in come inequality models. Urbanization, however, has displa yed a strong (linear) relationshi p with measures of economic income, using both the GDP ppp per capita as well as the log GDP ppp per capita. On a final note, as exhibited in the last scatterplot, this re lationship might not necessarily be a linear one, as shown with the cubic best fit line plotted, yieldi ng a superior correlation value. Finally, adding a geographical dimension, land-locked countries have been found to have a lower level of economic income

PAGE 137

137 Table 5-1. Model 1 regression results (Adjusted R2: 0.382; R2: 0.499) Missing Variables replaced with Mean (df: 30)^ Regression Type: Enter Dependent Variable: Gini Coefficient Unstandardized Coefficients (B) Standardized Coefficients (beta) t Sig. (Constant) 58.958**2.364.025 Before WTO 3.117.1981.281.210 Agriculture Value Added -.274-.407*-2.029.051 Total Agricultural Exports as percentage of merchandise exports -.195-.669**-2.219.034 Log of Agricultural Exports as percentage of GDP 1.809.2381.051.302 Log of GDP ppp per capita (international US$) 2.387.195.752.458 Children's Infant Mortality rate (Deaths per 1000 live births -.00167-.005-.030.976 Urbanization -.375-.550**-2.222.034 Note: *: significant at .10 leve l; **: significant at .05 level, **: significant at .01 level, ****: significant < 0.01 level ^: This model has an n=38, with a minimum of 2 data points per countr y, and a maximum of 3. : Calculated as the combination of Food E xports and Agricultural Raw Exports (as% of merchandise exports).

PAGE 138

138 Table 5-2. Model 2 regression results (Adjusted R2: 0.516; R2: 0.608) Missing Variables replaced with Mean (df: 30)^ Regression Type: Enter Dependent Variable: Gini Coefficient Unstandardized Coefficients (B) Standardized Coefficients (beta) t Sig. (Constant) 86.249***3.801.001 Before WTO 5.171.329**2.518.017 Agriculture Value Added -.425-.632****-3.427.002 Food Exports as percentage of merchandise exports -.176-.538****-3.846.001 Square Root of Agricultural Raw Materials Exports as percentage of merchandise exports 1.235.2171.485.148 Log of GDP ppp per capita (international US$) -1.343-.110-.466.645 Children's Infant Mortality rate (Deaths per 1000 live births -.00383-.116-.807.426 Urbanization -.270-.396**-2.058.048 Note: *: significant at .10 leve l; **: significant at .05 level, **: significant at .01 level, ****: significant < 0.01 level ^: This model has an n=38, with a minimum of 2 data points per country, and a maximum of 3. It is identical, with the exception of the lower data points, to Model 3.

PAGE 139

139 Table 5-3. Model 3 regression results (Adjusted R2: 0.403; R2: 0.489) Missing Variables replaced with Mean (df: 42) Regression Type: Enter Dependent Variable: Gini Coefficient Unstandardized Coefficients (B) Standardized Coefficients (beta) t Sig. (Constant) 75.210****3.972.000 Before WTO 3.634.248**2.168.036 Agriculture Value Added -.356-.580****-3.839.000 Food Exports as percentage of merchandise exports -.153-.431****-3.293.002 Square Root of Agricultural Raw Materials Exports as percentage of merchandise exports .651.110.880.384 Log of GDP ppp per capita (international US$) -.585-.047-.252.802 Children's Infant Mortality rate (Deaths per 1000 live births -.00783-.023-.168.867 Urbanization -.215-.320*-1.938.059 Note: *: significant at .10 leve l; **: significant at .05 level, **: significant at .01 level, ****: significant < 0.01 level : Excluding South Africa and substitu ting with the GDP ppp per capita, the R2 value drops to .457, with an adjusted R2 value of .361.

PAGE 140

140 Table 5-4. Regression resu lts Gini coefficient (Adjusted R2: 0.509; R2: 0.545) Missing Variables not replaced with Mean and Burkina Faso (df: 25) Regression Type: Stepwise Dependent Variable: Gini Coefficient Unstandardized Coefficients (B) Standardized Coefficients (beta) t Sig. (Constant) 61.00720.579.000**** Agriculture Value Added -.325-.539-3.582.001**** Food Exports as percentage of merchandise exports -.085-.320-.2126.044** Note: *: significant at .10 leve l; **: significant at .05 level, **: significant at .01 level, ****: significant < 0.01 level Table 5-5. Regression re sults GDP ppp per capita (Adjusted R2: 0.513; R2: 0.548) Missing Va riables replaced without Mean (df: 26) Regression Type: Stepwise Dependent Variable: GDP ppp per capita (inter national US$) Unstandardized Coefficients (B) Standardized Coefficients (beta) t Sig. (Constant) 7397.0126.130.000**** Value share of 3 of top 20 agricultural exports -54.402-.449-2.565.016** Agriculture Value Added -58.254-.363-2.072.048** Note: *: significant at .10 leve l; **: significant at .05 level, **: significant at .01 level, ****: significant < 0.01 level Variables not significant: Arable Land (ha pe r person), Food Exports as percentage of merchandise exports, Agricultural Raw Materials Exports as percen tage of merchandise exports, Before WTO

PAGE 141

141 Table 5-6. Regression resu lts GDP ppp per Capita (Adjusted R2: 0.484; R2: 0.554) Missing Variables replaced with Mean (df: 32) Regression Type: Enter Dependent Variable: GDP ppp per capita (international US$)^ Unstandardized Coefficients (B) Standardized Coefficients (beta) t Sig. (Constant) 8.811****15.825.000 Before WTO -0.01856-.015-.114.910 Value share of 3 of top 20 agricultural exports -.0153-.335**-2.258.031 Agriculture Value Added -.02927-.532****-3.466.002 Total Agricultural Exports (% of merchandise exports) 0.00361.151.918.365 Arable Land (ha per capita) .279.052.335.740 Note: *: significant at .10 leve l; **: significant at .05 level, **: significant at .01 level, ****: significant < 0.01 level ^: This regression model is based on the inequality models 1 and 2 in terms of the data set used, but uses the combined variable of food and agricultural raw mate rials exports, to tal agricultural exports. The model here uses the logged GDP ppp per capita, but even if not logged, the model changes minimally. Overall, these results are also extremely robust, as both agriculture, value added, and value share of 3 of the top 20 agri cultural exports always ranked significant, no matter if the regression model was stepwise or enter, replaced with missing values or not replaced, or log GDP ppp per capita or not logged.

PAGE 142

142 Table 5-7. Correlations of agricu ltural exports and inequality Gini Coefficient Agriculture Value Added Food Exports (% of total exports) Square-Root Agricultural Raw Material Exports (% of total exports) Pearson Correlation 1-.482**-.580** -.497** Gini Coefficient Sig. (2-tailed) ..000.001 .004 Pearson Correlation -.482**1.427* .402* Agriculture Value Added Sig. (2-tailed) .000..021 .025 Pearson Correlation -.580**.427*1 .514** Food Exports (% of total exports) Sig. (2-tailed) .001.021. .005 Pearson Correlation -.497**.402*.514** 1 Square-Root Agricultural Raw Material Exports (% of total exports Sig. (2-tailed) .004.025.005 ** Correlation is significant at the 0.01 level (2-tailed). Correlation is significant at the 0.05 level (2-tailed).

PAGE 143

143 Agriculture Value Added60 50 40 30 20 10 0 70 60 50 40 30 20Rsq = 0.2320 Figure 5-4. Scatterplot agriculture value added

PAGE 144

144 Food Exports as percentage of merchandise exports100 80 60 40 20 0 70 60 50 40 30 20Rsq = 0.3368 Figure 5-5. Scatterplot f ood exports as percentage of merchandise export

PAGE 145

145 Agricultural Raw Materials Exports as percentage of merchandis e 70 60 50 40 30 20 10 0 -10 70 60 50 40 30 20Rsq = 0.0003 Figure 5-6. Scatterplot agricultura l raw materials exports as per centage of merchandise exports

PAGE 146

146 SQTRADRA6 5 4 3 2 1 0 -1 70 60 50 40 30 20Rsq = 0.2758 Figure 5-7. Scatterplot square-r oot raw agricultural exports as percentage of merchandise exports

PAGE 147

147 Table 5-8. Correlations of agricu ltural exports and development GDPPPPCA LOGGDPPPAGVALUE TRADFOOD SQRTTRAW Pearson Correlation 1.883**-.543**-.339 -.066 GDPPPPCA Sig. (2tailed) ..000.000.072 .725 Pearson Correlation .883**1-.568**-.282 .131 LOGGDPPP Sig. (2tailed) .000..000.139 .481 Pearson Correlation -.543**-.568**1.427* .402* AGVALUE Sig. (2tailed) .000.000..021 .025 Pearson Correlation -.339-.282.427*1 .514** TRADFOOD Sig. (2tailed) .072.139.021. .005 Pearson Correlation -.066.131.402*.514** 1 SQRTTRAW Sig. (2tailed) .725.481.025.005 ** Correlation is significant at the 0.01 level (2-tailed). Correlation is significant at the 0.05 level (2-tailed).

PAGE 148

148 Log of GDP ppp per capita (international US$)9.5 9.0 8.5 8.0 7.5 7.0 6.5 6.0 70 60 50 40 30 20Rsq = 0.0392 Figure 5-8. Scatterplot of log of GDP ppp per capita

PAGE 149

149 Table 5-9. Correlations of arable land (ha per person) Gini Coefficient GDP ppp per capita (international US$) Log of GDP ppp per capita (international US$) Arable Land (ha per person) Pearson Correlation 1.273.198 .342** Gini Coefficient Sig. (2-tailed) ..055.168 .015 Pearson Correlation .2731.882 .152 GDP ppp per capita (international US$) Sig. (2-tailed) .055..000 .293 Pearson Correlation .198.8821 .057 Log of GDP ppp per capita (international US$) Sig. (2-tailed) .168.000. .692 Pearson Correlation .342**.152.057 1 Arable Land (ha per person) Sig. (2-tailed) .015.293.692 ** Correlation is significant at the 0.05 level (2-tailed)

PAGE 150

150 Arable Land (ha per person).7 .6 .5 .4 .3 .2 .1 70 60 50 40 30 20Rsq = 0.1173 Figure 5-9. Scatterplot arable land

PAGE 151

151 Value share of 3 of top 20 agricultural exports100 90 80 70 60 50 40 30 10000 8000 6000 4000 2000 0Rsq = 0.3007 Figure 5-10. Scatterplo t value share of three of the top 20 agricultural expor ts versus GDP ppp per capita

PAGE 152

152 Value share of 3 of top 20 agricultural exports100 90 80 70 60 50 40 30 70 60 50 40 30 20Rsq = 0.0438 Figure 5-11. Scatterplo t value share of three of the top 20 agricultural expor ts versus Gini coefficient

PAGE 153

153 Table 5-10. Correlations of value share of 3 of Top 20 agricultural exports Gini Coefficient GDP ppp per capita (international US$) Log of GDP ppp per capita (international US$) Value share of 3 of top 20 agricultural exports Pearson Correlation 1.273.198 -.209 Gini Coefficient Sig. (2-tailed) ..055.168 .145 Pearson Correlation .2731.882 -.548**** GDP ppp per capita (international US$) Sig. (2-tailed) .055..000 .000 Pearson Correlation .198.8821 -.502**** Log of GDP ppp per capita (international US$) Sig. (2-tailed) .168.000. .000 Pearson Correlation -.209-.548****-.502**** 1 Value share of 3 of top 20 agricultural exports Sig. (2-tailed) .145.000.000 **** Correlation is signif icant at < 0.01 level

PAGE 154

154 Table 5-11 Correlations of cash crops dummy variable Cash Crops Gini Coefficient Pearson Correlation 1-.358** Cash Crops Sig. (2-tailed) ..011 Pearson Correlation -.3581 Gini Coefficient Sig. (2-tailed) .011. ** Correlation is sign ificant at 0.05 level Cash Crops Dummy: 1 = The top export crop for a given year was coffee, cotton, cocoa or tea.

PAGE 155

155 Table 5-12. Correlations of agricultur al policies, inequa lity and growth Gini Coefficient GDP ppp per capita Log GDP ppp per capita Before WTO Pearson Correlation 1.273.198 .304* Gini Coefficient Sig. (2-tailed) ..055.169 .032 Pearson Correlation .2731.883 -.037 GDP ppp per capita Sig. (2-tailed) .055..000 .800 Pearson Correlation .198.8831 -.037 Log GDP ppp per capita Sig. (2-tailed) .169.000. .801 Pearson Correlation .304*-.037-.037 1 Before WTO Sig. (2-tailed) .032.800.801 ** Correlation is significant at the 0.01 level (2-tailed). Correlation is significant at the 0.05 level (2-tailed).

PAGE 156

156 Population Density (people per square kilometer)140 120 100 80 60 40 20 0 70 60 50 40 30 20Rsq = 0.0007 Figure 5-12. Scatterplo t of population density

PAGE 157

157 URBANIZA60 50 40 30 20 10 0 70 60 50 40 30 20Rsq = 0.0181 Figure 5-13. Scatterp lot of urbanization URBANIZA60 50 40 30 20 10 0 9.5 9.0 8.5 8.0 7.5 7.0 6.5 6.0Rsq = 0.5974 Figure 5-14. Scatterplo t of non-linear (cubic) relationship be tween urbanization and log of GDP ppp per capita

PAGE 158

158 Table 5-13. Correlations of urba nization, inequality and growth Gini Coefficient GDP ppp per capita Log GDP ppp per capita Urbanization (%) Population Density Pearson Correlation 1.273.198.135 -.027 Gini Coefficient Sig. (2tailed) ..055.169.352 .854 Pearson Correlation .2731.883.515** -.123 GDP ppp per capita Sig. (2tailed) .055..000.000 .394 Pearson Correlation .198.8831.619** -.103 Log GDP ppp per capita Sig. (2tailed) .169.000..000 .476 Pearson Correlation .135.515**.619**1 -.308* Urbanization (%) Sig. (2tailed) .352.000.000. .029 Pearson Correlation -.027-.123-.103-.308* 1 Population Density Sig. (2tailed) .854.394.476.029 ** Correlation is significant at the 0.01 level (2-tailed). Correlation is significant at the 0.05 level (2-tailed).

PAGE 159

159 Table 5-14. Correlations of landlocked dummy variable Landlocke d Log of GDP ppp per capita (international US$) Pearson Correlation 1-.387**** Landlocked Sig. (2-tailed) ..006 Pearson Correlation -.3871 Log of GDP ppp per capita (international US$) Sig. (2-tailed) .006. **** Correlation is signif icant at < 0.01 level

PAGE 160

160 CHAPTER 6 DISCUSSION Empirical Contributions The role of agriculture, as measured through production and exports in relationship to income inequality in sub-Saharan Africa has seld om been addressed by previous research. While previous studies have investigated the impact of inequa lity on growth, or vice versa, as well as the relationship between income inequality a nd other socio-economic variables, the findings developed in our regression models statistically significantly show that there appears a strong relationship between inequality and agricultural production and e xports in sub-Saharan Africa, confirming our main hypotheses. This main finding should not come as a surpri se. Given the complexities of the dependent variable, inequality, and the independent variable of interest, agriculture, it might strike us as a surprise, however, which specific variables were found to be signi ficant and insignificant in the models respectively. One might be surprised to find that agricultural variables still were the most significant, even when placed in a model with co nventional inequality-shaping variables, such as degree of urbanization, infant mortality, or GDP ppp per capita. Given that urbanization was significant in some models, the most surprising non-factor in the equa tion might arguably be infant mortality, which might have been expect ed to measure to some degree socio-economic inequalities as reflected in mortal ity figures according to Sen (1998). The least surprising of the inde pendent significant variables app ears to be the percentage of value added of agriculture to the GDP. Given the overall complexities of the agricultural sectors forces, this broad vari able seems well-fit to capture the overall strength and importance of agricultural production in a c ountry. While partially reminis cent of the Kuznets hypothesis, the countries with a higher share of value added of agriculture seem to enjoy a lower amount of

PAGE 161

161 income inequality. It is clearly difficult to pinpoint specifically to a single factor influencing this result, such as a heavier taxation of the sector immobility of labor due to lack of employment opportunities in other sectors and being tied to the land, or successful development aid policies targeting agriculture. Even though failing to highlight specific complexities, as shown in Chapter 4, the agricultural sector plays an important role in a majority of the citizens under study and thus appears to reflect well the importance it plays not only on their economic developments, through income and growth, but also on inequality. Consequently, similar to such case studies as Adams (1995), these findings put forward the cl aim that agricultural production matters in relationship to inequality. The other agricultural variab les found significant, agricultu ral food exports and total agricultural exports as percentage of total merchandise exports, allow us to delve below the surface of the previous variable, as it highlig hts the importance of agricultural exports. Consequently, while the previous variables m easured agricultural production, this variable measures agricultural production th at resulted in exports. The re gression models show that the share of agricultural exports is significantly related to the amount of inco me inequality observed in our countries. However, beyond that signif icant relationship, it also contributes to the growing body of existing scholarship on the impact of trade and exports (sometimes broadly labeled as globalization), by displaying a negative relationship between an increased share of agricultural exports and income in equality. This finding, compared to the previous variables empirical contribution, appears far more contentious, as it seems to support arguments in favor of an export-oriented agricultural sector in connect ion with reducing inequality. The slope of the variable appears to indicate th at an increase in agricultural e xport commodities, most likely cash crops, is correlated with a lower amount of inequa lity at the national scale. This argument is

PAGE 162

162 strengthened by the negative and significant correlation between cash crop exports in a given year and the Gini coefficient, as well as the robustness of the variable s importance. This appears to support Adams findings that cash cr ops are able to positively influence incomedistributions through th e creation of new income and em ployment opportunities (1995, 467). One of the countries that is displaying a high amount of exports and a low amount of inequality is for example Ethiopia. Recalling Ethiopias ag ricultural sectors stru cture, it is heavily dominated by coffee exports produced by smaller-scal e farmers. Consequently, given this more equal distribution of cash crop produ ctions in Ethiopia, as opposed to South Africa or Kenya, the country appears to be a prime example supporting this variables direction of the slope. It is important to note that sc ale is a key factor in our anal ysis. At the local and regional level the introduction of export pr oduction and/or cash crops might increase the gap between the richer and the poorer farmers. Since our study fo cuses on the national level, this would not be the case on this aggregated level for the followi ng reason. Given the lower starting point of wealth for the rural wealthier farmers, their increased share of income would be inequality reducing at the national scale, as they move upwards on the ladder from the bottom towards the middle tier, effectively reducing national inequality. Since the variable measures the share and not the absolute amount of agricultural food exports, it could be derived that, similarly to th e previous dependent va riable on agricultural GDP, that a large manufacturing export sector, on other hand, has an inequality increasing impact, placing South Africa at the other end of the strata. However, the formal statistical analysis undertaken on that vari able (as agricultural exports as percentage of GDP) was not significant. Nevertheless, while appearing to contribute to th e literature by highlighting the negative relationship between agricultural expo rts and income inequality, with inequality

PAGE 163

163 decreasing, there is a chance that the variable might just measure agricultur al exports vis--vis a weak manufacturing sector. However, given th e lack of correlation, the models support the empirical finding that this variable adds a uni que, significant dimension to our understanding of inequality. This finding of the importance of the expor t-orientation of agri cultural production gains further weight by the third agricultural variab le found to be significant: the WTO time dummy. As indicated by the positive slop e, countries income inequalitie s are statistically significantly lower after the end of the Uruguay Round and the creation of the WTO. While the variable certainly could measure other important factors, such as the wave of democratization and increasing stability that swept across the African landscape in the 1990s, even in its broader interpretation it appears to hold it s weight. In fact, as observed in our case studies, almost every country has undergone by the end of the 20th century a process of liberalization and/or privatization of its agricultural sector. These reforms included a reduction of export barriers, and benefiting from facilitated access to some main export markets, i.e. EU, in the aftermath of the implementation of the Uruguay Round. Of course, this transformation was by no means linear. Nevertheless, it appears to add stre ngth to the previous variables findings that increased export orientation is not related to increased, but overa ll decreased inequality. Observing the lack of correlations between the WTO dummy and measur es of development (GDP ppp per capita), it appears that the impact of the independent du mmy variable is applicable only to income inequality, as the experience in terms of economi c development are more diverse. Consequently, while not necessarily increasing the overall inco me, as identified in th e case of Uganda for example (Kayizzi-Mugerwa 2001), the post-1994 time dummy might hint at the ability of poorer farmers of benefiting from higher pri ces of their crops in the aftermath.

PAGE 164

164 The control variable, urbanizati on, holds also a negative slope in several of our models, as an increase in urbanization is related to a decrease in income inequality. While not as significant as the other agricultural variables, it appears to capture forces that are not increasing the urbanrural dualism per se. Consequently the direction of the slope coul d hint at the po tential decline in the urban bias in sub-Saharan Africa (Lipton 1977), as it also might run counter to Bates claim of the urban interests creat ing policies hostile to farmers (Bates 2005). Furthermore, it could hint at an overall increase in local dema nd by the new urbanites for agricultural products for consumption and manufactur ing (Tiffen 2003), the creation of backwardand forwardlinkages, and a subsequent increase in the av ailability of non-farm income opportunities, including remittances, further stabilizing the livelihoods of the rural poor. From a political perspective, the relationship betw een urbanization and income ine quality in our regression model could also stem from an observe d increase in bargaining power of farmers, something that was not likely to occur in previous decades (Bassett 2 001). It is this latter argument concerning the bargaining power that will be pursued in the next section concerning the policy implications of these findings. Overall though, by observing the in teractions between the urbanization variable and the Gini coefficient in our correlation analysis (Table 5-11), it surpri singly is not significant, indicating the weakness of a linear relationship on the one hand, while furthermore highlighting the very strong correlations between urbanizatio n and GDP ppp per capita, as an increase in urbanization is associated with a signif icantly higher level of development. Agriculture is also statistical ly significant in explaining pa tterns of GDP ppp per capita, as measured by the independent variables of the va lue share of three of the top 20 agricultural exports and agriculture value adde d. The latter has been shown to be highly significant, with a negative slope, indicating that an increase of the contribution of agriculture to the GDP is related

PAGE 165

165 with a decrease in the level of development, a finding that appears not to be surprising and seems to confirm previous findings, such as by the FAO (2004) and Grant and Agnew (1996). Similarly, the dependency variable appears to support the previous cl aim that a lack of diversity will hold a negative impact on economic development in the long-run. This variables relationship to GDP ppp per capita is unique, and no significant co rrelation has been identified with the Gini coefficient. While not found in the regression model, anothe r key variable, arable land, is correlated with the Gini coefficient, yielding a positive correlation value, as proved previously in the literature (Dollar and Kray 2002). It consequently appears to support than an increase in arable land (ha per person) is correlated with a higher Gini coefficient. This finding was expected after the literature review a nd thus only adds to the previous sc holarship. Recalling South Africas struggle with land redistribution, th e country is a case in point, as it actually holds around .4 ha per capita, well-above the mean of .25 ha per capita, and one of th e highest Gini coefficients in our sample around 60. Overall, the findings that agricultural variab les significantly explain a great degree of income inequality in our case studies greatly c ontributes to the prior re search and understanding on inequality in sub-Saharan Africa. While these findings appear to support the Kuznets hypothesis, GDP ppp per capita was not found to be significant in our income inequality models, highlighting that agriculture and the overall structure of the ec onomy appears to better capture the complexities in our sample. Consequently, given the lack of linear relationship between the Gini coefficient and GDP ppp per capita, even if logged, the findings appear to place an important question mark behind its contribution to wards affirming this age-old hypothesis, as it highlights the importance of the overall structure of the economy, as opposed to its gross output.

PAGE 166

166 Policy Implications Translating the above empirical contributions into policy imp lications, there are two major implications that stand out: strengthening the WTO negotiations and supporting pro-poor agricultural development strategi es focused on diversification. Strengthening the WTO negotiations Positive impacts of a succe ssful (and meaningful) WTO D oha Development Round have been widely discussed in relationship to improve d world commodity prices at the international level, increased export revenue at the national le vel trickling down and resulting in a subsequent reduction of poverty ultimately. While questions concerning the long-term viability and economic, environmental sustainability of such an agreement remain, its impact on inequality has rarely been discussed (Ledermann and Mose ley 2006). This studys findings consequently add another important variable to the negotiating table for discussi on. As an agreement is likely to increase the share of agricultures contribu tion towards the economy, especially exports, it seems clear that it subsequently will have a positive impact on reducing inequality. Thus, it appears to add another key argument to both the First and Third world ne gotiators toolset. Given the First world policymakers increasi ng interest in reducing inequality through development, it appears that this goal, which is usually target ed specifically through aid, could be greatly impacted by achieving a truly historic breakthrough towa rds more free and fair trade of agricultural goods, guided by the principle of equitable growth. This additional argument of decreasing inequality might hold little weight in the public arena of th e interest groups most likely to lose by a reduction of subsidies and tariffs in the developed worlds, such as farmers and agro-businesses. Nevertheless, a narrative could be constructed highlighting symbiotic relationship between lesser dependency on aid in Africa and increased ability to support

PAGE 167

167 domestic programs, such as farm payments made for ecological, not economic reasons, as is the case in Switzerland. From a Third world negotiators perspective, the finding that a potential increase in agricultural production and exports might have a positive impact on inequality further showcases the importance of achieving a successful agreement. Negotiators, with th eir smaller technical capacities, are certainly likely al ready at their limits and undert aking all steps possible. It however could substantially add another meaningf ul component to their bargaining power, as they could publicly increase awareness that a su ccessful outcome is not only likely to increase growth, reduce poverty, but also holding the potential to redu ce inequality. Given their governments increasing concerns at improving the livelihoods of a majority of people, a successful outcome consequently had the potenti al of increasing domestic support at home, as well as improve their standing in the face of international donor s and organizations by achieving greater success at reducing poverty. Even beyond th e connection with agriculture, any achieved reduction in inequality is likely to translate growth more efficien tly into poverty reduction results (Fosu 2005).36 Pro-Poor Development Strategies focu sed on Agricultural Diversification Given this described potential positive c ontribution towards the policy environment surrounding the Doha Round, it is important to place the significance of inequality and agriculture within the broader spectrum of development policy. First, lower inequality, by no means, is currently (nor either likely to be) the maxim of development, as reduced poverty and increased growth continue to dominate the polic y arena. Secondly, agricultural development per 36 This process of course w ill vastly differ, based upon the current co mmodity chains in place. As discussed in Chapter 4, Ethiopian coffee farmers, for example, are receiv ing a significantly smaller share of the world price than their Eastern African counterparts.

PAGE 168

168 se is not a panacea for development. As show n in the empirical study, an increase in the agricultural sectors sh are of the economy is related to a lower GDP ppp per capita, indicating the need for countries to divers ify outside of the sector in or der to achieve higher growth. Given these two important limitations, based upon our findings an important avenue for agricultural development appears to exist through the strengtheni ng of pro-poor growth policies that emphasize agricultural diversification ca tion (Dorward et al. 2004) Since a decreased dependency on a few crops has been found to hol d a positive relationship with GDP ppp per capita, and an overall increase in agricultural production holds a positiv e one with inequality, agricultural development polic ies focused on diversifying ex port production and promoting diversification through support and education to the local farmer s could ideally have both an inequality reducing and gr owth increasing impact. Any potential increase of development efforts in the agricultural sect or however should be undertaken in conjunction with a clear policy aimed at addressing the disparaging problems surrounding land inequality. While the land variab le has not been found to be statistically significant in our model, there is a clear need to address the issue in se veral of our case studies, including Kenya and South Africa, in order to achieve progress on land redistribution and equality. A lack of land equality might seri ously undermine any potential positive impact of both the WTO agreement or diversification effort s, as farmers are limited in their small land holdings to respond to the incentives created by higher world prices and lower export barriers, expand production and consequently reap the benefits of those policy achievements or changes.

PAGE 169

169 CHAPTER 7 CONCLUSION AND FURTHER AVENUES This study on inequality and growth stemmed from an intricate and growing curiosity to investigate the multiple roles played by the agri cultural sector in sub-Saharan Africa. As referenced in the introduction, a k een aspect of the rese arch is to highlight the way policies and policy makers are shaping this sector across all scales, from the global to the local. Consequently, while the thesis proximate goal was to evaluate the links between agriculture, inequality and growth, its ultimate aim is to tr anslate the findings into policy implications. Having established, both quantitatively and qual itatively that the agricultural production and exports in our case studies explai n a significant variati on in their patterns of income inequality, we have moved beyond the empirical findings by highlighting possible policy implications. In summary, the study not only builds upon previ ous studies on income inequality in subSaharan Africa, but takes a step further successfully arguing that a great degree of variation in income inequality in sub-Saharan Africa can be explained through differences in the agricultural sectors. Especially astonishing are the overall robustness of these agricultural variables across numerous regression models in lie u of other variables such as in fant mortality and urbanization, as well as their positive contribution towards redu cing pressures of greater income inequality at the national scale. It consequently highlights once again the far reaching importance of this primary sector beyond economic growth on th e path towards development into the 21st century. The lack of in-field work and primary data sources clearly looms as one of the brightest avenues for future research in th is case. Ideally, fieldwork could be undertaken that investigates the ten countries agricultural policies at greater depth at the national level. In addition, as highlighted by most of our case studies, scale pl ays a key role in the measuring inequality, as vast differences tend to exist be tween various regions, most prom inently rural and urban areas.

PAGE 170

170 Working within the limitations of our national-level dataset, on e clearly can only hint at the important role they play. Consequently, any fi eldwork on policy should ideally be supplemented with local, household-level case studies across multiple years in order to get an understanding of not only the decision-making and adap tation strategies at that scale, but also of the importance or prevalence of inequality (perceived and actual). It is within this context that one might not only acquire better data37 and a sense of the importance of scale, but also most importantly be able to support and articulate more tell ing empirical contributions. Addressing these limitations would ultimately allow for the achieving of the goal of contributing towards the shaping of the policie s. While such ambitious work would be extensive, laborious and certainly time-cons uming, it appears both necessary and rewarding given the potential implementation of a drastically liberalized trad e regime at the international scale in the aftermath of a su ccessful and meaningful end of the WTO Doha Trade Round in the coming years, and the need for a more coherent understanding and managing of its impacts from an African perspective. 37 One key variable that is of great interest for future i nvestigation is terms of trade, which could provide fertile grounds for a more poignant evaluation of the impacts of trade forces on inequality and growth.

PAGE 171

171 APPENDIX A NORMALITY Table A-1. Model 1:Test of normality Shapiro-Wilk Statistic df Sig. GINI .95638.138 PREWTO .63338.000 URBANIZA .95738.151 AGVALUE .96238.216 LOGGDPPP .85738.000 INFANTMO .97138.413 TOTAGEXP .88738.001 LNAGXGDP .86438.000 This is a lower bound of the true significance. a Lilliefors Significance Correction Table A-2. Model 2:Test of normality Shapiro-Wilk Statistic df Sig. GINI .95638.138 PREWTO .63338.000 URBANIZA .95738.151 AGVALUE .96238.216 TRADFOOD .92538.014 LOGGDPPP .85738.000 INFANTMO .97138.413 SQRTRARA .89238.002 This is a lower bound of the true significance. a Lilliefors Significance Correction Table A-3. Model 3:Test of normality Shapiro-Wilk Statistic df Sig. GINI .96650.156 PREWTO .63450.000 ARABLAND .83350.000 VALAGRI3 .94050.014 URBANIZA .96050.091 AGVALUE .96050.087 TRADFOOD .90350.001 TRADRAW .58450.000 GDPPPPCA .43650.000 POPDENSI .89750.000 INFANTMO .97250.287 This is a lower bound of the true significance. a Lilliefors Significance Correction

PAGE 172

172 Table A-4. Arable land test of normality Shapiro-Wilk Statistic df Sig. LOGARABL .93750.010 ARABLAND .83350.000 SQARABL .89150.000 a Lilliefors Significance Correction Table A-5. Population density test of normality Shapiro-Wilk Statistic df Sig. SQPOPDEN .93750.010 LNPOPDEN .94950.031 POPDENSI .89750.000 This is a lower bound of the true significance. a Lilliefors Significance Correction Table A-6. Value of Top 3 Exports Tests of Normality Shapiro-Wilk Statistic df Sig. VALAGRI3 .94050.014 LOGAGRI .87350.000 SQRAGRI .91250.001 a Lilliefors Significance Correction

PAGE 173

173 27 23 N =Before WTO1 0 70 60 50 40 30 20 11 13 2 Figure A-1. Before WTO

PAGE 174

174 GDPPPPCAHistogram20 10 0 Std. Dev = 1489.62 Mean = 1314.8 N = 50.00 Figure A-2. GDP per capita transformations Table A-7. GDP ppp per capita test of normality Shapiro-Wilk Statistic df Sig. SQGDPPPP .64250.000 GDPPPPCA .43650.000 LOGGDPPP .85650.000 a Lilliefors Significance Correction

PAGE 175

175 LOGGDPPP9.00 8.75 8.50 8.25 8.00 7.75 7.50 7.25 7.00 6.75 6.50 6.25 6.00Histogram14 12 10 8 6 4 2 0 Std. Dev = .59 Mean = 6.94 N = 50.00 Figure A-3. Histogram of log GDP ppp per capita transformation TRADRAW70.0 65.0 60.0 55.0 50.0 45.0 40.0 35.0 30.0 25.0 20.0 15.0 10.0 5.0 0.0Histogram30 20 10 0 Std. Dev = 9.76 Mean = 10.5 N = 50.00 Figure A-4. Histogram of raw trade agricultural goods

PAGE 176

176 SQTRADRA8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0Histogram30 20 10 0 Std. Dev = 1.25 Mean = 3.0 N = 50.00 Figure A-5. Histogram of ra w agriculture transformation Table A-8. Trade agricultural raw materials tests of normaliy Shapiro-Wilk Statistic df Sig. TRADRAW .58450.000 SQTRADRA .85050.000 a Lilliefors Significance Correction

PAGE 177

177 TRADFOOD85.0 80.0 75.0 70.0 65.0 60.0 55.0 50.0 45.0 40.0 35.0 30.0 25.0 20.0 15.0 10.0 5.0 0.0Histogram30 20 10 0 Std. Dev = 20.87 Mean = 43.9 N = 50.00 Figure A-6. Histogram trade food exports Table A-9. Post-Transforma tion Tests of Normality Shapiro-Wilk Statistic df Sig. GINI .96650.156 PREWTO .63450.000 LOGARABL .93750.010 VALAGRI3 .94050.014 URBANIZA .96050.091 AGVALUE .96050.087 TRADFOOD .90350.001 SQTRADRA .85050.000 LOGGDPPP .85650.000 LNPOPDEN .94950.031 INFANTMO .97250.287 This is a lower bound of the true significance. a Lilliefors Significance Correction

PAGE 178

178 APPENDIX B OUTLIERS Table B-1. Model 1 residual statistics Minimum Maximum Mean Std. Deviation N Predicted Value 38.764663.121349.98185.59221 38 Std. Predicted Value -2.0062.350.0001.000 38 Standard Error of Predicted Value 1.607814.701892.78931.62445 38 Adjusted Predicted Value 40.240765.497749.95936.05972 38 Residual -11.513614.8874.00005.60597 38 Std. Residual -1.8492.391.000.900 38 Stud. Residual -2.0092.563.001.999 38 Deleted Residual -13.586117.1021.02256.96381 38 Stud. Deleted Residual -2.1232.851.0101.036 38 Mahal. Distance 1.49420.1306.8163.612 38 Cook's Distance .000.164.031.041 38 Centered Leverage Value .040.544.184.098 38 a Dependent Variable: GINI

PAGE 179

179 Table B-2. Model 1 outliers Country Year GINI Mahal D Cook's D Leverage Burkina Faso 1994 58.904.07924.00168 .11025 Burkina Faso 1998 62.503.81768.12215 .10318 Cameroon 1996 50.806.84780.07044 .18508 Cameroon 2001 44.209.49318.00036 .25657 Cote d`Ivoire 1985 50.609.51945.00640 .25728 Cote d`Ivoire 1995 43.905.65703.00096 .15289 Cote d`Ivoire 1998 44.405.78623.00297 .15638 Ethiopia 1995 32.707.14803.04274 .19319 Ethiopia 2000 29.704.67063.09081 .12623 Gambia 1993 60.862.32193.00802 .06275 Gambia 1994 59.452.06468.00515 .05580 Gambia 1998 50.208.24548.04833 .22285 Ghana 1992 39.707.77637.03850 .21017 Ghana 1998 50.706.04525.04832 .16339 Guinea 1991 50.7010.34476.06216 .27959 Guinea 1994 55.106.67054.00373 .18028 Kenya 1992 59.906.18728.01969 .16722 Kenya 1994 44.304.34378.03943 .11740 Kenya 1997 44.503.79572.00503 .10259 Madagascar 1993 48.507.41296.00677 .20035 Madagascar 1997 44.002.34465.00722 .06337 Madagascar 2001 47.404.31458.01030 .11661 Malawi 1993 62.006.07669.10578 .16423 Malawi 1997 49.306.16731.00256 .16668 Mozambique 1997 40.007.76913.00004 .20998 Mozambique 2003 42.001.49400.01382 .04038 Nigeria 1985 53.806.35705.00056 .17181 Nigeria 1996 48.306.69889.01145 .18105 Nigeria 1997 50.2020.13034.16439 .54406 South Africa 1993 59.5013.68649.04597 .36991 South Africa 1997 60.1013.61450.00490 .36796 Tanzania 1991 58.903.74551.05208 .10123 Tanzania 1993 47.703.79126.00056 .10247 Uganda 1989 44.3011.23542.11089 .30366 Uganda 2000 46.907.86212.00371 .21249 Zambia 1993 51.105.79890.00397 .15673 Zambia 1996 54.808.15003.00176 .22027 Zambia 1998 57.407.53513.00829 .20365

PAGE 180

180 Table B-3. Model 2 residual statistics Minimum Maximum Mean Std. Deviation N Predicted Value 36.616563.656049.98186.1728138 Std. Predicted Value -2.1652.215.0001.00038 Standard Error of Predicted Value 1.426104.717582.45566.6047638 Adjusted Predicted Value 37.826166.158149.91876.3134438 Residual -12.034211.9001.00004.9594538 Std. Residual -2.1852.161.000.90038 Stud. Residual -2.3432.351.004.99338 Deleted Residual -13.843814.0848.06316.0730538 Stud. Deleted Residual -2.5492.559.0081.03138 Mahal. Distance 1.50726.1716.8164.28438 Cook's Distance .000.127.028.03438 Centered Leverage Value .041.707.184.11638 a Dependent Variable: GINI

PAGE 181

181 Table B-4. Model 2 outliers Country Year GINI Mahal D Cook's D Leverage Burkina Faso 1994 58.903.80324.00147 .10279 Burkina Faso 1998 62.5026.17143.04349 .70734 Cameroon 1996 50.807.69390.05344 .20794 Cameroon 2001 44.209.24690.00127 .24992 Cote d`Ivoire 1985 50.609.11601.01161 .24638 Cote d`Ivoire 1995 43.905.48210.00143 .14816 Cote d`Ivoire 1998 44.404.79566.00008 .12961 Ethiopia 1995 32.707.75715.02555 .20965 Ethiopia 2000 29.703.86274.10323 .10440 Gambia 1993 60.862.33784.00754 .06318 Gambia 1994 59.452.06511.00537 .05581 Gambia 1998 50.204.06209.02937 .10979 Ghana 1992 39.707.68583.05686 .20773 Ghana 1998 50.706.23507.10648 .16852 Guinea 1991 50.709.98365.04915 .26983 Guinea 1994 55.106.58664.00374 .17802 Kenya 1992 59.906.30099.01741 .17030 Kenya 1994 44.304.39199.04978 .11870 Kenya 1997 44.503.95964.00263 .10702 Madagascar 1993 48.505.12738.00635 .13858 Madagascar 1997 44.002.41769.00387 .06534 Madagascar 2001 47.403.65847.00094 .09888 Malawi 1993 62.004.76563.12680 .12880 Malawi 1997 49.308.20754.08383 .22183 Mozambique 1997 40.006.09805.00025 .16481 Mozambique 2003 42.001.50688.01938 .04073 Nigeria 1985 53.807.27245.00048 .19655 Nigeria 1996 48.306.01937.00451 .16269 Nigeria 1997 50.209.24111.01250 .24976 South Africa 1993 59.5012.93092.06865 .34948 South Africa 1997 60.1013.96243.04545 .37736 Tanzania 1991 58.903.90183.05724 .10545 Tanzania 1993 47.703.95451.00289 .10688 Uganda 1989 44.309.84909.04327 .26619 Uganda 2000 46.907.98479.00006 .21581 Zambia 1993 51.105.55977.00823 .15026 Zambia 1996 54.807.79237.00482 .21060 Zambia 1998 57.407.21171.00190 .19491

PAGE 182

182 Table B-5. Model 3 residuals statistics Minimum Maximum Mean Std. Deviation N Predicted Value 37.806061.391649.59285.16343 50 Std. Predicted Value -2.2832.285.0001.000 50 Standard Error of Predicted Value 1.376684.539902.21080.57068 50 Adjusted Predicted Value 35.710862.371649.38005.19473 50 Residual -12.696312.4716.00005.28143 50 Std. Residual -2.2262.186.000.926 50 Stud. Residual -2.3492.342.0151.005 50 Deleted Residual -14.137514.3111.21286.32884 50 Stud. Deleted Residual -2.4902.482.0171.029 50 Mahal. Distance 1.87430.0546.8604.658 50 Cook's Distance .000.416.027.062 50 Centered Leverage Value .038.613.140.095 50 a Dependent Variable: GINI

PAGE 183

183 Table B-6. Model 3 outliers Country Year Gini Mahal D Cook's D Leverage Burkina Faso 1998 62.530.054170.41623 0.61335 Burkina Faso 1994 58.94.303760.00625 0.08783 Cameroon 1996 50.88.942710.03521 0.1825 Cameroon 2001 44.210.649110.0038 0.21733 Cote d`Ivoire 1985 50.66.35530.00491 0.1297 Cote d`Ivoire 1986 49.92.335010.00042 0.04765 Cote d`Ivoire 1987 48.82.334180.00124 0.04764 Cote d`Ivoire 1988 45.92.472940.00479 0.05047 Cote d`Ivoire 1995 43.95.10760.0026 0.10424 Cote d`Ivoire 1998 44.45.027320.00085 0.1026 Ethiopia 1997 48.27.240360.10051 0.14776 Ethiopia 1995 32.76.451350.02639 0.13166 Ethiopia 2000 29.74.014990.07826 0.08194 Gambia 1998 50.25.08110.01863 0.1037 Gambia 1992 48.132.620440.01361 0.05348 Gambia 1994 59.452.232040.00873 0.04555 Gambia 1993 60.863.863350.02237 0.07884 Ghana 1998 50.77.087660.06663 0.14465 Ghana 1987 42.96.142810.00197 0.12536 Ghana 1992 39.76.009730.02059 0.12265 Ghana 1989 47.95.596230.0057 0.11421 Guinea 1994 55.17.507080.00049 0.15321 Guinea 1991 50.711.587090.03042 0.23647 Kenya 1992 59.96.351340.03043 0.12962 Kenya 1994 44.34.213430.01841 0.08599 Kenya 1997 44.54.76710.00206 0.09729 Madagascar 1993 48.54.623090.00144 0.09435 Madagascar 1997 442.745520.00353 0.05603 Madagascar 1999 43.14.162320.02848 0.08495 Madagascar 2001 47.44.154160.0008 0.08478 Malawi 1993 625.318150.10112 0.10853 Malawi 1997 49.39.145370.03267 0.18664 Mozambique 2003 421.873720.01603 0.03824 Mozambique 1997 406.754190.0052 0.13784 Nigeria 1980 538.307170.01478 0.16953 Nigeria 1985 53.811.496350.00015 0.23462 Nigeria 1992 54.24.83740.00027 0.09872 Nigeria 1996 48.35.817230.00733 0.11872 Nigeria 1997 50.212.814750.00025 0.26153 South Africa 1993 59.515.743070.01081 0.32129 South Africa 1997 60.116.755820.045 0.34196 Tanzania 1993 47.73.871760.0003 0.07902 Tanzania 1991 58.93.887680.04881 0.07934 Uganda 1989 44.37.318740.00914 0.14936

PAGE 184

184 Table B-6. Continued Country Year Gini Mahal D Cook's D Leverage Uganda 1992 395.829770.07284 0.11897 Uganda 2000 46.97.437760.00004 0.15179 Zambia 1991 59.38.639060.02242 0.17631 Zambia 1993 51.17.140740.00552 0.14573 Zambia 1996 54.88.493630.00224 0.17334 Zambia 1998 57.47.484370.00323 0.15274

PAGE 185

185 APPENDIX C NORMALLY DISTRIBUTED ERROR Regression Standardized ResidualHistogram Dependent Variable: GINI8 6 4 2 0 Std. Dev = .90 Mean = 0.00 N = 38.00 Figure C-1. Histogr am GINI Model 1 Normal Q-Q Plot of Standardized ResidualObserved Value3 2 1 0 -1 -2 2 1 0 -1 -2 Figure C-2. Residuals plot Model 1

PAGE 186

186 Table C-1. Normality test Model 1 Shapiro-Wilk Statistic df Sig. Standardized Residual .979 38 .690 This is a lower bound of the true significance. a Lilliefors Significance Correction Regression Standardized Residual2.00 1.50 1.00 .50 0.00 -.50 -1.00 -1.50 -2.00Histogram Dependent Variable: GINI10 8 6 4 2 0 Std. Dev = .90 Mean = 0.00 N = 38.00 Figure C-3. Histogram standa rdized residuals Model 2

PAGE 187

187 Normal P-P Plot of Regression Stand a Dependent Variable: GINIObserved Cum Prob1.00 .75 .50 .25 0.00 1.00 .75 .50 .25 0.00 Figure C-4. Regression pl ot residuals Model 2 Table C-2. Test of normality Model 2 Shapiro-Wilk Statistic df Sig. Standardized Residual .983 38 .814 This is a lower bound of the true significance. a Lilliefors Significance Correction

PAGE 188

188 Regression Standardized ResidualHistogram Dependent Variable: GINI14 12 10 8 6 4 2 0 Std. Dev = .93 Mean = 0.00 N = 50.00 Figure C-5. Histogram standa rdized residuals Model 3 Normal P-P Plot of Regression Stand a Dependent Variable: GINIObserved Cum Prob1.00 .75 .50 .25 0.00 1.00 .75 .50 .25 0.00 Figure C-6. Regression pl ot residuals Model 3

PAGE 189

189 Table C-3. Test of normality Model 3 Shapiro-Wilk Statistic df Sig. Standardized Residual .984 50 .706 a Lilliefors Significance Correction

PAGE 190

190 APPENDIX D HOMOSCEDASICITY Standardized Residual3 2 1 0 -1 -2 3 2 1 0 -1 -2 -3 Figure D-1. Scatterplot Model 1 Standardized Residual3 2 1 0 -1 -2 -3 3 2 1 0 -1 -2 -3 Figure D-2. Scatterplot Model 3

PAGE 191

191 APPENDIX E NONLINEARITY Table E-1. Model 1 nonlinearity GINI Unstandardized Residual N 3838 Mean 49.9818.0000000 Std. Deviation 7.918325.60596701 Minimum 29.70-11.51358 Maximum 62.5014.88744 Table E-2. Model 2 nonlinearity GINI Unstandardized Residual N 3838 Mean 49.9818.0000000 Std. Deviation 7.918324.95945197 Minimum 29.70-12.03422 Maximum 62.5011.90005 Table E-3. Model 3 nonlinearity N Minimum Maximum Mean Std. Deviation GINI 5029.7062.5049.5928 7.38610 Unstandardized Residual 50-12.6963012.47162.0000000 5.28142726

PAGE 192

192 APPENDIX F MULTICOLLINEARITY Table F-1. Correlations Model 1 GINI PREWTO AGVALUETOTAGEXPLNAGXGDPURBANIZA LOGGDPPPINFANTMO Pearson Correlation GINI 1.000.377-.442-.418-.180.072.235-.198 PREWTO .3771.000.081-.145-.022-.242-.070-.030 AGVALUE -.442.0811.000.361.069-.563-.689.323 TOTAGEXP -.418-.145.3611.000.721-.524-.169.185 LNAGXGDP -.180-.022.069.7211.000-.156.050-.075 URBANIZA .072-.242-.563-.524-.1561.000.644-.359 LOGGDPPP .235-.070-.689-.169.050.6441.000-.570 INFANTMO -.198-.030.323.185-.075-.359-.5701.000 Sig. (1tailed) GINI ..010.003.004.139.334.077.117 PREWTO .010..315.192.448.072.339.429 AGVALUE .003.315..013.341.000.000.024 TOTAGEXP .004.192.013..000.000.155.133 LNAGXGDP .139.448.341.000..175.383.326 URBANIZA .334.072.000.000.175..000.013 LOGGDPPP .077.339.0 00.155.383.000..000 INFANTMO .117.429.0 24.133.326.013.000. N GINI 3838383838383838 PREWTO 3838383838383838 AGVALUE 3838383838383838 TOTAGEXP 3838383838383838 LNAGXGDP 3838383838383838 URBANIZA 3838383838383838 LOGGDPPP 3838383838383838 INFANTMO 3838383838383838 Table F-2. Coefficients Model 1 Collinearity Statistics Model Tolerance VIF 1 (Constant) PREWTO .6971.436 AGVALUE .4152.408 TOTAGEXP .1845.432 LNAGXGDP .3253.074 URBANIZA .2733.667 LOGGDPPP .2484.027 INFANTMO .5901.696 a Dependent Variable: GINI

PAGE 193

193 Table F-3. Correlations Model 2 GINI PREWTO AGVALUETRADFOODSQRTRARALOGGDPPP URBANIZAINFANTMO Pearson Correlation GINI 1.000.377-.442-.489-.125.235 .072-.198 PREWTO .3771.000.081-.076-.225-.070 -.242-.030 AGVALUE -.442.0811.000.294.278-.689 -.563.323 TRADFOOD -.489-.076.2941.000.291-.196 -.491.158 SQRTRARA -.125-.225.278.2911.000.075 -.202.057 LOGGDPPP .235-.070-.689-.196.0751.000 .644-.570 URBANIZA .072-.242-.563-.491-.202.644 1.000-.359 INFANTMO -.198-.030.323.158.057-.570 -.3591.000 Sig. (1tailed) GINI ..010.003.001.228.077 .334.117 PREWTO .010..315.326.087.339 .072.429 AGVALUE .003.315..036.045.000 .000.024 TRADFOOD .001.326.036..038.119 .001.171 SQRTRARA .228.087.045.038..327 .112.367 LOGGDPPP .077.339.0 00.119.327. .000.000 URBANIZA .334.072.000.001.112.000 ..013 INFANTMO .117.429.024.171.367.000 .013. N GINI 383838383838 3838 PREWTO 383838383838 3838 AGVALUE 383838383838 3838 TRADFOOD 383838383838 3838 SQRTRARA 383838383838 3838 LOGGDPPP 383838383838 3838 URBANIZA 383838383838 3838 INFANTMO 383838383838 3838 Table F-4. Coefficients Model 2 Collinearity Statistics Model Tolerance VIF 1 (Constant) PREWTO .7651.306 AGVALUE .3852.599 TRADFOOD .6681.498 SQRTRARA .6131.632 LOGGDPPP .2354.254 URBANIZA .3542.826 INFANTMO .6351.574 a Dependent Variable: GINI

PAGE 194

194 Table F-5. Correlations Model 3 GINI PREWTO URBANIZA AGVALUE TRADFOOD LOGGDPPP INFANTMO SQTRADRA GINI Pearson Correlations 1.304.135-.482-.457.198-.122-.123 Sig.(2-tail) ..032.352.000.001.169.400.395 Sum of Squares 2673.255.3537.1-2100.3-3450.842.0-960.1-55.5 Covariance 54.51.111.0-42.9-70.4.857-19.6-1.1 PREWTO Pearson Correl .3041-.113.027-.107-.037-.047-.119 Sig.(2-tail) .032..435.850.459.801.747.409 Sum of Squares 55.312.4-30.78.1-55.1-.530-25.2-3.7 Covariance 1.129.253-.626.166-1.125-.011-.513-.075 URBANIZA Pearson Correl .135-.1131-.555-.475.619-.330-.208 Sig.(2-tail) .352.435..000.000.000.019.147 Sum of Squares 537.1-30.75964.3-3616.3-5358.8196.6-3883.7-140.3 Covariance 11.0-.626121.7-73.8-109.44.01-79.3-2.9 AGVALUE Pearson Correl -.482.027-.5551.319-.568.214.273 Sig.(2-tail) .000.850.000..024.000.135.055 Sum of Squares 2100.3 8.15-3616.371123928.8-196.82758.1200.9 Covariance -42.86.17-73.8145.180.2-4.0256.34.1 TRADFOOD Pearson Correl -.457-.107-.475.3191-.245.242.250 Sig.(2-tail) .001.459.000.024..087.091.079 Sum of Squares -3450-55.1-5358.83928.821335-147.05386.2319 Covariance -70.4-1.13-109.480.2435.4-3.0109.96.5 LOGGDPPP Pearson Correl .198-.037.619-.568-.2451-.549.036 Sig.(2-tail) .169.801.000.000.087..000.804 Sum of Squares 42.0-.53196.6-196.8-147.016.91-344.81.3 Covariance .857-.0114.013-4.016-3.0.345-7.036.026 INFANTMO Pearson Correl -.122-.047-.330.214.242-.5491.094 Sig.(2-tail) .400.747.019.135.091.000..515 Sum of Squares -960-25.2-3883.72758.15386.2-344.823288126 Covariance -19.6-.51-79.2656.3109.9-7.0475.32.6 SQTRADRA Pearson Correl -.123-.119-.208.273.250.036.0941 Sig.(2-tail) .395.409.147.055.079.804.515. Sum of Squares -55.5-3.7-140.3200.9319.41.291125.776.3 Covariance -1.134-.075-2.8634.1016.518.0262.5661.556 Correlation is significant at the 0.05 level (2-tailed). ** Correlation is significant at the 0.01 level (2-tailed).

PAGE 195

195 Table F-6. Correlations Model 3 Collinearity Statistics Model Tolerance VIF 1 (Constant) PREWTO .9321.072 AGVALUE .5321.878 TRADFOOD .7111.406 SQTRADRA .7791.283 URBANIZA .4452.247 LOGGDPPP .3572.799 INFANTMO .6411.560 a Dependent Variable: GINI Table F-7. Alternate st epwise regression model Model Variables Entered Variables Removed Method 1 AGVALUE Stepwise (Criteria: Probability-of-F-toenter <= .050, Probability-of-F-toremove >= .100). 2 TRADFOOD Stepwise (Criteria: Probability-of-F-toenter <= .050, Probability-of-F-toremove >= .100). 3 URBANIZA Stepwise (Criteria: Probability-of-F-toenter <= .050, Probability-of-F-toremove >= .100). 4 PREWTO Stepwise (Criteria: Probability-of-F-toenter <= .050, Probability-of-F-toremove >= .100). a Dependent Variable: GINI

PAGE 196

196 Table F-8. Alternate model summary Model R R Square Adjusted R Square Std. Error of the Estimate 1.482.232.2166.53977 2.578.334.3066.15253 3.653.426.3895.77321 4.692.479.4335.56210 a Predictors: (Constant), AGVALUE b Predictors: (Constant), AGVALUE, TRADFOOD c Predictors: (Constant), AGVALUE, TRADFO OD, URBANIZA d Predictors: (Constant), AGVAL UE, TRADFOOD, URBANIZA, PREWTO e Dependent Variable: GINI Table F-9. Alternate model ANOVA Model Sum of Squares df Mean SquareF Sig. 1Regression 620.2741620.27414.503.000 Residual 2052.8954842.769 Total 2673.16849 2Regression 894.0462447.02311.809.000 Residual 1779.1224737.854 Total 2673.16849 3Regression 1139.9893379.99611.401.000 Residual 1533.1804633.330 Total 2673.16849 4Regression 1281.0054320.25110.352.000 Residual 1392.1634530.937 Total 2673.16849 a Predictors: (Constant), AGVALUE b Predictors: (Constant), AGVALUE, TRADFOOD c Predictors: (Constant), AGVALUE, TRADFO OD, URBANIZA d Predictors: (Constant), AGVAL UE, TRADFOOD, URBANIZA, PREWTO e Dependent Variable: GINI

PAGE 197

197 Table F-10. Alternate model coefficients Unstandardized Coefficients Standardized Coefficients t Sig. Collinearity Statistics Model B Std. Error Beta Tolerance VIF 1 (Constant) 59.3252.71821.829.000 AGVALUE -.295.078-.482-3.808.000 1.0001.000 2 (Constant) 62.3942.80022.284.000 AGVALUE -.229.077-.374-2.979.005 .8981.113 TRADFOOD -.120.044-.338-2.689.010 .8981.113 3 (Constant) 75.6975.55713.621.000 AGVALUE -.338.083-.551-4.093.000 .6881.454 TRADFOOD -.166.045-.468-3.680.001 .7701.299 URBANIZA -.264.097-.394-2.716.009 .5931.687 4 (Constant) 71.8665.64712.727.000 AGVALUE -.333.080-.543-4.184.000 .6871.455 TRADFOOD -.149.044-.420-3.370.002 .7451.343 URBANIZA -.228.095-.340-2.395.021 .5741.742 PREWTO 3.4511.617.2352.135.038 .9531.049 a Dependent Variable: GINI

PAGE 198

198 Table F-11. Alternate M odel excluded variables Beta In t Sig. Partial Correlation Collinearity Statistics Model Tolerance VIF Minimum Tolerance 1 PREWTO .3172.660.011.362.999 1.001.999 TRADFOOD -.338-2.689.010-.365.898 1.113.898 SQTRADRA .009.068.946.010.926 1.080.926 URBANIZA -.192-1.272.210-.182.692 1.446.692 LOGGDPPP -.112-.725.472-.105.678 1.475.678 INFANTMO-.019-.148.883-.022.954 1.048.954 2 PREWTO .2822.478.017.343.984 1.016.885 SQTRADRA .071.560.578.082.896 1.116.859 URBANIZA -.394-2.716.009-.372.593 1.687.593 LOGGDPPP -.145-.997.324-.145.673 1.485.643 INFANTMO.044.348.730.051.921 1.086.867 3 PREWTO .2352.135.038.303.953 1.049.574 SQTRADRA .070.589.559.087.896 1.116.593 LOGGDPPP .026.170.866.025.538 1.857.474 INFANTMO-.023-.191.849-.0281.135.567 4 SQTRADRA .099.865.392.129 .881 .884 1.131.574 LOGGDPPP .011.071.943.011.537 1.862.458 INFANTMO-.006-.048.962-.007.877 1.141.547 a Predictors in the Model: (Constant), AGVALUE b Predictors in the Model: (Constant), AGVALUE, TRADFOOD c Predictors in the Model: (Consta nt), AGVALUE, TRADFOOD, URBANIZA d Predictors in the Model: (Constant), AGVALUE, TRADFOOD, URBANIZA, PREWTO e Dependent Variable: GINI

PAGE 199

199 Table F-12. Collinearity diagnostics EigenvalueCondition IndexVariance Proportions ModelDimension (Constant) AGVALUETRADFOOD URBANIZAPREWTO 1 1 1.9401.000.03.03 2 5.969E-025.702.97.97 2 1 2.8281.000.01.01.02 2 .1135.014.10.20.97 3 5.930E-026.906.89.79.01 3 1 3.6371.000.00.01.01 .00 2 .2473.834.00.03.17 .17 3 9.995E-026.033.00.43.59 .02 4 1.528E-0215.4291.00.53.22 .81 4 1 4.2171.000.00.00.01 .00.02 2 .4273.143.00.01.03 .00.85 3 .2464.144.00.03.15 .17.02 4 9.678E-026.601.00.47.56 .01.04 5 1.409E-0217.2991.00.49.25 .81.08 a Dependent Variable: GINI Table F-13. Collinearity diagnostics Model 1 Eigenvalue Condition Index Variance Proportions Dimension (Constant) PREWTO AGVALUE TRADFOOD SQTRADRA URBANIZA LOGGDPPP INFANTMO 1 7.0101.000 .00.01.00.00.00.00.00.00 2 .4573.918 .00.83.00.01.01.00.00.00 3 .2625.170 .00.02.03.12.01.11.00.00 4 .1157.791 .00.00.02.38.58.01.00.00 5 9.166E-028.745 .00.06.34.26.27.01.00.02 6 4.593E-0212.353 .00.00.24.10.00.12.00.42 7 1.726E-0220.155 .03.08.09.11.04.70.06.15 8 1.069E-0380.971 .97.00.28.00.09.06.94.40 a Dependent Variable: GINI Table F-14. Residuals statistics MinimumMaximumMean Std. Deviation N Predicted Value 38.434661.855149.59285.1130250 Std. Predicted Value -2.1822.398.0001.00050 Standard Error of Predicted Value 1.214972.563291.72193.3623450 Adjusted Predicted Value 36.848862.362549.54255.1615150 Residual -12.398012.3444.00005.3302450 Std. Residual -2.2292.219.000.95850 Stud. Residual -2.3452.335.0041.01150 Deleted Residual -13.719813.6601.05035.9362450 Stud. Deleted Residual -2.4752.463.0071.03350 Mahal. Distance 1.3589.4273.9202.05650 Cook's Distance .000.142.023.03550 Centered Leverage Value .028.192.080.04250 a Dependent Variable: GINI

PAGE 200

200 APPENDIX G INDEPENDENCE Table G-1. Model 1 summary R R Square Adjusted R Square Std. Error of the Estimate Durbin-Watson Model 1 .706.499.3826.22574 1.807 a Predictors: (Constant), INFANTMO, PREWTO, LNAGXGDP, AGVALUE, URBANIZA, LOGGDPPP, TOTAGEXP b Dependent Variable: GINI Table G-2. Model 2 summary Model R R Square Adjusted R Square Std. Error of the Estimate Durbin-Watson 1 .780.608.5165.50775 2.062 a Predictors: (Constant), INFANTMO, PREWTO, TRADFOOD, SQRTRARA, AGVALUE, URBANIZA, LOGGDPPP b Dependent Variable: GINI Table G-3. Model 3 summary Model R R Square Adjusted R Square Std. Error of the Estimate Durbin-Watson 1 .699.489.4035.70459 2.131 a Predictors: (Constant), INFANTMO, PREWTO, SQTRADRA, AGVALUE, TRADFOOD, URBANIZA, LOGGDPPP b Dependent Variable: GINI

PAGE 201

201 APPENDIX H COMPLETE DATASET Table H-1. Complete dataset COUNTRY YEAR GINI INCDEFN SOURCE1 Mozambique 2003 42.00 Consum ption Journal Article Gambia 1994 59.45 Consumption Dein inger & Squire, World Bank 2004 Cote d`Ivoire 1987 48.80 Consumption Deininger & Squire, World Bank 2004 Cote d`Ivoire 1986 49.90 Consumption Deininger & Squire, World Bank 2004 Cote d`Ivoire 1988 45.90 Consumption Deininger & Squire, World Bank 2004 Gambia 1992 48.13 Consumption Dein inger & Squire, World Bank 2004 Madagascar 1997 44.00 Consumption De ininger & Squire, World Bank 2004 Gambia 1993 60.86 Consumption Dein inger & Squire, World Bank 2004 Tanzania 1993 47.70 Consumption Dein inger & Squire, World Bank 2004 Tanzania 1991 58.90 Consumption World Bank Poverty Monitoring Database 2002 Ethiopia 2000 29.70 Consumption World Bank, World Development Indicators 2004 Madagascar 2001 47.40 Consumption World Ba nk, World Development Indicators 2004 Madagascar 1999 43.10 Consumption De ininger & Squire, World Bank 2004 Kenya 1994 44.30 Consumption World Ba nk Poverty Monitoring Database 2002 Burkina Faso 1994 58.90 Consumption De ininger & Squire, World Bank 2004 Madagascar 1993 48.50 Consumption De ininger & Squire, World Bank 2004 Kenya 1997 44.50 Consumption World Bank, World Development Indicators 2004 Nigeria 1992 54.20 Consumption Dein inger & Squire, World Bank 2004 Cote d`Ivoire 1998 44.40 Consumption World Bank, World Development Indicators 2004 Gambia 1998 50.20 Consumption World Bank, World Development Indicators 2004 Cote d`Ivoire 1995 43.90 Consumption Deininger & Squire, World Bank 2004 Malawi 1993 62.00 Consumption Worl d Bank, Africa Department Ghana 1989 47.90 Consumption Deininger & Squire, World Bank 2004 Nigeria 1996 48.30 Consumption Dein inger & Squire, World Bank 2004 Uganda 1992 39.00 Consumption World Ba nk Poverty Monitoring Database 2002 Ghana 1992 39.70 Consumption Deininger & Squire, World Bank 2004 Ghana 1987 42.90 Consumption Deininger & Squire, World Bank 2004 Kenya 1992 59.90 Consumption Deini nger & Squire, World Bank 2004 Cote d`Ivoire 1985 50.60 Consumption Deininger & Squire, World Bank 2004 Ethiopia 1995 32.70 Consumption Dein inger & Squire, World Bank 2004 Mozambique 1997 40.00 Consumption World Bank Poverty Monitoring Database 2002 Ghana 1998 50.70 Consumption Deininger & Squire, World Bank 2004 Zambia 1993 51.10 Consumption Dein inger & Squire, World Bank 2004 Ethiopia 1997 48.20 Consumption Dein inger & Squire, World Bank 2004 Uganda 1989 44.30 Consumption World Ba nk Poverty Monitoring Database 2002 Uganda 2000 46.90 Expenditure Dein inger & Squire, World Bank 2004 Zambia 1998 57.40 Consumption Dein inger & Squire, World Bank 2004 Guinea 1994 55.10 Consumption Deininger & Squire, World Bank 2004 Nigeria 1980 53.00 Consumption Dein inger & Squire, World Bank 2004 Zambia 1996 54.80 Consumption Dein inger & Squire, World Bank 2004 Zambia 1991 59.30 Consumption Dein inger & Squire, World Bank 2004 Cameroon 1996 50.80 Consumption / Expenditure Deininger & Squire, World Bank 2004 Malawi 1997 49.30 Consumption World Bank, World Development Indicators 2004 Cameroon 2001 44.20 Consumption / Expenditure Wo rld Bank, World Development Indicators 2004 Nigeria 1985 53.80 Consumption Dein inger & Squire, World Bank 2004 Guinea 1991 50.70 Consumption Deininger & Squire, World Bank 2004 Nigeria 1997 50.20 Consumption World Ba nk Poverty Monitoring Database 2002 South Africa 1993 59.50 Consumption World Bank Poverty Monitoring Database 2002 South Africa 1997 60.10 Income, Gross De ininger & Squire, World Bank 2004 Burkina Faso 1998 62.50 Consumption De ininger & Squire, World Bank 2004

PAGE 202

202 Table H-1. Continued COUNTRY YEAR QU AL PREWTO ARABLAND VALAGRI3 URBAN AGVALUE TRADFOOD TRAD -RAW GDPPPPCA Mozambique 2003 n/a 0 .23 73. 88 32.98 24.38 43.79 10.61 1128.00 Gambia 1994 3 1 .16 76.63 24.90 27.50 43.79 8.80 1463.00 Cote d`Ivoire 1987 2 1 .21 49.43 37.50 29.18 43.79 10.61 1338.00 Cote d`Ivoire 1986 2 1 .22 51.26 37.50 28.46 43.79 10.61 1339.00 Cote d`Ivoire 1988 2 1 .21 52.94 37.50 32.04 43.79 10.61 1355.00 Gambia 1992 3 1 .16 73.01 24.90 26.17 43.79 10.61 1456.00 Madagascar 1997 3 0 .19 62. 14 25.50 31.55 54.40 6.20 761.00 Gambia 1993 3 1 .15 84.14 24.90 25.24 43.79 3.00 1478.00 Tanzania 1993 3 1 .13 71. 09 20.00 48.11 43.79 10.61 440.00 Tanzania 1991 3 1 .13 68. 52 19.00 48.14 43.79 10.61 442.00 Ethiopia 2000 3 0 .16 86.06 14.90 47.70 70.60 18.70 636.00 Madagascar 2001 3 0 .17 84. 59 26.00 28.60 48.70 3.70 866.00 Madagascar 1999 3 0 .18 55. 53 25.50 30.03 35.90 6.10 788.00 Kenya 1994 3 1 .16 70.48 24.70 33.32 56.50 7.90 956.00 Burkina Faso 1994 3 1 .35 86.49 13.60 31.30 43.79 10.61 829.00 Madagascar 1993 1 1 .21 63. 55 23.60 28.71 66.30 3.80 749.00 Kenya 1997 3 0 .15 74.54 30.00 30.91 55.60 6.80 983.00 Nigeria 1992 3 1 .31 82.20 35.00 23.80 43.79 10.61 765.00 Cote d`Ivoire 1998 3 0 .19 57.63 41.70 24.12 59.40 11.20 1606.00 Gambia 1998 3 0 .18 82.45 26.40 28.42 76.80 9.70 1582.00 Cote d`Ivoire 1995 3 0 .20 85.96 41.70 24.73 59.10 16.10 1440.00 Malawi 1993 3 1 .19 90.36 12.62 48.90 43.79 10.61 475.00 Ghana 1989 3 1 .17 96.26 32.90 48.97 43.79 10.61 1238.00 Nigeria 1996 3 0 .27 76.18 39.50 30.70 43.79 1.60 793.00 Uganda 1992 2 1 .27 82.85 11.20 51.12 43.79 10.61 787.00 Ghana 1992 3 1 .17 92.97 36.50 44.78 50.90 14.70 1406.00 Ghana 1987 3 1 .18 97.04 32.90 50.60 43.79 10.61 1104.00 Kenya 1992 3 1 .17 71.20 24.70 28.74 43.50 5.50 936.00 Cote d`Ivoire 1985 2 1 .23 49.52 37.50 26.54 68.00 11.90 1354.00 Ethiopia 1995 2 0 .17 83.68 13.90 57.65 72.50 13.40 523.00 Mozambique 1997 3 0 .23 84. 30 28.00 37.18 71.70 13.80 718.00 Ghana 1998 3 0 .20 91.89 40.20 36.01 58.40 10.60 1746.00 Zambia 1993 3 1 .58 70.82 39.40 34.10 3.90 .90 815.00 Ethiopia 1997 3 0 .17 89.41 13.90 55.86 78.50 10.80 602.00 Uganda 1989 3 1 .29 96.54 9.90 56.79 43.79 10.61 671.00 Uganda 2000 2 0 .21 77.58 12.00 37.34 67.30 14.10 1244.00 Zambia 1998 3 0 .51 72.76 37.30 21.14 7.10 6.40 749.00 Guinea 1994 2 1 .11 76.37 25.30 21.71 43.79 10.61 1583.00 Nigeria 1980 3 1 .41 75.13 26.90 20.63 43.79 10.61 534.00 Zambia 1996 3 0 .54 80.85 37.30 17.57 6.40 1.90 751.00 Zambia 1991 3 1 .61 45.99 39.40 17.43 43.79 10.61 814.00 Cameroon 1996 1 0 .44 63.65 44.70 41.33 24.30 25.20 1601.00 Malawi 1997 3 0 .18 87.72 14.02 32.59 84.50 3.60 556.00 Cameroon 2001 3 0 .39 69.69 49.00 39.86 16.80 21.30 2022.00 Nigeria 1985 3 1 .36 90.20 30.70 37.31 2.20 .00 508.00 Guinea 1991 3 1 .11 71.13 25.30 24.15 43.79 10.61 1545.00 Nigeria 1997 3 0 .26 74.48 39.50 33.63 1.70 .10 806.00 South Africa 1993 3 1 .38 35. 05 48.80 4.17 6.50 2.80 7767.00 South Africa 1997 3 0 .36 34. 87 52.60 4.01 10.80 3.90 8752.00 Burkina Faso 1998 3 0 .36 89.37 15.20 34.48 16.40 68.10 940.00

PAGE 203

203 Table H-1. Continued COUNTRY YEAR LOGGDPPP POPDENSI INFANTMO LOGARABL SQTRADRA LNPOPDEN Mozambique 2003 7.03 24.30 104.00 -1.47 3.26 3.19 Gambia 1994 7.29 107.80 96.00 -1.83 2.97 4.68 Cote d`Ivoire 1987 7.20 35.80 103.00 -1.56 3.26 3.58 Cote d`Ivoire 1986 7.20 34.40 103.00 -1.51 3.26 3.54 Cote d`Ivoire 1988 7.21 37.10 103.00 -1.56 3.26 3.61 Gambia 1992 7.28 100.50 103.00 -1.83 3.26 4.61 Madagascar 1997 6.63 25. 50 95.00 -1.66 2.49 3.24 Gambia 1993 7.30 104.10 96.00 -1.90 1.73 4.65 Tanzania 1993 6.09 32.90 103.00 -2.04 3.26 3.49 Tanzania 1991 6.09 30.70 102.00 -2.04 3.26 3.42 Ethiopia 2000 6.46 68.50 116.00 -1.83 4.32 4.23 Madagascar 2001 6.76 28. 70 84.00 -1.77 1.92 3.36 Madagascar 1999 6.67 27. 00 84.00 -1.71 2.47 3.30 Kenya 1994 6.86 46.50 73.00 -1.83 2.81 3.84 Burkina Faso 1994 6.72 35.00 110.00 -1.05 3.26 3.56 Madagascar 1993 6.62 22. 60 95.00 -1.56 1.95 3.12 Kenya 1997 6.89 50.30 73.00 -1.90 2.61 3.92 Nigeria 1992 6.64 105.20 115.00 -1.17 3.26 4.66 Cote d`Ivoire 1998 7.38 50.30 115.00 -1.66 3.35 3.92 Gambia 1998 7.37 123.40 92.00 -1.71 3.11 4.82 Cote d`Ivoire 1995 7.27 46.40 110.00 -1.61 4.01 3.84 Malawi 1993 6.16 105.10 133.00 -1.66 3.26 4.65 Ghana 1989 7.12 66.20 78.00 -1.77 3.26 4.19 Nigeria 1996 6.68 117.10 120.00 -1.31 1.26 4.76 Uganda 1992 6.67 96.40 93.00 -1.31 3.26 4.57 Ghana 1992 7.25 71.90 78.00 -1.77 3.83 4.28 Ghana 1987 7.01 62.60 78.00 -1.71 3.26 4.14 Kenya 1992 6.84 43.90 63.00 -1.77 2.35 3.78 Cote d`Ivoire 1985 7.21 33.00 114.00 -1.47 3.45 3.50 Ethiopia 1995 6.26 60.00 123.00 -1.77 3.66 4.09 Mozambique 1997 6.58 21.40 145.00 -1.47 3.71 3.06 Ghana 1998 7.47 83.50 62.00 -1.61 3.26 4.42 Zambia 1993 6.70 12.20 102.00 -.54 .95 2.50 Ethiopia 1997 6.40 63.50 123.00 -1.77 3.29 4.15 Uganda 1989 6.51 86.90 93.00 -1.24 3.26 4.46 Uganda 2000 7.13 123.30 85.00 -1.56 3.75 4.81 Zambia 1998 6.62 13.80 102.00 -.67 2.53 2.62 Guinea 1994 7.37 29.60 129.00 -2.21 3.26 3.39 Nigeria 1980 6.28 75.20 108.00 -.89 3.26 4.32 Zambia 1996 6.62 13.20 102.00 -.62 1.38 2.58 Zambia 1991 6.70 11.60 101.00 -.49 3.26 2.45 Cameroon 1996 7.38 29.30 92.00 -.82 5.02 3.38 Malawi 1997 6.32 112.60 133.00 -1.71 1.90 4.72 Cameroon 2001 7.61 32.60 95.00 -.94 4.62 3.48 Nigeria 1985 6.23 86.10 115.00 -1.02 .00 4.46 Guinea 1991 7.34 26.30 145.00 -2.21 3.26 3.27 Nigeria 1997 6.69 120.10 120.00 -1.35 .32 4.79 South Africa 1993 8.96 32.80 45.00 -.97 1.67 3.49 South Africa 1997 9.08 37.10 45.00 -1.02 1.97 3.61 Burkina Faso 1998 6.85 39.00 107.00 -1.02 8.25 3.66

PAGE 204

204 Table H-1. Continued COUNTRY YEAR GINI MAH_1 COO_1 LEV_1 ZRE_1 RES_1 Mozambique 2003 42.00 1.87372 .01603 .03824 -1.39730 -7.97103 Gambia 1994 59.45 2.23204 .00873 .04555 .96431 5.50101 Cote d`Ivoire 1987 48.80 2.33418 .00124 .04764 -.35675 -2.03509 Cote d`Ivoire 1986 49.90 2.33501 .00042 .04765 -.20884 -1.19133 Cote d`Ivoire 1988 45.90 2.47294 .00479 .05047 -.68566 -3.91139 Gambia 1992 48.13 2.62044 01361 .05348 -1.12764 -6.43275 Madagascar 1997 44.00 2. 74552 .00353 .05603 -.56283 -3.21071 Gambia 1993 60.86 3.86335 .02237 .07884 1.21247 6.91666 Tanzania 1993 47.70 3. 87176 .00030 .07902 -.14064 -.80231 Tanzania 1991 58.90 3. 88768 .04881 .07934 1.78558 10.18598 Ethiopia 2000 29.70 4.01499 07826 .08194 -2.22563 -12.69630 Madagascar 2001 47.40 4. 15416 .00080 .08478 -.22154 -1.26381 Madagascar 1999 43.10 4.16232 .02848 .08495 -1.31875 -7.52290 Kenya 1994 44.30 4.21343 .01841 .08599 -1.05399 -6.01261 Burkina Faso 1994 58.90 4.30376 .00625 .08783 .60753 3.46571 Madagascar 1993 48.50 4. 62309 .00144 .09435 -.28075 -1.60155 Kenya 1997 44.50 4.76710 .00206 .09729 -.33078 -1.88695 Nigeria 1992 54.20 4.83740 .00027 .09872 .11923 .68015 Cote d`Ivoire 1998 44.40 5.02732 .00085 .10260 -.20669 -1.17906 Gambia 1998 50.20 5.08110 .01863 .10370 .96197 5.48763 Cote d`Ivoire 1995 43.90 5.10760 .00260 .10424 -.35849 -2.04503 Malawi 1993 62.00 5.31815 .10112 .10853 2.18624 12.47162 Ghana 1989 47.90 5.59623 .00570 .11421 .50455 2.87828 Nigeria 1996 48.30 5.81723 .00733 .11872 .56015 3.19542 Uganda 1992 39.00 5.82977 07284 .11897 -1.76314 -10.05798 Ghana 1992 39.70 6.00973 .02059 .12265 -.92132 -5.25575 Ghana 1987 42.90 6.14281 .00197 .12536 -.28153 -1.60600 Kenya 1992 59.90 6.35134 .03043 .12962 1.08471 6.18782 Cote d`Ivoire 1985 50.60 6.35530 .00491 .12970 .43554 2.48458 Ethiopia 1995 32.70 6.45135 02639 .13166 -1.00091 -5.70979 Mozambique 1997 40.00 6.75419 .00520 .13784 -.43252 -2.46738 Ghana 1998 50.70 7.08766 .06663 .14465 1.50310 8.57458 Zambia 1993 51.10 7.14074 .00552 .14573 -.43071 -2.45704 Ethiopia 1997 48.20 7.24036 .10051 .14776 1.82204 10.39401 Uganda 1989 44.30 7.31874 .00914 .14936 -.54564 -3.11264 Uganda 2000 46.90 7.43776 .00004 .15179 .03703 .21125 Zambia 1998 57.40 7.48437 .00323 .15274 .31994 1.82510 Guinea 1994 55.10 7.50708 .00049 .15321 -.12406 -.70773 Nigeria 1980 53.00 8.30717 .01478 .16953 -.64008 -3.65140 Zambia 1996 54.80 8.49363 .00224 .17334 -.24579 -1.40215 Zambia 1991 59.30 8.63906 .02242 .17631 .76824 4.38251 Cameroon 1996 50.80 8.94271 .03521 .18250 .94061 5.36580 Malawi 1997 49.30 9.14537 .03267 .18664 .89220 5.08964 Cameroon 2001 44.20 10.64911 .00380 .21733 -.27292 -1.55689 Nigeria 1985 53.80 11.49635 .00015 .23462 -.05183 -.29567 Guinea 1991 50.70 11.58709 .03042 .23647 -.72424 -4.13151 Nigeria 1997 50.20 12.81475 .00025 .26153 .06005 .34257 South Africa 1993 59.50 15. 74307 .01081 .32129 -.33159 -1.89157 South Africa 1997 60.10 16. 75582 .04500 .34196 .63635 3.63014 Burkina Faso 1998 62.50 30.05417 .41623 .61335 .84070 4.79585

PAGE 205

205 APPENDIX I REGRESSION RESULTS WITHOUT MISSING VARIABLES Table I-1. Model summary Model R R Square Adjusted R Square Std. Error of the Estimate Durbin-Watson 1 .680.463.4425.46325 2 .738.545.5095.12730 1.901 a Predictors: (Constant), AGVALUE b Predictors: (Constant), AGVALUE, TRADFOOD c Dependent Variable: GINI Table I-2. Coefficients Unstandardized Coefficients Standardized Coefficients t Sig. Collinearity Statistics Model B Beta Tolerance VIF 1 (Constant) 61.00720.579.000 AGVALUE -.410-.680-4.733.0001.0001.000 2 (Constant) 62.08121.954.000 AGVALUE -.325-.539-3.582.001.8041.244 TRADFOOD -8.479E-02-.320-2.126.044.8041.244 a Dependent Variable: GINI Table I-3. Collinearity diagonstics Eigenvalue Condition Index Variance Proportions Model Dimension (Constant) AGVALUE TRADFOOD 1 1 1.9371.000.03.03 2 6.261E-025.563.97.97 2 1 2.7851.000.01.01.02 2 .1554.234.20.05.91 3 6.010E-026.807.79.94.06 a Dependent Variable: GINI

PAGE 206

206 Table I-4. Residuals statistics Minimum Maximum Mean Std. Deviation N Predicted Value 37.206460.175447.85365.40041 28 Std. Predicted Value -1.9722.282.0001.000 28 Standard Error of Predicted Value 1.008942.501591.61632.46019 28 Adjusted Predicted Value 34.820860.386447.80655.49203 28 Residual -10.899710.9209.00004.93375 28 Std. Residual -2.1262.130.000.962 28 Stud. Residual -2.2452.358.0041.017 28 Deleted Residual -12.153613.3792.04715.52242 28 Stud. Deleted Residual -2.4612.619.0131.072 28 Mahal. Distance .0815.4631.9291.612 28 Cook's Distance .000.417.040.083 28 Centered Leverage Value .003.202.071.060 28 a Dependent Variable: GINI

PAGE 207

207 Scatterplot Dependent Variable: GINIRegression Standardized Residual3 2 1 0 -1 -2 -3 3 2 1 0 -1 -2 -3Rsq = 0.2223 Figure I-1. Residuals scatterplot

PAGE 208

208 APPENDIX J GDP REGRESSION Table J-1. Descriptive statistics Mean Std. Deviation N LOGGDPPP 6.9637.6474038 PREWTO .45.50438 AGVALUE 32.458211.7559638 TOTAGEXP 52.353927.1567338 Arable Land (ha per person) .2497.1218238 Value share of 3 of top 20 agricultural exports 75.245214.2009938 Table J-2. Model summary Model R R Square Adjusted R Square Std. Error of the Estimate Durbin-Watson 1 .744.554.484.46487 1.133 a Predictors: (Constant), Value share of 3 of top 20 agricultural exports, PREWTO, Arable Land (ha per person), AGVALUE, TOTAGEXP b Dependent Variable: LOGGDPPP Table J-3. ANOVA Model Sum of Squares df Mean SquareF Sig. 1 Regression 8.59251.7187.952.000 Residual 6.91532.216 Total 15.50837 a Predictors: (Constant), Value share of 3 of top 20 agricultural exports, PREWTO, Arable Land (ha per person), AGVALUE, TOTAGEXP b Dependent Variable: LOGGDPPP

PAGE 209

209 Table J-4. Coefficients Unstandardized Coefficients Standardized Coefficients t Sig. Collinearity Statistics Model B Std. Error Beta Tolerance VIF 1 (Constant) 8.811.55715.825.000 PREWTO -1.865E-02.164-.015-.114.910.8561.168 AGVALUE -2.927E-02.008-.532-3.466.002.5931.687 TOTAGEXP 3.605E-03.004.151.918.365.5141.947 Arable Land (ha per person) .279.833.052.335.740.5671.762 Value share of 3 of top 20 agricultural exports -1.525E-02.007-.335-2.258.031.6351.575 a Dependent Variable: LOGGDPPP Table J-5. Collinearity diagnostics Eigenvalue Condition Index Variance Proportions Model Dimension (Constant) PREWTO AGVALUE TOTAGEXP Arable Land (ha per person) Value share of 3 of top 20 agricultural exports 1 1 5.0141.000.00.01.00.00 .00.00 2 .5373.056.00.77.00.01 .01.00 3 .3293.905.00.00.01.12 .19.00 4 7.641E-028.101.01.05.54.38 .08.01 5 3.194E-0212.529.13.07.29.37 .53.24 6 1.140E-0220.977.86.10.16.11 .20.75 a Dependent Variable: LOGGDPPP

PAGE 210

210 Table J-6. Residuals statistics Minimum Maximum Mean Std. Deviation N Predicted Value 5.92818.31606.9637.48190 38 Std. Predicted Value -2.1492.806.0001.000 38 Standard Error of Predicted Value .12069.25768.18029.04074 38 Adjusted Predicted Value 5.78207.98936.9602.46641 38 Residual -.6753.9018.0000.43232 38 Std. Residual -1.4531.940.000.930 38 Stud. Residual -1.5562.054.0031.024 38 Deleted Residual -.83941.0907.0035.52704 38 Stud. Deleted Residual -1.5932.170.0121.051 38 Mahal. Distance 1.52010.3954.8682.646 38 Cook's Distance .000.275.038.061 38 Centered Leverage Value .041.281.132.072 38 a Dependent Variable: LOGGDPPP

PAGE 211

211 Regression Standardized ResidualHistogram Dependent Variable: LOGGDPPP10 8 6 4 2 0 Std. Dev = .93 Mean = 0.00 N = 38.00 Figure J-1. Histogram regression residuals

PAGE 212

212 APPENDIX K GDP REGRESSION WITHOUT MISSING VARIABLES Table K-1. Variables entered Model Variables Entered Variables Removed Method 1 VALAGRI3 Stepwise (Criteria: Probability-of-F-toenter <= .050, Probability-of-F-toremove >= .100). 2 AGVALUE Stepwise (Criteria: Probability-of-F-toenter <= .050, Probability-of-F-toremove >= .100). a Dependent Variable: GDPPPPCA Table K-2. Model summary Model R R Square Adjusted R Square Std. Error of the Estimate Durbin-Watson 1 .688.473.4541414.59880 2 .740.548.5131335.53674 2.410 a Predictors: (Constant), VALAGRI3 b Predictors: (Constant), VALAGRI3, AGVALUE c Dependent Variable: GDPPPPCA Table K-3. ANOVA Model Sum of Squares df Mean Square F Sig. 1 Regression 48566620.085148566620.08524.270.000 Residual 54029423.363272001089.754 Total 102596043.44828 2 Regression 56220925.707228110462.85415.760.000 Residual 46375117.741261783658.375 Total 102596043.44828 a Predictors: (Constant), VALAGRI3 b Predictors: (Constant), VALAGRI3, AGVALUE c Dependent Variable: GDPPPPCA

PAGE 213

213 Table K-4. Coefficients Unstandardized Coefficients Standardized Coefficients t Sig. Collinearity Statistics Model B Beta Tolerance VIF 1 (Constant) 7651.5626.018.000 VALAGRI3 -83.327-.688-4.926.0001.000 1.000 2 (Constant) 7397.0126.130.000 VALAGRI3 -54.402-.449-2.565.016.567 1.765 AGVALUE-58.254-.363-2.072.048.567 1.765 a Dependent Variable: GDPPPPCA Table K-5. Collinearity diagonstics Eigenvalue Condition Index Variance Proportions Model Dimension (Constant) VALAGRI3 AGVALUE 1 1 1.9781.000.01.01 2 2.157E-029.576.99.99 2 1 2.9231.000.00.00.01 2 6.159E-026.889.27.01.63 3 1.575E-0213.622.72.99.36 a Dependent Variable: GDPPPPCA Table K-6. Residuals statistics Minimum Maximum Mean Std. Deviation N Predicted Value -721.11155266.42921522.86211417.00042 29 Std. Predicted Value -1.5842.642.0001.000 29 Standard Error of Predicted Value 250.08817711.44415409.65642131.50801 29 Adjusted Predicted Value -1014.69654261.45611446.60931310.27176 29 Residual -1838.71923485.5710.00001286.95540 29 Std. Residual -1.3772.610.000.964 29 Stud. Residual -1.4533.084.0261.068 29 Deleted Residual -2083.76394866.568476.25281589.41194 29 Stud. Deleted Residual -1.4863.797.0561.164 29 Mahal. Distance .0166.9801.9311.949 29 Cook's Distance .0001.256.090.254 29 Centered Leverage Value .001.249.069.070 29 a Dependent Variable: GDPPPPCA

PAGE 214

214 Regression Standardized Residual2.50 2.00 1.50 1.00 .50 0.00 -.50 -1.00 -1.50Histogram Dependent Variable: GDPPPPCA10 8 6 4 2 0 Std. Dev = .96 Mean = 0.00 N = 29.00 Figure K-1. Histogram regression residuals

PAGE 215

215 APPENDIX L DEFINITIONS AND DATA RANGES Table L-1. Arable land (ha) per capita COUNTRY YEAR ARABLAND Zambia 1998 .51 Cameroon 2001 .39 Burkina Faso 1998 .36 South Africa 1997 .36 Nigeria 1997 .26 Mozambique 2003 .23 Uganda 2000 .21 Ghana 1998 .20 Cote d`Ivoire 1998 .19 Malawi 1997 .18 Gambia 1998 .18 Madagascar 2001 .17 Ethiopia 2000 .16 Kenya 1997 .15 Tanzania 1993 .13 Guinea 1994 .11 Definition: "Arable land (hectares per person) includes land defined by the FAO as land under temporary crops (double -cropped areas are counted once), temporary meadows for mowing or for pasture, land under market or kitchen gardens, and land temporarily fallow. Land abandoned as a result of shifti ng cultivation is excluded." Source: WDI

PAGE 216

216 Table L-2. Value share of top 3 agricultu ral exports (as % of top 20 agricultural exports) COUNTRY YEAR VALAGRI3 Ghana 1998 91.89 Burkina Faso 1998 89.37 Malawi 1997 87.72 Ethiopia 2000 86.06 Madagascar 2001 84.59 Gambia 1998 82.45 Uganda 2000 77.58 Guinea 1994 76.37 Kenya 1997 74.54 Nigeria 1997 74.48 Mozambique 2003 73.88 Zambia 1998 72.76 Tanzania 1993 71.09 Cameroon 2001 69.69 Cote d`Ivoire 1998 57.63 South Africa 1997 34.87 Definition: Calculated by a dding the value (in US$) of th e top 3 agricu ltural export commodities and dividing it by the value of th e top 20 agricultural exports. In cases where data was not available, FAO has estimated or corrected them, etc. Source: FAO

PAGE 217

217 Table L-3. Urbanization (% of urba n population) and population density COUNTRY YEAR URBANIZA POPDENSI South Africa 1997 52.6037.10 Cameroon 2001 49.0032.60 Cote d`Ivoire 1998 41.7050.30 Ghana 1998 40.2083.50 Nigeria 1997 39.50120.10 Zambia 1998 37.3013.80 Mozambique 2003 32.9824.30 Kenya 1997 30.0050.30 Gambia 1998 26.40123.40 Madagascar 2001 26.0028.70 Guinea 1994 25.3029.60 Tanzania 1993 20.0032.90 Burkina Faso 1998 15.2039.00 Ethiopia 2000 14.9068.50 Malawi 1997 14.02112.60 Uganda 2000 12.00123.30 Defintion: "Urban Population as a Percent of To tal Population is the pr oportion of a country's total national population that resides in urban areas. Any person not residing in an area classified as urban is counted in the rura l population. Definitions of urban populations vary slightly from country to country." Source: Earthtrends/UN

PAGE 218

218 Table L-4. Agriculture, value added (as % of GDP) COUNTRY YEAR AGVALUE Tanzania 1993 48.11 Ethiopia 2000 47.70 Cameroon 2001 39.86 Uganda 2000 37.34 Ghana 1998 36.01 Burkina Faso 1998 34.48 Nigeria 1997 33.63 Malawi 1997 32.59 Kenya 1997 30.91 Madagascar 2001 28.60 Gambia 1998 28.42 Mozambique 2003 24.38 Cote d`Ivoire 1998 24.12 Guinea 1994 21.71 Zambia 1998 21.14 South Africa 1997 4.01 Definition: "[P]roportion of tota l output of goods and services wh ich are a result of the value added by the agricultural sector. The agricultural sector corresponds to International Standard Industrial Classification (ISIC) divisions 1 and includes fore stry, hunting, and fishing, as well as cultivation of crops and lives tock production. Value added is th e value of the gross output of producers less the value of intermediate goods and services consumed in production. It is calculated without making deductions for deprecia tion of fabricated asse ts or depletion and degradation of natural resources." Source: Earthtrends/UN

PAGE 219

219 Table L-5. Food exports (as % of total merchandise exports) COUNTRY YEAR TRADFOOD Malawi 1997 84.50 Gambia 1998 76.80 Ethiopia 2000 70.60 Uganda 2000 67.30 Cote d`Ivoire 1998 59.40 Ghana 1998 58.40 Kenya 1997 55.60 Madagascar 2001 48.70 Guinea 1994 43.79* Mozambique 2003 43.79* Tanzania 1993 43.79* Cameroon 2001 16.80 Burkina Faso 1998 16.40 South Africa 1997 10.80 Zambia 1998 7.10 Nigeria 1997 1.70 *: Missing value replaced with mean of all data points Definition: "Components may not sum to 100% becau se of unclassified trade. Food comprises the commodities in SITC sections 0 (food and liv e animals), 1 (beverages and tobacco), and 4 (animal and vegetable oils and fats) and SITC di vision 22 (oil seeds, oil nu ts, and oil kernels). World Bank staff estimates from the COMTRADE database maintained by the United Nations Statistics Division. Source: WID

PAGE 220

220 Table L-6. Agricultural raw exports (as % of total merchandise exports) COUNTRY YEAR TRADRAW Burkina Faso 1998 68.10 Cameroon 2001 21.30 Ethiopia 2000 18.70 Uganda 2000 14.10 Cote d`Ivoire 1998 11.20 Guinea 1994 10.61* Mozambique 2003 10.61* Tanzania 1993 10.61* Ghana 1998 10.60 Gambia 1998 9.70 Kenya 1997 6.80 Zambia 1998 6.40 South Africa 1997 3.90 Madagascar 2001 3.70 Malawi 1997 3.60 Nigeria 1997 .10 *: Missing value replaced with mean of all data points Definition: "Components may not sum to 100% becau se of unclassified trade. Agricultural raw materials comprise SITC section 2 (crude mate rials except fuels) ex cluding divisions 22, 27 (crude fertilizers and minera ls excluding coal, petroleum, and precious stones), and 28 (metalliferous ores and scrap). World Bank staff estimates from the COMTRADE database maintained by the United Na tions Statistics Division." The main component of interest is division 26 textile fibres. For more information, visit: http://unstats.un.org/uns d/cr/registry/regcs.asp?Cl=14&Lg=1&Co=26 Source: WID

PAGE 221

221 Table L-7. Total agricultural exports (Food and Raw Ag) as % of total merchandise exports COUNTRY YEAR TOTALEXP Ethiopia 2000 89.30 Malawi 1997 88.10 Gambia 1998 86.50 Burkina Faso 1998 84.50 Uganda 2000 81.40 Cote d`Ivoire 1998 70.60 Ghana 1998 69.00 Kenya 1997 62.40 Guinea 1994 54.40* Mozambique 2003 54.40* Tanzania 1993 54.40* Madagascar 2001 52.40 Cameroon 2001 38.10 South Africa 1997 14.70 Zambia 1998 13.50 Nigeria 1997 1.80 *: Missing value replaced with mean of all data points Definition: Simple addition of both Food and Raw Ag exports. Source: WID

PAGE 222

222 Table L-8. GDP ppp per capit a & log GDP ppp per capita COUNTRY YEAR GDPPPPCA LOGGDPPP South Africa 1997 8752.009.08 Cameroon 2001 2022.007.61 Ghana 1998 1746.007.47 Cote d`Ivoire 1998 1606.007.38 Guinea 1994 1583.007.37 Gambia 1998 1582.007.37 Uganda 2000 1244.007.13 Mozambique 2003 1128.007.03 Kenya 1997 983.006.89 Burkina Faso 1998 940.006.85 Madagascar 2001 866.006.76 Nigeria 1997 806.006.69 Zambia 1998 749.006.62 Ethiopia 2000 636.006.46 Malawi 1997 556.006.32 Tanzania 1993 440.006.09 Definition: "GDP per capita based on purch asing power parity (PPP). PPP GDP is gross domestic product converted to international dolla rs using purchasing power parity rates. An international dollar has the same purchasing power over GDP as the U.S. dollar has in the United States. GDP at purchaser's prices is the sum of gross value added by all re sident 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 and degradation of natural resources. Data are in constant 2000 intern ational dollars." Source: WID

PAGE 223

223 Table L-9. Agricultural exports (% of GDP) COUNTRY YEAR AGEXPGDP Cote d`Ivoire 1998 25.42679325 Malawi 1997 17.74849714 Ghana 1998 17.69871505 Madagascar 2001 10.73272894 Kenya 1997 9.778089314 Burkina Faso 1998 9.647086372 Cameroon 2001 6.933639164 Uganda 2000 6.315187972 Ethiopia 2000 5.532443331 Zambia 1998 4.302988053 Gambia 1998 4.035912 South Africa 1997 3.068714499 Nigeria 1997 0.097177379 Tanzania 1993 Mozambique 2003 Guinea 1994 Definition: Adding Raw&Food Exports, multiply tim es merchandise export value, then divided by the total GDP, keeping US $ (in both cases) constant. Source: WID

PAGE 224

224 APPENDIX M ROBUSTNESS OF MODELS

PAGE 225

225Table M-1. Comparison of re gression models robustness MODEL 1 MODEL 2 CLASSICAL MODEL 3 REVISED MODEL 4 NEW MODEL 5 NEW MODEL 6 NEW MODEL 7 Comparison of Variables' Significance in Inequality Regression Models n=38/df=30n=38/df=30n=50/df=42 n=30/df =25 n=32/df=24 n=32/df=28 n=20/df=18 Before WTO nsig ** ** nsig ** *** nsig Agriculture Value Added (% of GDP) **** **** **** **** ** **** Food Exports (% of merchandise exports) **** **** ** **** ** Agricultural Raw Exports (% of merchandise exports)1 nsig nsig nsig nsig nsig Total Agricultural Exports (% merchandise exports)3 ** Total Agricultural Exports (% of GDP) nsig GDP ppp per capita2 nsig nsig nsig nsig nsig nsig nsig Children's Infant Mortality rate nsig nsig nsig nsig nsig nsig nsig Urbanization ** ** nsig nsig nsig Note: -: not included; nsig: not significant; *: significant at .10 leve l; **: significant at .05 leve l, ***: significant at .0 1 level, ****: significant < 0.01 level 1: Transformed (square-root) 2: Transformed (logged) 3: Transformed (logged)

PAGE 226

226 Table M-1. Continued Definition of Models Model 1 Model 2 M odel 3 Model 4 Model 5 Model 6 Model 7 Missing Variables Replaced x x x x x Not Replaced x x Regression Type Enter x x x x Stepwise x x x Countries Excluded Burkina Faso x South Africa 2 points/country (x) (x) x x x Notes Min.2,Max.3 data points/ country Min.2,Max.3 data points/ country Adjusted R2 0.382 0.516 0.403 0.509 0.471 0.443 0.483

PAGE 227

227 Table M-1. Continued NEW MODEL 8 NEW MODEL 9 NEW MODEL 10 NEW MODEL 11 HYBRID MODEL 12 HYBRID MODEL 13 MODEL 14 MODEL 15 Comparison Variables' Significance n=20/df=12 n=32/df=24 n=32/df=31 n=20/df=18 n=28/df=21 n=38/df=35 n=26/df=18 n=18/df=16 Before WTO nsig nsig ** nsig nsig **** nsig nsig Agriculture Value Added **** ** **** **** ** **** **** **** Food Exports **** Agricultural Raw Exports1 nsig nsig Total Agricultural Exports (% merchandise exports)3 ** nsig nsig nsig nsig nsig Total Agricultural Exports (% of GDP) nsig nsig nsig nsig nsig nsig GDP ppp per capita2 nsig nsig nsig nsig nsig nsig nsig nsig Children's Infant Mortality nsig nsig nsig nsig nsig nsig nsig nsig Urbanization nsig nsig nsig nsig nsig nsig ** nsig Note: -: not included; nsig: not signi ficant; *: significant at .10 level; **: significant at .05 level, ***: significant at .01 level, ***: significant < 0.01 level 1: Transformed (square-root) 2: Transformed (logged) 3: Transformed (logged)

PAGE 228

228 Table M-1. Continued Definition of Models Model 8 Model 9 Model 10 Model 11 Model 12 Model 13 Model 14 Model 15 Missing Variables Replaced x x x Not Replaced x x x x x Regression Type Enter x x x x x Stepwise x x x Countries Excluded Burkina Faso South Africa x x 2 points/country x x x x (x) (x) (x) x Notes Min.2,Max.3 data points/ country Min.2,Max.3 data points/ country Min.2,Max.3 data points/ country Adjusted R2 0.614 0.348 0.32 0.483 0.414 0.33 0.602 0.393

PAGE 229

229 LIST OF REFERENCES Ackerman, Frank, Neva R. Goodwin, La urie Dougherty, and Kevin Gallagher. 2000. The political economy of inequality Edited by N. R. Goodwin, Frontier issues in economic thought Washington, DC: Island Press. Adams, Richard H. 1995. Agricultural income, cas h crops, and inequality in rural Pakistan. Economic Development and Cultural Change 43 (3):467-491. Alesina, Alberto, and Dani Rodrik. 1994. Distributive politics and economic growth. The Quarterly Journal of Economics 109 (2):465-490. Baffes, John. 2004. The 'cotton problem' The World Bank. Available online at http://www.fao.org/es/ESC/common/ecg /47647_en_CottonProblem_Baffes.pdf (last accessed 21 February 2007). 2005. Cotton: Market setting, trade policies, and issues The World Bank. Available online at http://wwwwds.worldbank.org/servlet/WDSConte ntServer/WDSP/IB/2004/06/03/000009486_20040 603091724/Rendered/PDF/wps3218cotton.pdf (last accessed 21 February 2007). Barrett, Hazel R, Brian W Ilbery, Angela W Brown, and Tony Binns. 1999. Globalization and the changing networks of food supply: Th e importation of fresh horticultural produce from Kenya into the UK. Transactions of the Institute of British Geographers 24 (2):159174. Bassett, Thomas J. 2001. The peasant cotton revolution in We st Africa : Cte d'Ivoire, 18801995 African studies series: 101 Cambridge; New York: Cambridge University Press. Bates, Robert H. 2005. Markets and states in tropical Africa : the political basis of agricultural policies Berkeley: University of California Press. BBC. 2007. Timeline: Ethiopia and Somalia Available online at http://news.bbc.co.uk/ 2/hi/africa/6159735.stm (last accessed 25 January 2007). Bebbington, Anthony. 1999. Capitals and capabilit ies: A framework for analyzing peasant viability, rural livelihoods and poverty. World Development 27 (12):2021-2044. Beer, Linda, and Terry Boswell. 2001. The effects of globalizati on on inequality: A crossnational analysis Atlanta, Georgia: Emory Universit y. The Claus M. Halle Institute for Global Learning. Beghin, John C, Ndiame Diop, Holger Matt hey, and Mirvat Sewa deh. 2003. Groundnut trade liberalization: A south-south debate? Center for Agricultur al and Rural Development. Iowa State University.

PAGE 230

230 Beneria, Lourdes. 1999. Globaliz ation, gender and the Davos man. Feminist Economics 5 (3):6183. Binswanger, Hans P., and Klaus Deininger. 1997. Explaining agricultural and agrarian policies in developing countries. Journal of Economic Literature 35 (4):1958-2005. Block, S., and Patrick Webb. 2001. The dynamics of livelihood dive rsification in post-famine Ethiopia. Food Policy 26 (4):333-350. Bloom, David E., and Jeffrey Sachs. 1998. Ge ography, demography, and economic growth in Africa. Harvard Institute for International Development. Bloomer, Phil. 2004. Coffee: an empty cup for growers. International Herald Tribune 02/03/2004. Bourguignon, Francois, Sylvie Lambert, and Ak iko Suwa-Eisenmann. 2004. Trade exposure and income volatility in cash-crop exporting developing countries. European Review of Agricultural Economics 31 (3):369-387. Britannica. 2006. Britannica.com Available online at http://www.britannica.com (last accessed 12 January 2007). Camara, Oumou, and Ed Heinemann. 2006. Overview of the fertilizer situation in Africa African Fertilizer Summit. Abuja, Nigeria. June 9-13, 2006. Castello, Amparo, and Rafael Domenech. 2002. Human capital ine quality and economic growth: Some new evidence. The Economic Journal 112:187-200. Chauveau, J.P. 2003. How does an institution evol ve? Land, politics, intra-household relations and the institution of the tutorate between autochthons and immigrants in the Gban region of Cote d'Ivoire. Draft CIA. 2007. World Factbook CIA. Available online at https://www.cia.gov/library/publications /the-world-factbook/index.html (last accessed 2 February 2007). Cogneu, Denis, Thomas Bossuroy, Phillipe De Vreyer, Charlotte Guenard, Victor Hiller, Phillippe Leite, Sandrine Mesp le-Somps, Laure Pasquier-Doume r, and Constance Torelli. 2006. Inequalities and equity in Africa DIAL. Available online at http://www.dial.prd.fr/dial_publicati ons/PDF/Doc_travail/2006-11_english.pdf (last accessed 04 January 2007). Collier, Paul, and Jan Willem Gunning. 1999. E xplaining african economic performance. Journal of Economic Literature 37 (1):64-111. Cooksey, Brian. 2003. Marketing reform? The rise and fall of agricultura l liberalisation in Tanzania. Development Policy Review 21 (1):67-91.

PAGE 231

231 Cotlook. 2007. Seasonal indices Available online at http://www.cotlook.com/cdis/more_cotlook_ indices.php?from_day=1&from_month=1&f rom_year=2006&to_day=1&to_month=1&to_year=1998&view=3 (last accessed 27 March 2007). Cramer, Christopher. 1999. Can Africa industria lize by processing primary commodities? The case of Mozambican cashew nuts. World Development 27 (7):1247-1266. Crawford, Neta C. 1999. Trump card or theater? An introduction to two sanctions debates. In How Sanctions Work: Lessons from South Africa edited by N. C. Crawford and A. Klotz: Palgrave Macmillan. Deininger, Klaus, and John Okidi. 2003. Grow th and poverty reduction in Uganda 1999-2000: Panel data evidence. Development Policy Review 21 (4):481-509. Deininger, Klaus, and Lyn Squire. 1996. A ne w data set measuring income inequality. The World Bank Economic Review 10 (3). 1998. New ways of looking at old issues: inequality and growth. Journal of Development Economics 57:259-287. Diao, Xinshen, Sherman Robinson, Marcelle Thomas, and Peter Wobst. 2002. Assessing impacts of declines in the world price of t obacco on China, Malawi, Turkey, and Zimbabwe Available online at http://www.ifpri.org/divs/tmd/dp/papers/tmdp91.pdf (last accessed 2 February 2007). Diop, Ndiame, John C Beghin, a nd Mirvat Sewadeh. 2005. Groundnut policies, global trade dynamics, and the impact of trade liberalization. In Global Agricultural Trade and Developing Countries edited by M. A. Aksoy and J. C. Beghin. Washington D.C.: World Bank. Dollar, David, and Aart Kray. 2002. Growth is good for the poor. Journal of Economic Growth 7 (3):195-225. Dorward, Andrew, Jonathan Kydd, Jamie Morri son, and Ian Urey. 2004. A policy agenda for pro-poor agricultural growth. World Development 32 (1):73-89. DTI. 2007. Agriculture Department of Trade and Industry South Africa. Available online at http://www.thedti.gov.za/econdb/raportt/rapsiagcurrent.html (last accessed 22 March 2007). 2007. South African economy: Inequality Available online at http://www.dti.gov.za/saeconomy/inequality.htm (last accessed 22 March 2007). Earthtrends. 2003. Agriculture and Food Burkina Faso Available online at http://earthtrends.w ri.org/pdf_library/count ry_profiles/agr_cou_854.pdf (last accessed 03 February 2007).

PAGE 232

232 2003. Agriculture and Food Cote d'Ivoire Available online at http://earthtrends.w ri.org/pdf_library/count ry_profiles/agr_cou_384.pdf (last accessed 03 February 2007). 2003. Agriculture and Food Ethiopia Available online at http://earthtrends.w ri.org/pdf_library/count ry_profiles/agr_cou_231.pdf (last accessed 03 February 2007). 2003. Agriculture and Food Kenya Available online at http://earthtrends.w ri.org/pdf_library/count ry_profiles/agr_cou_404.PDF (last accessed 03 February 2007). 2003. Agriculture and Food Malawi Available online at http://earthtrends.w ri.org/pdf_library/count ry_profiles/agr_cou_454.PDF (last accessed 04 February 2007). 2003. Agriculture and Food Mozambique Available online at http://earthtrends.w ri.org/pdf_library/count ry_profiles/agr_cou_508.pdf (last accessed 03 February 2007). 2003. Agriculture and Food Senegal Available online at http://earthtrends.w ri.org/pdf_library/count ry_profiles/agr_cou_686.PDF (last accessed 03 February 2007). 2003. Agriculture and Food South Africa Available online at http://earthtrends.w ri.org/pdf_library/count ry_profiles/agr_cou_710.PDF (last accessed 03 February 2007). 2003. Agriculture and Food Uganda Available online at http://earthtrends.w ri.org/pdf_library/count ry_profiles/agr_cou_800.PDF (last accessed 03 February 2007). Easterly, William. 2006. Inequality does cause underdevelopment: Insights from a new instrument. Journal of Development Economics in press. Easterly, William, and Ross Levine. 1997. Africa's grow th tragedy: Policies and ethnic divisions. The Quarterly Journal of Economics 112 (4):1203-1250. Edwin, J., and W.A. Masters. 2005. Genetic im provement and cocoa yields in Ghana. Expl Agriculture 41:491-503. Ellis, Frank, and Ntengua Mdoe. 2003. Livelihoo ds and rural poverty reduction in Tanzania. World Development 31 (8):1367-1384. English, Philip, Steven M. Jaffee, and Julius Okello. 2004. Kenya: Exporting out of Africa Kenya's horticulture success story Shanghai, China. The World Bank Group. 5/27/2004.

PAGE 233

233 FAO. 2002. FAO/WFP crop and food supply assessment mission to Senegal Available online at http://www.fao.org/docrep/005/y8165e/y8165e00.htm (last accessed 14 March 2007). 2004. Cotton commodity notes FAO. Available online at http://www.fao.org/es/ESC/e n/20953/22215/highlight_28507en.html (last accessed 27 March 2007). 2004. The state of agricultural commodity markets. Commodities and Trade Division of FAO. FAO. 2005. Livestock sector brief: Burkina Faso Food and Agriculture Organization of the United Nations. Available online at http://www.fao.org/WAICENT/fa oINFO/AGRICULT/againfo/re sources/en/publications/ sector_briefs/lsb_BFA.pdf (last accessed 10 March 2007). 2006. Exports: Commodities by country Available online at http://www.fao.org/es/ess/toptrade/trade.asp (last accessed 5 February 2007). 2006. FAOSTAT FAO. Available online at http://faostat.fao.org (last accessed 10 December 2006). FAOSTAT. 2005. Country: Mozambique (last accessed 14 March 2007). Fosu, Augustin Kwasi. 2005. Inequality and the growth-poverty nexus: Evidence from subSaharan Africa ESPD Seminar Series. Addis Ababa. United Nations Economic Commission for Africa. Garson, Dave. 2007. Testing of assumptions Available online at http://www2.chass.ncsu.edu/garson/pa765/assumpt.htm (last accessed 5 April 2007). Ghana Export Promotion Council. 1995. Comparison of export performance of non-traditional products 1989-1999. Ghana Statistical Services. 2000. Quarterly digest of statistics. GSS. Accra. Goldberg, Pinelopi Koujianou, and Nina Pavcnik. 2003. The response of the informal sector to trade liberalization. Journal of Development Economics 72 (2):463-496. 2004. Trade, inequality, and poverty: What do we know? Brookings Trade Forum Goreux, Louis. 2005. Cotton in Burkina Faso. OECD. Government of Ghana. 1997. Budgetary stat ement. Ministry of Finance. Accra. 1998. Budgetary statement. Ministry of Finance. Accra. Government of Senegal. 1994. Declaration de Po litique de Developpement Agricole (DPDA). Ministry of Agriculture.

PAGE 234

234 Grant, Richard, and John Agnew. 1996. Represen ting Africa: The geography of Africa in world trade, 1960 1992. Annals of the Association of American Geographers 86 (4):729-744. Gray, Leslie C., and Michael Kevane. 2001. Evolving tenure right s and agricultural intensification in Southwestern Burkina Faso. World Development 29 (4):573-587. Grimm, Michel, and Isabel Gunther. 2004. Operationalising pro-poor growth: A country case study on Burkina Faso DFID.gov.uk. Available online at http://www.dfid.gov.uk/news/files/ propoorgrowthcasestudies.asp (last accessed 20 March 2007). Gunning, Jan Willem. 2000. Rethinking aid 12th Annual Bank Conference on Development Economics (ABCDE). Washington, DC. World Bank. April 18-20, 2000. Hertel, Thomas W., and L. Alan Winters, eds. 2006. Poverty & the WTO: Impacts of the Doha Development Agenda New York, NY: Palgrave Macmillan and The World Bank. IMF and IDA. 2001. Uganda: Poverty reduction st rategy paper progress report: Joint staff assessment. G. E. Gondwe, M. T. Hadjimichael, C. Madavo and K. Dervis. International Monetary Fund and the International Development Association. INTRACEN. 2001. Ethiopia's export strategy: Key feat ures and major influencing factors. Paper read at Regional Executive Forum on Nati onal Export Strategies at Nairobi, Kenya. IRIN. 2007. Senegal: Incumbent wins presidential election UN Office for the Coordination of Humanitarian Affairs. Available online at http://www.irinnews.org/Re port.aspx?ReportId=70496 (last accessed 14 May 2007). James, Robert C., Channing Arndt, and Kenneth R. Simler. 2005. Has economic growth in Mozambique been pro-poor Available online at www.ifpri.org/divs/fcnd/dp.htm (last accessed 12 March 2007). Kayizzi-Mugerwa, Steve. 2001. Globalisation, growth and income inequality: The African experience. OECD Development Centre Working Paper No.186:34. Kelly, Valerie, Bocar Diagana, Thomas Rear don, Matar Gaye, and Eric Crawford. 1996. Cash crop and foodgrain productivity in Senegal: Hi storical view, new survey evidence, and policy implications. MSU International Development Papers (20). Konandu-Agyemang, Kwadwo, and Sesime Ad anu. 2003. The changing geography of export trade in Ghana under structural adjustment programs: Some socioeconomic and spatial implications. The Professional Geographer 55 (4):513-527. Kristof, Nicholas. 2005. Year af ter year, grave after grave. New York Times October 11th, 2005, A.28. Krugman, Paul R. 1991. Geography and trade Gaston Eyskens lecture series Leuven, Belgium, Cambridge, Mass.: Leuven University Press, MIT Press.

PAGE 235

235 Krugman, Paul, and Anthony J. Venables. 1995. Gl obalization and the ine quality of nations. The Quarterly Journal of Economics 110 (4). Kurti, Daniel Balint. 2005. The Ivory Coast's cocoa war. The Christian Science Monitor December 14, 2005. Kuznets, S. 1955. Economic growth and income inequality. The American Economic Review 45 (1):1-28. Leach, Melissa, Robin Mearns, and Ian Scoones. 1999. Environmental entitlements: Dynamics and Iinstitutions in community-bas ed natural resource management. World Development 27 (2):225-247. Ledermann, Samuel, and William G. Moseley. 2006. The World Trade Organization's Doha Round and cotton: Continued peripheral status or a 'historical break through' for African farmers? Macalester College. Leichenko, Robin M., and Karen L. O'Brien. 2002. Th e dynamics of rural vulnerability to global change: The case of Southern Africa. Mitigation and Adaptation Strategies for Global Change 7:1-18. Lemarchand, Rene. 2003. The venomous flowers of Ivoirite. West African Research Association Newsletter Lipton, M. 1977. Why poor people stay poor Cambridge, MA: Harvard University Press. Mandel, Jennifer L. 2004. Mobility matters: Wome n's livelihood strategies in Porto Novo, Benin. Gender, Place and Culture 11 (2):257-287. Mayer, Susan E., and Ankur Sarin. 2003. An assessment of some mechanisms linking economic inequality and infant mortality Available online at http://harrisschool.uchicago.edu/About/ publications/worki ng-papers/pdf/wp_02_12.pdf (last accessed 07 April 2007). MCC. 2006. Ghana Millenium Challenge Corporation. Available online at http://www.mcc.gov/countries/ghana/index.php (last accessed 10 February 2007). Moradi, Alexander, and Joerg Baten. 2005. Inequali ty in sub-Saharan Africa: New data and new insights from anthropometric estimates. World Development 33 (8):1233-1265. Moss, Todd J. 2007. African development: Making sense of the issues and actors Boulder, London: Lynne Rienner. new-agri.co.uk. 2006. Country profile: Burkina Faso Available online at http://www.newagri.co.uk/07/01/countryp.php (last accessed 10 February 2007). Norton, Roger D. 2004. Agricultural development policy: Concepts and experiences Edited by FAO. West Sussex, England: John Wiely & Sons Inc.

PAGE 236

236 O'Brien, Karen L., and Robin M. Leichenko. 2003. Wi nners and losers in the context of global change. Annals of the Association of American Geographers 93 (1):89-103. Onafowora, Olugbenga A., and Oluewole Ow oye. 1998. Can trade liberalization stimulate economic growth in Africa? World Development 26 (3):497-506. Oxfam. 2002. Crisis in the bi rthplace of coffee. Oxfam. 2003. Ethiopia: Aver ting the crisis. Oxfam.org.uk Feb 2003. Paluzie, Elisenda. 2001. Trade polic y and regional inequalities. Papers in Regional Science 80 (1):67-85. Peacock, Christie. 2005. Africa: Farming sense Available online at http://www.oecdobserver.org/news/fullsto ry.php/aid/1617/Africa:_Farming_sense_.html (last accessed 30 March 2007). Persson, Torsten, and Guido Tabellini. 1991. Is inequality harmful fo r growth? Theory and evidence. National Bureau of Economics Research. Peters, Pauline E., and M. Guillermo Herrera. 1994. Tobacco cultivation, food production, and nutrition among smallholders in Malawi. In Agricultural Commercialization, Economic Development, and Nutrition edited by J. von Braun and E. Kennedy: John Hopkins University Press. Polanyi, Karl. 1957. The great transformation Boston, MA: Beacon Press. Reardon, Thomas, and J. Edward Tayler. 1996. Ag roclimatic shock, income inequality, and poverty: Evidence from Burkina Faso. World Development 24 (5):901-914. Richards, Paul. 2004. Controversy over recent West African wars Centre of African Studies. University of Copenhagen. January 2004. Sachs, Jeffrey. 2001. The strategic si gnificance of global inequality. The Washington Quarterly 24 (3):187-198. Sen, Amartya. 1998. Mortality as an indi cator of economic success and failure. Economic Journal 108 (446):1-25. Silva, Julie. 2007. Trade and income inequality in a less developed country: The case of Mozambique. Economic Geography 83:111-136. Southafrica.info. 2007. South African agriculture International Marke ting Council of South Africa, Available online at http://www.southafrica.info/doing_busines s/economy/key_sectors/agriculturalsector.htm (last accessed 10 February 2007). Statistics South Africa. 2004. Labour force survey March 2004. Statistics South Africa.

PAGE 237

237 Stiglitz, Joseph. 2006. Making globalization work London: Penguin Books. Stock, Robert. 2004. Africa South of the Sahara : Guilford Press. Thirtle, Colin, Lin Lin, and Jenifer Piesse. 2003. The impact of res earch-led agricultural productivity growth on poverty reduction in Africa, Asia, and Latin America. World Development 31 (12):1959-75. Thurow, Roger. 2003. Behind the famine in Et hiopia: Glut and aid policies gone bad. Wall Street Journal July 11th, 2003. Thurow, Roger, and Scott Kilman. 2005. Farmers, charities join forces to block famine-relief revamp. Wall Street Journal October 26th, 2005. Tiffen, Mary. 2003. Transition in sub-Saharan Af rica: Agriculture, urbanization and income growth. World Development 31 (8):1343-1366. UN. 2006. World Development Indicators Available online at http://web.worldbank.org/WBSITE/EXTER NAL/DATASTATISTICS/0,,contentMDK:2 0398986~menuPK:64133163~pagePK:64133150~piPK:64133175~theSitePK:239419,00 .html (last accessed 12 February 2007). UNAIDS. 2007. Mozambique UNAIDS. Available online at http://www.unaids.org/en/Regions _Countries/Countries/mozambique.asp (last accessed 27 March 2007). UNCTAD. 2006. Cotton: Companies Available online at http://r0.unctad.org/infocomm/a nglais/cotton/companies.htm (last accessed 10 February 2007). UNDP. 2006. Country fact sheets: Malawi Available online at http://hdr.undp.org/hdr2006/st atistics/countries/country_f act_sheets/cty_fs_MWI.html (last accessed 22 March 2007). 2006. Country fact sheets: South Africa Available online at http://hdr.undp.org/hdr2006/st atistics/countries/country_f act_sheets/cty_fs_ZAF.html (last accessed 22 March 2007). UNECA. 2003. Uganda A tale of two economies? In Economic Report on Africa 2003: Accelerating the pace of development edited by Economic Commission for Africa. Addis Ababa, Ethiopia: Economic Commission for Africa. UNU-WIDER. 2005. World Income Inequality Database: User guide and data sources Available online at http://www.wider.unu.edu/wiid/WIID2.pdf (last accessed 10 February 2007). USAID. 2005. USAID/Kenya annua l report FY 2005. USAID.

PAGE 238

238 USDA. 2002. African franc zone: Cotton regions dry Available online at http://www.fas.usda.gov/pecad2/high lights/2002/09/franc_zone/index.htm (last accessed 11 February 2007). Wade, Robert. 2002. Globalization, poverty and income distributi on: Does the liberal argument hold? Working Paper Series, Deve lopment Studies Institute (02-33). Wanyeki, L. Muthoni. 2007. Here comes the U.S. of Africa. The East African March 13, 2007. Wetzel, Hayden. 2003. Uganda country commercial guide FY 2004: Economic trends US and Foreign Commercial Service, US Depart ment of State. Available online at http://strategis.ic.gc.ca/epic/site/imr-ri.nsf/en/gr120296e.html (last accessed 10 February 2007). White, Phillip. 2005. War and food securi ty in Eritrea and Ethiopia, 1998-2000. Disasters 29 (1):92. Wines, Michael. 2005. Drought deepens poverty, starving more Africans. New York Times 11/2/2005. Wobst, Peter. 2003. The impact of domestic and global trade liberalisation on five southern African countries. The Journal of Development Studies 40 (2):70-92. Wood, Adrian. 1994. North-South trade empl oyment and inequality IDS Development Studies Series Oxford: Clarendon Press. 2002. Could Africa be like America? Annual Bank Conference on Development Economics Washington, DC. World Bank. April-May 2002. Wood, Adrian, and Jrg Mayer. 2001. Africa's expor t structure in a comp arative perspective. Cambridge Journal of Economics 25:369-394. Wood, Geof. 2003. Staying secure, stay ing poor: The "Faustian Bargain". World Development 31 (3):455-471. Woods, Dwayne. 2003. The tragedy of the cocoa pod: Rent-seeking, land a nd ethnic conflict in Ivory Coast. Journal of Modern African Studies 41 (4):641-655. World Bank. 2005. African development indicators 2005 World Bank Africa Database : World Bank Publications. 2005. Equity and development Edited by W. Bank, World Development Report 2006 New York, NY and Washington, DC: World Bank and Oxford University Press. ed. 1993. The East Asian miracle: Economic growth and public policy Edited by W. Bank. New York: Oxford University Press.

PAGE 239

239 World Institute for Development Economics Research. 2005. World Income Inequality Database V 2.0a June 2005 UN. Available online at http://www.wider.unu.edu/wiid/wiid.htm (last accessed 04 October 2006). Yager, Thomas R. 2004. The mine ral industry of Uganda. USGS.

PAGE 240

240 BIOGRAPHICAL SKETCH Growing up in Kuettigkofen, a small agricultu ral village in Switz erland, Saemi Thomas Ledermann has been intimately involved with the dynamic relationship of human-environment interactions on a daily basis. During his year s at the Gymnasium Solot hurn, where he graduated in 2000 with a Matura degree in economics a nd law, he observed the gradual yet powerful impact that national and internat ional policies had on the viability of local farms. At Macalester College, Saint Paul, MN, he expanded this curiosity to the global scale, majoring with honors in Political Science, Geography, and International Studies with a sp ecific focus on development in Africa. This focus on development culminated in the successful completion of an Honors Thesis entitled Agricultural Subsidie s and the Doha Round: A Histor ic Breakthrough?, highlighting the potential mediumand long-te rm socio-economic and environmental impacts of the World Trade Organizations Doha Round on sub-Saharan cotton farmers. During his graduate studies at the University of Florida, Saemi continued to expand upon his previous studies, deve loping a further expertise in development in Africa, with specific foci on inequality and agriculture. Entering his PhD studies, his ultimate aim is not only to provide new insights into some of the most pressing i ssues plaguing African farmers, but also speak towards the policy-makers. Consequently, Saemi is intent on gaining ch allenging, yet rewarding experiences in academia and beyond.