|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 AGRICULTURAL EXTENSIFICATION IN THE WESTERN HIGHLANDS OF KENYA: IMPACTS ON MAIZE PRODUCTION AND ADOPTION OF SOIL FERTILITY ENHANCEMENT PRACTICES By MARIA C. MORERA A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UN IVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010
2 2010 Maria C. Morera
3 To Christy Gladwin
4 ACKNOWLEDGMENTS I would like to thank the many individuals and org anizations that made this work possible. First and foremost, I must thank the chair of my supervisory committee, Christina H. Gladwin, for her unwavering support and tireless mentoring I must also thank my supervisory committee members, Peter H. Hildebr and, Willie Baber, and Hugh L. Popenoe. Special thanks are extended to Clare nce Gravlee and Allan Burns. I also wish to thank Frank Place for his kind help with the doctoral fiel dwork, as well as Steven Franzel Quresh Noordin, Steve Ruigu, John Were, and the staff of the World Agroforestry Centre. I am equally appreciative for the logistical support generously provided by Daniel Rotich and the Kenya Agricultural Research Institute Many thanks as well to the staff of Kenya Agricultural Productivity Prog ramme, Kenya Women Finance Trust, National Cere als and Produce Board, Kenya Farmer s Association, United States Agency for International Development, Republic of Kenya Ministry of Agriculture, and Agricultural Finance Corporation for their thoughtful coope ration. I am deeply grateful for the warm reception I was granted in Kakamega Municipality and Shinyalu Division. Above all, I am indebted to Henry Ngaira and Roselida Odhiambo for their selfless and ongoing cooperation and to the many people of Mutsulio, Shikusi, Shinakotsi, and Lugango who welcomed me into their homes. Research would have been impossible without their openness, sincerity, and trust. I am fieldwork assista nce and to the Sisters of Mary, the Pabaris, and the staff of Shikusi Dispensary for their hospitality and friendship.
5 Warm thanks go to Karen Jones and the wonderful staff of both the Department of Anthropology and the Office of Graduate Minority Progra ms at the University of Florida. I also wish to extend heartfelt thanks to Eric Tillman for his kind assistance across several continents, to Julio Cesar Fajardo, Alicia Peon, and Jennifer Hale Gallardo for opening their homes (and hearts) to me, and to A lfredo Rios for his statistical advice and camaraderie Very special thanks as well to Felix Perez Baez and Rachel Muiz. This work was made financially possible by a Dissertation Research Grant provided by the National Science Foundation, by an Auzenne D issertation A ward, and by a Graduate Supplemental Retention Award.
6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 9 LIST OF FIGURES ................................ ................................ ................................ ........ 11 LIST OF ABBREVIATIONS ................................ ................................ ........................... 13 ABSTRACT ................................ ................................ ................................ ................... 15 CHAPTE RS 1 INTRODUCTION: THE ROOTS OF EXTENSIVE AGRICULTURE ....................... 17 2 THEORETICAL FRAMEWORK: KENYAN SOIL FERTILITY MANAGEMENT IN CONTEXT ................................ ................................ ................................ ............... 25 Introduction ................................ ................................ ................................ ............. 25 Effects of Soil Degradation on Agricultural Productivity ................................ .......... 25 Consequences of Low Agricultural Productivity ................................ ...................... 35 Other Factors Affecting Agricultural Productivity ................................ ..................... 40 Macroeconomic Policies ................................ ................................ ................... 40 Export Strategies ................................ ................................ .............................. 41 Trade Policies ................................ ................................ ................................ ... 46 State Intervention ................................ ................................ ............................. 47 Infrastructure ................................ ................................ ................................ .... 49 Neo patrimonialism ................................ ................................ .......................... 50 Nonfarm Work ................................ ................................ ................................ .. 52 Disease ................................ ................................ ................................ ............ 55 Conclusion ................................ ................................ ................................ .............. 56 3 RESEARCH SETTING: THE LAND AND ITS PEOPLE ................................ ........ 58 Introduction ................................ ................................ ................................ ............. 58 Ecological Setting ................................ ................................ ................................ ... 58 Socioeconomic Setting ................................ ................................ ........................... 59 Agricultural Practices ................................ ................................ ........................ 62 Collaboration with the PLAR Project ................................ ................................ 63 Other Institutions Working in the Surrounding Areas ................................ ........ 66 Markets and Infrastructure ................................ ................................ ................ 67 Additional Livelihood Strategies ................................ ................................ ....... 68 Effects of Illness on Livelihoods ................................ ................................ ....... 69 Conclusion ................................ ................................ ................................ .............. 71
7 4 RESEARCH METHODS: UNDERSTANDING THE OPPORTUNITIES AND CONSTRAINTS AFFECTING SOIL FERTILITY MANAGEMENT .......................... 77 Introduction ................................ ................................ ................................ ............. 77 Research Objectives and Assumptions ................................ ................................ .. 79 Research Hypotheses ................................ ................................ ............................. 80 Data Collection ................................ ................................ ................................ ....... 86 Data Analysis ................................ ................................ ................................ .......... 93 Multiple Regression Analysis ................................ ................................ ............ 95 Frequency Distributions ................................ ................................ .................. 101 Script Analy sis ................................ ................................ ................................ 102 Farm History Analysis ................................ ................................ .................... 102 Feasibility Analysis ................................ ................................ ......................... 103 Concl usion ................................ ................................ ................................ ............ 103 5 RESEARCH RESULTS: THE MAKING OF A SUCCESSFUL FARMER ............. 105 Introduction ................................ ................................ ................................ ........... 105 Multivariate Statistics: Multiple Regression Analysis ................................ ........... 106 Production Equations ................................ ................................ ..................... 109 Maize production ................................ ................................ ...................... 110 Bean production ................................ ................................ ....................... 110 Input Equations ................................ ................................ .............................. 111 Fertilizer use ................................ ................................ ............................ 113 Oxen use ................................ ................................ ................................ .. 114 Hired labor ................................ ................................ ............................... 115 Maize seeding rate ................................ ................................ ................... 115 Bean seeding rate ................................ ................................ .................... 116 Wealth Equations ................................ ................................ ........................... 117 Wealth index ................................ ................................ ............................ 118 Landholding size ................................ ................................ ...................... 119 Cattle owned ................................ ................................ ............................ 119 Cumulative years of schooling ................................ ................................ 119 Additional Equations ................................ ................................ ....................... 120 Kale acreage ................................ ................................ ............................ 120 Number of months maize is purchased for consumption ......................... 121 Descriptive and Inferential Bivariate Statistics: Frequency Distributions and Two Tailed t Test for Difference of Means ................................ ......................... 121 Crops Sold ................................ ................................ ................................ ...... 122 Household Use of Remittances ................................ ................................ ...... 123 Most Important Cash Sources ................................ ................................ ........ 123 Most Important Expenditures ................................ ................................ .......... 124 Social Capital ................................ ................................ ................................ 124 Technological Dissemination ................................ ................................ .......... 126 Quality of Life, Farm, and Economy ................................ ............................... 126 What Would Improve Your Farm? ................................ ................................ .. 127 Inv estment Preferences ................................ ................................ .................. 127
8 The Making of a Successful Farmer ................................ ............................... 128 Script Analysis ................................ ................................ ................................ ...... 129 Farm History Analysis ................................ ................................ ........................... 130 ................................ ................................ ................................ ......... 130 ................................ ................................ ................................ .... 131 ................................ ................................ ............................ 131 Feasibility Analysis ................................ ................................ ............................... 131 Calculation 1 ................................ ................................ ................................ ... 132 Calculation 2 ................................ ................................ ................................ ... 132 Calculation 3 ................................ ................................ ................................ ... 133 Conclusion ................................ ................................ ................................ ............ 133 6 DIS CUSSION: EXTENSIVE FARMING AND DIVERSE LIVELIHOOD STRATEGIES ................................ ................................ ................................ ....... 176 Introduction ................................ ................................ ................................ ........... 176 Extensive Farming Strategies ................................ ................................ ............... 177 Diverse Livelihood Strategies ................................ ................................ ............... 178 Off farm Nonagricultural Work ................................ ................................ ........ 179 On farm Nonagricultural Work ................................ ................................ ........ 180 Off farm Agricultural Work ................................ ................................ .............. 181 The Matter of Wealth ................................ ................................ ............................ 182 The Role of Gender ................................ ................................ .............................. 184 Achievements and Failures of the PLAR Project ................................ .................. 186 Conclusion ................................ ................................ ................................ ............ 188 LIST OF REFERENCES ................................ ................................ ............................. 191 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 201
9 LIST OF TABLES Table page 3 1 Geographical coordinates of Lugango, Mutsulio, Shikusi, and Shinakotsi villages ................................ ................................ ................................ ............... 73 3 2 Agricultural Luhya terms and their English translations ................................ ...... 73 5 1 Key of terms used in multiple regression analyses ................................ ........... 135 5 2 Multiple regression analysis for Y = maize yield per acre cropped in maize and beans ................................ ................................ ................................ ......... 136 5 3 Multiple regression analysis for Y = bean yield per acre cropped in maize and beans ................................ ................................ ................................ ................ 137 5 4 Multiple regres sion analysis for Y = total fertilizer per acre cropped in maize and beans ................................ ................................ ................................ ......... 138 5 5 Multiple regression analysis for Y = total fertilizer per acre cropped in maize and beans ................................ ................................ ................................ ......... 139 5 6 Multiple regression analysis for Y = oxen use per acre cropped in maize and beans ................................ ................................ ................................ ................ 140 5 7 Multiple regression analysis for Y = oxen use per acre cropped in maize and beans ................................ ................................ ................................ ................ 141 5 8 Multiple regression analysis for Y = hired labor per acre cropped in maize and beans ................................ ................................ ................................ ......... 142 5 9 Multiple regression analysis for Y = hired labor per acre cropped in maize and beans ................................ ................................ ................................ ......... 143 5 10 Multiple regression analysis for Y = maize seeding per acre cropped in maize and beans ................................ ................................ ................................ ......... 14 4 5 11 Multiple regression analysis for Y = maize seeding per acre cropped in maize and beans ................................ ................................ ................................ ......... 145 5 12 Multiple regression analysis for Y = bean seeding per acre cropped in maize and beans ................................ ................................ ................................ ......... 146 5 13 Multiple regression analysis for Y = bean seeding per acre cropped in maize and beans ................................ ................................ ................................ ......... 147 5 14 Multiple regression analysis for Y = wealth index ................................ ............. 148
10 5 15 Multiple regression analysis for Y = total acres owned ................................ ..... 149 5 16 Multiple regression analysis for Y = total cattle owned ................................ ..... 150 5 17 years of schooling ................................ ................................ ............................. 151 5 18 Multiple regression analysis for Y = kale acreage ................................ ............ 152 5 19 Multiple regression analysis for Y = number of m onths per year maize is purchased ................................ ................................ ................................ ......... 153 5 20 General farming calendar for western Kenya ................................ ................... 154 5 21 stern Kenya ................................ .................... 155 5 22 ................................ ...................... 155 5 23 ................................ ........ 156 5 24 ................................ .... 157 5 25 ................... 158 5 26 Calculation 1: Feasibility of Tithonia diversifolia vs. diammonium phosphate (DAP) ................................ ................................ ................................ ................ 159 5 27 Calculation 2: Feasibility of three alternatives for obtaining a sack of maize ... 160 5 28 Calculation 3: Feasibility of maize production through a comparison of input and output prices during 2004 and 20 07 in Shinyalu Division .......................... 160
11 LIST OF FIGURES Figure page 3 1 Road map indicating location of Shinyalu, adapted from Nelles Map: Kenya (Source: Nell es Verlag 2005) ................................ ................................ .............. 74 3 2 Legend for road map indicating location of Shinyalu, adapted from Nelles Map: Kenya (Source: Nelles Verlag 2005) ................................ ......................... 75 3 3 Map of Kakamega District indicating location of Shinyalu Division ( Source: United Nations 2009) ................................ ................................ .......................... 76 5 1 Frequency distribution of crops sold by all farmers ................................ ........... 161 5 2 Frequency distribution of crops sold by project participants and non participants ................................ ................................ ................................ ....... 161 5 3 Frequency distribution of crops sold by male h eaded and female headed households ................................ ................................ ................................ ....... 162 5 4 Frequency distribution of household use of remittances by all farmers ............ 162 5 5 Frequ ency dis tribution of household use of remittances by project participants and non participants ................................ ................................ ...... 163 5 6 Frequency distribution of household use of remittances by male headed and female headed households ................................ ................................ .............. 163 5 7 Frequency distribution of most important cash sources for all farmers ............. 164 5 8 Frequency distribution of mos t important cash sources for project participants and non participants ................................ ................................ ......................... 164 5 9 Frequency distribution of most important cash sources for male headed and female headed households ................................ ................................ .............. 165 5 10 Frequency distribution of most important expenditures for all farmers ............. 165 5 11 Frequency distribution of most important expenditur es for project participants and non participants ................................ ................................ ......................... 166 5 12 Frequency distribution of most important expenditures for male headed and female headed households ................................ ................................ .............. 166 5 13 Frequency distribution of social capital among all farmers ............................... 167 5 14 Frequency distribution of social capital among project participants and non partici pants ................................ ................................ ................................ ....... 167
12 5 15 Frequency distribution of social capital among male headed and female headed households ................................ ................................ .......................... 168 5 16 Frequency dist ribution of technological dissemination among all farmers ........ 168 5 17 Frequency distribution of technological dissemination among project participants and non participants ................................ ................................ ...... 169 5 18 Frequency distribution of technological dissemination among male headed and female headed households ................................ ................................ ....... 169 5 19 Frequency distribution of res ponses regarding quality of life, farm, and economy now compared to five years ago among all farmers .......................... 170 5 20 Frequency distribution of responses regarding quality of life now compared to fiv e years ago among project participants and non participants ....................... 170 5 21 Frequency distribution of responses regarding quality of life now compared to five years ago among male headed and female headed households .............. 171 5 22 Frequency distribution of responses regarding farm now compared to five years ago among project participants and non participants ............................. 171 5 23 Frequency distribution of responses regarding farm now compared to five years ago among male headed and female headed households ..................... 172 5 24 Frequency dis tribution of responses regarding economy now compared to five years ago among project participants and non participants ....................... 172 5 25 Frequency distribution of responses regarding economy now comp ared to five years ago among male headed and female headed households .............. 173 5 26 Frequency distribution of open s ................................ .................. 173 5 27 Frequency distribution of open participants ... 174 5 28 headed and female headed households .................. 174 5 29 Relative fre preferences ................................ ................................ ................................ ....... 175 5 30 making of a successful farmer ................................ ................................ .......... 175
13 LIST OF ABBREVIATION S ACT African Conservation Tillage Network AFC Agricultural Finance Corporation AIDS acquired immune deficiency syndrome CARLs countries with abundant rural labor DAP d iammonium phosphate a fertili zer FAO Food and Agriculture Organization FAOSTAT Food and Agriculture Organization Statistical Database FHH female headed household H hypothesis Ha hectare HH household HIV human immunodeficiency virus ICRAF International Centre for Research in Agroforest ry IEA Institute of Economic Affairs IIRR International Institute of Rural Reconstruction IMF International Monetary Fund KARI Kenya Agricultural Research Institute KFA k g kilogram KIT Royal Dutch Institute for Tropical Agricultu re KWFT Kenya Women Finance Trust MHH male headed household MoARD Ministry of Agriculture and Rural Development NCPB National Cereals and Produce Board
14 OFW off farm work P P value P2O5 phosphorous pentoxide, a f ertilizer PLAR Participatory Learning and Act ion Research Q quantity SAPs structural adjustment programs SID Society for International Development UNAIDS The Joint United Nations Programme on HIV/AIDS USAID United States Agency for International Development WHO World Health Organization X independent variable Y dependent variable
15 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy AGRICULTURAL EXTENSIFICATION IN THE WE STERN HIGHLANDS OF KENYA: IMPACTS ON MAIZE PRODUCTION AND ADOPTION OF SOIL FERTILITY ENHANCEMENT PRACTICES By Maria C. Morera December 2010 Chair: Christina H. Gladwin Major: Anthropology So il fertility enhancement is a form of agricu ltural intensifica tion W ithout sufficient returns to agricultural investment i ncluding labor farmers will not bear its costs. I f necessary, farmers will search for alternatives to agriculture. This study set out to identify factors a management p ractices in western Kenya and gauge trends in agricultural productivity. In doing so, the study addressed the following questions. Is agriculture in western Kenya intensifying or extensifying? Does off farm work lead to an expansion or a contraction of on farm inve stments? Does limited resource access lead to labor affect tech nological choice? Do agricultural extension efforts to disseminate soil fertility enhancing technologies have lasting effe cts on farming practices? How do factors originating outside the farm at local and national levels affect farming decisions? fertility manageme nt in rural western Kenya in its attempt to view agricultural decisions from the perspective of the farmer while locating those decisions within a broader context. Using ethnographic field methods and applying both
16 quantitative and qualitative analyses th e study tested the interrelation of variables and improved fall maize production. The study found that larger landholders in western Kenya, rather than incre ase the productivity of staple crops, substitut e cash crop production for maize and bean production while land poor farmers, rather than intensify food production, apply their labor towards agricultural wage work or nonfarm work. Thus larger landholders do not invest their wealth in inorganic soil fertility enhancements, such as mineral fertilizers, while smallholders do not invest their labor in organic soil fertility enhancements, such as green fertilizers or improved fallows. On average, farmers substitu te land for fertilizer. Meanwhile, soil fertility enhancement and aggregate agricultural productivity remain low in this region, calling into question the efficacy of resource conservation and agricultural development efforts and pointing to a need for alt ernate food policies if economic structural transformation the transformation of an agrarian economy to a diver sified economy is to take place in Kenya
17 CHAPTER 1 INTRODUCTION: THE R OOTS OF EXTENSIVE AG RICULTURE In 1985, Piers Blaikie wrote that soil e rosion is not purely an environmental issue; it is also a political economic issue. Individuals, not environmental forces, accelerate and intensify soil erosion through practices that are influenced by the local, national, and global structure of society. The deeper causes and wider consequences of soil erosion are located within a broad context. In turn, not only the environment, but human populations as well, are adversely affected by soil erosion. Blaikie and Brookfield (1987: 1) suggested that land d egradation, by definition, is a social problem insomuch as Thus, to understand the fac tors that affect soil degradation one must inevitably examine the factors that impact human behavior. Often, this involves investigating the political ecology of farming decisions. A political ecology perspective takes into account both political economic and human ecological analyses when addressing natural resource management (St onich 1993). It examines the interacting roles that social institutions (international, national, regional, and local) play in providing constraints and possibilities that affect human decisions that in t urn affect those institutions as well as the natura l environment" ( Stonich 1993: 25). Because soil degradation is often linked to agricultural p ractices, a political ecology approach to the study of soil management entails researching the opportunities and constraints faced by farmers. This requires looking at the social factors shaping farming decisions, as well as the economic and pol itical policies at the local national, and international levels that
18 set the stage for these trends. Human resources and physical infrastructure also impinge upon farming decisions and indirectly affect soil management. The impo rtance of a political ecology approach is that location specific solutions to soil degradation do not work if broader location non specific politica l and socioeconomic factors are not taken into consideration (Stonich 1993; Blaikie and Brookfield 1987; Blaikie 1985). For example, research indicates that even when agricultural development and extension programs widely disseminate soil conservation tec hnologies to farmers, macroeconomic disincentives to agricultural intensification essentially nullify those efforts (Morera and Gladwin 2006; Morera 1999). Similarly, the present study will illustrate that extensive research, development, an d disseminatio n of various soil fertility enhancement technologies in western Kenya have been a failure because soil infertility is not the sole factor limiting increased agricultural production. Many farmers cannot afford even the labor costs of ameliorating soil infe rtility through organic methods and low maize prices render unfeasible the rising costs of inorganic fertilizer. Without incentives for agr icultural intensification, soil fertility enhancement technologies are for naught because farmers rather plant exten sively with minimum applic ation of inputs, including soil fertility enhancement practices. Ironicall y, it has been suggested that soil fertility depletion is the chief cause of declining per capita food production in Africa (Sanchez et al. 1997 ). That is, even when other conditions such as seed production, research and extension services and enabling government policies are remedied, agricultural productivity continues to decr ease unless soil fertility depletion is addressed (Sanchez et al. 1997: 3). Howe ver, the presen t study will indicate that soil fertility is just one of several causes of declining
19 per capita food production in Kenya Even when soil fertility is addressed, as it has been in areas where agricultural extension projects have temporarily ach adoption of soil fertility enhancement practices, a disabling policy environment will effectively discourage ongoing increased agricultural production. Because soil fertility enhancement is a form of agri cultural intensification without sufficient returns to agricultural investment including labor farmers will not bear its costs. If necessary, farmers will search for alternatives to agriculture. This study explores the context of soil management in Kenya. It addresse s the reasons why f armers in Kenya have incentives to degrade their soils. Perhaps more importantly, the study also addresses the reasons why agricultural productivity in Kenya The impetu s for the study was ignited by a host of publications in which a strong case was made for the replenishment of soil nutrient deficiencies across Africa through the application of improved fallow and biomass transfer systems (Place et al. 2001; Franzel et a l. 2000; Pisanelli et al. 2000; Sanchez 1999; Kwesiga et al. 1999; Swinkels et al. 1997; Buresh et al. 1997). Improved fallow systems were built on African farming techniques by improving traditional fallows with fast growing tree or shrub legume species, such as Sesbania sesban, Crotalaria grahamiana, and Tephrosia vogelii which fix nitrogen from the air and return it to the soil. Biomass transfer systems were also built on African farming by utilizing hedges around homesteads and traditional live fences of common species such as Tithonia diversifolia and Lantana camara for green manure. Together, these systems could provide small holders with labor intensive fertility enhancing alternatives to cash expensive industrial,
20 inorganic f ertilizers. The technologies were extensively researched and disseminated in western Kenya during the 1990s by the International Centre for Research in Agroforestry (ICRAF). wi th improved fallow and biomass transfer systems were compelling. Nonetheless, my own work in Honduras in 1997 (Morera 1999) led me to conclude that economic constraints, particularly resulting from structural adjustment programs (SAPs), inhibit smallholde rs from investin g in labor intensive, fertility enhancing technologies despite their proven capacity to ameliorate soil degradation and raise maize yields. Moreover, economic trends in Kenya indicated that the production growth of maize remained lower tha n population growth, rendering maize an importable commodity even after liberalization from 1992 onward (FAOSTAT 2004). With all of this in mind and amidst a and corre spondingly low agricultural productivity, I embarked on my own investigation of farming patterns among smallholders in western Kenya. management practices in western Kenya and to gauge trends in agricultural productivity. In doing so, the study addressed the following guiding questions: 1) Is agriculture in western Kenya intensifying or extensifying? 2) Does off farm work lead to an expansion or a contraction of on far m investments? 3) Does limited resource access, particularly to land and cash, lead to labor practices? 4) How does gender affect technological choice? 5) Do research and development efforts to disseminate soil fertility enhancing technologies have lasting
21 effects on farming practices? 6) How do factors originating outside the farm at both the community and national levels affect farming decisions? In order to address these research questions, the study gathered both emi c etic Kenya in its attempt to view agricultural decisions from the perspective of the farmer while locating those decisions within a general context. Using ethnographic f ield methods and applying both quantitative and qualitative analyses the study tested six hypotheses: H1: Farmers in western Kenya are intensifying agricultural production by increasing the availability of soil nutrients to their staple crops, thus increa sing crop yields per area. H2: Nonfarm income generation leads to increased on farm agricultural investments while agricultural off farm income generation leads to decreased on farm agricultural investments. H3: The wealthier a household, the more it inv ests in agricultural productivity. H4: There is no difference between male headed and female headed household investment in one or more of the following: kg/acre of green manure, square meters/acre of improved fallows, kg/acre of mineral fertilizer, days /acre of hired oxen, days/acre of hired labor, and/or kg/acre maize and bean seed. H5: There is no difference in soil fertility enhancement practices among farmers living in nearby villages where labor intensive soil fertility enhancement technologies hav e been disseminated in some villages and not in others. H6: If the international price of oil rises and driving time between eastern and western Kenya increases, the price of imported inputs (i.e. fertilizer) will rise faster than the price of outputs (i. e. maize) and farmers will not invest in the intensification of their maize production through the application of fertilizer. In the chapters ahead the results of the study will largely support all but the first hypothesis and indicate that in western Keny a, land poor farmers, rather than intensify production, apply their labor towards agricultural wage work or nonfarm work while larger landholders, rather than increase agricultural productivity, substitute kale
22 production for maize and bea n production Th us smallholders do not invest their labor in organic fertilizers or improved fallows while larger landholders do not invest their wealth in inorganic fertilizers. On average, farmers substitute land for fertilizer. Meanwhile, soil fertility enhancement a s well as aggregate agricultural productivity remains low in this region of western Kenya, calling into question the efficacy of resource conservation and agricultural development efforts and pointing to a need for alternate food policies if economic struc tural transformation the transformation of an agrarian economy into an industrial economy -is to take place in Kenya. Chapter 2 sets the stage of the study by synthesizing and analyzing the literature on agricultural productivity and soil fertility managem ent. The chapter illustrates that without incentives for agr icultural intensification, soil fertility enhancement technologies are for naught because farmers rather plant extensively with minimum applic ation of inputs, including soil fertility enhancement practices. It argues that if food prices fail to rise, marginal producers will not invest in the increased costs of intensification which, essentially are the costs associated with fertility enhancement. Instead, they will cultivate until they degrade t he land and must abandon the area or abandon agriculture. Chapter 3 explores the region of western Kenya, noting the characteristics of the land, customs of its people, and nature of the various institutions and projects engaging it. By weaving primary data collected through participant observation together with secondary data collected from published materials, the chapter presents a current, as well as historical, view of the ecological and socioeconomic features characterizing the ting. In doing so, the chapter provides a context in which the
23 theoretical framework, research methods, and empirical results of this study can be understood. Chapter 4 discusses the motivations, assumptions, objectives, hypotheses, ethnographic technique s, and analytical tools involved in carrying out the empirical work of this study. Using ethnographic field methods and combining both quantitative and qualitative analyses, the study examines the nature of soil fertility management in western Kenya, inve stigating its role in agricultural productivity and exploring how it is affected by technological adoption, gender, off farm work, and other off farm factors The study seeks to establish whether past institutional efforts to disseminate improved fallows whether o ff farm work and gender affect soil fertility enhancement practices and, in turn, agricultural intensification. Chapter 5 presents the results of quantitative and qualita tive analyses of ethnographic data collected from 120 households located throughout the villages of Mutsulio Shikusi Lugango and Shinakotsi in western Kenya. In doing so, the chapter addresses the six research questions and hypotheses raised at the lau nch of the study concerning agricultural productivity, soil fertility management, technological adoption, and the role of gender, off farm work, and other off farm factors on these. It provides the evidence that is use d in Chapter 6 to reject the first hy pothesis while supporting the rest. Chapter 6 discusses the five major findings of the research study and examines their implications. Because results indicate that farmers in this part of western Kenya are producing maize and beans extensively through th e application of inadequate
24 amounts of soil fertility enhancement measures, the study concludes that farmers are degrading their lands and failing to increase the agricultural productivity of their staple crops. The study also concludes that no amount of a gricult ural extension can achieve long term farmer adoption of soil fertility enhancement practices when macro po licies provide disincentives to agricultural intensification. However, the consequence of staple crop extensification is low aggregate agricult ural productivity and delayed economic development. Because raising aggregate agricultural productivity remains the most important development strategy for agrarian economies increased efforts to intensify staple crop production will be necessary if stru is to be achieved.
25 CHAPTER 2 THEORETICAL FRAMEWOR K: KENYAN SOIL FERT ILITY MANAGEMENT IN CONTEXT Introduction This chapter explores the context of soil management in Kenya Incorporating a political ecology analyti cal framework, it examines the interrelated processes at local, national, and international levels that contribute to soil fertility depletion, identifying the reasons why farmers in Kenya have incentives to degrade t heir soils. The chapter discusses the effects of soil degradation on agricultural productivity, explains the importance of agricultural productivity to economic development, and reviews key factors that affect agricultural land manageme nt and production. In addressing the obstacles to soil fe rtility replenishment and agricultural intensification in Kenya, the chapter also addresses the issues that Effects of Soil Degradation on Agricultural Productivity Soil quality directly affects crop yields by l imiting the amount of moisture and nutrients plants can derive from the soil. Soils are composed of inorganic particles, organic matter, air, and water. The inorganic particles comprise larger mineral fragments which are imbedded in finer c olloidal mater ials (Brady 1974: 40). The ratio between the larger and finer materials determines whether the soil is gravelly or sandy, or whether it is clayey (40). This relative proportion of particle sizes is known as soil ndy soil remains sandy and clay soil remains aggregates and can be altered by farm practices such as tillage, cultivation, liming, and manuring (40). Organic matt er
26 matter, soil structure, and soil texture all affect the pore space which is occupied by air and water. Soils degrade in a variety of ways, either by soil particle movement or soil nutrient depletion (Eionet). Natural degradation results from wind, river, and sheet erosion as well as landslides and nutrient leaching (IIRR and ACT 2005). Human induced degradation accelerates these natural processes through human activities. For example, tillage disturbs soil structure and increases the likelihood of nutrient leaching (IIRR and ACT 2005). Hills ide farming increases the likelihood of landslides and runoff, speeding up erosion. Continuous cultivation depletes the soil of its nutrients. In western Kenya, human induced soil degradation usually results from continuous manual agricultural cultivation on small farms in two ways. First, whether land is tilled with a hoe by hand or with a plow by oxen, vegetation and crop residues are turned into the soil (or they are burned or used for fodder), accelerating the decomposition of organic matter. Because organic matter helps soils retain water, nutrients, and organisms, the elimination of organic matter hastens water evaporation, leaching, erosion, weed and pest infestation, and the destruction of soil structure (IIRR and ACT 2005). Moreover, hoeing and plowing both result in soil compaction and also destroy soil structure (IIRR and ACT 2005). Second, continuous cultivation mines the soil of nutrients. While nutrient depletion is location specific and depletion rates vary with soil properties (Sanchez et al. 1997), monocropping robs the soil of the same nutrients year after year, resulting in specific deficiencies. Throughout western Kenya, where the majority of smallholders mostly
27 cultivate maize and beans, soils are particularly deficient in nitrogen, p hosphorous, and potassium. Acc ording to Sanchez et al. (1997: Busia, and Kisumu Districts are severely deficient in P (<5 mg bicarbonate extractable P kg 1 soil), and most are deficient in N when P deficiency is o production is particularly constrained by nitrogen and phosphorous deficiencies, maize yields can drop by more than 30% over a 20 year period without nitrogen and phosphorous inputs (Sanchez et al. 1997; Qureshi 1991). In the pas t, traditional farming throughout this region normally included land fallowing. However, land is commonly transferred through inheritance resulting in shrinking farms and decreased fallows over time. Moreover, a rising population growth rate resulting in 500 1200 people per kilometer throughout the highlands of western Kenya (Sanchez et al. 1997) has resulted in even smaller landholdings, often making it difficult to take land out of cultivation altogether. Thus, crop yields cannot be sustained without ad equate soil fertility replenishment, outside a farm. If the inputs come from farm, it is merely nutrient cycling and not replenishment (Sanchez et al. 1997). Mineral fertilizers provide the highest concentration of nutrients on a dry weight basis. They can be easily transported to and dispersed throughout the field. Yet mineral fertilizers are commonly at least twice their international price throughout rural Africa (Sanchez et al. 1997). Poor infrastructure increases transport costs, as do taxes and import duties.
28 Over the last decad e, a variety of organic input options have been researched and developed, providing farmers with alternatives to mineral fertilizers. Yet their nutrient concentration is far lower than mineral fertilizers on a dry weight basis. This means that very large quantities of organic fertilizers are necessary to provide adequate nutrient replenishment and they can be difficult to transport to and disperse throughout the field. Improved fallow systems are another option that has been extensively researched and dis seminated. Fast growing species of leguminous trees and shrubs, such as Sesbania sesban Crotalaria grahamiana and Tephrosia vogelii have been tested for smallholders as a substitute for nitrogen fertilizers (Amadalo et al. 1998). Because the legumes f ix nitrogen from the air and transport it to the soil through its roots, they provide farms with an external, not recycled, source of nitrogen. Improved fallows involve planting the fast growing shrub or tree species into cropland that is fallowed for at least nine months, including two rainy seasons. The selected species enhance soil fertility by either bringing up nutrients from deeper soil levels and/or by fixing atmos pheric nitrogen (Amadalo et al. 1998). Nonetheless, Quiones et al. (1997) of the S asakawa Global 2000 Project maintain that mineral fertilizer use will remain pivotal to increased productivity across Africa while low input technologies, such as organic fertilizers and improved fallows, can only be supplemental. Organic fertilizers, the y argue, can increase food production by 2% at best (Quiones 1997:83; Hayami and Ruttan 1985). Yet agricultural productivity in Africa needs to increase by 5 6% annually in order to exceed its population rate of 3% and assure food security (Quiones 1997 :83).
29 Data collected in this study show that small scale farmers in western Kenya are investing in neither mineral fertilizers nor replenishment alternatives to mineral fertilizers. They plant a minimum amount of maize seed over a broad area and apply onl y minimal quantities of mineral fertilizer. Essentially, they are substituting land for fertilizer and practicing extensive agriculture Extensive agriculture can be understood as a food supply system in which a minimum amount of crop is harvested per uni t of land, or in which a minimum amount of labor and capital is invested per unit of crop produce d. According to Boserup (1981: 18), at its least intensive, a food supply system may simply comprise an area used only for gathering. An area planted with one or two successive crops followed by a 15 25 year fallow would represent a more intensive form of land use. If the fallow was shortened to 5 years, the area would then represent a still more intensive form of land use. ich the land is 13). Boserup (2005: 12) contrasts her perspective with the classical economic conception of agricultural expansion whereby agricultural output is raised in production at the so called extensive margin, by the creation of new fields, and the expansion of production by more intensive cultivation of ex 12) argues that the classical economic conception of agricultural expansion, whic h inherently distinguishes between cultivated and were writing at a time when the almost empty lands of the Western Hemisphere were gradually taken under cultivation by Eur opean settlers, and it was therefore natural that they should stress the importance of the reserves of virgin land and make a sharp
30 special conditions for agriculture in th e Western Hemisphere in preceding centuries cannot be applied to the current conditions for agriculture in developing countries where methods in 005: 13). Thus Boserup dismisses from her definition of agricultural growth the classical economic implication that production can be expanded at the so called extensive area of land and time and the latter refers to a food system of lesser yields per area of land and time. That is, extensive agriculture no longer refers to a metho d of increasing agricultural production. More recently, extensive agriculture, or extensification has been defined as a less intensive use of farming that, using fewer chemical fertilizers and leaving uncultivated areas at the edges of fields, allows low er yields from the same area of farmland (AgricultureDictionary.com). According to Netting (1993: 28), the key characteristic of intensive agriculture is the manipulation of nutrients, water, and sunlight for increased supplies that support more biotic grow th for longer periods of time. Naturally, this involves replenishing elements that become exhausted. It is a means of coping with limite d space and time (Netting 1993: 102 ). Nonetheless, Boserup (1981: 15), and later Netting (1993: 56), were careful to dis tinguish between intensive/extensive and traditional/modern. That is, a very traditional food supply system involving the simplest of tools can be highly intensive while a very modern system involving highly sophisticated technology, such as tractors
31 and combine harvesters, can be extensive a nd even wasteful (Netting 1993: 56). Moreover, lumping together an intensive short fallow system and an extensive long between de mographic conditions and method s of food supply (Boserup 1981: 15). According to Boserup (1981), extensive agriculture is a feature of food supply systems characterized by abundant land and/or sparse population. However, western Kenya has neither abundant land nor sparse population. Instead, western Kenya is characterized by one of the highest population densities in the world, a shrinking land supply, and annual cropping/multi cropping. Naturally, it is now also characterized by declining fertility. Yet curiously, it remains uncharacterized by the use of fertility enhancement technologies. Nonetheless, Boserup (1981: 26) believed the problem of soil fertility would be as in high technology countries; and (3) labor intensive practices, as in densely populated al change is and income distribution that are associated with economic growth (Binswanger et al. 1978: 3; Hayami and Ruttan 1970). In the case of western Kenya, it follow s that a shrinking land base resulting from population pressure, and declining yields resulting from continuous cultivation, should motivate a technological shift in agriculture towards either labor intensive fertility enhancement practices (e.g. organic f ertilizers and improved fallows) or mineral fertilizer use. Yet Blaikie and Brookfield term this line of
32 variety of social and ecological circumstances and arises even in cases of decreased population densitie s (Blaikie and Brookfield 1987: 4). In turn, the issue of extensification in western Kenya not only calls into question the applicability of the theory of induced innovation, but also calls into question the efficacy of the Green Revolution. While Norman Borlaug firmly maintains that increased agricultural productivity in Africa cannot be achieved without mineral fertilizers (Thurow and Kilman 2009; Quiones et al. 1997), it remains unclear how this will take place. Af rica has the lowest rates of use in the world (Morris et al. 2007; Crawford et al. 2006; FAOSTAT). In 2002, fertilizer use in sub Saharan Africa averaged 8 kg/ha compared to 78 kg/ha in Latin America, 101 kg/ha in South Asia, and 96 kg/ha in East and Sout heast Asia (Morris et al. 2007: to its weak bargaining position as a low volume importer and to the high cost of transportation resulting from poor infrastructure (Quiones et al. 1997). More over, the Green Revolution itself has been criticized for its inability to benefit smallholders (Shiva 1991). Nonetheless, from 1962 to 1982, fertilizer use in sub Saharan Africa grew at the relatively same rate it did in both Latin America and East and So utheast Asia 1 (Morris et al. 2007: 17). Only after 1982 did fertilizer use in sub Saharan Africa plummet (Morri s et al. 2007). Gladwin (1991: 197) argues that SAP mandated fertilizer subsidy removal programs implemented throughout sub Saharan Africa during the 1980s resulted in 1 Fertilizer use grew annually at 8.71% in sub Saharan Africa, 7.70% in Latin America, 7.64% in East and Southeast A sia, and 13.19% in South Asia during 1962 1982; however from 1982 to 2002, it grew at 0.93% in sub Saharan Africa, 3.06% in Latin America, 3.39% in East and Southeast Asia, and 4.99% in South Asia (Morris et al. 2007: 17; FAOSTAT).
33 decreased fertilizer use because smallholders, unable to overcome cash constraints and access credit markets, could not afford fertilizer at higher prices. However, Minot (2009) points out that fertilizer subsidy removal programs in sub Saharan Africa had mixed results during the 5 year period following subsidy removal: fertilizer use decreased in some countries (Nigeria, Ghana, Cameroon, Senegal, and Tanzania) and increased in others (Benin, Togo, Mali, and Madagascar). In Kenya, where fertilizer distribution was more market based than in other sub Saharan countries, fertilizer consumption continued to increase despite liberalization in the early 1990s (Minot 2009; Crawford et al. 2006 ). Minde et al. (2008: 18) report that small sc ale farmers across western Kenya are currently using fertilizer on maize at dose rates of roughly 163 kg/ha annually, comparable to fertilizer use throughout Latin America and Asia. Yet data collected in this study indicate that small scale farmers across 4 villages in western Kenya applied fertilizer on maize during the long rains season of 2007 at an average rate of 69 kg/ha. It is unlikely that their annual average dose rate approaches the amount reported by Minde et al. (2008). Despite the controversy policy makers have recently reconsidered fertilizer subsidies. Minot (2009) attributes the renewed interest to four factors: 1) Jeffrey Sachs (2005) recommends localized, temporary fertilizer subsidies as a way to jumpstart small scale agricultural pro duction, particularly in villages where disease and hunger make it difficult for producers to climb out of a poverty trap. 2) In Malawi, where fertilizer markets were never fully liberalized, its Agricultural Input Subsidy Programme is credited with maki ng the country an exporter of maize in 2005. 3) In 2006, the case in favor of fertilizer subsidies was brought up during the Africa Fertilizer Summit held in
34 Abuja, Nigeria (Morris et al. 2007). 4) In 2008, the upward spike in oil, fertilizer, and food prices focused attention on food production. Nonetheless, Blaikie and Brookfield argue that the remedies for land degradation quently make assumptions about the way in which a society operates and changes in discussing the causes and implicati ons vity. Ultimately, however, there may simply be more profitable alternatives to intensive small scale farming on degraded land. agriculture to move onto marginal land, the price of food must rise to cover the increased costs of production, resulting in an unearned income for labor inputs on land t process applies to degraded land which, of course, has become marginal land. It has a lower productivity potential that must be addressed with higher labor and/or capital inputs than superior land. Blaikie and Brookfield ( 1987: therefore, directly consumes the product of labour, and also consumes capital inputs into production; other things being equal, the product of work on degraded land is less than that on the same land without degradatio Arguably, if food prices fail to rise, marginal producers will not invest in these increased costs which, essentially, are the costs associated with fertility enhancement. Instead, they will cultivate until they degrade the land and must abandon the a rea or
35 abandon agriculture. Boserup (1981) suggests that without agricultural intensification, densely populated areas must rely on imported food. And that is what is occurring in many African countries where the price of imported maize is lower than th e price small scale farmers can fetch for their marginally produced maize. Farmers are degrading their land and finding alternatives to farming. Meanwhile, aggregate agricultural productivity remains low. Consequences of Low Agricultural Productivity At the household level, low agricultural productivity can translate into higher levels of malnutrition and illness, lower levels of education and health care, and increased urban migration to slum areas already teeming with people and disease. Without a surp lus to sell, subsistence farmers have few opportunities for generating the cash needed to pay for additional foodstuffs, agricultural investments, medications, and school fees. Low agricultural productivity renders individuals food insecure and vulnerable (Roberts 2008). Their margin of survival is smaller than that of surplus producers during periods of crisis brought on by drought, war, and pandemics (Tomich et al. 199 5; Blaikie and Brookfield 1987: 2). Moreover, the importance of agricultural productivi ty extends beyond the farm. At the national level, low aggregate agricultural productivity leads to reduced agricultural exports and also leads to a demand for food that requires food imports. Decreasing exports and increasing imports both lead to reduct ions in foreign exchange and compound the fiscal crises most sub Saharan countries are already grappling with (Ayittey 2005). While some economists argue that food imports, per se are not d etrimental (Timmer et al. 1990: 272), they require already scarce f oreign exchange. Multiplying this
36 effect, the opportunity to earn foreign exchange through the sale of surplus food is lost. Again, some economists argue that this is not prob lematic (Thomson and Metz 1997: 22). The theory of comparative advantage, first proposed by David Ricardo, suggests that countries fare better economically by specializing in the commodities they produce best while trading freely for those commodities produced more efficiently in other countries (Roberts 2008:115; Thomson and Metz 19 97: 22). Yet Lewis (1954) notes that many developing countries have been wrongly advised to allow their fledgling industries to be destroyed by cheap imports because the law of comparative costs is only valid if written in marginal, and not average, terms. produce 3 units of food in a more developed country and 1 unit of food in a less developed country. Yet higher wages in the more developed country will render food ess developed country, food production (Lewis 1954). Another problem with the application of the theory of comparative advantage is that the majority of farmers in Kenya for example, know best how to grow maize a commodity that is under priced in the international market as a result of subsidies originating in the U nited States European Union, and Japan. For instance, in 2005 American taxpayers covered the difference b etween the $1.85 per bushel world market price of corn and the $3 per bushel it cost the U nited S tates to grow it course, only rewarded [American] farmers for overproducing in 2008: 121; Hanrahan et al. 2006). The low international price of maize acts as a disincentive for maize production in countries where IMF and World Bank lending have
37 restricted governments from providing their farmers those same subsidies. On the other hand, growing alternate crops is problematic in developing countries where agricultural extension, infrastructure, and market demand are limited. For example, in implementing structural adjustment programs (SAPs) during the 1980s, most of Central America followed the agricultural development advice provided by the United States Agency for International Development (USAID) which included financial support for the development of new non traditional agricultural commodities, such as broccoli, melons, and flowers (Stonich 1993). Although Central America producers had no experience farming these cash crops nor was there any prior infrastructure support for their development, USAID provided the funds for their promotion in order to facilitate an increase in agricultural exports throughout the region. Ye traditional crops throughout Central America lay in the U S Bumpers Amendment to the Foreign Assistance Act of 1961 which prohibits USAID from promoting abroad the development of a crop for expor t that competes wi th a similar crop grown in the United States assistance to its producers would have naturally benefited the greatest portion of the rural sector. However, US AID could not promote basic grain production for export in Central America because the U nited S tates itself is a major basic grain producer. The resulting regional emphasis on new non traditional cash crops replaced support for basic grain production despi te its inability to benefit the majority of Central scale producers were unable to substitute basic grains with cash crops because the international market places great emphasis on
38 quality control a requirement that is di fficult for small scale producers to meet (Farnsworth et al. 1996). Additionally, the infrastructure was limited and the market for the new crops inadequate and unstable. Once produced, many farmers had trouble both transporting the crops to market and s elling them. Thus, it is doubtful that basic grain producing populations in developing countries will easily switch to cash crops particularly those not consumed domestically or increase basic grain productivity in the face of low international prices enab led by subsidies in developed countries. Indeed, agricultural growth over the last half century has been slowest in Africa (Anderson and Masters 2009: predominantly agricultural countries are located. In Kenya, agriculture gre w at a rate of 2.3% annually from 1980 to 2 004 (Anderson and Masters 2009: 9) yet its total population grew at an average rate of over 3% annually during the same period (Winter N elson and Argwings Kodhek 2009: 257). Thus food production growth rates did no t keep up with population growth rates. Clearly, increased agricultural productivity remains essential to development. Tomich et al. (1995) argue that developing countries must develop their subsistence sectors in order to industrialize. In order to reac h the structural transformation turning point, the point "when the absolute size of the agricultural work force begins to decline," developing countries must increase low agricultural labor productivity (output per farm worker) while simultaneously absorbi ng a significant proportion (4 6%) of the total labor force into the non agricultu ral sector (Tomich et al. 1995: 9). Tomich et al. (1995: 2) illustrate this point by explaining that countries with abundant rural labor (CARLs) have a large agricultural sec tor, where more than 50% of
39 the total population is employed, and a small non agricultural, industrial sector. For 2004 in Kenya was 75%. This means that the industrial capitalist sector is small and t he potential for profits, savings, and investment, which lead to capital incre ase, is also small (Lewis 1951: 12). In turn, limited. Tomich et al. (1995: 14) clarify this by i ndicating that the rate of growth in the force and by the rate of growth in the total labor force. Thus the key point is that the rate of growth in the nonagricultura l workforce can be stagnant for decades if most of the total labor force is predominantly in agriculture and population rates are high. To overcome these constraints, agricultural productivity must rise for a country to transform its predominantly rural e conomy to a predominantly industrial economy 2 This also explains why the structural and demographic characteristics of a developing country constrain its developme nt options (Tomich et al. 1995: 2). Tomich et al. (1995) suggest that the key to encouraging this shift is investment in agriculture and agricultural specialization. Bates (1981) also argues that the key to development lies in providing proper incentives for farming. Similarly, Arthur Lewis (1954) notes industrialization is dependent on agricul tural improvement. He argues that i t is not profitable to produce a growing volume of manufactures unless agricultural production is growing simultaneously. This is also why industrial and agrarian revolutions always go together, and why economies in wh ich agriculture is stagnant do not show industrial development. 2 Tomich et al. (1995:14 ) express the timing of the structural transformation point with the identity L a L t L a ) 1/( L a / L t ) + L n where L a represents the rate of growth in the agricultural labor force, L t represents the rate of growth in the total labor force, L n represents the rate of growth in the nonagricultural work, and L a / L t represents agri
40 Hence, if we postulate that the capitalist sector is not producing food, we must either postulate that the subsistence sector is increasing its output, or else conclude that the expansion of the capitalist sector will be brought to an end through adverse terms of trade e ating into profits [Lewis 1954:20] In sum, the consequence of low aggregate agricult ural productivity in developing countries is delayed development. Other Factors Affecting A gricultural Productivity Therefore, aggregate agricultural productivity is the most important factor affecting development while soil fertility is but one of the many factors affecting agricultural productivity. Yet ultimately, both agricultural productio n and soil fertility management are determined by price signals. In turn, price signals are affected by macroeconomic policies, trade and export production policies, investment in agricultural research and extension, infrastructure, and b anking. Timmer et al. (1990: unfavorable macroeconomic environment will ultimately erode even the best plans for plex relationships Brookfield 1987: 17) is indispensable in understanding trends in agricultural land management and production. Macroeconomic Policies Neo liberal economic theory holds that government policy alterations of the so wage rates, interest rates, land rental rates, foreign exchange rates, and rural urban terms of trade can result in distorted prices that fail to signal the actual abundance (or scarcity) of domestic factors of production to farmers (Timmer et al 1990). Particularly during the 1960s and 1970s, governments in developing
41 consumption at the expense of rural production. For example, governments adjusted foreign exchange r ates because overvalued domestic currencies resulted in cheaper imports, needed or desired in urban sectors, even if they taxed the agricultural sector by bringing down the domestic price of food. Similarly, they subsidized the development of the industri al sector at the expense of the agricultural sector by providing inadequate prices (below the international value) to growers for their commodities in order to finance the formation of domestic manufacturing firms (Bates 1981). Normally, an expansion of th e industrial sector increases the demand for food and, in turn, the price of food (Lewis 1954: 20). Yet urban employees, averse to relinquishing their wages to high cost food, and industrial firms, averse to raising wages, both pressure governments politic ally in order to mainta in food prices low (Bates 1981: 30). disincentives that in the long run also retard the industrial sector because the industrial sector is neve r stimulated by the internal demand of the large subsistence sector (de Janvry 1981). They retard the backward and forward linkages between the subsistence and industrial sectors which are key to industrial expansion and are facilitated with proper incent ives for farming and agricultural specialization. Export Strategies International trade provides a critical outlet for abundant resources that can be traded for scarce resources. Moreover, exports are the most important source of foreign exchange for any country and integral to its balance of payments. Yet developing countries are often advised to adopt export strategies that do not necessarily support their internal productive and consumptive capacities. Such was the case in Central America when USAID p romoted new non traditional exports that were difficult to
42 produce and to market and had no consumptive value for most of the population. De Janvry (1981) notes that when key sectors only produce export or luxury goods that are not consumed by workers, pr oduction and consumption capacities are not in accord; capitalist profits and workers' wages are not rising and falling in sync. The resulting structural condition is what he terms "social disarticulation." In other words, the consumption needs of the po pulation are not linked to or met by its production; wages are not linked to profits. Emphasizing export production in this manner results in market distortions because scarce resources (credit and extension services) are channeled to a minority of export producers. In turn, finance and technology neglect the needs of the majority of subsistence producers. This results in a bimodal agrarian structure, where a small number of large and mechanized producers crop the majority of total land while most of the population subsists meagerly off a small proportion of total land. Governments in developing countries may favor a bimodal agrarian structure because it is easier for the state to tax output that is concentrated in a few large units than it is to ta x many small units (Lewis 1954: 17). However, a bimodal agrarian structure makes it difficult to move at least 50% of the rural labor force out of agriculture and into industry and services. Only through a broad based strategy can the bulk of the rural sector b e addressed and the subsistence sector developed. For most developing countries, this means addressing financially and technologically a configuration of abundant labor and scarce land. 1971), by drawing on the idea of induced technological innovation introduced by John Hicks
43 hrough market prices, [which steer] in 86). According to Tomich et al. (1995), the contrasting development paths taken towards structural transformation by the U nited S tates innovation hypothesis claims that relative factor prices (the prices of land, labor, and other factors of agricultural production) reflect relative factor scarcities and, in turn, influence patterns of producti vity growth by inducing innovations that save scarce, expensive factors and use abundant, epitomized by the successful yet contrasting patterns of agricultural productivity growth exemplified by the U nited S tates and Japan and their contrasting factor endowments. In short in the 1880s the United States was endowed with abundant land and scarce labor, while Japan was endowed with abundant labor and scarce land. These scarcities were reflected in t he market: in the United States land was inexpensive and labor was expensive while in Japan, land was expensive and labor was inexpensive. Yet technological innovations responded to market price signals in b oth cases. In the United St ates "the scope for cost reduction was greatest through...technology that substituted for hu man labor" (Tomich et al. 1995: 86). In Japan, it was through technology that substituted for land. Hence, in accordance with market price signals, mechanical inn ovati ons in the United States (tractors) and biological innovations in Japan (high yielding, fertilizer responsive crop varieties) emerged, allowing each country to increase agricultural productivity.
44 In Japan, the growth of factor productivity followed a "labor using, land saving pattern of agricultural development," based on intensifying production of sc arce land. In the United States productivity followed a "labor saving, land using pattern of agricultural development" based on expanding production of abu ndant land (Tomich et al. 1995: 81). The success of each of these countries in achieving structural transformation along different development paths implies that there is no single right path to agricultural development because resource endowments may v ary from country to country. Therefore, current day developing countries should avoid the pitfall of adopting an agricultural development pattern that fails to reflect their actual relative factor endowments and scarcities -such as favoring a small export producing segment of the agricultural sector and neglecting the productive capacity of the bulk of the rural labor force. Because most current day developing countries have scarce (and expensive) land and abundant (and cheap) labor, they resemble Japan in its pre structural transformation years. It makes sense that they should use what resources they have and save what resources they lack, thereby supporting the design of labor intensive and land saving agricultural innovations (e.g., high yielding, ferti lizer responsive crop and prescribed ends) or generating foreign exchange at the expense of economic efficiency and/or aggravating poverty by promoting/sustaining a bi modal agrarian structure that responds to a minority and fails to respond to the majority. Even so, Tomich et al. (1995: small farms are included in development strategies, there must be a sacrifice of
45 economi to the social advantages of broad based development. Instead, Tomich et al. (1995: based strategies to promote a unimodal agrarian structure offer bi gger economic, as well as social, advantages in [developing countries] than the dualistic approach [where productivity growth is restricted to a narrow range of Their advocacy of a unimodal strategy of development is not limited to an argument in favor of equity; it is more so based on an argument in favor of efficiency. rationale for this argument stems from the distinction between economic efficiency and technical efficiency. Economic efficiency comprises both technical efficie ncy and allocative efficiency. Technical efficiency indicates that an operating unit (i.e. a farm) uses the least amount of necessary inputs in order to produce a given output. For example, let us assume there are two farms, each producing an output of 1 ton of plantains and the only input required for each is fertilizer. Farm A produces its 1 ton using 0.5 tons of fertilizer and Farm B produces its 1 ton using 1 ton of fertilizer; therefore Farm A is technically superior to Farm B and Farm B is technica lly inefficient. Allocative efficiency means that production decisions are made in such a way that the resulting increase in production (Tomich et al 1995: 122). For exam ple, a farm that spends an additional 2 dollars on fertilizer in order to produce an additional pound of output should at least obtain 2 dollars for the sale of the additional pound of product. If the farm only obtains 1 dollar in the sale of the addition al pound of product, then the farm is allocatively inefficient. The farm is allocatively more efficient if it produces less output at fewer costs but obtains an equal revenue.
46 Therefore, the argument made by Tomich et al. implies that a farm can be techni cally efficient and not economically efficient. According to the authors, arguments made in favor of dualistic strategies are based on the mistaken notion that economies of scale are prevalent in agriculture. That is, the notion assumes that large farms are allocatively and economically more efficient resulting in policies that favor large farm production. As mentioned earlier, an additional problem with policies that follow dualistic strategies, especially in order to obtain fast cash through the export of cash crops, is that they de link domestic production and consumption capacities. De Janvry (1981) demand for exports originated abroad and were de linked from local cons umption demands within Latin American countries. Again, this decelerates structural transformation because the demands of the bulk of the rural sector cannot articulate and stimulate the nonfarm sector, and thus overall development. Trade Policies To this day, trade policy remains a hotly debated area. Even amongst economists, there is disagreement about optimal policies for development. While economic theory holds that liberalizing trade is good for welfare and growt h, Stiglitz and Charlton (2005: 17) ar Latin America grew rapidly during the years many of its countries adopted import substitution strategies where remaind er produced domestically (2005: 19). The neo liberal view maintains that import
47 (2005:21), yet Stiglitz and Char lton (2005: 21) point out that a ccording to South Centre (1996: 42), the economic decline had less to do with import substitution than it did with the combined effects of global recession, oil price shocks, and debt policies. Stiglitz and Charlton also note tha ez 16). Many East Asian countries protected their industries until they were ready for international competition while governments remedied market failures and provided requisite physical and institutional infrastructure. State Intervention The successes enjoyed by the U nited S tates and Japan in transforming their agrarian economies resulted from more than just the free hand of the market. In both cases, the state pur sued broad based development strategies, whereby the needs of the collective, i.e. the needs of farmers, determined the direction of research and dev elopment. Tomich et al. (1995:67, 48) claim that "both countries were pioneers in government action that in creased not only the opportunities for farmers but also their ability to seize those opportunities...[and this was critical because] without the right policies, even the richest endowment of natural resources will not lead to structural transformation." Th e U nited S tates and Japan succeeded because government seized the day and provided an enabling environment for agricultural productivity growth. In both cases, government took the initiative and built key public institutions and infrastructure including r ural schools, agricultural research facilities, and extension services. According to Tomich et al., they reached structural transformation by making concrete instit utions, and initiative. Sim ilarly, Djurfeldt et al. (2005:
48 in Asia as a state driven, market mediated and small farmer based strategy to increase the national self For many developing countries, ho wever, the role of the state has been undermined by IMF and World Bank lending conditions. Wherever implemented, SAPs have weakened the role of the state and its ability to act as a key intermediary. SAPs demand decreased government spending, privatizati employment in the public sector, and the removal of food and input subsidies (Gladwin 1991). In western Kenya, for example, government cutbacks have grounded extension services. Insufficient funding prevents extensionists fro m visiting individual fields and farmers are expected to gain technical knowledge solely at the extension office. Nyang'Oro and Shaw (1998) note that SAPs have so demised state interventions that civil society has increasingly played a larger role in assur ing basic needs throughout developing countries. Technically, reduced state spending should not affect good policy making but it does. When SAPs call for privatization and a dismantling of state sponsored agencies, including agricultural extension agenci es, the natural result is that former public institutions become privatized, profit seeking enterprises. Privately owned agricultural and development institutions develop technology for whomever can pay for it namely the wealthy minority. In turn, a dual istic strategy ensues. For this reason, Tomich et al. (1995) emphasize that collective action by farmers often needs publicly operated administrative structures. Otherwise, private entrepreneurs will usually curtail research efforts when they are unable to capture a full return to their investment. Without state intervention, a bimodal agrarian structure will distort the price
49 signals which lead to induced innovations that bank on abundant resources and save on scarce resources. Infrastructure Nowhere is the role of the state more critical than in the provision of infrastructure. Where there are no roads, agricultural goods cannot be transported to market and where there are poor roads, the costs of transport make goods expensive. In Kenya, the price of fertilizer doubles, if not triples, from port to farm gate (Sachs 2005; Sanchez et al. 1997). Yet Lewis (1954: 13) points out roads, viaducts, irrigation channels and buildings can be created by human l countries constructional activity, which lends itself to hand labour, is as much as 50 or 60 percent of gross fixed investment, so it is not difficult to think of labour creating capital without using any but the simplest tools. In terms of agricultural development, roads provide high returns to investment. Governmental investment in infrastructure was key to the economic successes enjoyed by the U nited S tates and Japan Sachs (2005: road network has contributed to its rising per capita food production during recent pr oduction. Tomich et al. (1995: superior transport networks and a ccess to consumer goods and services and concluding that investment in rural infrastructure is a priority in Africa Stiglitz and Charlton (2005: government in providing infrastructure. None theless, even as a public investment, roads often reflect special interests. This study will indicate that in Kenya, the road system is noticeably better near the capital and regions of commercial agriculture while in decrepit conditions in regions of
50 sub sistence agriculture and/or areas populated by ethnic groups lacking in power. In this way, the road network reflects the bimodal agrarian structure adopted by the state as well as its system of patronage. Neo p atrimonialism Systems of patronage can be un derstood as social structures embodying rules and patterns of res ource allocation. Orvis (1997: 69) argues that rules of reciprocity and redistribution commonly underlie rural social structures in Africa. Reciprocity underpins most social relationships so providing some futur 69). Redistribution implies that the wealthy will provide some goods and services to the poor and powerless. Naturally, these relationships are n ot egalitarian and Orvis (1997: 69) notes that patron client ties are integral to surviva l and accumulation (Orvis 1997: 75). Constituting more than defense mechanism s, Hyden (Hyden a nd Peters 1991: 305) argues these networks or both horizontally and vertically, they have a social logic of their own and are beyond market and state ra tionalities, particularly because markets and states are weak in Africa (Hyden and Peters 1991). Beyond the household level, systems of patronage often involve corruption. ( 2007: uption. However, Mdard (2002: 380) argues that the term corruption cannot be used in the case of traditional patrimonialism, such as identified by
51 Max Weber, because there is no clear distinction between public and private domains. The term can, however, be used in the case of neo patrimonialism where the public and private sectors are formally differentiated and corruption occurs when the distinction is disrespected. In Africa, the term neo patrimonialism is used to characterize the toge ther (2002: 380). Both Ake (1996) and Mdard (2002) explain that corruption in Africa stems from the economically and politically insecure position of African elites after independence. To an extent, this remains the case. State power is the only leverage they control and economic accumulation comes necessarily through the state (Mdard 2002; Ake 1996). In turn, political support is accumulated through the redistribution of state resources. Mdard (2002: 383) argues that t he art of governing is not only the art of extracting resources, but also of redistribution: it is the only way of legitimizing power, in the absence of ideological legitimacy. In this way, corruption fits into the logic of the accumulation of political econom ical goods, within the framework of survival strategies where the economy and politics are closely articulated. Additionally, both Ake (1996) and Reno (1998) note that political power involves not only accumulating political economic goods but also denying resources to political rivals. This view would explain why state sponsored infrastructure seems to rarely make its way to certain regions in Kenya, for example, where ethnicities are divided by territories and political rivalries occur along ethnic lines By definition, neo patrimonialism reduces economic efficiency. As a result, important state controlled institutions lose credibility and fail to perform the functions they were designed fo r. For example, Ayittey (2005: 282) complains that people resort to saving money under mattresses if they cannot trust banks with their deposits. In turn,
52 me applies to currencies. People will keep their savings in physical assets and foreign currencies if they lose confidence in the national currency which in turn le ads to inflation (Ayittey 2005: 282). Neo patrimonialism affects agricultural productivity i n the same manner as do price distortions. Factor prices fail to reflect actual resource abundance and scarcity because elite segments of the population direct credit, investments, and infrastructure preferentially. In turn, necessary channels for agricu ltural production and marketing are distorted or nonexistent. For example, if roads in a particular region are purposely market and the price of their inputs will b e higher as a result of the additional transportation costs associated with poor road systems. Likewise, some farmers may be unable to access privileged lending, extension services, and subsidies. Fertilizer and improved seeds may be subsidized in one re gion and not another, as western Kenyan farmers have pointed out for this study. Nonfarm Work Nonfarm work, which refers to either off farm nonagricultural activities or nonfarm activities carried on in the homestead (Tomich et al. 1995: 201) can impact agr icultural productivity both positively and ne gatively. Tomich et al. (1995: 201) cite nonfarm activities as the key link between agriculture and industry. Nonfarm activities can account from 40 to 80 percent of output and employment in the rur al economy ( Tomich et al. 1995: 201). Moreover, nonfarm incomes can smooth total rural household incomes over the year and reduce income inequalities in rural areas (Tomich et al.
53 1995: 203). Orvis (1997) maintains that nonfarm income is critical to agricultural inves tments, improvements, and expansion. On the other hand, nonfarm work can reduce available labor for agricultural production. Labor intensive agricultural activities, such as soil conservation, cannot be undertaken. In Kenya, where off farm employment is usually reserved for men (Orvis 1997: 63), women are left to manage all farm activities on their own. And while labor is usually the most abundant and inexpensive resource throughout deve loping countries (Tomich et al. 1995), poor farmers often experience household labor shortag es and bottlenecks (e.g. Morera 1999; FAO 1998; Ayieko 1995; Zimmerer 1993) for the following reasons. Poor farmers cannot afford to hire additional labor during peak seasons when labor demands are very high because cash flows are se asonal, increasi ng only after a harvest (Morera 1999). Young families have more household con sumers than producers (Chayanov 1966). Labor is tied up in alternate incom e generating activities (Morera 1999; FA O 1998; Ayieko 1995; Zimmerer 1993). Nonetheles s, the role of off farm activities in diversifying, stabilizing, and increasing rural incomes has been given significant at tention (de Janvry and Sadoulet 2001; Ruben and Van den Berg 2001; Reardon et al. 2001). De Janvry and Sadoulet (2001) parallel the rising importance of off farm work with the deep economic changes brought on by trade liberalization, decentralization, elimination of public subsidies and reduction of parastatal services to agriculture. Increased costs of production, decreased access to rural services, and natural resour ce degradation lead farmers off farm to generate income through multiple activities ( Morera and Gladwin 2006; Morera
54 narrow, already favo red stratum of the 257), off farm income also subst itutes for formal credit (Orvis 1997). Even so, Ruben and Van den Berg (2001) maintain that the importance of off farm work has been largely neglected by rural developme nt programs intent on increasing rural incomes through improved agricultural productivity. Land and agricultural labor have often been viewed as the only assets controlled by the rur al poor (de Janvry and Sadoulet 2001). In turn, rural poverty has been a ddressed through the implementation of redistributive land reforms and integrated rural development programs in an effort to raise the productivity of these assets (de Janvry and Sadoulet 2001). its promotion of off t has become clear that raising the capacity of the poor to participate in non based on findings indicating that 1) off farm activities are fundamental for the land poor; 2) nonagricultural incomes are far large r than agricultural wage incomes; 3) off farm incomes help mitigate income inequalities; and 4) off farm work helps overcome credit failures and smooth risk man agement (de Janvry and Sadoulet 2001; Ruben and Van den Berg organizations (as opposed to state supported programs) applying participatory methodologies (as opposed to the application of economic incentives). It is characterized by a decentralized, demand driven allocation of public re sources (de Janvry and Sadoulet 2001: 1).
55 farm work lacks a discussion of its potential relationship to land degradation and gender. Tradeoffs occur within households when limited labor is appli ed towards off farm work. The fruits of off farm income generation do not flow back to the farm consistently. In cases where men migrate long distances and for long periods, the remittances they send back to the farm may be insufficient to make up for th e loss of labor and unfulfilled gender roles their absences constitute. Even though household consumption requirements are lessened by male migration, women are left to undertake all agricultural activities and labor intensive soil conservation practices are not undertaken if household labor is limited and cash is unavailable for hired labor. farm work also lacks a discussion of its potential relationship to gender and disease. Tradeoffs within households also occur when families are separated for long periods of time. In Kenya, where polygamy is the norm, men may acquire additional wives during their off farm migrations. For women, farm income and an increased threa t for HIV infection. Disease The omnipresence of disease has often been characterized as a symptom of underdevelopment that contributes t o the poverty trap Disease disproportionately affects the poor and the oppressed worldwide. In turn, development is st unted by disease. Sachs (2005: disease. He notes that virtually everyone in sub Saharan Africa contracts malaria at least once a year (Sachs 2005: 197). This translates to several days or wee ks of
56 project in its tracks, whether a new mine, farm region 197). DS is the leading cause of death in sub Saharan Africa, where over 22 million people are HIV positive [and] most victims are between 15 and 49 years of age (Hock 2010: 289; UNAIDS/WHO 2007). Because HIV/AIDS affects people in the prime of their lives, Afr ica is losing its most productive individuals. The costs of the disease are reflected in all sectors of the economy. Africa is losing its teachers and doctors, its civil servants and farmers, its disarray from massive medical costs for workers, relentless absenteeism, and an avalanch e of worker deaths [ Sachs 2005: 201] Moreover, the co occurrence of tropical diseases in Africa compounds the severity of e systems are impaired, they are more susceptible to additional infections. The burden of disease impacts agricultural productivity in a number of ways. At the household level, labor availability is reduced, nonfarm income that could otherwise be used for agricultural intensification or education must be diverted towards health care, and productive assets such as cattle must be sold. Over time, a household may also experience technological reversal if it loses its knowledgeable membe rs (Sachs 205: 55). At the national level, disease places a greater toll on the health care system and reduces productivity across all sectors of the economy. Conclusion This chapter has set the stage for the empirical work to follow in the next four chapters. By synthesizing and analyzing the literature on agricultural productivity and
57 soil fertility management, this chapter has illustrated that without incentives for agr icultural intensification, soil fertility enhancement technologies are for naught because farmers rather pl ant extensively with minimum applic ation of inputs, including soil fertility enhancement practices. If food prices fail to rise, marginal producers will not invest in the increased costs of intensification which, essentially are the costs associated with fertility enhancement. Instead, they will cultivate until they degrade the land and must abandon the area or abandon agriculture. The chapters that follow will indicate that in western Kenya, farmers are degrading their land and finding alternatives to s taple crop production. Land poor farmers, rather than intensify production, apply their labor towards agricultural wage work or nonfarm work while larger landholders, rather than increase agricultural productivity, substitute kale production for maize and bea n production Thus smallholders do not invest their labor in organic fertilizers or improved fallows while larger landholders do not invest their wealth in inorganic fertilizers. On average, farmers substitute land for fertilizer. Meanwhile, soil fe rtility enhancement as well as aggregate agricultural productivity remains low in this region of western Kenya, calling into question the theory of induced innovation as well as the efficacy of the Green Revolution and pointing to a need for increased effo rts if agricultural research and development are to be successful and structural transformation in Kenya attained
58 CHAPTER 3 RESEARCH SETTING: T HE LAND AND ITS PEOP LE Introduction This chapter describes the ecological and socioeconomic features of the s research setting. It weaves primary data collected through participant observation together with secondary data collected from published materials in order to present a current, as well as historical, glimpse into this region of western Kenya. By examining the land, people, and institutions interacting in Shinyalu Division of Kakamega District in the Western Province of Kenya, the chapter provides a context in which the theoretical framework, research methods, and empirical results of this study ca n be understood. Ecological Setting The empirical work of this study took place throughout four villages located within Shinyalu Division of Kakamega District in the Western Province of Kenya. Spanning almost 333 square miles, Shinyalu Division borders th e Nandi Escarpment and encompasses a portion of Kakamega Forest National Rese rve 1 as illustrated in Figure 3 ies are illustrated in Figure 3 3 Its land is of high agricultural potential and its population of 103,948 is pr edominantly Luhya second largest ethnic group (Kenya 2001) 2 The four villages Mutsulio, Shikusi, Lugango and Shinakotsi are located close to one another within an area that spans from the Yala River in the south to the 1 ller and Mburu 2009). While it has received government protection since 1933, protection varies by area. M ller and Mburu (2009) note that the region around Kakamega Forest is among the most densely populated in the world and predict that forest clearing will increase particularly in areas with lower protection status, especially near roads and market centers. However, they also predict cons iderable forest clearing in the strictly protected National Reserve. 2
59 unpaved road in the north connecting Highway A1 to Highway C39 (see Figure s 3 1 and 3 2 ). The geographical coordinates of each vil lage entrance appear in Table 3 1. The area encompassing the four villages lies at an altitude of 1500 to 1800 meters above sea level on Archaean Kavi rondian Group rock formati ons (Kenya 1962) and is characterized by temperatures ranging from 14 o to 26 o Celsius (M athu and Davies 1996 and phosphorous deficiencies, and it s vegetation includes scattered woodlands. Its annual rainfall of 1200 to 1900 millimeters is distributed across two rainy seasons. The first of these is known as the long rains season, which runs from March through July, and is the more reliable of the two seasons. The second rainy season is known as the short rains season and runs from August through November. The rains are known to fail periodically. Socioeconomic Setting The area encompassing Mutsulio, Shikusi, Lugango and Shinakotsi villages overl aps what used to be known as Isukha Location, or Isukhaland an area peopled by the Abesukha a sub group of the Luhya of western Kenya, who are also known as the Western Abaluyia (Nakabayashi 2003). According to oral tradition, the ancestor of the Abesuk ha who spoke an Oluluyia dialect, moved into this region from eastern Uganda, finding the region unpopulated upon his arrival (Mwayuuli 2003; Were 1967). One of his sons established the Abesukha sub tribe. Thus all the clans of the Abesukha originated i n this region (Were 1967). Moreover, they have not migrated since (Mwayuuli 2003; Were 1967). Abesukha elders interviewed at Isukha Location in 1964 traced their genealogy back 13 generations to their common ancestor (Were 1967). To this day, the people of this region identify themselves as Abesukha
60 The Abesukha like the rest of the Luhya are patriarchal. Land is passed from the father to the sons, each son inheriting a partition of the original landholding. Sons move into their own houses once they are old enough to marry. However, they inherit their own piece of land only several years after marriage. Thus, when a woman first marries, in law and helps work her father in Only after a period of this cohabitation does the son According to tradition, Luhya marriages are arranged by the parents of the bride and groom with the help of Wanjira (Wako 1985 ; A list of additional Luhya terms and their translations ap pear in Table 3 2). A young woman is chosen from a different clan (or different sub tribe or different tribe) on the basis of her price, usually in the form of cattle, which must also be agreed upon. Only then does the bride journey, in the company of her for the marital feast (Wako 1985 ). Surprisingly, many of these traditions remain in place. During my field work season in 2007, I witnessed the distress of a family w hose daughter arranged her own marriage. A bride price was never paid nor a marital feast held in the village. As such, gone wrong in her marriage. Also during the field work season of 2007, a young Luhya man explained to me the importance of establishing a family on inherited land. The Luhya consider it
61 shameful for a man to leave idle the land his father has passed on to him. Yet the Luhya also encourage their sons to migrate in search of off farm work. Thus Luhya later leaving their wives to tend the land while they migrate in search of nonagricultural jobs. It is common for Luhya women to bear entirely the responsibility of farming and feeding their families. Women are responsible for fetching water and firewood, plowing by hand, planting, weeding, and caring for the young tasks that must be completed on a regular basis. Fo r example, many of the women interviewed for this study walked long distances every day to fetch water from the Yala River and collect firewood from the scattered woodlands 3 Men are often responsible for managing cash crops, maize harvests, and heavy wor k such as land clearing, carpentry, and oxen plowing. However, these jobs are often not necessary for long periods of time. Thus most women interviewed in this study actually preferred to have their husbands employed in an off farm job rather than residi ng and working on the homestead. Off farm jobs are usually better paid than agricultural work and although women only receive a portion of the income in the form of remittances, these are an important source of cash that is often applied towards school fe es and towards farming and household expenses. 3 used stream s/rivers as their source of water from 1989 1999 declined from 36% to 21%, while the use of other sources of water such as bore holes and roof water increased from 16% to 54%. In most cases, this should have translated as decreased workloads and improved c 2003 Demographic and Health Survey (2004:43) indicates only 13.3% of women interviewed in Western Province enjoyed full or joint decision making in all decisions regarding healthcare, purchases, visits, and cooking. Mo reover, illiteracy was 22.5% for women, whereas it was 15% for men (2004:29 30). The in Western Province was 80 deaths per 1000 live births and under f ive mortality was 144 per 1000 (2004:116). Trends in the nutritional status of children indicate 30% of children under five in Western Province were stunted (2004:167). Data were unavailable at district, division, and village levels.
62 Agricultural Practices The farms I visited throughout the villages of Mutsulio, Shikusi, Lugango and Shinakotsi during the field work season of 2007 averaged a little over a hectare (2.83 acres) in size. Mos t were cropped predominantly in maize and beans, with smaller portions of the landho ldings devoted to sugarcane, na pier grass, kale, tea, and cattle. The steeper farms, in Mutsulio especially, had t erraces and live barriers of na pier grass on their hillsi des while many Lugango farms specialized in zero grazing cattle rearing. During the short rains season, most farmers left a portion of their maize fields fallow. Fallowing for soil nutrient replenishment has long been practiced in this region while terrac ing for soil conservation as well as zero grazing for more efficient cattle rearing are relatively new technologies disseminated by institutions such as the Kenya Development (MoARD), and the International Centre for Research in Agroforestry (ICRAF). Over the years, these institutions have also disseminated planting and fertilizing technologies. During one interview in Mutsulio in 2007, an informant explained to me that prio broadcasted, rather than spaced, their maize seed and use very little fertilizer. Kale production, too, is a relatively recent agri cultural practice that is increasingly disseminated through extension effo rts. Although Kenyan farms have long grown a variety of African vegetables for subsistence, kale ( Brassica oleracea ) is a nontraditional species that is now grown in large quantities for market (Nekesa and Meso 1997) The success of kale production lies in its high domestic consumption. Its local name, sukuma wiki a Swahili term for (Parkinson et al. 2006), implies its ubiquity in the diet and, like traditional African vegetables, is typically served with ugali
63 epared maize meal. Thus the niche kale has filled existed previously and kale is quickly replacing African vegetables in much the same way that maize has largely replaced the production and consumption of sorghum and millet in this region. Collaboration w ith the PLAR Project The PLAR project, as it came to be known in Mutsulio and Shikusi villages, was a collaborative effort between several institutions to scal e up the work on soil fertility enhancing technologies ICRAF had achieved during the 1990s in Vih iga District, a district lying just south of Kakamega District. Together with KARI, MoARD, and the Royal Dutch Institute for Tropical Agriculture (KIT), ICRAF sought to disseminate their soil fertility enhancement technologies in Mutsulio from 1998 to 199 9 using the Participatory Learning and Action Research (PLAR) approach (Place et al. 2005; Baltissen et al. 2000). During the year 2000, the technologies were also disseminated in Shikusi (Baltissen et al. 2000). The PLAR approach, as outlined by Baltisse n et al. (2000), involves four phases that include farmer input, farmer to farmer exchanges, and farmer interaction with extension staff across all phases. The first phase of the PLAR approach entails diagnosis and analysis. Community representatives are invited to a meeting and asked to outline the territory, organization, and current natural resource management of the village, developing a set of criteria indicative of good soil fertility management. Later, village committees are formed to coordinate a nd monitor the PLAR process. Members of the committee include both men and women of varying resource bases. During the second phase of the PLAR approach, extension visits and workshops are organized, introducing farmers to alternative management practices During the
64 third phase, technologies are experimented with. Key farmers arrange field visits to their demonstration plots in order to share their experiences with the technologies and disseminate them to others. The fourth and last phase of the PLAR a pproach involves evaluation of the activities implemented at the community level and assessment of experiments carried out by committee members, appraising their successes and failures. From 1998 through 2000, the PLAR project worked with Mutsulio and Shik usi villagers to disseminate improved fallows, biomass transfer, and their optimal combinations with recommended doses of rock phosphate (Baltissen et al. 2000) As will be explained in Chapter 4 improve d fallowing is a soil fertility enhancing agrofores try technology that involves sowing legumes within maize rows during the long rains season, leaving the maize field fallow during the short rains season, and turning the legumes back into the soil as the following long rains season begins (Kamiri et al. 19 97). B iomass transfer is another soil fertility enhancing technology that involves cutting the leaves and soft twigs of Tithonia diversifolia or Lantana camara from hedges and live fences, chopping them into small pieces, carrying them to the field, and i ding to Baltissen et al. (2000: 8), farmers in Mutsulio conducted 449 experiments with soil conser ving and organic soil fertility enhancing technologies over the period of 1998 to 1999. Farmers in Shikusi conducted 55 of these experiments from March to August of 2000. However, as will be discussed in the following chapters, by my fieldwork season of 2007, improved fallows and biomass transfer technologies had all but vanished from
65 this re gion. The technologies were discontinued in Mutsulio and Shikusi villages and never reached Lugango and Shinakotsi villages. Farmers from Mutsulio and Shikusi villages interviewed in this study pointed out that although the technologies improved their pr oduction of maize, the implementation of the technologies took up too much land and labor. Nonetheless, the participatory methods and farmer to farmer exchanges emphasized by the PLAR project resulted in long term benefits for its participants that became apparent during my fieldwork season of 2007. As wi ll be discussed in Chapters 5 and 6 PLAR project participants exhibited greater access to social capital, such as labor sharing and money saving groups, than did non participants. A higher number of PLAR project participants than non participants also reported the sale of cash and staple crops as well as a better quality of life and economy. Finally, PLAR project participants applied fertilizer in greater quantities and purchased maize for consumption in lesser quantities than did non participants. In Shikusi for example, fertilizer was applied at an average rate of 88 kg/ha, compared to an average rate of 69 kg/ha applied across all four villages during the long rains season of 2007. Similarly, while P LAR project participants reported purchasing maize for consumption at an average rate of 3.1 months per year, non participants did so at a rate of 4.5 months per year. Shinakotsi non participants purchased maize for consumption at an average rate of 5.3 m onths per year. Thus the PLAR project, while failing to achieve the long term adoption of improved fallow and biomass transfer technologies in Mutsulio and Shikusi seems to have significantly improved the overall wellbeing of its project participants 4 4 Moreover, fieldwork had been economically better off to begin with than non participants. In fact, the PLAR project targeted
66 Ot her Institutions Working in the Surrounding Areas In addition to the institutions that collaborated to bring the PLAR project to Mutsulio and Shikusi there are a number of institutions located throughout Kakamega District that provide an array of services for farmers living in Shinyalu and other surrounding divisions The most frequently mentioned of these during my fieldwork season of 2007 was Kenya Women Finance Trust (KWFT). KWFT is an independent microfinance organization founded to address the finan cial needs of women. Loans, commonly ranging from 10,000 to 100,000 Kenyan shillings ( $150 to $1500 ), are provided to groups and to individuals, using household items and cattle as collateral. reason, KWFT has outreach programs and targets. Perhaps as a result, farmers interviewed for this study had only positive things to say about KWFT. Another institution working in the surrounding area is the National Cereals and purchase, store, ma rket and generally manage cereal grains and other produce in t ). The NCPB was transformed in 1993 when the grain sector of Kenya was fully liberalize d. The NCPB continues to purchase, store, and produce to willing buyers at market driven prices for different regions depending o n m ore marginal farmers. Participants, on average, had less land and fewer head of cattle than non participants.
67 ). Because th e NCPB is located at the outskirts of Kakamega Municipality, farmers interviewed for this study naturally preferred to sell their maize locally, in the Shinyalu area, rather than absorb the transport costs of selling it to the NCPB. Moreover, smallholders find it difficult to meet the standards of quality communication). Other institutions serv ing the farmers of Shinyalu interviewed in this study in 200 7 had experienced few exchanges with most of these institutions. Only thirteen percent of the farmers interviewed had ever visited the Shinyalu extension office. Markets and Infrastructure Although the roads surrounding the villages of Mutsulio, Shikusi, Shinakotsi, and Lugango are not paved, they constitute important networks for market transactions. Large trucks transporting molasses often traverse the Shinyalu area, traveling along the wider routes and stopping to sell molasses. Village women often s ell maize and vegetables at busy junctions. K iosks that carry bread, tea, soda, candy, etc. can also be found along the roads, even in the in terior reaches of the villages. Despite their importance, the roads in this region have been hard to maintain During the fieldwork season of 2004, two wheel drives had a difficult time crossing from A1 to C39 along the unpaved Shinyalu road (see Figure 3 1) From 2004 to 2 007, the condition of A1 between Kakamega and Kisumu also worsened. Moreover, while it took
68 me approximately 6 hours to travel by bus from Kisumu to Nairobi in 2004, the same trip took over 8 hours in 2007 Yet c ertain portions of the route between western and eastern Kenya remained well paved, such as those surroundi ng Nakuru and Nairobi. Addi tional Livelihood Strategies In addition to agricultural production, households throughout Mutsulio, Shikusi, Lugango and Shinakotsi engage in a number of alternate livelihood strategies. The most prevalent of these are agricultural wage work and brewing liquor ( ) or beer ( busaa ) for sale. Agricultural wage work is commonly undertaken by younger farmers, men and women alike, as the work is too strenuous for older farmers. On the other hand, and busaa brewing are commonly undertaken by w omen of all ages, though less commonly by men. As such, brewing is an important source of income for women, particularly land constrained women. Although home brewing has been illegal in Kenya since Moi took office in 1979, thus providing a marginal live lihood, it is better remunerated than agricultural wage work whose compensation is the lowest in the rural sector. Home brewing also represents an important form of nonfarm specialization in the rural economy which, ac cording to Tomich et al. (1995: 201), is the key link between agriculture and industry. It is important to note that not all of the ingredients used to brew are produced on the farm. which can be obtained locally from p ressed sugarcane or, more commonly, purchased in the form of molasses. However, informants interviewed in this study explained that substituting regular sugar for rock sugar or molasses produces a clearer, higher quality During the fieldwork se ason of 2007, I confirmed that made from sugar fetched a higher price than did made from molasses.
69 Throughout my time in Kenya, I was fortunate enough to catch glimpses of being brewed on several occasions. According to one of my informants, 5 liters of requires about 4 kg of maize, 1 kg of millet, and 3 kg of regular sugar. The process takes 11 or 12 days. The grains are fermented during approximately 7 days after which the sugar is added and the mixture fermented f or another 4 or 5 days. The mixture is then boiled and distilled. I calculated in 2007 that the ingredients necessary to brew 5 liters of cost approximately 370 Kenyan shillings. In turn, 5 liters of could be sold for 600 Kenyan shilli ngs in Khayega 1200 Kenyan shillings in Kisumu and 1700 Kenyan Shillings in Nairobi The cost of traveling from Khayega to Kisumu roundtrip, in 2007 was approximately 200 Kenyan Shillings. Busaa too, is often brewed with ingredients not produced on th e farm, since finger millet a key component of busaa is no longer as widely cultivated as it was before the introduction of maize to this region. Busaa unlike which is made with sugar and requires distillation, is made strictly with grains. Thu s busaa is a beer, rather than liquor, and has a much lower alcoholic content than In addition to home brewing and agricultural wage work, farmers throughout this region also engage in retail in an effort to supplement their incomes generated thr ough agricultural production. The most common forms of retail involve the buying and reselling of maize, vegetables, fish, paraffin, and medicine. Retail is most often undertaken by women. Effects of Illness on Livelihoods Illness has affected the agric ulture, economy, and social structure of Mutsulio Shikusi, Lugango and Shinakotsi in a variety of ways. At the very least, as in the case of malaria, illness annually robs several weeks of work from every farmer in the area.
70 At worst, as in the case of HIV/AIDS, illness depletes household assets, alters age and sex ratios, and transforms the very fabric of the community. As one representative of Kakamega agricultural production in this regi on because virtually every family has had to deal with disease Nonetheless, the effects of disease on farming did not become apparent to me until my fieldwork season of 2007, when I returned to Mutsulio and Shikusi after a hiatus of 3 years. At that time, I began to realize that the composition of many households interviewed in 2004 had evolved. What were once male headed households, were now female headed households. During my first fieldwork season in 200 4, I had difficulty locating female headed households for this study. In fact, the main reason I included Shikusi in the sample was that I could not locate a sufficient number of female headed households in Mutsulio one of the villages involved in the PL households was to be disaggregated by both PLAR project participation and gender, I had to include one more PLAR project involved village in the study in order to randomly select a total of 30 female headed ho useholds from a list that combined households from both villages. During my second fieldwork season in 2007, however, I had the opposite problem. I returned to Mutsulio and Shikusi once again combining lists of farmers from both villages in order to rand omly select 30 female headed and 30 male headed households previously involved in the PLAR project 5 What I found was that many households 5 Qualitative results reflect analysis of data collected during all three fieldwork seasons. Quantitative results reflect analysis of data colle cted in 2007.
71 which were male headed in 2004 had become female headed by 2007. Many of the initial random selections were discard ed when, upon arriving at the homestead, I discovered the husband of the household had passed away. In addition to the emotional pain and financial costs incurred by medical treatments and funeral arrangements, the loss of a male head of household can tran slate to a loss of off farm income, reduced schooling, and increased work load for the remaining household. Because men, more typically than women, work off farm and far m improvements and expansion. Thus young de facto female headed households have fewer opportunities than others to achieve food security and the overall betterment of their families. Sadly, many homesteads visited during 2007 were dotted with graves. The Luhya typically bury relatives at the entrance of the homestead. Thus by the time I entered a household, I was painfully aware of the misfortune endured by the family. One informant had buried six relatives over the last year alone. The only way she co uld now meet the demands of her farm was through hired labor. This unfathomable loss tragically illustrated the effects of illness on rural productivity. Conclusion This chapter has examined the ecological and socioeconomic features that characterize th e region in which the research study took place. By describing the nature of t he land, the ways of the people, Western Province, this chapter has sought to g round the theoretical framework dis cussed in the pre ceding chapter as well as contextualize the empirical work to follow in the remaining chapte rs. Thus the next chapter now turns to the research methods
72 used to examine the nature of soil fertility management in this region and its role in agricultural pro ductivity, investigating how it is affected by technological adoption, gender, off farm work, and other off farm factors
73 Table 3 1. Geographical coordinates of Lugango, Mu tsulio, Shikusi, and Shinakotsi villages Village Geographical Coordinates Kilome ters from Khayega tarmac Shinakotsi 00 12.43 N 2.6 34 47.68 E Lugango 00 12.24 N 3.2 34 47.89 E Shikusi 00 12.22 N 3.4 34 47.94 E Mutsulio 00 11.52 N 34 48.81 E 5.6 Coordinates recorded at each village entranc e using a global positioning system (GPS) receiver Table 3 2. Agricultural Luhya terms and their English translations Luhya term English translation Luhya term English translation Amamondo M oney Khuraka Planting Amasika funerals (or tears) K hwabitsa t op dressing Amasingo cow dung K ikabicchi Collards Amatsi W ater Cattle Bandu many people Likhubi Cowpeas Bulimi agriculture (or farming) Likondi Sheep Chichirishi ox Lipata Duck Eshikulu hill Lusuu nap ier grass Ifula rain Mabere milk (or fin ger millet) ifula yo murotso long rains Majani Tea ifula yo mshibwe short rains M akanda Beans Ikanisa church Matuma Maize Imbolea fertilizer Mbuli Goat Ingokhoo chicken Miro African vegetable (produces small white seeds) Ingulume pig Mukhali Wife I ntzu house Mulimi field (or farmer) Ipunda donkey Mundu one person Irotso long Murere African vegetable (produces small black seeds) Isimbwa dog M usatsa Husband Isukari sugar O bulwale Illness Isukuma kales O khufwa Death Itwasi cow O muchera River khu birira second weeding O mukhonye Sugarcane khu sembera first weeding O musala tree (or drugs) Khufuna harvest O mwana Child Khukombola hiring Shibwe Short Khukonera idle Shimiyu Khulia eating Tsikwui Firewood Khulima plowing or dig ging Compiled with the assistance of Sarah Anyolo and Mildred Machika
74 Figure 3 1. R oad map indicati ng location of Shiny alu, adapted from Nelles Map: Kenya (Source: Nelles Verlag 2005)
75 Figure 3 2 Legend for r oad map indicati ng location of Sh iny alu, adapted from Nelles Map: Kenya (Source: Nelles Verlag 2005)
76 Figure 3 3 Map of Kakamega District indicating location of Shinyalu Division 6 ( Source: United Nations 2009) 6 Map provided courtesy of the UN Office for the Coordination of Humanitarian Affairs The boundaries and names shown and the designations used on this map do not imply official endorsement or acceptance by the United Nations.
77 CHAPTER 4 RESEARCH METHODS: U NDERSTANDING THE OPPORTUNITIES AN D CONSTRAINTS AFFECTIN G SOIL FERTILITY MAN AGEMENT Introduction The impetus for this study was ignited by a host of publications in which a strong case was made for the replenishment of soil nutrient deficiencies across Africa through th e application of improved fallow and biomass transfer systems (Place et al. 2001; Franzel et al. 2000; Pisanelli et al. 2000; Sanchez 1999; Kwesiga et al. 1999; Swinkels et al. 1997; Buresh et al. 1997). Improved fallow systems were built on African farmi ng techniques by improving traditional fallows with fast growing tree or shrub legume species, such as Sesbania sesban, Crotalaria grahamiana, and Tephrosia vogelii which fix nitrogen from the air and return it to the soil. Biomass transfer systems were also built on African farming by utilizing hedges around homesteads and traditional live fences of common species such as Tithonia diversifolia and Lantana camara for green manure. Together, these systems could provide smallholding farmers with labor int ensi fertility enhancing alternatives to expensive industrial, inorganic fertilizers. The technologies were extensively researched and disseminated in western Kenya during the 1990s by the International Centre for Research in Agrof orestry. Improved fallows involve sowing legume seeds in maize rows either at planting (if using Sesbania sesban ) or after the second weeding (if using Crotalaria grahamiana or Tephrosia vogelii ) during the long rains season (Kamiri et al. 1997). The long rains season in western Kenya, as the name implies, is the longer of two rainy seasons. The long rains usually extend from March through July and the short rains from August through November. Many farmers plant during both seasons. However, improved
78 fa llow systems require leaving a field fallow after the long rains maize harvest and letting the legumes grow during the short rains season. By the time the next long rains season is about to begin, the legumes have grown into young shrubs and returned to t he soil anywhere from 50 to 150 kilograms (kg) of nitrogen per hectare (Kamiri et al. 1997), at which time the young shrubs should be turned back into the soil. Biomass transfer involves cutting the leaves and soft twigs of Tithonia diversifolia or Lantana camara from hedges and live fences, chopping them into small pieces, (ICRAF 1997). The mixture should be left to decompose for at least one week befo re planting a crop (ICRAF 1997: 6). In western Kenya, an application on maize fields of 5 tons of Tithonia diversifolia (dry matter) per hectare yields comparable results to an application of 50 kg/ha of inorganic P 2 O 5 and 60 kg/ha of inorganic nitrogen (ICRAF 1997: 7). The r with improved fallow and biomass transfer systems were compelling. Nonetheless, my own work in Honduras in 1997 (Morera 1999) led me to conclude that economic constraints, particula rly resulting from SAPs, inhibit smallholders from investin g in labor intensive, fertility enhancing technologies despite their proven capacity to ameliorate soil degradation and raise maize yields. Moreover, economic trends in Kenya indicated that the pr oduction growth of maize remained lower than population growth, rendering maize an importable commodity even after liberalization from 1992 onward (FAOSTAT relatively low co nsumption of inorganic fertilizer and correspondingly low agricultural
79 productivity, I embarked on my own investigation of farming patterns among smallholders in western Kenya. Research Objectives and Assumptions This study set out to identify factors affe practices throughout western Kenya and to gauge trends in agricultural productivity. In doing so, the study addressed the following guiding questions: 1) Is agriculture in western Kenya intensifying or extensifyi ng? 2) Does off farm work lead to an expansion or a contraction of on farm investments? 3) Does limited resource access, particularly to land and cash, lead to labor How does gender affect technological choi ce? 5) Do research and development efforts to disseminate soil fertility enhancing technologies have lasting effects on farming practices? 6) How do factors originating outside the farm at both the community and national levels affect farming decisions ? In order to address these questions using a political ecology approach, the study sought to integrate information from various levels of Kenyan society local, national, and international. At the local level, primary data were obtained through interviews with farmers, community leaders, and representatives of local government agencies, organizations, banks, and businesses. At the national and international levels, secondary data were gathered from national archives and online publications. The informati and social resource access and this access is embedded within a larger framework of exchanges and interconnections across horizontal and
80 vertical levels of society, facilitated or limited by social rules of varying flexibility. The study gathered both emic etic management in rural western Kenya in its attempt to view agricultural decisions from the perspective of the farmer while locating those decisions within a general contex t. Thus the methods of the study were: 1) to collect household wealth, education, employment, and farming data from 120 West Kenyan households; 2) to statistically test the association between these variables using multiple regression; 3) to collect sever al farming routines and biographical vignettes in order to elucidate the opportunities and constraints faced by farmers of varying access to social and physical capital; 4) to develop ethnographic descriptions of West Kenyan farmers in order to depict thei r relationships to their social and physical environments, thereby contextualizing the results of the study; and lastly 5) to collect background information on national and international agricultural policies and trends from Kenyan archives, gray literatur e at ICRAF, ICRAF personnel, and Kenyan administrative personnel in order to further contextualize the ground level results of the study. Research Hypotheses According to Sanchez et al. (1997), the region of western Kenya has one of the highest population densities in the world and, as a result, shrinking agricultural landholdings and declining soil fertility. Because Boserup (1981) argues that in the face of population pressure, soil fertility is dealt with either through labor intensive or industrial soi l fertility enhancement, it follows that farmers in western Kenya should be either applying green manure (biomass transfer), using improved fallows, or applying mineral fertilizer in order to arrest the decline of agricultural yields. As such, this means farmers in western Kenya should be intensifying agricultural production because, as
81 Netting (1993: 102) notes, the key characteristic of intensive agriculture is the manipulation of nutrients, water, and sunlight for increased plant growth during longer per iods of time as well as the replenishment of elements that become exhausted. Intensification is a means of coping with limited space and time. Thus, the following two H1: Farmers t hroughout western Kenya are intensifying agricultural production by increasing the availability of soil nutrients to their staple crops, thus increasing crop yields per area: a) in kg of green manure per acre of land, acres of improved fallows per total acres of land, and/or kg of mineral fertilizer per acre of land) will be positively related to landholding size (measured in acres); b) r acre of land) will be positively related to landholding size (measured in acres). On the other hand, rejection of H1 would imply that farmers in western Kenya are extensifying agricu lture as described in Chapter 2 of this study, which eliminates the classic economic distinction between cultivated and uncultivated land as well as the notion of raising agricultural output through expansion of production at the so called extensive margin this study adopts the more recent definition of extensification as a means of decreasing agricultural yields per area (AgricultureDictionary.com). Rejection of H1 would imply that farmers in western Kenya, on average are decreasing agricultural output per area of farmland through fewer applications of soil fertility enhancements. This would apply to both largeholders and smallholders. That is, H1 would be rejected landho per area of farmland and landholding size.
82 Because farms throughout western Kenya tend to be less than 2 hectares (4.9 acres) in size (Amadalo et al. 1998), households must supplement farm production with off farm work in order to subsist (IEA/SID 2001; Orvis 1997; Ayieko 1995). Previous research indicates rural Kenyan households allocate labor in response to soil fertility, rainfall, and cash availability: households with m arginal farmland and limited cash flows tend to invest less labor in on farm production and more labor in off farm work (Place et al. 2001; Orvis 1997; Ayieko 1995). Consequently, poorer farmers are less likely to apply labor intensive soil fertility enha ncing practices because these represent a risky use of available household labor that can instead be used off farm to gain immediate cash. On the other hand, research throughout Latin America (De Janvry and Sadoulet 2001; Reardon et al. 2001; Ruben and va n den Berg 2001) illustrates correlations between non farm generated income and on farm improvements. Similarly, Orvis (1997) correlates ongoing off farm employment and on farm re investment in agriculture throughout Kenya. In either case, however, the fi ndings fail to make the distinction whether off farm agricultural jobs can generate the same level of on farm improvements that nonfarm jobs can generate. The distinction is critical for countries whose nonfarm sectors provide for less than 50% of the lab or force. Because agricultural jobs tend to yield lower incomes t han nonfarm jobs (Tomich et al. 1995), they should result in lower levels of household re investments in agricultural productivity such as soil fertility enhancement. Thus, the following tw o part hypothesis addressed whether the kind of off farm income generation results in an expansion or contraction of agricultural productivity.
83 H2: Nonfarm income generation leads to increased on farm agricultural investments while agricultural off farm i ncome generation leads to decreased on farm agricultural investments: a) The greater the number of household members employed in nonfarm work, the more a household invests in one or more of the following: kg/acre of green manure, square meters/acre of improv ed fallows, kg/acre of mineral fertilizer, days/acre of hired oxen, days/acre of hired labor, and/or kg/acre maize and bean seed; b) The greater the number of household members employed in agricultural wage work, the less a household invests in one or more of the following: kg/acre of green manure, square meters/acre of improved fallows, kg/acre of mineral fertilizer, days/acre of hired oxen, days/acre of hired labor, and/or kg/acre maize and bean seed. Wealth, in general, is often correlated with higher leve ls of agricultural productivity. Wealthier households can afford to use more agricultural inputs, such as labor, seed, and fertilizer per area than can poorer households. Research throughout Yuscarn, Honduras ( Morera and Gladwin 2006; Morera 1999) indica tes that resource access is significantly tied to agricultural intensification measures. Research throughout some villages in western Kenya also indicates a positive link between wealth and the uptake of labor intensive soil fertility enhancing technologi es (Place et al. 2005) However, labor intensive soil fertility enhancing technologies in western Kenya were researched and developed for lower resource farmers unable to afford mineral fertilizers. The original assumption made by Sanchez et al. (1997) wa s that the technologies would be of particular importance to land and cash constrained farmers grappling with declining yields. Nonetheless, the assumption made in this study is that labor intensive soil fertility enhancing technologies are a form of inte nsification requiring labor inputs that wealthy not poor farmers can better afford. Moreover, because landholding size often denotes wealth irrespective of its direct effect on agricultural inputs and outputs as described in H1, it was incorporated into the following hypothesis as well.
84 H3: The wealthier a household, the more it invests in agricultural productivity: Household wealth indicated by greater landholding size, number of cattle owned, cumulative years of schooling, and the receipt of bank loans leads to greater investment in one or more of the following: kg/acre of green manure, square meters/acre of improved fallows, kg/acre of mineral fertilizer, days/acre of hired oxen, days/acre of hired labor, and/or kg/acre maize and bean seed. Because se veral proxies for wealth, such as land ownership, cattle ownership, and off farm nonagricultural work in Kenya tend to be controlled by men, it follows that wealth mediated investments in agricultural productivity should be higher for male headed household s than for female headed households. That is, if it is the wealthier households that can invest in greater inputs of labor, seed, and fertilizer, then these should also tend to be male headed households. On the other hand, because off farm nonagricultura l work generally involves the out male(s) (Orvis 1997), de jure female headed households, like wealthy male headed households, will also have greater resources to invest in inputs of labor, seed, and fertilizer. The overall effe ct should be such that there is no difference between male headed and female headed household investments in agricultural productivity, including technological choice. H4: There is no difference between male headed and female headed household investment i n one or more of the following: kg/acre of green manure, square meters/acre of improved fallows, kg/acre of mineral fertilizer, days/acre of hired oxen, days/acre of hired labor, and/or kg/acre maize and bean seed. An important assumption of this study is that labor intensive soil fertility enhancing technologies are no more affordable to resource constrained farmers than are mineral fertilizers because labor inputs, long believed to be abundant, inexpensive, and the only factor of production controlled by the poor (Tomich et al. 1995; Marx 1859) can always be channeled into more remunerative activities if the payoff for labor investments in soil
85 fertility enhancement are uncertain. For this reason, wealthier farmers should be better able to invest in them in the same way they are hypothetically better able to invest in mineral fertilizers. That is why the above hypotheses do not differentiate on the basis of technology because whether labor intensive or cash intensive, where labor and cash units are inter changeable, all forms of soil fertility enhancement represent measures of intensification. Nonetheless, labor intensive soil fertility enhancement technologies remain an alternative or complement to mineral fertilizers which farmers may be unaware of. Thu s an important objective of this study was to determine whether national or international agricultural institutional efforts to disseminate labor intensive soil fertility enhancement technologies have been long lasting. Previous research experience in Hon duras (Morera and Gladwin 2006; Morera 1999) indicates that technical knowledge is a necessary but insufficient condition of technological adoption. Farmers are often well versed in labor intensive soil fertility enhancing technologies but nonetheless cho ose to channel their labor into alternative livelihood strategies. Because resource access is a limiting factor in the uptake of agricultural intensification measures, there should be no difference in technological practices between nearby villages where agricultural institutions have disseminated labor intensive soil fertility enhancing measures in some and not in others. Thus the current study tested the following hypothesis where the sample data comprised farmers from two villages where labor intensive soil fertility enhancement technologies were disseminated 10 years ago as well as farmers from another two villages where labor intensive soil fertility enhancement technologies were not disseminated.
86 H5: There is no difference in soil fertility enhancem ent practices among farmers living throughout nearby villages where labor intensive soil fertility enhancement technologies have been disseminated in some villages and not in others. In addition to agricultural institutional efforts to affect farming pract ices, there are other factors originating beyond the farm that may provide incentives or disincentives to agricultural intensification. The most obvious of these is prices. If input prices are high and output prices are low, a simple mathematical equatio n will indicate whether an agricultural endeavor is feasible. Moreover, a cursory examination of input and output markets will reveal possible explanations for their prices. For this study, a simple comparison of agricultural input and output prices durin g 2004 and 2007 as well as a firsthand comparison of transportation conditions between eastern and western Kenya and documentation of oil prices during the same period patter ns. The following two part hypothesis incorporates basic indicators to gauge the likelihood of intensification or extensification in western Kenya. H6: If the international price of oil rises and driving time between eastern and western Kenya increases a s a result of poor road conditions, the price of imported inputs (i.e. fertilizer) will rise faster than the price of outputs (i.e. maize) and farmers will not invest in the intensification of their maize production through the application of fertilizer. D ata Collection This study predominantly took place throughout four farming villages in western Kenya, differentially involved in the past with a local soil fertility management project. The project disseminated organic soil fertility enhancement technolog ies, particularly improved fallows and biomass transfer, to both male and female farmers. The project was collaboratively implemented by the International Center for Research in Agroforestry (ICRAF) now known as the World Agroforestry Centre -and the Kenya
87 limited farmers with farming practices that promised to reduce their need for expensi ve fertilizers (Place et al. 2005). Of the four villages selected, two had been involved with the project. Thus, the villages of Mutsulio and Shikusi were selected for their involvement with the PLAR project while the villages of Lugango and Shinakotsi w ere selected for their lack of involvement with the PLAR project. All four villages are located near one another within Shinyalu Division in the P rovince of Western Kenya and share similar ecological and infrastructural conditions. Nonetheless, Mutsulio l ies on a somewhat steeper ridge and furthest from the Khayega tarmac, which runs perpendicular to the unpaved road leading to the selected villages, as described and illustrated in Chapter 3 Village selection was based on the assumption that participation with the PLAR project would have resulted in familiarization with and practice of soil fertility enhancement practices. It was assumed that at least some farmers in Mutsulio and Shikusi would still be practicing soil fertility enhancement practices at th e time of this study. On the other hand, it was assumed that a lack of project involvement would have affected knowledge of the technologies, thereby resulting in fewer (although not necessarily significantly fewer) farmers practicing the soil enhancement technologies throughout the other two villages, Lugango and Shinakotsi which had not participated in the PLAR project. Data were collected for this study during three fieldwork seasons in order to gain a longer term perspective than would be possible had data been collected during a single
88 consecutive fieldwork season. The first fieldwork season took place during a period of 6 months, from early February through late July of 2004. The second fieldwork season took place during a period of 9 months, from late March through late December of 2007. The third and final fieldwork season took place during a period of 2 months from mid June through mid August of 2008. The first and second fieldwork seasons focused on household interviews with farmers living in t he selected four villages of western Kenya. The third fieldwork season focused on interviews with personnel employed by nearby institutions and organizations affecting the farmers of Mutsulio, Shikusi, Lugango and Shinakotsi The third fieldwork season also focused on gathering secondary data from national archives in Nairobi, fieldwork season was to contextualize the information gathered from the household interviews. A stratified sample of 120 households across the four villages was selected. The first selection consisted of thirty male headed households (MHH) and thirty female headed households (FHH) randomly chosen from a compilation of Mutsulio and Shikusi lists. The second selection consisted of thirty MHH and thirty FHH randomly chosen from a compilation of Lugango and Shinakotsi farmers were compiled with the assistance of Shinyalu s assistant chief and a community leader, both of whom had worked as liaison farmers with the PLAR project and were well acquainted with Shinyalu s farmers. The assistant chief and community leader also located two fieldwork assistants who acted as language interpreters for this study duri ng 2004 and 2007, respectively.
89 The fieldwork assistants were fluent in English, Swahili and Luhya mother tongue and were familiar with the location of the households to be interviewed. A third fieldwork assistant, independently located in Ki sumu, Kenya, was also hired during 2007 for her language proficiency and previous experience working with diverse peoples. She moved to Kakamega, a large town located a few kilometers north of Shinyalu, and commuted each day to the selected villages for t he duration of the 2007 fieldwork season. Each of the three fieldwork assistants was paid a wage of 300 Kenyan Shillings per day of fieldwork, which typically stretched from 9 am to 5 pm. The wage was typical for specialized non rural labor in this area a nd was negotiated with the help of the assistant chief and community leader. In 2007, the wage was six times the wage for unspecialized agricultural labor. During the 2004 fieldwork season, I lived in a convent in Mukumu a small township located within b iking distance of the selected villages. The convent was composed of Kenyan and Ugandan nuns, mostly hailing from nearby villages. The large house in which they lived featured spare rooms, electricity, running water, and security. Given my frequent need to work on a laptop, it was recommended I rent a room in the convent and commute by bicycle to the villages. In my spare time, I accompanied the nuns to local events, such as village celebrations, holiday feasts, and church gatherings. These extracurric ular activities provided an unexpected source of comradeship and inclusion as well as a privileged view of rural life in western Kenya. On fieldwork days, I took a boda boda (bicycle taxi) from the convent to Shikusi where the first fieldwork assistant w ould be waiting. Together we would head out from there by foot to the selected households to conduct interviews.
90 During the 2007 and 2008 fieldwork seasons, I lived in a convent in Kakamega with the same order of nuns. On fieldwork days, together with th e third fieldwork assistant, I commuted by matatu (small, privately owned buses serving as public transportation) to the Khayega tarmac which runs perpendicular to the unpaved road leading to the selected villages. From the tarmac, the third fieldwork ass istant and I would take two boda bodas to Shikusi, where the second fieldwork assistant would be waiting. From there the three of us would head out by foot to the selected households to conduct interviews. Thus, each of the 120 selected households was vis ited in the company of the hired fieldwork assistants who helped interpret the interviews. Consent forms were obtained for all interviews and research participants were compensated for their time. In 2004, compensation consisted of a small Polaroid famil y portrait. In 2007, compensation consisted of either a small bag of tea leaves or half a kilogram of sugar. Three ethnographic field methods were used throughout the individual visits to the 120 selected households: The ethnographic interview, the quest ionnaire, and participant observation. These were based on the method employed by Spradley (1979) in The Ethnographic Interview The ethnographic interviews consisted of informal dialogues between the farmers, the research assistants, and me. The interv iews conducted in 2004 were in depth, semi structured, and open ended. Heads of households were asked to discuss their experiences with the PLAR project, their sentiments regarding the recommended technologies, farming routines, livelihood strategies, and household dynamics. The first twenty five interviews conducted in 2007 also took place in this manner. A survey questionnaire having a similar objective was
91 created after approximately 40 such interviews, during the 2007 fieldwork season. The questionn aire was updated whenever new questions arose during a household interview. The questionnaire continued to be refined throughout the entire first half of the 120 household visits. Yet once the questionnaire was developed, interviews became more structure d. During the second half of the 120 household visits, the interviews were strictly a survey. When these were completed, the first 60 households were visited a second time in order to elicit any questions not asked during the earlier data collection phas e. Participant observation also took place during the interviews. On several different occasions, interviews were conducted in the midst of farming activities. Several interviews occurred in the field as the soil was tilled, on the floor of a home as v egetables were prepared for sale, and even on the grass as maize kernels were removed from cobs. The technique involved participating in the activities of the informant and learning while observing (Spradley 1979). Conceptually, the difference between obs ervation and participant observation is in the research partic 34) compares these two modes of respectively. Spradley emphasizes that treating people as informants emphasizes the perspec tive. Similarly, Harris (1979: 32) distingui shes between emic knowledge and etic may never be entirely grasped by an outsider, ethnographic interviews and participant
92 observation are critical in designing meaningful and relevant questions about an treating them as actors, or even subjects and respondents (Spradley 1979:34). Bernard (2006: questions, subjects are the subject what they think you need to know According to Bernard (2006: 196) there are two kinds of informants: specialized informants and key info dgeably about it (Bernard 2006: 196). For example, because this study sought to gain an understanding of agricultural decision making in western Kenya, as a researcher I interviewed western Kenyan farmers. In this case, the farmers interviewed were specialized informants. On the other hand, key own, willing to share all their kno 196). Key informants can explain to an outsider the meaning of a broad range of phenomena in their own culture. Bernard also notes that a researcher does not choose key informants key informants and the resear cher choose each other over time as a result of ea sy conversation (Bernard 2006: 196). For this study, several individuals graciously served as key informants. The community leader who had worked as a liaison farmer for the PLAR project and helped compile me in order to help me make sense of my surroundings. He explained to me the agricultural technologies PLAR recommended to the farmers during the 1990s. He also
93 provided a history of the area and offered descriptions of its customs. Most importantly, he helped me construct questions in ways that would be locally acceptable. One of the nuns at the convent also ensured questions, such as those regarding income, would not be inadvertently presented in offensive ways and frequently met with me to chat about throug hout the hillsides of Shinyalu Division I had the great fortune of enjoying many stor ies from my fieldwork assistants who were pivotal in providing background information regarding Kenyan farming, education, social life, politics, and the economy from a Kenyan perspective. Data Analysis Because this study set out to identify factors affect ing soil fertility management practices and staple crop productivity in western Kenya, the agricultural inputs invested and outputs yielded by a sample of farmers throughout four western Kenyan villages during a full growing season were each to be measured as dependent variables 1 These dependent variables were to be calculated in acres or kg per total acres of agricultural land as follows: acres of improved fallows, kg of biomass transfer, kg of chemical fertilizer, days of oxen hire, days of labor hire, kg of maize and bean seed, and kg of maize and bean yields. The independent variables included: Gender: Male Headed Household (MHH) or Female Headed Household (FHH) (measured as a categorical variable and coded as 1 or 0 respectively ) Number of househo ld members Total household land (measured in acres as a continuous variable ) 1 Measurements were self reported. Although Bernard (2006:245) notes that self reported behavior and self reported measurements of environmental circumstances result in errors, the logistical and budgetary eports of socioeconomic data.
94 Cattle ownership (measured as a continuous variable) Access to credit and bank loans (measured as a categorical variable where an affirmative response was coded as 1 and a negativ e response was coded as 0 ) informants were asked whether they had access to credit or received a bank loan over the last 5 years Off farm employment (measured as a categorical variable and coded according to whether household members were engaged in off fa rm agricultural work, on farm nonagricultural work, or off farm nonagricultural work) P articipation in the PLAR P roject (measured as a categorical variable) Receipt of project incentives (measured as a categorical variable) informants were asked if they ha d received payments of any kind in order to experiment with a technology, such as bags of fertilizer or seed Receipt of remittances (measured as a categorical variable) Months of maize purchase (number of months per year during which maize must be purchase d as a result of insufficient household production) Membership in community organization s (measured as a categorical variable) Cumulative schooling of children (in number of years of schooling per child) However, by the end of the fieldwor k season of 2007, it became clear that none of the sampled households were practicing improved fallows and only three were practicing biomass transfer This meant that, for improved fallows and biomass transfer, there was no variation in the dependent var iable against which the effects of the independent variables could be analyzed. On the other hand, most farmers used at least some chemical fertilizer. Thus, chemical fertilizer use became the only measure of soil fertility enhancement. Additionally, t he long rains season of 2007 became the frame of reference for most of the independent variables. For example, farmers were asked how many kg of diammonium phosphate (18 46 0) (DAP) chemical fertilizer were applied during the
95 long rains season of 2007, ho w many agricultural laborer days were hired during the long rains season of 2007, how many household members were employed in off farm jobs during the long rains season of 2007, how many head of cattle were owned during the long rains season of 2007, how m any kg of maize were harvested at the end of the long rains season of 2007, etc. This was done in an effort to systematize the observations, making them comparable, and to obtain an entire agricultural season of socioeconomic data. The long rains season was selected because most farmers plant during this period of abundant rainfall whereas not all farmers plant during the short rains season when rainfall is less dependable. Multiple Regression Analysis Multiple regression analysis in Excel was used in ord er to test the first five hypotheses of this study and determine how much each of the independent variables contributes to the prediction of agricultural inputs invested and outputs yielded in the production of staple crops by the sample of western Kenyan farmers during the long rains season of 2007. Multiple regression analysis was used because it estimates an equation that accounts for these relationships as well as the interrelationships among the indep endent variables (Bernard 2006: 661), providing the best fitting line between is accounted for by a series of independent variables after taking into account all of the overlap in variances accounted for across the in depe 2006: 661). It also provides a measure of association between the dependent variable and each of the independent variables, holding the other independent variables constant. This type of analysis was feasible because in this study the dependent variable was always a continuous, quantitative variable while the independent variables
96 were either continuous quantitative variables or categorical variables coded as 1 or 0 Had the dependent variable been a categorical variable, mul tiple regression analysis would have been inappropriate. The independent variables included in each multiple regression equation consisted solely of those variables theorized to affect the dependent variable, whether or not their associations with the depe ndent variable resulted significant once the regression was the independent variable that correlates best with the dependent variable and then adds in [independent] v ariables one at a time, accounting for more and more variance, until all the specified variables are analyzed, or until variables fail to enter because (Bernard 2006:667). While stepwise regression omits from the multiple regression equation those variables whose associations with the dependent variable are not significant, it can result in omitted variable bias, particularly when independent variables synergistically affec t the dependent variable. To build a multiple regression equation that tested the first hypothesis of this study (H1), which posits that farmers in western Kenya are intensifying agricultural production by increasing the availability of soil nutrients to t heir staple crops, production and input functions were estimated for each of the staple crops and for each of the main inputs farmers invest in to produce their staple crops, respectively. Therefore, production functions were estimated for maize yield per acre ( Q mzyield/acre ) and for bean yield per acre ( Q bnyield/acre ). Input functions were estimated for fertilizer use per acre ( Q fert/acre ),
97 oxen use per acre ( Q oxen/acre ), labor hired per acre ( Q labor/acre ), maize seeding per acre ( Q mzseed/acre ), and bean seeding per acre ( Q bnseed/acre ) 2 The independent variables included in the production functions were limited to the inputs western Kenyan farmers required to produce maize and beans, identified through ethnographic interviews. For example, during one in terview, a farmer explained that she used part of the funds gained through a bank loan to purchase fertilizer for her maize and beans. In another interview, a farmer pointed out that he improved his maize and bean yields by applying the seeding rates and distances recommended by the all included in the set of independent variables of which maize production and bean production are functions, as illustrated below. The sam e set of independent variables was used to estimate both functions because maize and beans are grown together. The production function for maize yield per acre ( Q mzyield/acre ) was estimated as: Q mzyield/acre = f ( MHHorFHH, TotalHHMembers, ProjectParticipa tion, Mz&BnAcreage, VisitExtOffice, TrainingBaraza, GroupBorrow, BankLoan, NonAgOFW, AgWageWork, Brewing, OxenUse/Acre, HiredLabor/Acre, MaizeSeedingRate, BeanSeedingRate, TotalFert/Acre) The production function for bean yield per acre ( Q bnyield/acre ) was estimated as: Q bnyield/acre = f ( MHHorFHH, TotalHHMembers, ProjectParticipation, Mz&BnAcreage, VisitExtOffice, TrainingBaraza, GroupBorrow, BankLoan, NonAgOFW, AgWageWork, Brewing, OxenUse/Acre, HiredLabor/Acre, MaizeSeedingRate, BeanSeedingRate, TotalFert /Acre) On the other hand, the independent variables selected for each of the input functions were those hypothesized to affect the application of intensification measures, 2 To increase the availability of soil nutrients to crops, farmers normally increase plowing and seeding rates in addition to increasing fertilizer use or other forms of soil fertility enhancement.
98 such as gender, project participation, physical and social capital, off farm work, a nd the application of other inputs. For example, the input function for fertilizer use per acre ( Q fert/acre ) was estimated as: Q fert/acre = f ( MHHorFHH, TinRoof, CattleOwned, AcresOwned, ProjectParticipation, GroupBorrow, BankLoan, NonAgOFW, AgWageWork, B rewing, OxenUse, HiredLabor, LaborShareOffFarm, LaborShareOnFarm, MaizeSeedingRate, BeanSeedingRate ) The same set of independent variables was included in the functions estimated for oxen use per acre ( Q oxen/acre ), labor hire per acre ( Q labor/acre ), maize seeding per acre ( Q mzseed/acre ), and bean seeding per acre ( Q bnseed/acre ). Moreover, because gender, project participation, physical and social capital, and off farm work are the subjects of H2 through H5 of this study (listed on pages 6 through 9), in ad dition to testing H1, the input functions also tested H2 through H5. Because wealth was hypothesized in H3 to affect soil fertility management and staple crop productivity, an aggregate wealth index based on the visible conditions of a building materials, roof, floor, furniture, etc.), land, and livestock was created during the first half of the 2007 fieldwork season and a function estimated for it in order to assess its relationship to agricultural input investments as well as other pr oxies for wealth. Because gender, social capital, and alternate livelihood strategies were theorized to affect wealth, these were also included in the set of independent variables. The function was estimated as: W ealth = f (MHHorFHH, TotalHHMembers, TinR oof, CattleOwned, AcresOwned, TrainingBaraza, BankLoan, OxenUse/Acre, HiredLabor/Acre, TotalFert/Acre, NonAgOFW, AgWageWork, Brewing, QMoMzPurch)
99 variables because, as the ind ex was created only for Mutsulio and Shikusi villages, the data set in this case solely comprised project participants. Additional functions that could be applied towards the entire data set were estimated for land ( Q acresowned ) and cattle ( Q cattleowned ) a s wealth proxies in order to assess their relationships to alternate livelihood strategies and socioeconomic indicators. Thus the set of independent variables included the three main alternative livelihood strategies, off farm nonagricultural work, agricu ltural wage work, and home brewing, as well as socioeconomic indicators such as loan receipts and participation in merry go rounds and the PLAR project. The following function was estimated for landholding size ( Q acresowned ): Q acresowned = f ( MHHorFHH, Ti nRoof, CattleOwned, ProjectParticipation, MerryGoRound, GroupBorrow, BankLoan, NonAgOFW, AgWageWork, Brewing ) A similar function using the same set of independent variables shown above was estimated for total cattle owned ( Q cattleowned ). During the course of ethnographic interviewing, several informants pointed out the manner in which income generated from alternative livelihood strategies was used, elucidating the interrelationships between nonfarm work and farming an important objective of this study. T hey explained, for example, that remittances received from farm nonagricultural work were often applied towards school fees, while the income generated from the sale of brewed beer and liquor was often applied towards maize purchases. As a r esult, additional functions were estimated to predict
100 household consumption. Cumulative years of schooling ( Q yrsschooling ) were estimated as: Q yrsschooling = f ( MHHorFHH, Ac resOwned, CattleOwned, BankLoan, NonAgOFW, AgWageWork, Brewing, Mutsulio, Shikusi, Shinakotsi ) Months of annual maize purchase for household consumption ( Q monthsmzpurch ) were estimated as: Q momthsmzpurch = f ( MHHorFHH, TotalHHMembers, TinRoof, CattleOwned, ProjectParticipation, TrainingBaraza, BankLoan, KaleAcreage, MaizeAcreage, OxenUse/Acre, HiredLabor/Acre, TotalFert/Acre, MzYield/Acre, NonAgOFW, AgWageWork, Brewing ) During the course of ethnographic interviewing it also became evident that kale producti on was an important source of income for many farmers. As an alternative livelihood strategy, it was particularly relevant to this study as it appeared that some farmers were substituting, rather than complementing, their maize production with kale produc tion. It became important to assess what kind of farmer would be more likely to switch from staple crop production to cash crop production. For this reason, the set of independent variables included physical and social capital indicators, such as land, c attle, and participation in training barazas and the PLAR project, as well as alternative livelihood strategies, such as off farm nonagricultural work, agricultural wage work, and home brewing. Therefore, a function was estimated for the amount of acres d evoted to kale production ( Q kaleacreage ) as follows: Q kaleacreage = f ( MHHorFHH, TotalHHMembers, TinRoof, CattleOwned, MaizeAcreage, ProjectParticipation, TrainingBaraza, BankLoan, NonAgOFW, AgWageWork, Brewing, EnoughMzHarvest )
101 In sum, while the first set of functions analyzes factors affecting the production of maize and beans, the second set of functions analyzes factors affecting inputs towards maize and bean production, including soil fertility management measures. Together these functions test H1 thr ough H5 of the study. The third set of functions analyzes the role of wealth in soil fertility management and agricultural productivity. The fourth set of functions analyzes the interrelationships between off farm work, farming, and investment priorities while the final equation analyzes the relationship between cash crop production and staple crop production. In total, the functions estimated specialization. Frequency Dist ributions Data were also analyzed using frequency distributions in order to capture trends in household nonnumeric responses to semi structured or open ended questions. For example, frequency counts were tabulated for household data on social capital, inc ome and expenses, and attitudes towards quality of life, agriculture, and the economy. For data on social capital, frequency counts were tabulated for total households barazas (commun ity workshops on agricultural technologies), agricultural labor sharing groups, other social groups, and agricultural extension visits. For data on income and expenses, frequency counts were tabulated for the most common sources of income, the most common expenses, the most common crops sold, and whether yields provided sufficient food or whether food had to be purchased to supplement household production. For data on attitudes towards quality of life, agriculture, and the economy, frequency counts were t abulated for households observing either improvements or
102 economy. Script Analysis A script analysis was made of various agricultural routines depicted through fa rming calendars. Schank and Abelson (1977) define individuals' routine patterns of activities as "scripts." Scripts represent unconscious detailed knowledge that allows individuals to do less mental processing during frequently experienced events. In th is farming practices an emic distinction informants often made during open ended ethnographic interviews when commenting on the nature of farming routines. Thus, farming calendars were elicited from several informants and also compiled from the many in depth and semi structured interviews conducted throughout all four villages. Normative themes regarding idealized and constrained farming were incorporated into the calenda rs. Farm History Analysis Brief farm histories were elicited from several informants in order to provide a longitudinal backdrop to current agricultural and nonagricultural livelihood strategies observed during 2007. The histories bring to life the farmin g opportunities and constraints leading to current agricultural practices examined in this study, thereby providing a qualitative context in which quantitative results can be conceptualized. Informants were asked to describe their lives over the last deca de with particular reference to changes in agricultural technologies and income generation.
103 Feasibility Analysis Three simple calculations were made in order to compare 1) organic and inorganic fertility management practices, 2) on farm and off farm liveli hood strategies, and 3) 2004 and 2008 input and output prices involved in the production and sale of maize. The first calculation documents the labor required in the biomass transfer of an amount of Tithonia diversifolia approximately equivalent to an app lication of 50 kg/ha of diammonium phosphate (18 46 0) (DAP). The calculation then converts the labor units to cash units and compares the cash cost of biomass transfer to the cash cost of 50 kg of DAP. The second calculation compares the labor and/or ca sh requirements of three alternatives for obtaining a 90 kg bag of ma ize using 2007 Shinyalu Division prices. In one scenario, the household purchases the maize. In a second scenario, the household produces the maize using biomass transfer. In a third s cenario, the household produces the maize using inorganic fertilizer. Finally, the third and last calculation tests the sixth hypothesis (H6) of this study and compares the prices of inputs and outputs involved in producing and selling maize i n Shinyalu D ivision from 2004. Together the three calculations test the feasibility of soil fertility management and agricultural intensification in western Kenya. Conclusion This chapter has discussed the motivations, assumptions, objectives, hypotheses, ethnographi c techniques, and analytical tools involved in carrying out the empirical work of this study. By combining quantitative and qualitative methods, the study examined the nature of soil fertility management in western Kenya, investigating its role in agricul tural productivity and exploring the effects of technological adoption, gender, off farm work, and other off farm factors on both. The study sought to establish whether
104 institutional efforts to disseminate improved fallows and biomass transfer had lasting farm work and gender affected soil fertility enhancement practices and, in turn, agricultural intensification. The following chapter now turns to the research results themselves.
105 CHAPTER 5 RESEARCH RESULTS: THE MAKING OF A SUCCESSFUL FARM ER In t roduction This chapter presents the results of quantitative and qualitative analyses of ethnographic data collected from 120 households located throughout the villages of Mutsulio Shikusi Lugango and Shina kotsi in western Kenya. Five sets of analyses were applied to the sample data: 1) multivariate statistics 2) descriptive and inferential bivariate statist ics, 3) script analysis, 4) farm history analysis, and 5) feasibility analysis. Multivariate stati ethnographic questions regarding their most recent farming season. Descriptive and ended responses to general et hnographic questions regarding social, financial, and agricultural activities. Script analysis was used to analyze farme history analysis was used to understand the conditions leading up to, and the results obtained from, a dopting various technological and livelihood strategies. Lastly, feasibility analysis was used to assess the economic feasibility of several farming alternatives by calculating various costs and benefits derived from each using real world prices. Statisti cal analysis was applied to responses elicited systematically of all farmers during the entire fieldwork season. On the other hand, script analysis was applied to agricultural routines compiled early in the field work season from open ended ethnographic i nterviews during which farmers commented on their daily routines and nature of farming. Similarly, farm histories were compiled early in the fieldwork season from open ended, in depth interviews during which farmers commented on their past experiences and choices leading up to their current lives on the farm Lastly, feasibility
106 analysis was applied to hypothetical alternatives based on several technological and livelihood options recommended by agricultural institutions or witnessed throughout the fieldw ork season. Together, the five sets of analyses were used to address the six research qu estions discussed in Chapter 4 concerning agricultural productivity, soil fertility management, technological adoption, and the role of gender, off farm work, and other off farm factors on these. Results of multivariate statistics were used to test the first five hypotheses listed in Chapter 4 regarding the effects of landholding size, off farm work, wealth, gender, and PLAR project participation on the per acre applica tion of soil fertility enhancement measures and other intensification measures as well as on overall maize and bean production per acre. Results of descriptive and bivariate statistics were used to illustrate the effects of gender and project participatio n on farming activities. Res ults of script analysis and farm history analysis were used to contextualize and substantiate quantitative results. Results of feasibility analysis were used to test the sixth hypothesis listed in Chapter 4 regarding the effec ts of off farm factors on soil fertility management. Multivariate Statistics: Multiple Regression Analysis The results of multiple regression analy sis are illustrated in Tables 5 2 through 5 1 9. Each multiple regression corresponds to a linear equation, where y=a+bx 1 +bx 2 +bx 3 n + and where is an error term. The dependent y variables for each of these appear in blue at the top left hand corner of each of the tables. Each of the independent x variables are listed in the first column of each table under X Variables after Intercept The b coefficients for each of the x variables are listed in the second column under Coefficients for each of the tables. The first of these is the coefficient for the Intercept
107 representing a in the equation. As an e xample, the linear equation co rresponding to Table 5 2 is y=269.80 13.87x 1 +1.36x 2 58.67x 3 32.94x 4 +80.77x 5 45.22x 6 +30.85x 7 +10.41x 8 0.01x 9 37.72x 10 12.08x 11 +0.25x 12 +35.59x 13 +3.37x 14 +11.08x 15 +1.47x 16+ where y= maize yield per acre cropped in maize and beans, x 1 = MHH or FHH, x 2 = TotalHHMembers, x 3 = Project/NonProject, etc. Table 5 1 provides a legend for the codes used to identify each of the variables included in the multiple regression analyses. As exp lained in Chapter 2 because western Kenya reportedly has one of the highest population growth rates in the world (Sanchez et al. 1997), I expected to find households increasing the productivity of their maize and bean farming through the application of so il fertility enhancing measures such as fertilizer, biomass transfer, and improved fallows as well as other intensification measures such as oxen use and hired labor. I also expected to find households with larger landholdings, more cattle, and more nonfa rm income relative to agricultural off farm income to be applying greater amounts of these measures per acre than households with smaller landholdings, fewer cattle, and less nonfarm income relative to agricultural off farm income. On the other hand, I di d not expect to find any differences in soil fertility management and other intensification measures between male headed and female headed households or between PLAR participant and non participant households. Therefore, I expected multiple regression anal ysis to result in a significant and positive association between landholding size and each of the intensification measures applied per acre, as suggested in the first hypothesis. I also expected significant,
108 positive associations between each of the inten sification measures applied per acre and both on farm nonagricultural work ( brewing ) and off farm nonagricultural work ( nonagofw ), as suggested in the second hypothesis. By the same token, however, I expected significant, negative associations between eac h of the intensification measures applied per acre and agricultural off farm wage work. As suggested in the third hypothesis, I expected significant, positive associations between intensification measures applied per acre and each of the wealth proxies, m easured as acres owned, On the other hand, I did not expect multiple regression analysis to result in significant associations between any of the intensification measur es and either gender of the household head (MHH or FHH) or PLAR project participation, as suggested in the fourth and fifth hypotheses. However, it should be kept in mind that while fertilizer use, biomass transfer, and improved fallows were all to be meas ured as soil fertility enhancement practices, fieldwork revealed that by 2007 none of the 120 sample households were applying improved fallows and only three reported the use of biomass transfer. Instead, most households applied small quantities of fertil izer to their land, their sole means of enhancing its soil fertility. This meant that, for improved fallows and biomass transfer, there was no variation in the dependent variable against which the effects of the independent variables could be analyzed. T hus, the use of improved fallows and biomass transfer could not be included in multiple regression analysis. Fertilizer application per acre was the only measure of soil fertility enhancement which, together
109 with oxen use per acre, hired labor per acre, a nd seed per acre were measured as indicators of intensification. Production Equations Production equations were estimated in order to assess sample household outputs of maize and beans per acre, thereby examining agricultural productivity and determining w hether households were intensifying or extensifying staple crop production. As stated in the first hypothesis, I expected farmers to be intensifying their maize and bean production. Therefore, I expected the production regressions to result in significan t and positive associations between landholding size and quantities of maize and bean yields. That is, if households were intensifying staple crop production, the acre. However, the production regressions instead resulted in negative associations between landholding size and maize and bean yields per acre, with only the amount of maize yield per acre significantly associated with landholding size. The association betwee n landholding size and the amount of bean yield per acre was not significant. The results of the production regressions, the multiple regression analyses of household production of maize per acre of land cropped in maize and beans ( y=maize yield per acre c ropped in maize and beans ) and household production of beans per acre of land cropped in maize and beans ( y=bean yield per acre cropped in maize and beans ), appear in Tables 5 2 and 5 3, respectively. As noted earlier, each y variable appears in blue at t he top left hand corner of each table. Significance F corresponding to the overall significance of each equation, appears towards the center of each table. X variables, whose associations with the dependent variable are significant, are marked
110 in red. X variables, whose associations with the dependent variable are only somewhat significant, are marked in orange. Maize production Table 5 y=maize yield per acre cropped in maize and beans ), the dependent varia ble, is significantly and negatively associated (P=.04) with land size ( maize and beans acreage ). On the other hand, the positively associated with oxen used ( oxen used per acre of land cropped in maize and beans ) (P=.004), bean seeding rate ( bean seeding per acre of land cropped in maize and beans ) (P=.03), and fertilizer used ( total fertilizer used per acre of land cropped in maize and beans ) (P=.11). Project participation ( PLAR project participation/nonparticipation ) is somewhat significantly and negatively associated with the dependent variable (P=.18). Bean production Table 5 y=bean yield per acre cropped in maize and beans ), the de pendent variable, is negatively but not significantly associated with project participation (P=.03) and brewing of or busaa for sale ( brewing ) (P=.09), in both cases significantly and positively associated with the ability to borrow from a group (P=.07). Yet it is only somewhat significantly and positively associated with bean seeding rate (P=.18).
111 Input Equations Inpu t equations were estimated in order to assess sample household applications of inputs per acre of land cropped in maize and beans for oxen used, hired labor, fertilizer used, and maize and bean seeding rates. Similar to production equations, input equatio ns were estimated as a way of examining agricultural productivity and determining whether households were intensifying or extensifying staple crop production. Because I expected households to be intensifying their maize and bean production, I also expecte d the input regressions to result in significant and positive associations between landholding size and quantities of inputs applied per acre. However, the input regressions instead resulted in negative associations between landholding size and quantities of inputs applied per acre, with only fertilizer u se (Table 5 5) and bean seeding rate (Table 5 13) significantly associated with landholding size. Oxen u se (Table 5 7) and maize seeding rate (T able 5 11) were somewhat significantly associated with landh olding size. The selection of independent variables included in the input equations also allowed me to assess whether the kind of off farm work affected staple crop production, as suggested in the second hypothesis listed in Chapter 4 Because I expected that greater amounts of nonagricultural off farm income generation would lead to greater amounts of agricultural inputs applied to staple crop production, I also expected a significant and positive association between off farm nonagricultural work and each of the inputs applied. On the other hand, because I expected that greater amounts of agricultural wage work lead to smaller amounts of agricultural inputs applied to staple crop production, I expected a significant and negative association between agricu ltural wage work and each of the inputs applied. The results of the input regressions
112 supported these hypotheses in five cases. Off farm nonagricultural work resulted significantly and positively ass ociated with oxen use (Tables 5 6 and 5 7) while agricu ltural wage work resulted somewhat significantly and negatively associated with hired labor (T able 5 9) and with bean seeding rate (Tables 5 12 and 5 13). Finally, the selection of independent variables included in the input equations also allowed me to as sess whether gender and project participation affected the application of soil fertility enhancement practices and other intensification measures. As suggested in the fourth and fifth hypothese s listed in Chapter 4 (the third hypothesis is addressed by th e wealth equations discussed further ahead), I expected no gender or project participation differences in the application of soil fertility enhancement practices or other intensification measures. Thus, I expected the input regressions would not yield any significant associations between any of these. However, the input regressions indicate project participation is somewhat significantly and positively associated with fertilizer use (Table 5 4) and significantly but negatively associated with maize seedin g rate (Table 5 11). It is also significantly and negatively associated with oxen use. The input regressions also yielded a significant and positive association between female headed households an d hired labor (Tables 5 8 and 5 9). Nonetheless, because none of the 120 sample households were using improved fallows and only 3 were using biomass transfer, the effects of gender and project participation on labor intensive soil fertility enhancement measures were nil. The results of the input regressions, mul tiple regression analyses for each of the agricultural inputs farmers used on their fields throughout the growing season in order to produce outputs of maiz e and beans, appear in Tables 5 4 through 5 13. The five
113 inputs analyzed are: Fertilizer used, oxe n used, labor hired, maize seeding rate, and bean seeding rate. Two tables are presented for each input used per acre of land cropped in maize and beans, for a total of 10 tables. The first table for each input presents a multiple regression analysis of all the variables hypothesized to affect the application of that input, including the application of the remaining inputs. That is, the use of one input is hypothetically affected by the use of another input. For example, it is unlikely for a farmer to i nvest in the mineral fertilization of a field that is not properly plowed. However, it is possible that including highly correlated variables in a multiple regression analysis results in multicollinearity Accordi ng to Agresti and Finlay (1997: e certain important predictors are in the model, the addition of other variables often provides only a small boost in R 2 particularly when the additional variables have sma multiple regressions was run for each of the inputs analyzed, this time omitting the remaining input variables from the equation in an effort to control for multicollinearity. These are presented in a second table for each input. Fertilizer use The results of the first multiple regression analysis for fertilizer use as the dependent variable ( y = total fertilizer per a cre cropped in maize and beans ) appears in Table 5 associated with having a tin roof (P=.04), receiving a bank loan (P=.02), with oxen use (P=.02), hired labor (P=.04), participation in an off farm labor sharing group ( labor share off farm ) (P=.04), and bean seeding rate (P=01). Fertilizer use is also somewhat
114 significantly and positively associated with project participation (P=.12) while it is somewhat significantly bu t negatively associated with participation in an on farm labor sharing group ( labor share on farm ) (P=.14). The results of the second multiple regression analysis for fertilizer use as the depen dent variable appear in Table 5 5. The results are similar to the first regression analysis except that, in this case, fertilizer use is significantly associated (P=.04) with landholding size ( total acres owned ). The association is negative As mentioned earlier, the associations between landholding size and each of the inpu ts applied per acre are negative throughout all the multipl e regression analyses. Table 5 5 also indicates that fertilizer use here is significantly and positively associated with cattle owned ( total cattle ) (P=.04) and the ability to borrow fr om a group (P=.04). However, a analysis. Oxen use Table 5 6 presents the first of two multiple regression analyses for oxen use as the dependent variable ( y = oxen use per acre cropped in maize and beans ). Results indicate that oxen use is significantly and positively associated with cattle owned (P=.04), the ability to borrow from a group (P=.06), off farm nonagricultural work ( nonagofw ) (P=.05), maize seeding rate ( maize seeding per acre cropped in maize and beans) (P=.01), and fertilizer use (P=.02). On the other hand, oxen use is significantly but negatively associated with project participation (P=.07). Table 5 7 presents the second multiple regression analysis for oxen use as the dependent variable. The results are similar to those of the previous analysis except that here, oxen use is somewhat significantly and negatively associated with
115 landholding size (P=.15) and tin roof (P=.16). Also, cattle ownership is more highly a predictor of oxen use (P=.01). Hired labor The first of two multiple regression analyses for hired labor as the dependent variable ( y = hired labor per acre cropped in maize and beans ) appears in Table 5 8. Results indicate that hired la bor is significantly and negatively associated (P=.02) with whether a household head is male or female ( male headed household or female headed household ). Because male headed households were coded as 1 and female headed households were coded as 0, the neg ative sign of the coefficient indicates hired labor is significantly and negatively associated with a household head being male but significantly and positively associated with a household head being female. Results also indicate hired labor is significant ly and positively associated with cattle owned (P=.01), participation in an on farm labor sharing group (P=.002), bean seeding rate (P=.02), and maize seeding rate (P=.04). On the other hand, hired labor is significantly but negatively associated with par ticipation in an off farm labor sharing group (P=.10). The second multiple regression analysis for hired labor as the depend ent variable appears in Table 5 9. The results are the same as those for the previous analysis except that with the other agricultu ral inputs omitted from the regression, the association between hired labor and cattle owned is highly significant in this case (P=.002). The significance of the association between hired labor and gender of the household head remained the same (P=.02). M aize seeding rate Table 5 10 presents the first of two multiple regression analyses for maize seeding rate as the dependent variable ( y = maize seeding per acre cropped in maize and
116 beans ). Results indicate that maize seeding rate is significantly and pos itively associated with oxen use (P=.01) and bean seeding rate (P=.002). Maize seeding rate is also somewhat significantly and positively associated with participation in an on farm labor sharing group (P=.18) and somewhat significantly but negatively ass ociated with participation in an off farm labor sharing group (P=.16). Table 5 11 presents the second multiple regression analysis for maize seeding rate as the dependent variable. With the other agricultural inputs omitted from the regression in this cas e, the maize seeding rate is significantly and positively associated with the ability to borrow from a group (P=.06) and significantly but negatively associated with having a tin roof (P=.08). Maize seeding rate is somewhat significantly and positively as sociated with cattle owned (P=.18). Yet it is somewhat significantly but negatively associated with landholding size (P=.11). Bean seeding rate The last two input regressions ap pear in Tables 5 12 and 5 13. Table 5 12 presents the first of two multiple r egression analyses for bean seeding rate as the dependent variable ( y = bean seeding per acre cropped in maize and beans ). Results indicate that bean seeding rate is significantly and positively associated with hired labor (P=.02), maize seeding rate (P=. 002), and fertilizer used (P=.01). On the other hand, bean seeding rate is significantly but negatively associated with the receipt of a bank loan (P=.06). Landholding size (P=.19), the ability to borrow from a group (P=.15), and agricultural wage work ( agwagework ) (P=.13) are all somewhat significantly and negatively associated with the bean seeding rate. Table 5 13 presents the second multiple regression analysis for bean seed seeding rate as the dependent variable. With the other agricultural inputs o mitted from
117 the regression, the bean seeding rate in this case is significantly and negatively associated with landholding size (P=.02), its best predictor. Bean seeding rate also results somewhat significantly and negatively associated with having a tin roof (P=.14) and somewhat significantly and positively associated with cattle owned (P=.17) in this case. The negative association between bean seeding rate and agricultural wage work remains relatively the same. Wealth Equations Wealth equations were est imated as a means of assessing whether wealth indicators were associated with greater applications of soil fertility enhancement measures and other intensification measures, as suggested in the third hypothesis listed in Chapter 4 Because I expected weal th indicators such as landholding size, cattle ownership, and the receipt of bank loans to lead to greater applications of soil fertility enhancement measures and other intensification measures, I expected multiple regression analysis to result in signific ant and positive associations between each of the wealth variables and the application of agricultural inputs. The previous input regressions, overall, support these hypothes es as does Table 5 14, which treats wealth as an aggregate dependent variable and indicates that it is significantly and positively associated with fertilizer use per acre and oxen use per acre. Table 5 14 also indicates that wealth is significantly and negatively associated with the number of months throughout the year that maize mus t be purchased. Moreover, the wealth regressions support the hypotheses regarding nonfarm income generation and agricultural wage work. That is, off farm nonagricultural work resulted significantly and positively associated with landholding size (Table 5 15) as well as with cumulative years of schooling received by a house
118 17). On the other hand, agricultural wage work resulted significantly and negatively associated with landholding size (Table 5 15) and with cumulative years of schooling received by 17). farm income received as remittances. Several fema le the responsibility of the husband, noting that nonagricultural off farm income was applied towards school fees. In addition to cumulative years of schooling, off farm nonagricultural work is only significantly and positively ass ociated with oxen use (Tables 5 6 a nd 5 7) and landholding size (5 15). Multiple regression analyses of three measures of wealth appear in Ta bles 5 14 through 5 ate measure furniture, etc.), land, and livestock. The measure was created during the first half of the 2007 fieldwork season and corresponds to 54 households inter viewed throughout the villages of Mutsulio and Shikusi On the other hand, the second and third measures of and correspond to the entire data set. Wealth index The resu variable ( y = wealth index ) appear in Table 5 14. Results indicate that wealth is significantly and positively associated with fertilizer use (P=.04), oxen use (P=.10), and landholdin g size ( total acres owned ) (P=.10). It is somewhat significantly and positively associated with participation in training baraza s (P=.14). On the other hand, wealth is
119 significantly but negatively associated with the number of months per year maize must be purchased ( quantity of months/year maize is purchased ) (P=.07). Landholding size Table 5 15 presents the results of multiple regression analysis for landholding size as the dependent variable ( y=total acres owned ). Results indicate that landholding siz e is significantly and positively associated with male headed households (P=.02), cattle owned (P=.0003), and off farm nonagricultural work (P=.007). It is somewhat significantly and positively associated with the ability to borrow from a group (P=.16). On the other hand, landholding size is significantly and negatively associated with agricultural wage work (P=.08) and brewing and busaa for sale (P=.03). Cattle owned The results of multiple regression analysis of total cattle owned as the depend ent variable ( y=total cattle owned ) appear in Table 5 16. Results indicate that cattle owned is significantly and positively associated with male headed households (P=.08), landholding size (P=.0003), and the receipt of a bank loan (P=.09). It is somewha t significantly and positively associated with having a tin roof (P=.16) and with brewing and busaa for sale (P=.13). Cumulative years of schooling The final wealt h regression appears in Table 5 17, which presents the multiple regression analysis children ( ). Results indicate that cumulative years of schooling are significantly and positively associated with off farm nonagricultural work (P=.0 5) and with the receipt of a bank loan (P=.10). It is also somewhat significantly and positively associated with total cattle owned (P=.19).
120 However, cumulative years of schooling are significantly and negatively associated with agricultural wage work (P =.07). Additional Equations Additional multiple regressions were run in order to clarify the factors associated with the production and sale of kale as well as and busaa topics that came up frequently during ethnographic interviews. Informants reported kale, and busaa were their most importan t sources of cash (see Figure 5 7). Yet kale, on the one hand, constitutes a cash crop while and busaa on the other hand, constitute locally manufactured nonfarm goods. Moreover, the ci rcumstances under which kale production or and busaa brewing were undertaken seemed to differ. Informants who reportedly grew kale for sale, tended to be better off overall while informants who reportedly brewed and busaa for sale tended to be worse off overall. The results of the following multiple regression analyses, presented in Tables 5 18 and 5 19, support th ese deductions (as does Table 5 15 which indicates that brewing and busaa for sale is associated with smaller landho ldings). Kale acreage Table 5 18 presents the multiple regression analysis for kale acreage as the dependent variable ( y=kale acreage ). Results indicate that kale acreage is significantly and positively associated with the number of individuals in a house hold ( total household members ) (P=.02), with off farm nonagricultural work (P=.07), and with the production of sufficient maize to last an entire year ( enough maize harvest ). It is also somewhat significantly and positively associated with total cattle ow ned (P=.20) and with participation in training barazas (P=.11). Kale acreage is significantly and negatively associated with maize and bean acreage (P=.09).
121 Number of months maize is purchased for consumption Table 5 19 presents the multiple regression an alysis for the number of months per year a household purchases maize for consumption as the dependent variable ( y=number of months per year maize is purchased ). Results indicate that the number of months of maize purchase is significantly and positively a ssociated with the brewing of and busaa for sale (P=.01) and with the number of individuals living in the household (P=.08). On the other hand, the number of months of maize purchase is significantly but negatively associated with total cattle ow ned (P=.05), project participation (P=.01), the receipt of bank loans (P=.03), maize and bean acreage (P=.002), and maize yield per acre (P=.07). Of these, maize and beans acreage best predicts, negatively, the number of months per year maize must be purc hased for consumption. Descriptive and Inferential Bivariate Statistics: Frequency Distributions and Two Tailed t Test for Difference of Means Descriptive qualitative responses to the s expenses, off farm work, social capital, and imp acts of structural adjustment. More crop production, cash sourc es, recurring expenses, social group membership, technological use, farming ideals, investment preferences, and attitudes towards quality of life, farm, and economy. The information is conveyed using bar graphs and pie graphs, illus trated in Figures 5 1 t hrough 5 30. Most trends are presented as an aggregate whole as well as segregated by PLAR project participation and by gender, resulting in three graphs per trend, with numbers of farmers appearing on the y axes
122 axes (Figures 5 1 through 5 28). In these cases, inferential bivariate statistical tests were conducted in order to ascertain any significant differences between project participating households and non participating households and between male headed hou seholds and female headed households. Where a two tailed t test for a difference of means resulted in a significant P value value is illustrated on the bar graph. Crops Sold All farmers interviewed were asked what crops they most often sell Figure 5 1 summarizes in a bar graph their most commonly sold crops. Each bar represents the number of farmers reporting the sale of the crop labeled on the x axis. Nonetheless, each of the crops illustrated are grown for both sale and consumption. A lmost 20% of all farmers interviewed produce strictly for consumption. In Figure 5 2 the crops sold are segregated by PLAR project participation. As shown, a significantly larger number of project participating farmers sell kale, cabbage, and collards (P= .0002), maize and beans (P=.01), and tea leaves (P=.004) than do non participating farmers. A significantly larger number of non participating farmers sell no crops at all (P=.009). In Figure 5 3 the crops sold are segregated by gender. As shown, a signi ficantly larger number of female headed households sell cowpeas (the leaves) and other green leafy vegetables (P=.06) such as pumpkin leaves and mirro (a local leafy vegetable ) than do male headed households. A significantly larger number of female heade d households also sell other crops (P=.002) such as cassava, pineapple, avocado, sweet potato, onion, and papaya.
123 Household Use of Remittances All farmers interviewed were asked whether anyone from the household was currently employed outside the farm and sending back remittances. Though not illustrated, nearly all household members working outside the farm were employed in nonfarm jobs. On the other hand, most household members employed in agricultural wage work did not live outside the homestead. Thus remittances were chiefly composed of nonfarm earnings. Figure 5 4 illustrates the most common expenditures remittances are used for. Each bar represents the number of farmers who apply their remittances to the expenditure listed on the y axis. Figure 5 4 also indicates that almost half of all farmers interviewed were rece iving no remittances. Figure 5 6 indicates that most of these were male headed households. Female headed households most commonly received remittances from husbands and sons, as it is more common for males to migrate in search of off farm work. Figure 5 5 indicates that there is no significant difference between PLAR project participant and non participant usage of remittance s. On the other hand, Figure 5 6 indicates that a significant ly higher number of female headed households apply remittances towards business such as retail and brewing (P=.04), farm inputs (P=.0007), and school fees (P=.06). As mentioned, a significantly higher number of male headed households receive no remittance s (P<.0001). Most Important Cash Sources All farmers interviewed were asked to name their most important sources of cash. Fig ure 5 7 illustrates their most common responses. Each bar represents the number
124 of farmers who mentioned the source of income lis ted on the x axis as an important source of cash. Figure 5 8 indicates that a significantly higher number of PLAR project participants than non participants reported that the sale of kale provides an important so urce of cash (P=.02). Figure 5 9 indicates that a significantly higher number of female headed households reported that the sale of bananas (P=.001), leafy greens (P<.10), and retail (P=.006) provide important sources of cash. On the other hand, a significantly higher number of male headed househo lds reported that the sale of tea (P=.06), milk (P=.06) and paid labor (P=.02) provide important sources of cash. Most Important Expenditures All farmers interviewed were asked to name their most im portant expenditures. Figure 5 10 illustrates their most common responses. Each bar represents the number of farmers who mentioned the expenditure listed on the y axis as an important expenditure. Figure 5 11 indicates that a significantly higher number of PLAR project participants report that farming inputs ar e their most important expenditure (P=.01) while a significantly higher number of non participants report that sugar is their most important expenditure (P=.001). Figure 5 12 indicates that a somewhat significantly higher number of female headed than male headed households reported that school fees are their most important expenditures (P=.11). Social Capital All farmers interviewed were asked about the social and financial groups they belong to as well as the agricultural groups and training they have par ticipated in. For example, farmers were asked whether they belong to a merry go round a local financial
125 group to which members contribute money monthly and each month one member keeps the entire amount contributed. Farmers were also asked whether they b elong to other money. They were also asked whether they had ever received a bank loan. Regarding agricultural groups, farmers were asked whether they regularly participat ed in on farm or off farm labor sharing groups. They were also asked about their interactions with local agricultural extension offices in order to clarify whether they had ever been visited by or sought advice from an extension officer. Finally, farmers were asked whether they have ever received inputs such as fertilizer or seed as an incentive to experiment with a technology and whether they currently practice improved fallowing or biomass transfer. Figure 5 13 illustrates all these trends. Figure 5 14 indicates that a significantly higher number of PLAR project participants than non participants can borrow from a social group they belong to (P=.04), have participated in a training baraza -a community meeting in which an agricultural technology is expla ined and illustrated (P=.0004), have received extension visits to their farm (P<.0001), have visited their local extension office (P=.02), and have received seed or fertilizer as an incentive to experiment with an agricultural technology or increase agricu ltural productivity (P<.0001 ). On the other hand, Figure 5 14 illustrates that there is no significant difference in the practice of improved fallowing and biomass transfer between PLAR project participants and non participants. Figure 5 15 indicates that a higher number of female headed than male headed households belong to a merry go round (P=.02) and can borrow money from a social group th ey belong to (P=.09). Figure 5 15 also indicates that there is no significant
126 difference in the practice of improve d fallowing and biomass transfer between male headed and female headed households. Technological Dissemination Figures 5 16 through 5 technolog y related activities. Figure 5 16 indicates that wh ile half of all farmers interviewed have been trained in the application of improved fallows and biomass transfer through a training baraza less than 4% practice the technologies The graph also indicates that a little over a third of all farmers intervi ewed have received an extension visit to their farms while approximately only a tenth have visited their local extension office. Figure 5 17 provides a close up illustration of the significant differences between PLAR project participant and non participa nt agricultural technology related activiti es also illustrated in Figure 5 14. Figure 5 18 provides a close up illustration of the lack of significant differences between male headed and female headed household agricultural technology related activiti es a lso illustrated in Figure 5 15. Quality of Life, Farm, and Economy All farmers interviewed were asked about the current quality of their lives, farms, and national economy as compared to 5 years ago. They were asked to rate each according to whether it wa s worse, the same, or better. Figure 5 19 captures their responses and indicates that more than half of all farmers feel that the quality of their lives, farms, and national economy are all worse now than 5 years ago. Yet Figure 5 20 indicates that a sig nificantly higher number of PLAR project participants reported that their lives were better now than 5 years ago (P=.06) while a significantly higher number of non participants reported that their lives were worse now than 5 years ag o (P=.01).
127 Similarly, Figure 5 24 indicates that a significantly higher number of PLAR project participants reported that the national economy was better now than 5 years ago (P<.0001) while a significantly higher number of non participants reported that the national economy wa s worse now than 5 years ago (P<.0001 ). On the other hand, Figure 5 22 indicates there is no significant difference in the responses made between PLAR project participants and non participants regarding the quality of their farms now compared to 5 years a go. Figures 5 21 and 5 25 also indicate there is no significant difference between male headed and female headed household responses regarding the quality of their lives and national economy. Yet Figure 5 23 indicates that a significantly higher number o f female headed than male headed households reported that their farms are the same now as 5 years ago. What Would Improve Your Farm? All farmers interviewed were asked to name the single most important thing that woul d improve their farm. Figure 5 26 capt ures their open ended responses and illustrates that fertilizer was their most common response. Overall, the frequencies of the responses appearing on the y axis of Figure 5 26 were equally distributed between PLAR project participants and non participant s and between male headed a nd female headed households, as is also illustrated in Figures 5 27 and 5 28. T he only significant differ ence is illustrated in Figure 5 28 where a higher number of male headed households than female headed households reported t hat more land would provide the single most important improvement to their farms (P=.06). Investment Preferences All farmers interviewed were asked, hypothetically, what they would do with 5000 Kenyan Shillings (approximately $ 70) if they could obtain a ba nk loan. Their open
128 ended responses were grouped into 6 main ca tegories and appear in Figure 5 29 The brewing and selling mandazis ally, the loan to improve or expand their farms, 44% would apply it towards nonfarm activitie s and 60% would invest the funds in alternatives to crop production. The Making of a Successful Farmer All farmers interviewed were asked what they believe makes a successful farmer. Their open ended responses were grouped into 5 main ca tegories and appea r in Figure 5 30 s thirds of all farmers feel that resource
129 access is the key to successful farming while another 21% feel that proper procedures, whose understanding is often inaccessible, make a successful farmer. Only 7% of all farmers feel successful farming depends solely on effort. Script Analysis A script analysis was made of various agricultural routines depicted through farming calendars. Schank and Abelson (1977) define individuals' routine patterns of activities as "scripts." Scripts represent unconscious detailed knowledge that allows individuals to do less mental processing during frequently experienced events. In this case, scripts farming practices a comparison farmers often made when commenting on the nature of farming routines during open ended ethnographic interviews. Farming calendars were elicited from sever al farmers and also compiled from the many in depth and semi structured interviews conducted throughout all four villages. Normative themes regarding idealized and constrained farming were incorporated into the calendars. These are i llustrated in Tables 5 21 and 5 22. A more generalized, all encompassing calendar is featured in Table 5 20. The main difference between the three calendars lies in the timing of farming harvesti farming season, sometimes altogether skipping a prescribed farming chore such as a second plowing, gapping (replanting where germination did not take place), or a second weeding
130 they rent out their land. Of course, an underlying and understated factor in the tim ing of farming activities is resource access, particularly to cash and labor. Farm History Analysis Tables 5 23 thr ough 5 25 present the brief farm histories of three farmers living in three of the four sampled villages. The stories bring to life the farm ing opportunities and constraints examined in this study, thereby providing a qualitative context in which previous quantitative results can be conceptualized. Thei r stories were selected for farm history analysis because of the varied circumstances of th farming strategies. Yet it is important to note that despite the variations, none of the three farmers practiced labor intensive soil fertility enhancement strategies. Two of the with improved fallow and biomass had never heard of them. Moreover, the role of resource access across the three stories suppor ts the hypotheses of Chapter 4 In the f irst story, nonfarm remittances and the receipt of a bank loan led to farming improvements, particularly increased fertilizer use and larger harvests. In the second story, sufficient land and maize production go hand in hand with kale production. In the third story, agricultural wage work led to the neglect of on farm activities. Table 5 23 presents Her story indicates that her household is female headed. She receives remittances from her husband who lives and works in Nairobi. She also receives remittances from her sister. Yohana recently improved her farming through the receipt of a bank loan. Prior to that, she was unable to invest sufficient fertilizer in the production of maize and beans. And although she is
131 fa miliar with improved fallows and biomass transfer, she does not apply the technologies because she feels they take up too much of her land. Table 5 2 4 presents farm history. His story indicates his household is male headed. He f ormerly participated in an agricultural project that disseminated improved fallow and biomass transfer technologies. His fields were used as demonstration plots. He has four fields, two in which he cultivates maize and bea ns, one in which he cultivat es k ale, and a fourth on which he keeps cattle. Mr. James is not currently applying improved fallows and biomass transfer. He instead improves the fertility of his soil with fertilizer. Table 5 25 presents farm history. Her story indicates her household is male headed. She has very little land and supplements her maize and bean production with agricultural wage work. However, her wage work results in the neglect of her farm. She has never heard of improved fallows an d biomass transfer. Feasibility Analysis Tables 5 26 through 5 28 present three calculations that compare the feasibility of various farming strategies using real world prices and labor requirements. In doing so, the calculations clarify why farmers tend to choose inorganic fertilizer over organic fertilizer. By providing the cash equivalents of labor units and illustrating that labor and cash are interchangeable at any time, the results of the calculations illustrate that organic, labor intensive measure s can be more expensive than inorganic fertilizer. The results of the calculations also clarify why farmers tend to substitute land for fertilizer by
132 illustrating that between 2004 and 2008, the price of fertilizer rose far more quickly than the price of maize. Calculation 1 C alculation 1, illustrated in Table 5 26 was created in order to document the labor required in the biomass transfer of an amount of Tithonia diversifolia approximately equivalent to an application of 50 kg/ha of diammonium phosphate (DAP). The calculation also converts those labor units into cash equivale nts using 2007 Shinyalu Division prices. The results of the calculation indicate that the labor required to collect an amount of Tithonia diversifolia equivalent to a 50 kg bag of D AP is 300 labor days. That is, it would take 1 laborer 300 days to complete the task. If a household lacked the necessary labor, the cash required to hire the labor would amount to 15,000 Kenyan shillings. On the other hand, a 50 kg bag would cost 2000 Kenyan shillings. Calculation 2 C alculati on 2, illustrated in Table 5 27 was created in order to compare the labor and/or cash requirements of three alternatives for obtaining a 90 kg bag of ma ize using 2007 Shinyalu Division prices. The first alternativ e involves purchasing the 90 kg of maize and presents the amount of casual agricultural wage work required to raise the necessary cash: 30 labor days to raise 1600 Kenyan shillings. The second alternative involves producing the 90 kg of maize through the application of green manure (biomass transfer) and presents the amount of labor and its cash equivalent required to do so: 60 labor days or 3000 Kenyan shillings to hire the labor. The third alternative involves purchasing the inorganic fertilizer neces sary to produce the 90 kg of maize and presents the amount of casual agricultural wage work required to raise the necessary cash: 8 labor days to raise 400 Kenyan shillings.
133 Calculation 3 Finally, C alculation 3, illustrated in Table 5 28 was created in o rder to compare the prices of inputs and outputs involved in producing and se lling maize in Shinyalu Division from 2004 to 2007. The results of the calculation support the 6 th hypothesis listed in Chapter 4 in which I expected a greater rise in the price of fertilizer than in maize, and illustrates the consequences of the international and national factors affecting the price of fertilizer, discussed in the resear ch setting chapter, Chapter 3 Moreover, by illustrating that the price of fertilizer rose 2 50% during the period between 2004 and 2008 in comparison to a rise of only 80% in the price of maize during the same period, the results of the calculation help explain why farmers use smaller quantities of fertilizer per larger quantities of land, as ill ustrated throughout the multiple regression results of this study. Conclusion This chapter has presented the extensive results of this study. Though various tools of analysis were applied, both quantitative and qualitative, results indicate some important consistencies. Households are no longer applying improved fallows and biomass transfer technologies and the larger their landholdings, the smaller their applications of fertilizer and other inputs. Nonfarm income is associated with on farm improvements such as increased oxen use, more schooling, and more land yet off farm agricultural work is associated with less land, decreased on farm labor, and less schooling. Kale production and brewing emerge as important sources of income, the former associated wi th wealth indicators and the latter associated with small landholdings. And finally, a comparison of labor requirements and price equivalents for soil fertility enhancement measures and other agricultural inputs substantiates these
134 trends by indicating th at biomass transfer is an expensive technology and the price of fertilizer is rising far faster than the price of maize. The following chapter will now turn to the discussion of these results in more detail.
135 Table 5 1. Key of terms used in multiple reg ression analyses AgWageWork: Whether or not the household head regularly works for wages as an agricultural laborer BankLoan: Whether or not the household head or spouse has received a bank loan (within the last 10 years) BeanSeedingRate : Bean seeding (in kg) per acre of land cropped in maize and beans BnYield/Acre Bean yield (in kg) per acre of land cropped in maize and beans BorrowFromAGroup: Whether or not the household head can borrow from one of the social groups he or she belongs to Brewing: Whether o r not the household head or spouse brews either or bus a a for sale EnoughMzHarvest: without a need to purchase maize for consumption HiredLabor/Acre: Amount of hired labor (in number of laborers multiplied by number of days hired) per acre of land cropped in maize and beans KaleAcreage: Amount of land (in acres) c ropped in kale LaborShareOffFarm: Whether or not the household head participates in an off farm agricultural lab or sharing group LaborShareOnFarm: Whether or not the household head participates in an on farm agricultural labor sharing group MHH or FHH: Male headed or female headed household MzYield /Acr e: Maize yield (in kg) per acre of land cropped in maize and bea ns MerryGoRound: Whether or not the household head participates in a merry go round Mutsulio: Whether or not the household is located in the village of Mutsulio Mz&BeansAcreage: Amount of land (in acres) cropped in maize and beans MaizeSeedingRate : Maize s eeding (in kg) per acre of land cropped in maize and beans NonAgOFW: Whether or not the household received remittances from a household member employed in a n off farm nonagricultural job OxenUse/Acre: Amount of oxen used (in oxen set multiplied by number o f days hired) per acre of land cropped in maize and beans Project/NonProject: Whether or not the household participated in the PLAR project QMo/YrMzPurchased: Number of months per year maize is purchased for household consumption Shinakotsi: Whether or not the household is located in the village of Shinakotsi Shikusi: Whether or not the household is located in the village of Shikusi TinRoof: Whether or not the household has a tin roof (as opposed to thatch) TotalCattle: Total number of cattle owned TotalAcr esOwned: Total number of acres owned TotalFert/Acre: Total amount of fer tilizer and urea (in kg ) used per acre of land cropped in maize and beans TotalHHMembers: Total number of individuals residing in the household TrainingBaraza: Whether or not the hou sehold head has participated in a training baraza (within the last 10 years) VisitToExtOffice: Whether or not the household head has visited the extension office (within the last 10 years)
136 Table 5 2 Multiple regression analysis for Y = maize yield per a cre cropped in maize and beans
137 Table 5 3. Multiple regression analysis for Y = bean yield per acre cropped in maize a nd beans
138 Table 5 4 Multiple regression analysis for Y = total fertilizer per ac re cropped in m aize and beans
139 Table 5 5 Multiple regression analysis for Y = total fertilizer per acre cropped in maize a nd beans
140 Table 5 6 Multiple regression analysis for Y = oxen use pe r acre cropped in m aize and beans
141 Table 5 7 Multiple regression analysis for Y = oxen use p er acre cropped in maize and beans
142 Table 5 8 Multiple regression analysis for Y = hired labor per acre c ropped in maize a nd beans
143 Table 5 9 Multiple regression analysis for Y = hired labo r per acre cropped in maize and beans
144 Table 5 10 Multiple regression analysis for Y = ma ize seeding per acre cro pped in maize and beans
145 Table 5 11 Multiple regression analysis for Y = maize seeding per acre cropped in maize and beans
146 Table 5 12 Multiple regression analysis for Y = bean see ding per acre cr opped in maize and beans
147 Table 5 13 Multiple regression analysis for Y = bean seeding per acre cropped in maize and beans
148 Table 5 14 M ultiple regression analysis for Y = wealth index
149 Table 5 15 Multiple regression analysis for Y = total acres owned
150 Table 5 16 Multiple regression analysis for Y = total cattle owned
151 Table 5 17 Multiple regression analysis f cumulative years of schooling
152 Table 5 18 Multiple regression analysis for Y = kale acreage
153 Table 5 19 Multiple regression analysis for Y = number of months per year maize is purchased
154 Table 5 20 General farming calendar for western Kenya January: Plow land by oxen or by hand Burn if grass is high February: Second plowing Begin planting Sell cows in order to pay school fees Mar ch: Continue planting April: Easter celebrations; schools closed Circumcisions; weddings Begin first weeding Purchase maize for consumption May: Labor Day celebrations Begin second weeding Purchase maize for consumption Top dress June: Begin roasting early maize Begin eating tender beans Second weeding July: Bean harvest August: Maize harvest Buy and sell maize Schools close for one month Circumcisions Begin plowing the land for short rains season September: Continue harvesting m aize Dry maize Continue land preparation for short rains season Begin planting maize, beans, cowpeas, and pumpkin for short rains season October: Send cards wishing success in exams Form 4 exams Build houses and repair houses in preparation for December holidays Young men over 18 set up their own houses Weeding November: Send cards wishing success in exams Standard 8 exams Continue preparations for December holidays Sew dresses in preparation for December holidays Continue selling mai ze Harvest cowpeas and pumpkin leaves December: Short rains harvest Weddings Holidays; feasting Begin preparing land for long rains season at the end of the month
155 Ta ble 5 21 January: Clear and burn land First plowing February: Second plowing Begin planting March: Gapping (whatever did not germinate is replanted) April: First weeding May: Second weeding June: Top dres s (with urea) July: Begin long rains harvest August: Continue harvest Prepare maize for storage Begin plowing the land for short rains season September: Begin planting maize, beans, cowpeas, and pumpkin for short rains season October: Weeding November : Begin short rains harvest December: Continue harvest Begin preparing land for long rains season Table 5 22 January: Continue short rains harvest February: Clear, burn, and plow land March: Begin planting April: Continue planting May: First weeding June: Second weeding (if at all) July: Bean harvest August: Maize harvest Buy and sell maize September: Continue harvesting maize Clear and plow land for short rains season Begin planting maize, beans, cowpeas, and pumpkin for short rains season October: Continue planting November: Weeding December: Short rains harvest Compiled with the assistance of Sarah Anyolo, Mildr
156 Table 5 missing. Her husband began its construction years ago but never fi nished it. I ask Yohana whether her husband, in time, will complete the house but she does not know. She explains that her husband lives in Nairobi with a second wife and Yohana has not seen him in two years. Nonetheless, he sends remittances of 2000 or 3000 Kenyan shillings whenever he is employed. permanent house. It is constructed of clay and dung and has a tin roof. The interior, filled with sofas and chairs, is comfortable. One wall is covered in pasted magazine and newspaper clippings. Two small children greet us. Just to the right side of the house is a small garden, bordered by flowering bushes. Beyond the homestead are her fields, which slope down all the way to the Yala River. Her property sits up high on the hillside with breathtaking views of the countryside, neat fields of maize pouring down into the valley and river. I begin the interview with a general question about how Yohana learned to farm and how her farming has changed over the last 5 years. She explains that her parents taught her how to farm, planting maize and beans. She continues to use the same tools her parents used, such as bangas and jembes yet now she also plants additional cro ps, such as potatoes, vegetables, then. She is now 42 years old. She notes that her life became better 4 years ago when she called for a 5000 Ksh loan from an a gricultural financial institution in Kakamega Formerly, she Prior to the loan, she used fewer materials, such as fertilizer, and missed out on producing more maize. With the money the loan provided, she was abl e to invest in her farm. The financial institution allows her to make loan payments in maize, each harvest. Her sister, who used to be a teacher but now runs her own business as a seamstress in Busia helps her make loan payments if her harvest is not en ough. The sister also sends remittances of 1000 to 1500 Kenyan shillings when Yohana asks her for help. I ask Yohana whether she has ever participated in an agricultural project that teaches new planting techniques and she replies that a few years back a project worked with the community, helping it install terraces, which she implemented on her fields. She has two on the steeper slope and another further below, bordered by a live barrier of napier grass. She walks us out back to the field and shows us the channels and points to the darker shade of green in the center maize and then shows us how the outer maize is not as dark because she has run out of fertilizer towards that side. I ask whether Yohana has ever let her l and rest and she answers, no, that she plants it continuously. I also ask whether she has ever planted any agroforestry species with her maize and she again answers, no, because the agroforestry species flood the soil and invade the maize and beans. Fina If there is no rain, there is no harvest. She also has difficulty obtaining fertilizer because its price has gone up over the last few years.
157 Table 5 24. Mr. James is a Shikusi elder whom I was introduced to during a baraza at the beginning of my field work season in 2007. His home is a semi permanent structure of typical size. A calf grazes on his lawn and pigeons move about the kitchen as we sit in his living room chatting. I have a view of his tea bushes from where I am sitting. I ask Mr. James to tell me a little about his farming. He explains tha t he has four fields. He plow s his 1 hectare f ield near the house by hand, hiring laborers, because it is steep and has terraces bordered by live barriers of na pier grass. Yet he plow s his two smaller fields with oxen, attaching either 4 small ox en or 2 large oxen to the plow A hectare, he notes, requires a st pl plow nd Mr. James employs the same amount of oxen. He generally applies 8 kg of maize, 6 kg of beans, and 50 kg of diammonium phosphate per hectare when planting maize and beans. He keeps two of his fields planted in maiz e and beans and a third in kale as I glance Before being introduced to Mr. James, I was told he is regarded a good farmer. In the past, Mr. area teaching farmers about improved fallows and biomass transfer. However, Mr. James no Nonetheless, he still has a live fence of Tithonia diversifolia the species recommended for biomass transfe r. But most farmers do not use the technologies, Mr. James says, because they shamba
158 Table 5 25. Mary Odhiambo has been married since 1999. She moved from her par mother in fertile and her family never applied fertilizer. At her mother in applications of both fertilizer and cattle manure togeth fertility. In 2004, she and her husband moved into their own home after her husband inherited a I ask whether Mary has ever planted an improved fallow or practiced biomass transfer, applying green manure to her field. However, she replies that she has never heard of the technologies. She adds that an agricultural project worked recently in her community, instead teaching farmers how to keep chick ens and plant potatoes and kale I ask Mary to describe her annual farming routine. She explains that she hoes her shamba twice throughout January and February, plants during March, and weeds during April. However, she often neglects her w eeding because April is also a peak month for agricultural wage work, which she undertakes in order to make ends meet. In May, she plants potatoes and vegetables in her garden. Throughout June, she again engages in agricultural wage work, harvesting othe r harvests her maize and also works for other farmers helping prepare their maize for storage. Throughout September Mary prepares her own shamba for the short rains season but again domestic wage work, getting paid to wash clothes for other households. Finally, during ion for the holidays. She explains that people normally touch up the interior walls of their houses during the holiday season by smearing them with a layer of clay and dung. At the end of our interview, Mary notes that if she had enough land, she would n ot work off the farm. She complains that her maize gets spoiled because rather than re weeding it, she shambas sham ba
159 Table 5 26 Calculation 1: Feasibility of Tithonia diversifolia vs. diammonium phosphate (DAP) According to ICRAF (1997: application of 5 tonnes [5000 kg] of T ithonia (dry matter) per hectare that is comparable to the yield obtained from applying the recommended rate of inorganic fertilizer 50 kg/ha of P 2 0 5 and Tithonia diversifolia dry matter enhances the fertility of western Kenyan soil in much the same way as does a 50 kg bag of diammonium pho sphate (DAP). Yet ICRAF (1997: 3) notes, the leaves of Tithonia diversifolia have a moisture content of 84%. Therefore, after drying, 5000 kg of fresh tithonia is reduced to 800 kg of dry matter. Thus, a farmer m ust collect 31,250 kg of fresh Tithonia chop it, transport it to the field(s), and spread it over a hectare of land in order to obtain a maize yield comparable to the yield obtained from applying a 50 kg bag of DAP. Fu rthermore, according to ICRAF (1 997: mi nutes to collect 1 kg of fresh Tithonia of fresh Tithonia per hour. If the laborer works 7 hours per day, 100 kg of Tithonia can be collecte d per day per laborer. Further calculation indicates that the labor requi red to harvest 31,250 kg fresh Tithonia is at least 300 labor days (or 30 laborers for 10 days). In the Shinyalu Division area in 2007, 30 laborers @ 50 Kenyan shilli ngs per day (pl us a meal ) for 10 days would cost at least 15000 Kenyan shillings. A 50 kg bag of DAP would cost 2000 Kenyan shillings.
160 Table 5 27. Calculation 2: Feasibility of three alternatives for obtaining a sack of maize Trials in western Kenya researching mai ze yields from continuous maize cropping without fertilization indicate farmers obtain approximately 600 kg/ha of maize in the absence of soil fertility enhancement measures (Amadalo et al. 1998). Thus, a family of 2 adults and 4 children living on a 0 .5 h ectare plot of land in western Kenya can expect to produce approximately 300 kg of maize annually if the family spends no cash on agricultural inputs and plants maize only during the long rains season, leaving the land fallow during the short rains season. However, a family of 2 adults and 4 children can al so expect to consume at least 39 0 kg of maize annually. Therefore, such a family would need to secure at least 9 0 kg more of maize annually. That is, the family would need to find a way to obtain appro ximately one more 90 kg sack of maize per year. If the family chose to purchase the additional sack of maize, in 2007 a 90 kg sack of maize was priced at approximately 1600 Kenyan sh illings in the Shinyalu Division area prior to harvest. In order to obta in the cash to purchase the sack, someone in the family would need to engage in agricultural wage work for approximately 30 days @ 50 Kenyan shillings per day. If the family chose to produce the additional sack of maize through the application of biomass transfer green manure to their field, someone in the family would need to collect, chop, dry, and spread approximately 6000 kg of fresh Tithonia diversifolia the nutrient equivalent of applying 10 kg of DAP to raise maize yields by approximately 90 kg (acc ording to Quiones et (1997:11) estimate it would take one laborer approximately 60 days to collect 6000 kg of fresh Tithonia diversifolia If the family chose to hire the labor required to collect the green manure, it would cost 3000 Kenyan shillings. The labor required to chop, dry, and spread it would be additional. If the family chose to purchase 10 kg of DAP and apply it to their field in order to raise m aize yields by approximately 90 kg, it would cost them 400 Kenyan shillings. To raise that amount of cash, someone in the family would need to engage in agricultural wage work during 8 days. Table 5 28. Calculation 3: Feasibility of maize production t hrough a comparison of input and output prices during 2004 and 2007 in Shinyalu Di vision Most farm ers throughout Shinyalu Division apply inputs of animal power, hired labor, seed, and fertilizer in order to obtain outputs of maize. Below are listed the real world prices of inputs and outputs involved in the production and sale of maize. Farmers ideally strive to store maize after harvest in order to sell it at peak pre harvest prices during May and June. As the tables indicate, the price of fertilizer rose 250% from 2004 to 2008. Yet the price of pre harvest maize only rose 80%. Outputs & Inputs 2004 Prices 2008 Prices Hired oxen: 500 Ksh/day (plus meal) 500 Ksh/day (plus meal) Hired l abor: 50 Ksh/day (plus meal) 70 Ksh/day (plus meal) 50 kg DA P fertilizer: 1600 Ksh 4000 Ksh 90 kg pre harvest maize: 1600 Ksh 2000 Ksh
161 Figure 5 1 Frequency distribution of crops sold by all farmers Figure 5 2 Frequency distribution of crops sold by project par ticipants and non participants
162 Figure 5 3 Frequency distribution of crops sold by male headed and female headed households Figure 5 4 Frequency distribution of household use o f remittances by all farmers
163 Figure 5 5 Frequency dis tribution of household use of remittances by project p articipants and non participants Figure 5 6 Frequency distribution of household use of remittances by male head ed and female headed households
164 Figure 5 7 Frequency distribution of most important cash sources for all farmers Figure 5 8 Frequency distribution of most important cash sources for pro ject participants and non participants
165 Figure 5 9 Frequency distribution of most important cash sources for male headed and f emale headed households Figure 5 10 Frequency distribution of m ost important expenditures for all farmers
166 Figure 5 11 Frequency distribution of most imp ortant expenditures for project participants and non participants Figure 5 12 Frequency distribution o f most importa nt expenditures for male headed and female headed households
167 Figure 5 13 Frequency distribution of social capital among all farmers Figure 5 14 Frequency distribution of social capital among project participants and non participants
168 Figure 5 15 Frequency distribution of social capit al among male headed and female headed households Figure 5 16 Frequency distributio n of technological dissemination among all farmers
169 Figure 5 17 Frequency distribution of technological dissemination among project participants and non participants Figure 5 18 Frequency dis tribution of technological dissemination among male headed and female headed households
170 Figure 5 19 Frequency distribution of responses regar ding quality of life, farm, and economy now compared to five years ago among all f armers Figure 5 20 Frequency distribution of responses regardi ng quality of life now compared to five years ago among project participants and non participants
171 Figure 5 21 Frequency distribut ion of responses regardi ng quality of life now compared to five years ago among male headed and female headed households Figure 5 22 Frequency distribution of responses rega rding farm now compared to five years ago among pr oject participants and non participants
172 Figure 5 23 Frequency distribution of responses rega rding farm now compared to five years ago among male headed and female headed households Figure 5 24 Frequency distribution of responses re garding economy now compared to five years ago among project participants and non participants
173 Figure 5 25 Frequency distribution of responses re garding economy now compared to five y ears ago among male headed and female headed households Figure 5 26 Frequency distribution of open ended r
174 Figure 5 27 Frequency distribution of open ended r ng project participants and non participants Figure 5 28 Frequency distribution of responses to th improve headed and female headed households
175 Fig ure 5 29 Relative frequency distribution of a investment preferences Figure 5 30 Relative frequency distribution of all f arding the making of a successful farmer
176 CHAPTER 6 DISCUSSION: EXTENSI VE FARMING AND DIVER SE LIVELIHOOD STRATE GIES Introduction The results of this study point to five major findings. The first and most important of these is that farmers substitute lan d for fertilizer. The larger their landholdings, the less fertilizer per hectare farmers apply. Because fertilizer represents the quintessential measure of agricultural intensification, western Kenyan farmers are largely decreasing, rather than increasin g, the productivity of their maize and bean agriculture. A second major finding is that despite their inability to invest in fertilizer, farmers have not adopted improved fallows and biomass transfer. Instead of increasing the yield of maize and beans wi th soil fertility enhancement measures, farmers supplement production with nonfarm work, agricultural wage work, and kale production. Yet a third major finding of this study is that off farm work differentially affects the farm according to the kind of of f farm work. Off farm nonagricultural work leads to on farm improvements while off farm agricultural work leads to the neglect of on farm activities. A fourth major finding is that / busaa brewing for sale constitutes the most important nonfarm sou rce of income, particularly for land constrained farmers, while kale production constitutes the most important cash crop, particularly for wealthy farmers. and busaa brewers produce insufficient quantities of maize and beans and, in turn, purchas e maize for consumption with cash earned through brewing. Kale farmers produce sufficient maize and beans and, in turn, substitute kale cultivation for maize and bean cultivation on spare land. Finally, a fifth major finding substantiates these tren ds by comparing the labor and/or cash requirements of soil fertility enhancement measures and other
177 agricultural inputs and illustrates that the prices of some inputs far exceed the value of the additional maize that would be produced. Extensive Farming Strateg ies growth rates (Sanchez et al. 1997) and the theories put forth by Boserup (1981) and Netting (1993) which posit that farmers intensify production in the fac e of decreasing fertility, it w as hypothesized i n Chapter 4 (H1a and H1b) that households throughout western Kenya would be intensifying the production of their staple crops through soil fertility enhancement measures such as improved fallows, biomass transfer, and fertilizer use and through increased a mounts per hectare of oxen use, hired labor, and seeding rate. Thus it was unexpected, although not entirely surprising, that the results of this study reject these hypotheses by indicating households in the study area in western Kenya are producing maize and beans extensively not intensively. That is, agricultural inputs including soil fertility enhancement measures are negatively associated with landholding size. The larger a landholding, the smaller the amounts of fertilizer applied, oxen used, labor hired, and seeding rate. Moreover, no one is using improved fallows and only three farmers reported using biomass transfer, out of a sample of 120 households. Extensification is most clearly indicated by the productio n and input equations (Tables 5 2 thr ough 5 13) presented in Chapter 5 The productio n regressions (Tables 5 2 and 5 3) indicate that per acre yields of maize are significantly and negatively though not quite significantly associated with landholding size. The input regressions (Tables 5 4 through 5 13) indicate consistently that all agricultural inputs applied per area of land in the production of maize and beans are negatively associated with
178 landhold ing size. Every single agricultural input is applied at a rate that is negatively associated with the amount of land owned by the household Fertilizer use (Table 5 5) and bean seeding rate (Table 5 13) are significantly and negatively associated with la ndh olding size. Oxen use (Table 5 7) and maize seeding rate (Table 5 11) are somewhat significantly and negatively associated with landholding s ize. Only hired labor (Table 5 9) is not quite significantly, yet still negatively, associated with landholdin g size. Thus, if households in western Kenya are applying agricultural inputs at rates that are negatively associated with the amount of land they own and are, in turn, producing maize and bean yields that are negatively associated with the amount of lan d they crop, it can be safely conc luded that households in western Kenya are producing their staple crops extensively Households are not applying their agricultural inputs intensively particularly fertilizer, the soil fertility enhancement measure. Thei r maize and bean yields are not positively associated with their maize and b ean acreage. Farmers in the study area in western Kenya are not increasing the agricultural productivity of maize and bean production. Instead, they are extensifying, rather than intensifying, their staple crop production Diverse Livelihood Strategies Because research (Morera and Gla dwin 2006; IEA/SID 2001; Orvis 1997; Ayieko 1995) indicates rural households supplement farm production with off farm work and this often leads to on farm labor shortages, it was hypothesiz ed in Chapter 4 (H2a and H2b) that the kind of off farm work would determine whether off farm income generation led to on farm improvements or led to the neglect of the farm. Nonfarm work was hypothesized to lead to greater applications of agricultural inputs and agricultural wage
179 work was hypothesized to lead to fewer applications of agricultural inputs. Overall, the results of this study support these hypotheses. Off farm Nonagricultural Work With the exception of /busaa brewing (discussed in the next section) and a few isolated cases of nonagricultural retail (i.e. paraffin, sweaters, and medicine), nonagricultural work was largely carried out at long distances from the farm, with only a fraction of the in come it generated reaching households in the form of remittances. Nonetheless, the results of this study indicate that nonagricultural off farm earnings support households in ways neither agricultural wage work nor brewing can. Four equations, in particul ar, illustrate the uses of nonagricultural off farm income. The first of these is i llustrated in Table 5 7 and indicates off farm nonagricultural work is significantly and positively associated with oxen use per acre of land cropped in maize and beans (P= .07), meaning nonfarm remittances are applied towards maize and bean production. The second e quation, illustrated in Table 5 18, indicates off farm nonagricultural work is significantly and positively associated with kale acreage (P=.07), implying nonfarm remittances are applied towards kale production. The third equation, illustrated i n Table 5 15, indicates off farm nonagricultural work is significantly and positively associated with total acres owned (P=.007), implying nonfarm remittances are applied t owards the purchase of agricultural land. Finally, the fourth equation is illustrat ed in Table 5 17 and indicates off farm nonagricultural work is significantly and nonf arm remittances are applied towards school fees. Results of inferential bivariate statistics and farm history analysis also illustrate the importance of nonfarm remittances. Figure 5 6 indicates female headed households, in
180 particular, apply nonfarm earni ngs towards agricultural inputs. The case study presented in Table 5 23 illustrates a female household head, which has improved her farming through the receipt of a loan, making loan payments with nonfarm remittances received from her husband and her sist er. In sum, quantitative and qualitative results indicate nonfarm income is used towards on farm improvements and the betterment of the household, suppo rting H2a, listed in Chapter 4 On farm Nonagricultural Work On farm nonagricultural work refers to non farm entrepreneurial activities that are research site is brewing. Farmers, mostly female, produce liquor for sale. In most cases, they purchase rather than pr oduce the ingredients that are necessary to brew and distill the liquor. Many farmers, including male heads, report that Busaa beer is also brewed for sale, although not as commonly as aa The illegality of brewing, however, makes it a marginal enterprise. The variable most closely associated with and busaa brewing is the number of months per year maize must be purchased for consumption as a result of insufficient maize product ion. The association is significant (P=.005) and positive as illustrated in Table 5 19. The next variable most closely associated with /busaa brewing is acres owned. Again, the association is significant (P=.03) yet negative as illustrated in T able 5 15. Not surprisingly, /busaa brewing is also significantly (P=.09) and negatively associated with bean yi eld, as illustrated in Tables 5 3. Together, these three equations reveal that /busaa brewers are land constrained. They mus t purchase maize for consumption because they own very little
181 land and harvest insufficient amounts of maize and beans to sustain them throughout the year. They earn cash through the sale of and busaa and purchase maize with th e cash. Interestin gly, Table 5 16 indicates they also purchase cattle with the cash. /busaa brewing is somewhat significantly (P=.13) and positively associated with total cattle owned. This last equation indicates /busaa brewing for sale is relatively lucr ative, although clearly not as much so as off farm nonagricultural work, but more so than agricultural wage work, as illustrated in the next section. Because results indicate that the nonfarm income generated through /busaa brewing not only feeds the farm, but expands its assets through investment in cattle, H2a is once again supported. Off farm Agricultural Work The nature of agricultural wage work is best understood by observing that its association with most inputs applied and outputs produced i n maize and bean farming is negative, as illu strated in Tables 5 2 through 5 13. In fact, its association with most wealth indicators is also negative, as illus trated in Tables 5 14 through 5 17. It is negatively and significantly associated with total a cres owned (Table 5 15; P=.08) and able 5 17; P=.07), and it is negatively and somewhat significantly assoc iated with hired labor (Table 5 9; P=.20) and bean seeding rate (Table 5 13; P=.11). Of course, the reason for all these negative associations is that agricultural wage work is both time consuming and poorly remunerated. Its wages are the lowest in the rural sector. Agricultural wage work tends to be undertaken by younger farmers as the work is too stre nuous for older farmers. Moreover, older farmers usually have older children who contribute to their livelihoods. In turn, younger farmers are more likely to have
182 recently inherited a partitioned, and thus small, landholding whose productive capacity mus t be supplemented with alternate livelihood strategies. Younger farmers also tend to have young children who are in their early years of schooling. The results of farm history and script analyses also illustrate the role of agricultural wage work for smal lholders. Table 5 25 presents the life story of a woman who neglects her own farming through engagement in agricultural wage work. Her agricultural wage work in order to ma ke ends meet. However, the work interferes with the results of script analysis, i llustrated in Tables 5 21 and 5 22 and discussed in Chapter 5 compare the calendar timing is linked to resource access. These qualitative results, together with the quantitative results discussed above, support H2b, listed in Chapter 4 The Matter of Wealth Five equations in Chapter 5 estimated for wealth index, landholding size, cattle owned, cumulative years of schooling, and kale acreage illustrate the nature of wealth throughout the sampled villages of western Kenya. Four of these are discussed in the ealth Equa and illus trated in Tables 5 14 through 5 5 and illustrated in Table 5 18. Together the equations point out several importa nt features of wealth. The first important feature of wealth is that fertilizer use per acre is the most significant and positive predi ctor of wealth index in Table 5 14 (P=.04). The next most significant predictor of wealth, negatively, is number of mont hs per year maize is
183 purchased for consumption (P=.07). That is, farmers who do not produce sufficient maize, such as those who engage in agricultural wage work or brew and busaa for sale, are not wealthy. Naturally, total acres owned is positiv ely and significantly associated with wealth index (P=.10). Together these features indicate that wealthier households tend to have more land, use more fertilizer, and purchase less maize annually for consumption than poorer households. Another important feature of w ealth, as indicated by Tables 5 15 and 5 16 is that total acres owned and total cattle owned, chief proxies for wealth in the countryside, are both controlled by men. For this reason, both are significantly and positively associated with mal e headed households (P=.02 and P=.08, respectively). Total acres owned, in Table 5 15, is also significantly and positively associated with off farm nonagricultural work (P=.007). On the other hand, total acres owned is significantly and negatively associ ated with both agricultural wage work and brewing (P=.08 and P=.03, respectively, in Table 5 15). That is, large landowners and cattle owners tend to be male headed households or households receiving nonfarm remittances while smallholders tend to be agric ultural wage workers and brewers. Interestingly, brewing is most often undertaken by women. Rei nforcing these results, Table 5 schooling, another proxy for wealth, are positively and significantly associate d with off cumulative years of schooling are also positively and somewhat significantly associated with total cattle (P=.19). On the other hand, schooling is negatively a ssociated with agricultural wage work.
184 The equation estimated for kale acreage a cash crop sheds further light on the characteristics of wealth by indicating that it is positively and significantly associated with the production of sufficient maize (P= .03) as illustrated in Table 5 18. Interestingly, kale acreage is also significantly yet negatively associated with maize and bean acreage (P=.09). This means that larger landholders can afford to substitute kale cultivation for maize cultivation because th ey produce more than enough maize year round to begin with. Moreover, kale acreage is positively and significantly associated with off farm nonagricultural work (P=.07), further reinforcing the important role and lucrative nature of nonfarm income generat ion. These five equations provide insight in to the role of wealth in this region of western Kenya. In turn, a sixth equation estimated for months of maize p urchase, illustrated in Table 5 apter 5 aptly illustrates what wealth is not Table 5 19 indicates that individuals who must purchase maize for consumption tend not to have much land, cattle, bank loans, fertilizer, nor participate in projects. Together, the six equations s upport H3, listed in Chapter 4 which posits that household wealth, indicated by greater landholding size, cattle owned, cumulative years of schooling, and the receipt of bank loans, is associated with greater investment in agricultural inputs and on farm improvement s. The Role of Gender The management of the western Kenyan farm is characterized by a gender e land by hand is usually undertaken by women while plowing the land using oxen is usually undertaken by men. These culturally assigned gender roles are evident across this
185 example, the significant differences between male and female headed households in the sale o f crops (Figure 5 3) reflect these gender divisions of labor. The sale of fruits and vegetables, such as avocados, bananas, and local greens (including those that are purchased for retail) tend to be carried out by women. Men, on the other hand, because they are more likely to manage cattle, are also more likely to carry out the sale of milk (Figure 5 9). Additionally, men are more likely to benefit from the sale of tea (Figure 5 9) because men are more likely to inherit tea land from their fathers. The significant difference between male and female headed households in the receipt of remittances also reflects culturally assigned gender divisions of labor. Males a re more likely to migrate in search of off farm work. Therefore, female headed households are more likely to receive remittances from husbands working off farm while male headed households are more likely to receive no remittances at all (Figure 5 6). Ye t a significantly higher number of male headed than female headed households report that the sale of labor constitutes their most important source of income, reflecting the obvious fact that remittances constitute but a fraction of off farm earnings. That is, a receiving only a portion of it in the form of remittances. Despite these gender differences in farming activities and income generation, results of multiple regres sion analysis do not reflect any significant gender differences in the application of most agricultural inputs, including soil fertility enhancement measures such as fertilizer use. Only in the case of hired labor is there a s ignificant difference (Tables 5 8 and 5.9; P=.02) in its application between male and female headed
186 households. The likely reason for the positive association between hired labor and female headed households is that oxen use is managed by males. Results of multiple regression analys is also do not reflect any significant gender differences in the production of maize and beans. Thus, overall, results s upport H4, listed in Chapter 4 which posits no difference between male headed and female headed household investment in agricultural i nputs. Achievements and Failures of the PLAR Project The results of this study indicate that the PLAR project, described in detail in the research setting chapter, provided long range benefits to its participants desp ite its failure to achieve long term ad option of improved fallows and biomass transfer among farmers. Results of bivariate statistical analysis show that a significantly higher number of PLAR project participant farmers sold cash crops and staple crops than did non participant farm ers, as illu strated in Figure 5 2. PLAR project participant farmers also had greater access to social capi tal, as illustrated in Figure 5 14, and reported a better quality of life and economy, as il lustrated in Figures 5 20 and 5 24 Nonetheless, multiple regressi on analysis indicates that maize and bean yields are actually negatively associated with project participation (P=.18 and P =.03, respectively, in Tables 5 2 and 5 3) as are oxen use and maize seeding rate (P=.10 and P=.03, re spectively, in Tables 5 7 and 5 11). The likely reason for this is that the PLAR project was aimed at marginal farmers. Particularly the village of Mutsulio lies on a steep ridge that constrains farming. On the bright side, project participation is somewhat significantly and positive ly associat ed with fertilizer use (Table 5 4; P=.12) and significantly and negatively associated with the number of months per year that maize is pur chased for consumption (Table 5 19; P=.02). Thus, project participation seems to have led to
187 increased app lication of a soil fertility enhancement measure and, perhaps consequently, fewer months of maize purchase for consumption. Ultimately, however, project participation did not lead to the adoption of improved fallows and biomass transfer despite significant gains in ag ricultural extension activities Figure 5 17 indicates that a significantly higher number of project participant farmers than non participant farmers participated in training barazas (P=.0004), received visits from extensionists (P<.0001), vis ited their extension office (P=.02), and received seed or fertilizer as an incentive to experiment with a technology (P<.0001). Yet the same table shows absolutely no difference between participant and non participant farmer adoption of either improved fa llows or biomass transfer. Furthermore farm history analysis in Tables 5 23 and 5 2 4 provides two examples of the progression by which farmers participated in the PLAR project, experimented with improved fallows and biomass transfer, and later rejected t he technologies. Thus, overall, quantitative and qualitative r esults s upport H5 in Chapter 4 which posits that there is no difference in the soil fertility enhancement practices of project participant and non participant farmers. The reason for the PLAR pr term adoption of improved fallows and biomass transfer is partially explained by the results of feasibility analysis, particularly Tables 5 26 and 5 27 These indicate that the labor requirements of biomass transfer are s o large as to render the technology expensive, especially when one considers that labor and cash are interchangeable units in the rural economy of western Kenya. Labor constrained households would need to pay more for biomass transfer than for mineral fer tilizer. Moreover, households with abundant labor would need to think twice before investing in biomass transfer which would still not guarantee
188 a harvest given the inherent risks of farming and its vulnerability to the vagaries of weather and pests when wage labor provides instant remuneration. This reasoning also helps to partially explain why farmers are substituting land for fertilizer and producing maize extensively, rather than intensively. Although Tables 5 26 and 5 27 indicate that fertilizer use is more feasible than biomass trans fer in raising maize yields, Table 5 28 nonetheless indicates that fertilizer prices have been rising far faster than maize prices over th e last 5 years. As Chapter s 2 and 3 explain the international rise in the price o f oil coupled with the degeneration of infrastructure between eastern and western Kenya have contributed to the rising price of fertilizer. Yet maize prices have not risen proportionally, explaining why farmers are unwilling to invest in the agricultural inputs necessary for its intensification. Thus, the results of feasibility analysis support H6 as listed in Chapt er 4 which posits that farmers will not invest in fertilizer use if the international price of oil and driving time between eastern and weste rn Kenya increase, causing fertilizer prices to rise faster than maize prices. Conclusion The results of this study have important implications for resource conservation and agricultural development efforts in Kenya Because farmers in this region of west ern Kenya are producing maize and beans extensively through the application of inadequate amounts of soil fertility enhancement measures, it can be safely concluded they are degrading their lands and failing to increase the agricultural productivity of the ir staple crops. Land poor farmers, rather than intensify production, apply their labor towards agricultural wage work or nonfarm work while larger landholders, rather than increase agricultural productivity, substitute cash crop production for maize and bean pro duction Thus smallholders do not invest their labor in organic fertilizers or improved
189 fallows while larger landholders do not invest their wealth in inorganic fertilizers. Instead, all farmers substitute land for fertilizer. Meanwhile, soil fe rtility enhancement as well as aggregate agricultural productivity remains low in this region of western Kenya. A more subtle implication of the results of this study is that no amount of agricult ural extension can achieve long term farmer adoption of soil fertility enhancement practices when macro policies provide disincentives for agricultural intensification. Without incentives for agr icultural intensification, soil fertility enhancement technologies are for naught because farmers cannot afford to apply them. In this case, a combination of international factors, such as the price of oil, and domestic factors, such as poor infrastructure, conspire to raise agricultural input prices faster than maize prices rendering staple crop production less attractive than alternate livelihood strategies, such as kale and liquor production. In and of itself, substituting kale and liquor production for staple crop production is not a problem. The demand for kale and liquor in Kenya is domestic, indicating that the needs of the population are in sync with the production of these commodities. Moreover, the production of kale and liquor represents specialization, an important facet of agricultural development. Nonetheless, increased aggregate agricultural productivity by remains essential to structural transformation and hence economic development. If food prices fail to rise, marginal producers will not invest in the increased costs of intensificati on which, essentially are the costs associated with fertility enhancement. In most cases, they will cultivate until they degrade the land and must abandon the area or abandon agriculture. In this region of western Kenya, results imply that farmers are
190 de grading their land and finding alternatives to staple crop production. In turn, the consequence of staple crop extensification is low aggregate agricultural productivity and delayed economic development. Because aggregate agricultural productivity remain s the most important factor affecting Kenyan development, increased efforts to intensify is to be achieved.
191 LIST OF REFERENCES Agr esti, Alan and Barbara Finlay 1997 Stat istical Methods for the Social Sciences. Upper Saddle River: Prentice H all. Ake, Claude 1996 Democracy and Development in Africa Washington DC: The Brookings Institution. Amadalo, B., A. Ni ang, S. Gathumbi, and B. Jama 1998 Us ing Short Duration Improved Fallows in Western Kenya fo r Soil Fertility and Crop Yield Improvement. Maseno: National Agroforestry Research Centre. Anderso n, Kym and William A. Masters 2009 Introdu ction and Summary. In Distortions to Agricultural Incentives in Africa. Kym Anderson and William A. Masters, eds. Pp. 3 67. Washington, DC : The World Bank. Aune, J.B. 1993 Ecological and Economic Requireme nts for Sustainable Land Use in S ub Saharan Africa Foru m Development Studies 2: 211 219. Ayieko, Monica Awuor 1995 Household Allocation of Labor Time in Two Types of Smallholder Farming systems in Rural Kenya. PhD Dissertation. Urbana : University of Illinois. Ayittey, George B.N. 2005 Africa Unchained. New York : Palgrave Macmillan. Bates, Robert 1981 Markets and States in Tropical Africa: The Political Basis of Agricultural Policies. Berkeley : The University of California Press. Bernard, H. Russell 2006 Research Metho ds in Anthropology: Qualitative and Quantitative Approaches. Lanham: AltaMira Press. Binswanger, Hans P and Vernon W. Ruttan 1978 Induced Innovation: Technology, Institutions, and Development. Baltimore : The Johns Hopkins University Pr ess. Blaikie Piers 1985 The Political Economy of Soil Erosion in Developing Countries. London: Longman Development Studies.
192 Blaikie Piers and Harold Brookfield 1987 Land Degradation and Society. New York : Methuen. Boserup, Ester 1981 Population and Technologic al Change: A study of Long Term Trends. C hicago : University of Chicago Press. Brady, Nyle C. 1974 The Nature and Properties of Soils. New York : Macmillan. Buresh, R., P. A. Snche z, and F. Calhoun, eds. 1997 Rep lenishing Soil Fertility in Africa. SSSA Special Publication no. 51. Madison: Soil Science Society of America and American Society of Agronomy. Chayanov, A.V. 1966 The Theory of Peasant Economy. D. Thorner, B. Kerblay, and R.E.F. Smith, eds. Homewood: Richard D. Irwin. Crawford, Eric W., T.S Jayne, and Valerie A. Kelly 2006 Agricultural and Rural Development Discussion Paper 22. Washington, DC: The World B ank. de Janvry, Alain 1981 The Agrarian Question: Agrarian Reform in Latin America Baltimore: Johns Hopkins University Press. de Janvry, Alain and Elisabeth Sadoulet 2001 Income Strategies Among Rural Households in Mexico: The Rol e of Off farm Activities. World Developmen t 29(3): 467 480. Djurfeldt, Gran, Hans Holmn, Magnus Ji rstrm, and Rolf Larsson 2005 African Food Crisis The Relevance of Asian Experiences. In The African Food Crisis: Lessons from the Asian Green Revolution. G. Djurfeldt, H. Holmn, M. Jirstr m, and R. Larsson, eds. Pp. 1 8. Oxon: CABI Publishing. Eionet N.d. European Environment Information and Observation Network. http://www.eione t.europa.eu/ accessed December 3, 2010. FAO 1998 The State of Food and Agriculture 1998: Rural Non Farm Income in Developing Countries. Rome: Food and Agriculture Organization of the United Nations (FAO).
193 FAOSTAT Statistical d a tabase. http://faostat.fao.org/ accessed December 3, 2010. Farnsworth, Steve, Michael Conroy, Do uglas Murray and Peter Rosset 1996 A Cautionary Tale: Failed U S Development Policy in Central America Boulder : Lynne Rienner. Franzel, Steven, Richard Coe, Peter Cooper, Frank Place and Sara J. Scherr. 2000 Assessing the Adoption Potential of Agroforestry Practices in Sub Saharan Africa. Agricultural Systems 69(1 2)37 62. Gladwin, Christina H 1991 Fertilizer Subsidy Removal Programs and Their Potential Impacts on Women Farmers in Malawi and Cameroon In Structural Adjustment and African Women Farmers. Christina H. Gladwin, e d. Pp. 191 216. Gainesville : Un iversit y of Florida Press. Gyllsrom, Bjorn 1991 State Administered Rural Change: Agricultural Cooperatives in Kenya. New York : Routledge. Hanrahan, Charles, Beverly A. Banks, and Carol Canada 2006 U S Agricultural Trade: Trends, Composition, Direction, and Policy. Congressional Research Service Report for Congress, September 25, 2006. http://fpc.state.gov/documents/organization/74919.pdf accessed Decem ber 3, 2010. Harris, Marvin 1979 Cultural Materialism: The Stru ggle for a Science of Culture. N ew York: Vintage Books. Hayami, Yujiro and Vernon W. Ruttan 1971 Agricultural Development: An International Perspective, 1 st e d ition Baltimo re: Johns Hopkins University Press. 1985 Agricultural Development: An International Perspective, 2 nd ed ition Baltimore: Johns Hopkins University Press. Hayami, Yujiro and Vernon W. Ruttan 1970 Agricultural Productivity Differences Among Countries. A merican Economic Review 60(1970):895 911. Hicks, John R. 1932 The Theory of Wages. London : Macmillan.
194 Hock, Roger R. 2010 Human Sexuality. Upper Saddle River : Prentice Hall. Hyden, Goran and Pauline E. Peters 1991 Debate on the Econom y of Affection: Is It a Useful Tool for Gender Analysis? In Structural Adjustment and African Women Farmers. Christina H. Gladwin, e d. Pp. 303 335. Gainesville : University of Florida Press. ICRAF 1997 Using the Wild Sunflower, Tith onia, in Kenya Nairobi: International Centre for Research in Agroforestry. IEA/SID 2001 Kenya at the Crossroads: Scenarios for Our Future. Nairobi : Institute of Economic Affairs/Society for International Development. IIRR/ACT 2005 Cons ervation Agriculture: A Manual for Farmers a nd Extension Workers in Africa. Nairobi : International Institute of Rural Reconstruction (IIRR)/African Construction Tillage Network (ACT). Kamiri, James, Stanley Gathumbi, Georg Cadisch, an d Ken Giller 1997 Mixed Species Improved Fallows for Western Kenya. Recommendations from a DFID (Natural Resources Systems Programme, Project No. R7056)/EU (Pilot Project) Funded Project. Kenya 2004 Kenya Demographic and Health S urvey 2003. Calverton: Central Bureau of Statistics, Ministry of Health, and ORC Macro. Kenya 1962 Kenya Geological Map. Nairobi: Survey of Kenya. Kenya 1985 Kenya Soils Map. Nairobi: Survey of Kenya. Kenya 2001 The 1999 Population and Housing Census: Count ing Our People for Development, Volume I. Nairobi: Central Bureau of Statistics Ministry of Finance and Planning. 2001 The 1999 Population and Housing Census: Counting Our People for Development, Volume II. Nairobi: Central Bur eau of Statistics Ministry of Finance and Planning.
195 Kwesiga, F.R., S. Franzel, F. Place D. Phiri, and C.P. Simwanza 1999 S esbania sesban Improved Fallows in Eastern Zambia: Their Inception, Development, and Farmers Enth usiasm. Agrofo restry Systems 47: 49 66. L ambsdorff, Johann Graf 2007 The Institutional Econo mics of Corruption and Reform: Theory, Evidence, and Policy. New York : Cambridge University Press. Lewis, W. Arthur 1954 Economic Development with Unlimited Suppl ies of Labour. The Manchester School. 1 30. Marx, Karl 1970 (1859) A Contribution of the Critique of Political Economy. New York: International Publishers. Mathu, E.M. and T.C. Davies 1996 Geology and the Environment in Kenya. J ournal of African Earth Sciences 23(4):511 539. Mdard, Jean Franoise 2002 Corruption in the Neo Patrimonial States of Sub Saharan Africa In P olitical Corruption: Concepts and Contexts. Arnold J. Heidenheimer and Michael Jo hnston, eds. Pp. 379 401. New Brunswick : Transaction Publishers. Minde, Isaac, T.S. Jayne, Eric Crawford, Joshua Ar iga, and Jones Govereh 2008 Evidence from Malawi, Zambia, Knowledge Support System (ReSAKSS) Working Paper No. 13. Pretoria: ReSAKSS. Minot, Nicholas 2009 Po licy Symposium of the COMESA African Agricultural Markets Programme (AAMP). Washington, DC : International Food Policy Research Institute. Morera, Maria C. an d Christina H. Gladwin 2006 Does Off Farm Work Discourage Soil Conservation? Incentiv es and Disincentives Throughout Two Honduran Hillside Communities. Human Ecology 34(3): 355 378. Morera, Maria C. 1999 At the Crossroads: Soil Conservation and Livelihood Strategies on the Gainesville : University of Florida.
196 Morris, Michael, Valerie A. Kelly, Ron J. Kopicki, and De rek Byerlee 2007 Fertilizer Use in African Agriculture: Lessons Learned and Good Practice Guidelines. Washington, DC : The International Bank for R econstruction and Development/The World Bank. M ller, Daniel and John Mburu 2009 Forecasting Hotspots of Forest Clearing in Kakamega Forest, Western Kenya. Forest Ecology and Management 257(2009):968 977. Mwayuuli, M.S. 2003 The History of the Isukha and Idakho Clans. Nairobi: Catholic University of Eastern Africa Publications. Nakabayashi, N. 2003 Mwayuuli and the History of Clans: An Introduction. In The History of the Isukha and Idakho Clans. M.S. Mwayuuli. Pp. 11 3 1. Nairobi: Catholic University of Eastern Africa Publications. National Cereals and Produce Board N.d. History of NCPB. National Cereals and Produce Board. http://www.ncpb.co.ke/index.ph p accessed October 7, 2010. Nekesa, P. and B. Meso 1997 Tra ditional African Vegetables in Kenya : Production, Marketing and Utilization. In Traditional African Vegetables. Promoting the Conservation and Use of U nderutilized and Neglected Cr ops. 16. Proceedings of the IPGRI International Workshop on Genetic Resources of Traditional Vegetables in Africa: Conservation and Use, 29 31 August 1995, ICRAF HQ, Nairobi, Kenya. L. Guarino, ed. Rome: International Plant Geneti c Resources Institute. Nelles Verlag 2005 Nelles Map: Kenya. Mnchen: Nelles Verlag GmbH. Netting, Robert McC. 1993 Smallholders, Householders: Farm Families and the Ecology of I ntensive, Sust ainable Agriculture. Stanford: Stanford Univer sity Press. Nyang'Oro J ulius and Timothy Shaw 1998 The African State in the Global Economic Context. In The African State at a Critical Ju ncture. Leonardo Villalon and Phillip Huxtable, eds. Boulder: Lynne Rienner. Orvis, Stephen 1997 T he Agrarian Question in Kenya Gainesville: University Press of Florida.
197 Parkinson, T om, Matt Phillips, Wi ll Gourlay and Caroline Evans 2006. Kenya Footscray: Lonely Planet Publications Pty Ltd. Pisanelli, A., S. Franzel, J. De Wolf, R. Ro mmelse, and J. Poole 2000 The Adoption of Improved Tree Fallows in Western Kenya: Farmer Practices, Knowledge, and Perception. Submitted to Agricultural Systems. Place, Frank, Michelle Adato, Paul H ebinck, and Mary Omosa 2005 The Impact of Agrofo restry Based Soil Fertility Replenishment Practices on the Poor in Western Kenya. International Food Policy Research Institute (IFPRI) Research Report 142. Washington, DC : IFPRI. Place, Frank, Steven Franzel, Joris DeWolf, Ralph Rommelse, Freddie Kwesiga, Amadou Niang, and Bashir Jama 2001 Agroforestry for Soil Fertility Replenishment: Evidence on Adoption P rocesses in Kenya and Zambia. In Natural Resources Management in African Agriculture: Understanding and Improving Current P ractices. Christopher B. Barrett, Frank Place, and Abdillahi A. Aboud, eds. Pp. 155 168. Wallingford : CAB International. Quiones, Marco A., Norman E. Borlaug, and C hristopher R. Dowswell 1997 A Fertilizer Based Green Revolution for Africa. In Replenishing Soil Fertility in Africa. SSSA Special Publication no. 51. Pp. 81 95. Madison : Soil Science Society of America and American Society of Agronomy. Qureshi, J.N. 1991 The Cumulative Effects of N P Fertilizers, Manure, and Crop Residues on Maize and Bean Yields and Some Soil Chemi cal Properties at Kabete. In KARI Annual Scientific Conference. Pp. 160 167. Nairobi : Kenya Agricultur al Research Institute (KARI). Reardon, Thomas, Julio Berdegue and German Escobar 2001 Rural Nonfarm Employment and Incomes in Latin America: Overview and Policy Implications. World Development 29(3): 395 410. Reno, William 1998 War lord Politics and African States. Boulder : Lynne Rienner. Ricardo, David 2005 (1951) The Works and Correspondence of David Ricardo, 11 Vols. Piero Sraffa, ed. Indianapolis : Liberty Fund. Roberts, Paul 2008 The End of Food. Boston : Hought on Mifflin Company.
198 Ruben, Ruerd a nd Marrit van den Berg 2001 Nonfarm Employment and Poverty Alleviation of Rural Farm Households in Honduras World Development 29(3): 549 560. Sachs, Jeffrey D. 2005 The End of Poverty: Econom ic Possibiliti es for Our Time. N ew York: The Penguin Press. Sanchez, Pedro A. 1999 Improved Fallows Come of Ag e in the Tropics. Agroforestry Systems 30: 5 55. Sanchez, Pedro A., Keith D. Shepherd, Meredith J. Soule, Frank M. Place, Roland J. Bur esh, Anne Marie N. Izac, A. Uzo Mokwunye, Fred R. Kwesiga, Cyrus G. Ndiritu, and Paul L. Woomer 1997 Soil Fertility Replenishment in Africa: An Investment in Natural Resource Capital. In Replenishing Soil Fertility in Africa. SSSA Special Publication no. 51. Pp. 1 46. Madison: Soil Science Society of America and American Society of Agronomy. Schank, Ro ger and Robert Abelson 1977 Scripts Plans Goals and Understanding. Hillsdale: Lawrence Erlbaum Associates. Shiva, Vandana 1991 The Violen ce of the Green Revolution: Third World Agriculture, Ecology and Politics. London : Zed Books, Ltd. South Centre 1996 Liberalization and Globalization: Drawing Conclusions for Development. Geneva : South Centre. Spradley, James P. 1979 Ethn ographic Interview. Fort Worth : Harcourt Brace Jovanovich College Publishers. Stigli tz, Joseph E. Andrew Charlton 2005 Fair Trade for All: How Trade Can Promote Development. New York: Oxford University Press, Inc. Stonich, Susan C. 1993 "I am Destroying the Lan d!": The Political Ecology of Poverty and Environmental Destruction in Honduras. Boulder : Westview Press. Swinkels, R., S. Franzel, K. Sheph erd, E. Ohlsson, and J. Ndufa 1997 The Economics of Short Rotation Improved Fallows: Evi dence from
199 Areas of High Population Density in Western Kenya Agricultural Systems 55: 99 121. Thomson, Anne and Manfred Metz 1997 Implications of Economic Policy for Food Security: A Training Manual. Rome : Food and Agricult ure Organization of the United Nations (FAO). T hurow, Roger and Scott Kilman 2009 : PublicAffairs. Timmer, C. Peter, Walter P. Falcon and Scott R. Pearson 1990 Food Policy Analys is. Baltimore : Johns Hopkins University Press. Timmer, C. Peter 1988 The Agricultural Transformation. In Handbook of Development Economics Volume 1. Hollis Chenery and TN Srinivasan. Pp. 276 331. Amsterdam: North Holland. Tomich, Thomas, Pe ter Ki lby, and Bruce Johnson 1995 Transforming Agrarian Economies: Opportunities Seized, Opportunities Missed. Ithaca : Cornell University Press. UNAIDS/WHO 2007 AIDS Epidemic Update. Geneva : Joint United Nations Programme on H IV/AIDS (UNAID S) and World Health Organization (WHO). United Nations 2009 Kakamega Base Map. Geneva: United Nations. http://www.reliefweb.int/rw /fullMaps_Af.nsf/luFullMap/14A6905F99640EF98525766A 0065CCB6/$File/map.pdf?OpenElement accessed October 7, 2010. Wako, Daniel M. 1985 The Western Abaluyia and their Proverbs. Nairobi: Kenya Literature Bureau Were, Gideon S. 1967 Western Kenya Histori cal Texts. Nairobi: East Africa Literature Bureau. Winter Nelson, Alex a nd Gem Argwings Kodhek 2009 Kenya In Distortions to Agricultural Incentives in Africa. Kym Anderson and William A. Masters, eds. Pp. 253 282. Washington, DC : The World Bank. World Bank 2002 African Development Indicators 2002. Washington DC : World Bank.
200 Zimmerer, Karl S. 1993 Soil Erosion and Labor Shortages in the Andes with Special Reference to Bolivia, 1953 Dev elopment 21(7):1659 1675.
201 BIOGRAPHICAL SKETCH Maria C. Morera has an MA in a nthropology from the University of Florida She also holds a BS in environmental studies and a BA in liberal s tudies from Florida International Univer sity. She investigates the interface between rural livelihood strategies a nd natural resource management She has worked with highland farmers in Kenya and Honduras, examining socioeconomic factors affecting their adoption of agroforestry and soil conserv ation practices. She is currently researching socioeconomic incentives and disincentives to agr icultural intensification in western Ke n ya.