1 UNDERSTANDING THE LINKS BETWEEN WATER, LIVELIHOODS AND POVERTY IN THE NYANDO RIVER BASIN, KENYA By JILLIAN JENSEN A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2009
2 2009 Jillian Jensen
3 To the people of the Nyando River basin
4 ACKNOWLEDGMENTS I thank all of you who supported me on this ne w journey: Peter Hildebrand for his kind heart and passion; Sandra Russo for remember ing me and putting the wheels in motion; and Renata Serra for her insight and guidance. I also thank Nancy Johnson and CIAT for putting me on the SCALES team and supporting this re search. Tremendous gratitude goes out to the field personnel with the World Agroforestry Cent re who facilitated my fieldwork in Kenya, in particular my driver, Erick. I wish to thank Mark Brown, Ignacio Porz ecanski and the entire AM:W3 IGERT Program, which has taught me the meaning and importance of watersheds. Without their perspective, I would have misse d what is most salient to this story. Thank you to my parents who never waver in thei r support of this significant change in my lifes direction. I wish to send my heartfelt love to Heather, Janice and my Brazilian sisters, Bianca and Ana Leticia, who helped me through this process and never stopped believing in me. Lastly, to the glorious scalaw ag whose voice of courage and in spiration came to me when I needed it most viva la revolu tion! and to Jamie, who helped me to trust my own. xo
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ............................................................................................................... 4LIST OF TABLES ...........................................................................................................................7LIST OF FIGURES .........................................................................................................................9ABSTRACT ...................................................................................................................... .............11 CHAPTER 1 MAKING INTEGRATED WATERSHED MANAGEMENT RESEARCH-FORDEVELOPMENT INCLUSIVE AND PRO-POOR ..............................................................132 DEVELOPING A FRAMEWORK FOR UNDERSTANDING THE LINKAGES BETWEEN WATER, POVERTY AND LIVELIHOODS ACROSS WATERSHED SCALES ........................................................................................................................ .........30Background .................................................................................................................... .........30The SCALES Framework .......................................................................................................35More about Poverty Traps and Their Links to a Systems Perspective ...................................40Synthesis ..................................................................................................................... ............493 STEPS ONE, TWO, THREE METHODOLOGY! ...........................................................53Step One: Picking a Research Design ....................................................................................53Step Two: Data Collection and Sampling ..............................................................................56Step Three: Data Analys is and Interpretation .........................................................................624 POVERTY DYNAMICS AND DRIVERS OF CHANGE IN LIVELIHOODS AND WELLBEING ACROSS THE NYAN DO RIVER BASIN (1977-2002) ..............................66Trends in Poverty Dynamics across the Nyando River Basin ................................................67Livelihood Strategies across the Nyando River Basin ...........................................................72Water, Poverty and Livelihoods in Chepkemel ...............................................................75Water, Poverty and Livelihoods in Miolo .......................................................................82Statistical Results on the Relationshi p between Livelihoods and Poverty ......................86Livelihood Summary .......................................................................................................87Geographical Factors and the Spatial Dist ribution of Poverty and Livelihoods ....................90Institutional Factors behind Poverty and Livelihoods in the Nyando River Basin ................935 POVERTY DYNAMICS AND DRIVERS OF CHANGE IN LIVELIHOODS AND WELLBEING: EVIDENCE FROM HOU SEHOLD LEVEL ANALYSIS (1977-2002) ....112Stages of Progress in the Nyando River Basin .....................................................................113
6 Household-level Conditions and Behaviors Correlated with Poverty Dynamics .................115Household-Level Correlates of Poverty Dynamics: Chepkemel ...................................117Household-Level Analysis of Poverty: Miolo ...............................................................1276 SUMMARY, CONCLUSIONS AND TR AGEDY: THE WATERSHED BECOMES BLOODSHED ......................................................................................................................141Synthesis and Conclusions ...................................................................................................142Opportunities for Collective Action via Water Management to Alleviate Poverty ..............146Epilogue: Violence Erupts in Western Kenya ......................................................................149 APPENDIX A LOGISTICAL FRAMEWORK ............................................................................................151B STATISTICAL RESULTS ON LIVELI HOOD ACTIVITY PARTICIPATION ...............153C REASONS FOR POVERTY ................................................................................................158D OPPORTUNITIES FOR COLLEC TIVE ACTION: SYNTHESIS .....................................159LIST OF REFERENCES .............................................................................................................164BIOGRAPHICAL SKETCH .......................................................................................................173
7 LIST OF TABLES Table page 2-1 Key features of the povert y traps conceptual model. .............................................................523-1 SCALES Project sample villag es in Nyando River basin, Kenya. .........................................653-2 Sampled household cases in two village subunits of analysis. ...............................................654-1 Economic transition matri x, Nyando River basin (1977 2002). ..........................................994-2 Economic transition matri x, Nyando River basin (1992 2002). ..........................................994-3 Economic transition matri x, Nyando district (1992 2002) ..................................................994-4 Economic transition matri x, Kericho district (1992 2002) ..................................................994-5 Trends in the percentage of poor households disaggregated by village. ..............................1004-6 Poverty dynamics (1992 2002) for cas e study villages, Chepkemel and Miolo ...............1014-7 Typical planning calendar for Chepkemel sma llholder farm. ..............................................1014-8 Comparison of maize productivity and inputs into crop farming between Chepkemel and Miolo. ........................................................................................................................1024-9 Typical planting calendar for Miolo smallholder farm. .......................................................1024-10 Percentage changes in odds of being poor and odds of entering poverty.. .........................1034-11 Percentage impact identified livelihood strategies have on the probability of being poor. ......................................................................................................................... ........1035-1 Stages of progress for Ch epkemel and Miolo villages. ........................................................1385-2 Regression results on reasons identifie d for poverty and whethe r household was poor or not poor. ...........................................................................................................................1385-3 Percentage influence of specifi ed reason on probability of being poor. ..............................139A-1 Logistical framework ..................................................................................................... ......152B-1 Livelihood activity participation rates, 1992 and 2002. ......................................................154B-2 Livelihood activity part icipation rates, Chepkemel and Miolo (1992 2002). ...................155B-3 Chepkemel logistic regression results modeling the relationshi p between livelihood activities and poverty status (poor versus not poor). .......................................................156
8 B-4 Miolo logistic regression results modeling the relationshi p between livelihood activities and poverty status (poor versus not poor) ........................................................................156B-5 Chepkemel logistic regression results modeling the relationshi p between livelihood activities and entering into poverty ..................................................................................157B-6 Miolo logistic regression results modeling the relationshi p between livelihood activities and entering into poverty. ................................................................................................157C-1 Reasons identified for poverty st atus based on household interviews. ................................158D-1 Opportunities and constr aints for collective action. ............................................................160
9 LIST OF FIGURES Figure page 1-1 Map of the Nile River basin ...................................................................................................251-2 Population density across the Nile River basin ......................................................................261-3 Map of land cover across the Nile River basin .......................................................................271-4 Map of the twelve sub-catchments, including Nyando, of the Lake Victoria ........................281-5 Hydrologic map of the Nyando River basin ...........................................................................292-1 The SCALES Framework .......................................................................................................502-2 Graphic model of the poverty traps framework. ....................................................................513-1 Map of sample villages across the Nyando River basin .........................................................633-2 Map of political districts and sample d village locations fo r SCALES project. ......................633-3 Multiple-case (embedded) rese arch design for SCALES (Kenya) .........................................644-1 Poverty dynamics, Kericho di strict, for sample villages ......................................................1044-2 Poverty dynamics, Nyando district, for sample villages. .....................................................1044-3 Intensive farming of maize and sweet potatoes on smallholder farm, Chepkemel. .............1054-4 Livelihood system of a t ypical high-level equilibrium state household in Chepkemel ........106 4-5 Livelihood system of a downwardly-m obile household on a pathway toward a lower equilibrium state in Chepkemel. ......................................................................................107 4-6 Livelihood system of a low-leve l equilibrium household in Chepkemel. ...........................108 4-7 Slope farming in Chepkemel. ............................................................................................. .109 4-8 Miolo environs showing the soil erosi on and loss of land of the local vulnerability context. ...................................................................................................................... .......109 4-9 Miolo farm plot after previous nights heavy rains and flash flooding. ..............................110 4-10 Miolo farmers plan t sisal on plot boundaries ....................................................................110 4-11 Earth harvested from local gully .......................................................................................111 5-1 Prototypical Stages of Progress ......................................................................................... ..139
10 5-2 Dense intercropping, Miolo farmer .....................................................................................14 0 5-3 Firewood sales. ....................................................................................................................140
11 Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science UNDERSTANDING THE LINKS BETWEEN WATER, LIVELIHOODS AND POVERTY IN THE NYANDO RIVER BASIN, KENYA By Jillian Jensen May 2009 Chair: Peter Hildebrand Major: Interdisciplinary Ecology Implementation of integrated watershed mana gement policies is often challenged by the presence of poor, marginalized people in upper catchments who typically depend on the natural resource base for their livelihoods. Upstr eam resource management practices, however, commonly produce water transitions degrading environmental c onditions and livelihood support of lowland areas, where the resulting hydrological ly-vulnerable situation traps people in waterrelated poverty. This generalized account aptly describes the Nyando River basin of western Kenya. The basin is a significant source of sediment loading into Lake Victoria providing strong impetus for conservation-oriented land management practices in the upper catchment. This is complicated by the widespread poverty conditions of the basin such that the livelihoods at stake must be considered. An optimal goal is to impl ement a management strate gy that can achieve the desired environmental outcome while simultane ously improving livelihoods and alleviating poverty. Social equity conditions would also be me t if this goal is inclusive of all households. The current study is part of the intern ational Challenge Program on Water and Food, Theme 2 project, SCALES, which is testing the hy pothesis that investment s in building social capital and fostering collective action around wa ter management can assist households in escaping poverty traps. This study s main goal is to assess the potential for change in livelihood
12 outcomes that could result from any new coope rative efforts in water management by probing the dynamic relationship between water, liveli hoods and poverty over the period from 1977 to 2002. The analysis finds that demographic and geographical conditions are principal drivers behind the dynamic patterns of poverty within the basin. Institutional changes consequential to the economic reforms of the 1990s combined with exogenous climatic shocks to deepen poverty substantially over the decade. The conclusion reached is that while local group action is important for alleviating the most immediate cond itions of poverty, wider structural reform is necessary to arrest the conditions that will otherwise reproduc e poverty into the future.
13 CHAPTER 1 MAKING INTEGRATED WATERSHED MANAGEMENT RESEARCH-FORDEVELOPMENT INCLUSIVE AND PRO-POOR This research project is carri ed out as part of the Cha llenge Program on Water and Food (CPWF), an international, multiinstitutional, and agriculture-related research-for-development initiative of the Consultative Group on International Ag ricultural Research (CGIAR). Purported in 2005 to be the largest research effort on water, food, and the e nvironment underway in developing countries, the rationale behind the CPWF is growing conditions of water scarcity associated with agriculture, where it is claimed to consume 70 -90% of available water supply either directly or through contamination (CPWF, 2005). Thusly, the heart of the CPWF revolves around the issue of water productivity, with the central goal being to produce more food using less water. By reducing scarcity and providing sufficient water in terms of quality and quantity, the CPWF anticipates positive impacts on househol d food security, sustainable rural livelihoods (poverty alleviation), human health, and enviro nmental security and sustainability. The CPWF states its overall impact target as follows: Global diversions of wa ter to agriculture are maintained at the level of the year 2000, while increasing food production, to achieve internationally adopted targets for decreasing malnourishment and rural poverty by the year 2015, particularly in rural and pe ri-urban areas in river basins with low average incomes and high physical, economic or environmental water scarc ity or water stress, with a specific focus on low-income groups within these areas (CPWF, 2005, p.15). Solving issues of water scarcity is not as simple as using le ss. Drivers of scarcity can be categorized as physical, economic or social and are the result of processes, policies and decisionmaking occurring at multiple scales. Physical scarci ty represents a lack of sufficient water of acceptable quality, commonly described as an issu e of the size of the water pie. Economic scarcity represents a lack of infrastructure investment in th e development and distribution of
14 water resources where such increased capacity co uld also increase the size of the water pie. Social scarcity occurs where there is a lack of social adaptive capacity to changes in water supply and demand (CPWF, 2005, p.1) and relates to issues of share s of the water pie. Increasing water productivity en tails both aspects of water-sav ing and water-sharing through actions taken also at multiple scales. Initial emphasis for improving wa ter management focused on decentralized, local collective action1 for example, through the development of local wateruser groups increasingly supported by new st atutory laws to strengthen legitimacy and enforcement however, there has been increasin g recognition of the need for other forms of collective action involv ing government agencies, non-governme ntal organizations and private firms (Swallow et al. 2006). This imperative to link colle ctive action from local to regional levels stems from the diversity of interests in resource management across users and uses particularly in times of scarcity, and thus ly, the potential for negative environmental externalities, including net water losses, reduc ed flows, groundwater depletion, or ecosystem pollution or degradation, to flow between scales necessitating c oordination and conflict mitigation. However, integrating collective action for water management across scales to achieve desired environmental outcomes is challenged, esp ecially in developing c ountries, in how to be inclusive of the poor and margin alized, commonly, systemically excluded from other political and social process ( ibid p.361). Building knowledge to resolv e this tension, this ambiguity between collective action and poverty alleviation is the ultimate goal of the research presented herein. This reorientation toward an integrat ed understanding and management of water holistically across a landscape is central to Integrated Water Resources Management (IWRM), 1 Following Ostrom (1990, p.25), collective action is define d as a situation where a group of principals who are in an interdependent situation can organize and govern themselves to obtain continuing joint benefits.
15 the most favored governance tool for water management in the world today. Beginning in 1977 the international community introduced IWRM as a process that aims to ensure the coordinated development and management of water, land an d related resources by maximizing economic and social welfare without compromising the sustaina bility of vital envir onmental systems (GWP, 2003, p. ii). Moreover, IWRM reflects wider sustai nable development goals by aiming to strike a balance between the use of re sources for livelihoods and cons ervation of the resources to sustain its function for future generation (ibid, p. vi), a perspective adopted by the CPWF. Yet, 30 years into its history and despit e all of the efforts by the international community to codify the principles and policies of IWRM, there are few reported successes of its actual translation into institutions2 and practical applications (GWP, 2003; Rahaman & Varis, 2005; Kerr, 2002; Perez & Tschinkel, 2003). One of the primary reasons put forward to explain why is the uneven distribution of costs and benefits across socioeconomic groups due to the pers istently poor understanding of the relationship between wate r and livelihoods (Ke rr, 2002). Reporting on numerous watershed management projects in India, Kerr (2002) finds that other IWRM goals of efficiency (productivity) and environmental su stainability (conservation) are confounded by project activities which exacerbate social inequity by enacting change s that pay little attention to the role of water and other common property re sources in the liveli hoods of the poor. Often, changes in resource management institutions to satisfy conservation or environmental objectives have a poverty trade-off by proscribing access or uses that are central to livelihood security (Kerr, 2002; FAO, 2006). Therefor e, understanding the relationshi p between water, poverty and 2 Institutions defined by Oakerson (1992, cited in Swallow et al ., 2006, p.366-7, emphasis in original) are the decision-making arrangements (rules) or authority relationships that specify who decides what in relation to whom Institutions can either be customary (traditional) or statutory.
16 livelihoods is emerging as a critical first step to achieving wider IWRM goals, the same goals expressly pursued by the CPWF. The CPWF has organized and coordinated its global research program along thematic and geographical (spatial) lines. Firstly, five them es were identified to focus research on different aspects and scales ( e.g., ecosystem, river basin, and global) of water productiv ity issues, ranging from locally enhancing crop and fisheries perfor mance to understanding th e role and importance of macroeconomic factors. This theme-based approach is intended to narrow principle focus onto the endogenous factors behind wate r productivity issues that can be directly investigated and impacted by CPWF research. The research outp uts are expected to include a variety of agricultural, environmental, inst itutional, and policy innovations to address the needs of the rural poor through increased water productivity. For example; CPWF research seeks higher crop yields, more income, more employment opportu nities, improved liveli hood quality, or some combination of these, for each cubic meter of water exploited. (CPWF, 2005, p.9) Three categories of solutions to improve the livelihood s of the poor are identified: economic solutions generating higher incomes for each cubic meter of water used; social solutions creating more jobs and higher food security for each cubic meter of water used; and environmental solutions obtaining greater resilience of ecosystems for each cubic meter of water present (CPWF, 2005). These solutions are to yield from innovations in technologies ( e.g ., crop varieties and irrigation systems), institutional mechanisms ( e.g ., payment for environmental services programs to upstream residents to alter their resource manage ment behavior for the benefit of downstream inhabitants, the resignification of customary, cooperative institutions, and multi-stakeholder processes), and, finally, policy recommendations for national and/or transboundary water-use efficiency or water-sharing agreements (ibid ).
17 The second organizing principle is based on recognition of the multi-scalar, dynamic and interacting qualities of watershe ds, meaning the basin scale is the most sensible to focus upon when compiling, synthesizing and implementing knowledge about increasing water productivity. Hence, nine benchmark basins, representing a wide variety of food and water challenges, were selected from across Asia, Africa and Latin America to conduct CPWF research and development efforts within. The so-named Basin Focal Projects (BFPs) are tasked with this knowledge-integrating function and their products are to be used in guiding development plans for the target basins. The research presented here is situated as part of basin-scale Theme 2, Water and People in Catchments, and is locate d in the Nile Basin of Africa. The particular project this research contribut es to is Project Number 20 (PN 20), Sustaining Inclusive Collective Action that links across Economic a nd Ecological Scales in Upper Watersheds (SCALES), which conducted a comparative study in the Andes Basin, specifically Colombia, as well as this research on the Nyando sub-basi n of western Kenya, located in the most southwestern tip of th e (White) Nile Basin. Water catchments are areas in which rainfall collects and eventually flows into creeks, rivers, lakes, oceans, or groundwater ( www.waterandfood.org). Further described, they are integrated systems where the complex relations hip between people, land and water are played out ( ibid ). Specifically, how people manage the la nd, forest and water resources of upper catchments to generate their own local livelihoods can produce what are term ed lateral flows, for example, of water, soil, nutrients or sedi ment, across the landscape tying landscape users together across scales (Swallow et al ., 2006). Water management can also produce water transitions, defined as negative im pacts (alt. externalities) on th e quality, quantity and timing of water for people, their livelihood options a nd welfare downstream in the catchment ( ibid ).
18 Downstream residents may accept these water tr ansitions passively or may counter through reverse flows of political, financial, social or other resources to influence the behavior of upstream resource users. Finding ways to be tter manage these upstream downstream interdependencies is the principle objective of Theme 2 research, echoing the shift in resource management thinking away from the local farmor plot-scale to the basi n scale. Theme 2 views the spillover effects3 between resource management and li velihood outcomes across scales as a special type of coordination problem, which can be solved through improved understanding of how the prevailing social and biophy sical systems interact over a watershed. To this end, a new conceptual framework has been laid out to li nk institutional and biophysical processes, which also serves as a tool to identify key issues affecting collective action within and across scales (Swallow et al ., 2006). Theme 2 projects have as their central goal to obtain sust ainable livelihoods4 across a basin, but, in particular, in upper catchments where, prototypically, the poorest and most marginalized and at-risk people live. Because the poor of upper catchments often have a narrowly-prescribed livelihood system and usuall y depend heavily on the natural resource base, social equity demands that environmentally-nece ssary changes in resource management of upper catchments recognize their dependence and vulnerability to avoid an unfair di stribution of costs. Learning from earlier IWRM programs, the CP WF acknowledges the importance to understand the link between water and poverty so that it can avoid prescribi ng perhaps the most efficient 3 Spillovers are defined by Barrett (200 7, p. 7) as special cases of coor dination failures, which result whenever externalities affect not only the welfare of others, but also their behaviors. This means an individual or households livelihood decisions are based not only on their own assets, but in consideration of other agents asset stocks and livelihood choices. 4 For clarity, the following definitions are adopted: A livelihood system comprises the capabilities, assets (including both material and social resources) and ac tivities required for a means of living. A livelihood is sustainable when it can cope with and recover from stresse s and shocks and maintain or enhance its capabilities and assets both now and in the future, while not undermining th e natural resource base. (Carney, 1998, p.2 cites by de Haan & Zoomers, 2005, p. 31)
19 solutions for improved resource management ( i.e. increasing overall economic returns to water) meanwhile resulting in inequitable outcomes or trade-offs regarding the welfare of the poor or the marginalized contrary to its motivations. Th erefore, one of the main activities of Theme 2 projects is to answer the following research questions: Where are th e poor? What is their relationship with environmental degradation? ( www.waterandfood.org ) To reinforce the intention to view systems and management holis tically as a watershed unit, these additional questions are posed: What is the nature of wa ter-related poverty in primary catchments? To what extent can it be alleviated through better water management? ( ibid ) As part of Theme 2, the SCALES Project asks these same questions to develop an understanding of the relationship between water and poverty in its two target watersheds. To do so, the project adapts to a watershed contex t the poverty traps concep tual framework, linking land use and scale to poverty indicated by vicious cycles of un der-investment, low productivity and environmental degradation (Barrett & Swallow, 2006). With poverty alleviation as the projects primary impact target, SCALES ar gues the extent to which improved watershed management can alleviate poverty hinges on whethe r increased availability or productivity of water can help the poor undertake new livelihood strategi es that will lead them out of poverty traps. The SCALES project intends to take this knowledge generated about the prevailing poverty traps to determine the pot ential for collective act ion to be poverty-alleviating in part by resolving the coordination problems found in wate rsheds. The SCALES proj ect seeks to fill the gap that purportedly exists betw een what is known generally about how to foster local collective action and the unique challenges pos ed by the hierarchical, scale-ne sted structure of watersheds to scale-up action to involve stakeholders, sectors and interests basin-wide.
20 The SCALES project articulate s several objectives relating to its goal, and promises several research outputs Appendix A contains a synthesizing logical framework (Table A-1) detailing the narrative summaries of the CPWF, Theme 2 projects, and the SCALES project. Table A-1 also displays the narrative particul ars about the present research and shows how it nests within and supports each ascending resear ch initiative. The key problem I attempt to address derives from the need for equity in water management by attempting to understand the relationship between water and poverty in the Nyando catchment in the upper (White) Nile Basin. This research project, with the admittedly banal yet apt title, Understanding the links between water, livelihoods and poverty in th e Nyando River basin, is designed to fulfill SCALES (Kenya) research output no. 2: The id entification and analysis of poverty traps at different scales. This means identifying the livelihood strategies5 of the poor and the factors determining their success or failure, with special attention paid to the role of water. I use the conceptual framework as a guide to link significant processes at multiple scales to poverty and to interpret what opportunitie s exist for future collective action. Altogether, this research is envisioned as undertaking the sc oping phase of the basic CPWF research process (CPWF, 2005). I state the basic research questi ons as follows: What are the live lihood systems and strategies of the poor and non-poor in the Nyando Basin? What are the poverty traps at di fferent scales in the Nyando Basin? What role does water play in the livelihood systems of the poor? Answers to these questions will be of value to SCALES project partners, non-governmental organizations and government agencies in designing inclus ive processes for watershed management. 5 A livelihood strategy is defined as the way people tran sform assets into income, services and products that ultimately become transformed into welfare outcomes (Bar rett & Swallow, 2005; see also Scoones, 1998; Chambers & Conway, 1991).
21 Altogether, the Nile Basin covers an area of 3.3 million square kilometers and involves ten riparian countries (Figure 1-1). The current population of 280 million is expected to grow to nearly 600 million by 2025, an average population growth rate of 2.5-3.0% per annum, with an average population density of 95.5 persons per square kilomete r (Figure 1-2) (www.cpnile.net ). Poverty, food security, land degradation, water po llution, and general water scarcity are major challenges for each of the riparian st ates with the average GINI index6 score greater than 50% and more than half the total population living be low the international $1 per day poverty line. Food security into the future as national populati ons continue to grow is a major problem facing the riparians, with eight out of ten countries now scori ng lower on per capita food and agricultural production indices compared to 1982 (Egypt and Uganda are the only two exceptions). In Rwanda, Sudan, Tanzania and Ethi opia, the food security situation has declined an average of 17 percent. In res ponse, major expansions in irrigated agriculture are planned, with Egypt and Sudan expected to add more than 4.9 million hectares, Uganda, Tanzania and Kenya to add nearly 400,000 hectares, and over 265,000 ne w hectares are expected in Ethiopia ( www.cpnile.net ), making likely future conflict over the distribution of water across the basin. Rainfall in the summer months on the Ethiopian highlands and the catchment areas of the southern equatorial lakes re gion, including Lake Victoria, is the source of Nile water ( www.cpnile.net ). The remainder of the basin tends toward arid and semiarid conditions (Figure 1-3 reflects water scarcity in land use); areas which are now experiencing conditions of acute water stress or scarcity over part of the year producing situations of serious environmental degradation and rural outmigration. The CPWF is working with stakeholders and partners across the Nile Basin to reverse these trends, prioritizing research on issues su ch as enhancing rainfed 6 The GINI index is a standard measure of inequa lity; for a detailed explanation, see Todaro (1977).
22 agriculture in the upper basin; developing wate r saving technologies; promoting sustainable fisheries; and control of aquatic weeds ( www.waterandfood.org ). Of key importance is work on institutionalizing regional coope ration for managing and developing the water resources of the Nile Basin to avert conflict and to obtain equ itable, efficient and environmentally sustainable solutions. Kenya has approximately 10% of its national area within the Nile Basin, in the equatorial lakes region of East Africa located at the very so uthern end of the (White) Nile Basin. Labeled domestically as the Lake Victoria drainage basin (Figure 1-4), it includes the next smaller scale, the Nyando River basin actually a catchment by the definition offere d here, but I will stay with the convention in the literature of calling it a basin. However, as th e Lake Victoria basin is also some of the wettest area in an otherwise semi-a rid or dry sub-humid country more than 40% of the countrys population resides there. The SCA LES Project selected the Nyando River basin (Figure 1-5) with an area of 3517 square kilomete rs it is the smallest ye t most environmentally significant of the catchments to drain into Lake Victoria as the African focal basin of its comparative case study. There are thr ee reasons that make it an inte resting and relevant selection. Firstly, it provides an opportunity to apply the conceptual framewor k in what are described as the east African water towers, ar eas where the prototypical distri bution of poor people in upland areas and wealthier people downstr eam who can balance the flow of water with reverse flows of money and power do not prevail (Swallow et al. 2006). In the Nyando River basin, the situation is that water and power are mutually constitu tive, flowing downstream together. Secondly, the Nyando River basin is very representative of the type of deforestation an d slope cultivation that has led to heavy soil erosion, biodi versity loss and sedimentation with in the Nile Basin at large, all significant environmental reas ons signaling the need for an in tegrated management approach.
23 The Nyando River basin has the highest average sl ope and sediment transport index score within the Lake Victoria basin; biophysic al factors explaining why it has become a major contributor to the eutrophication of the Lake. It is estimated th at approximately 61% of the Nyando River basin is a sediment source with an average erosion rate of 43 metric tons per hectare per annum (t/ha/yr), translating into an average soil loss to the lake of 3.2 million metric tons per annum since 1963 (World Agroforesty Center, 2006). The consequential environmental impacts in Lake Victoria over the last 40 years include extended spells of anoxia ; five times the concentrations of algal biomass; 67 percent decrease in lake tran sparency; and explosions in invasive aquatic species, most notably water hyacinth (Odada et al. 2004; Mungai et al. 2002). Thirdly, the Nyando River basin exhibits the same characteris tics as found across the N ile Basin in terms of high population density (Figure 1-2) and poverty prevalence. In 2000, the average population density of the Nyando basin was 214 persons per square kilometer ranging to as high as more than 1200 persons per square kilometer, with an approximate total popul ation in 2003 of 746,000 (Olaka, 2005). In 1997, the preval ence of poverty across the Nyando River basin ranged from 58 to 66% compared with the national poverty rate of 53% (KNBS, 2007). Given the similarity of conditions in the Nyando River sub-basin and the N ile River basin at large, studying this smaller basin may lay some foundations, build knowledge, and develop innovations to help CPWF as it begins to negotiate integrated water ma nagement at the next larger scale. This thesis proceeds as follows. Chapter 2 presents a review of the literature on understanding the relationship be tween water and poverty in catch ments. There is also a presentation of the new conceptual framework proposed by Theme 2 and further developed by the SCALES Project, which links so cial and biophysical systems in watersheds to make explicit the multi-scale processes and interactions that re sult. The conceptual framework of poverty traps
24 is also presented and discussed as one device to make the link between water and poverty in this analysis. Chapter 3 details the various methodologi es for sample selection and data collection and analysis employed in this research. Chapte r 4 analyzes the poverty dynamics and livelihood strategies data to characterize the water, pove rty and livelihoods relationship across the Nyando River basin and to begin to construct an unde rstanding of poverty traps at multiple scales. Chapter 5 builds on the relationship between pover ty and livelihoods estab lished in Chapter 4 by analyzing household-level conditio ns and behavior to interpret local-level poverty traps. This information is integrated into answering the question whether improved water productivity and management in the Nyando River basin might be poverty-alleviating by faci litating participation in more remunerative livelihood activities. Ch apter 6 concludes with some reflection and recommendations for future research.
25 Figure 1-1. Map of the Nile Rive r basin. (Retrieved March, 2009, from www.nilebasin.org)
26 Figure 1-2. Population density across the Nile River basi n. (Retrieved March, 2009, from www.nilebasin.org)
27 Figure 1-3. Map of land cover across the Ni le River basin (Retrieved March 2009, from www.nilebasin.org)
28 Figure 1-4. Map of the twelve sub-catchments, including Nyando, of the La ke Victoria basin in the upper White Nile basin. (Source: Douglas et al. 2006)
29 Figure 1-5. Hydrologic map of the Nyando Ri ver basin with boundary outlined in green. (Source: Teyie et al ., 2006)
30 CHAPTER 2 DEVELOPING A FRAMEWORK FOR UNDE RSTANDING THE LINKAGES BETWEEN WATER, POVERTY AND LIVELIHOO DS ACROSS WATERSHED SCALES Background The primary purpose of the CPWF is to increas e the productivity or efficiency of water within river basins as a means to simultaneous ly achieve improved environmental and social outcomes; the latter inclusively defined as wellbeing to capture multidimensional gains in, for example, health, food security and poverty alleviation. One key challenge for the CPWF thus becomes to properly understand the dialectic be tween water and livelihood outcomes to ensure development interventions are incl usively designed and targeted to satisfy the welfare needs of the rural poor1 and to bring potential pove rty environment tradeoffs into full focus. In 2003, the Global Water Partnership (GWP), the internationa l organization established to administer the policy framework IWRM, stated in re sponse to the necessity to better account for social equity in water and watershed management that The curr ent analytical failure to capture the critical importance of water in livelihood provision and protection needs to be remedied. (GWP, 2003, p.13) In particular, the GWP expressed the need to expand the focus of research beyond the heretofore primary concerns over drinking wate r and sanitation toge ther defined as the domestic uses of water and a principal concern of the Millennium Development Goals to look at the implications of environm ental quality and management, including but not limited to water, on the wellbeing of those whose livelihood systems are interdependent with the state of the natural environment, for example, smallholde r farmers and fishers. Minus an holistic and integrated understanding thre atens the chance of congruently achieving the three goals of sustainable development broadly and IWRM in particular economic growth or efficiency, 1 The CPWF also targets the peri-urban poo r as beneficiaries; however, this study is focused on only the rural poor and rural livelihoods, and fra mes its analysis accordingly.
31 environmental security, and social equity as global studies have confirmed ( e.g ., Perez & Tschinkel, 2003). Similarly, the CPWF recognizes that the role of water in wellbeing is more complex than the provision of drinking water or sanitation se rvices, instead acknowledg ing the critical role water plays in satisfying multiple productive and domestic needs ( e.g ., van Koppen et al., 2008; Moriarty et al ., 2004). Increasing water productivity (WP) as its goal, therefore, signals the turn away from principal or exclusive focus on domestic uses of water to its impact on production and livelihoods more broadly. Extensive and varied work is being produced by CPWF scientists trying to improve WP, basically defined as the return per unit of water consumed (CAWMA, 2007). This definition is reflected in the simple, standard measure for WP as a ratio of some measure of output in the numerator and some me asure of water availability or use in the denominator (Vosti et al ., 2006, p. 1). One common approach to try to maximi ze WP is to match water allocation to agroecological conditions. Such water productivity maps are intended to act as guides to policymakers in developing resource realloca tion and diversion plans to engineer water distribution to the areas of highest agricultural potential (e.g., Din Ahmad et al. 2008; Vosti et al. 2006). However, the output of these studies, as presaged by the WP formula above, tends to be based solely on biophysical systems data wit hout sufficient analysis and integration of human systems reflected in historic cultural and inst itutional conditions. This is troublesome, should these outputs be used as the sole basis fo r watershed management, based on the growing evidence that water-related poverty is less a c onsequence of physical scarcity (insufficient quantity) but more the result of economic and soci al scarcity associated with poor governance, competing demands, and inequitable access (Mulligan et al. 2008; Sullivan et al. 2008).
32 Growing case study evidence includes that from northern Lao PDR where Gurung et al. (2008) derive a list of the determinants of pove rty, yet water availability was not amongst those that significantly correlated with improved agricultural productivity or allevi ation of poverty. In assessing the relevance of usi ng WP with respect to fisheries and aquaculture, Nguyen-Khoa et al. (2008) conclude that the concep t must be further developed to integrate social and ecological conditions except in those applications where wate r is the only limiting factor in aquatic resource production. So, whilst the stated objectives of enhanced WP are increased food production, poverty alleviation and ecosystem c onservation, they, like the goals of IWRM generally, also are not necessarily congruent when based on implem enting the WP concept as currently defined (Nguyen-Khoa et al. 2008). Accordingly, some CPWF pr ojects have been developing and experimenting with a variety of conceptual models and tools in efforts to capture the multiplex relationship between water, poverty and liveli hoods by looking beyond a singular focus on the biophysical determinants of low WP ( e.g ., rainfall variability, erosi on or low soil fertility, and use of traditional crop varieties) to try to capture other contextu al factors and social conditions which also influence WP and poverty. Some models have maintained a fairly narro w focus on the localor household-scale. For example, one study deployed a farming systems simulation model, OLYMPE, in the Olifants River basin, South Africa, to perform a microec onomic analysis of smallholder farmer income subject to constraints of capit al, labor, land, water availabili ty and market price dynamics (Magombeyi & Taigbenu, 2008). Results show that livelihood strategies based on livestock and crop diversification each alrea dy well-established as the liveli hood strategies critical to the wellbeing of sub-Saharan smallholders are most likely to stabilize inco me and food security and are associated with house holds displaying the most resi lience in coping with extreme
33 exogenous events such as cyclones. What this me thod of analysis does is draw attention to the relationship between a smallholder households constraints and thei r capacity to adopt or achieve the comparable performance levels of the most remunerative livelihood strategies. It is a positive attempt to link a households livelihood outcome to both its agroecological context and microeconomic conditions. However, this common analytical focus on householdor farm-scale productivity constraints as the sole locus of action for pove rty alleviation denies the momentum toward broader units of analysis and intervention, including the landscape, catchment and watershed (German et al. 2005, p.1). It can also spark anew the poverty-environment tradeoff debate with one side, often natural scientists and conserva tionists, arguing that focusing on poverty and equity issues narrows the focus too much away from larger system needs and functions. For example, Perez & Tschinkel (2003) argue that the poverty approach tends to re strict the unit of intervention to the farmor plot -level with the peril of missing more important environmental threats and their sources to the watershed at la rge. They state: A concentration on poverty alleviation does not encourage development orga nizations to think in terms of economic or ecological systems larger than farms, seriou sly undermining their capacity to contribute to sustainable, broad-base d development and natural resource management (Perez & Tschinkel, 2003, p.2), which, instead, entails a shift in fo cus from technology generation to system regeneration. (German et al. 2005, p.1) This opinion is strengthe ned by studies th at find WP is not an explicit goal of farmers or fishers for example, Dayal (2008) reports a positive correlation between technology a doption by farmers and chances of economic benefit but not between technology adoption and potential wate r savings but is determined by other
34 socioeconomic and agroecological incentives or c onstraints which factor into an individual or households livelihood decisions (Maneta et al ., 2008). Social scientists tend to argue oppositely in favor of the pove rty approach because of the environmental consequences associated with the behavior of the poor and their resource use decisions (Reardon & Vosti, 1997). Further, the poverty approach has become not incongruent with a systems perspective through the application of integrated, multi-scale systems models that capture important contextual factors from the human system, which may be more significant for livelihood and environmental outcomes. For instance, Maneta et al. (2008) report findings from research in the So Francisco Ri ver basin, Brazil, which developed three interrelated models of water-agriculture-poverty links at plot, sub-catchment and basin sc ales to assess the effects of policy changes on farmer behavior, farmer inco me, rural poverty, and water resources. This integrated approach found that under some agroecological and socioec onomic conditions waterpoverty links do exist, and, in these cases, policies that aff ect access to water can reduce poverty. (Maneta et al ., 2008, p. 150) In other contexts wher e there are found to be no direct links between water and poverty, an alysis determined that allevi ating persistent rural poverty will depend more on policies relating to cr edit access and technology dissemination, which encourage a change in the product mix and improve ments in agricultural productivity leading to improved and more secure livelihood outcomes (Maneta et al ., 2008). These changes nevertheless can generate indirect gains in environmental quality. Social scientists argue theo retically that a focus on pove rty alleviation can resolve spillover effects and coordination problems by envisioning watershed management as a framework for enhancing collective action and equity in natural resource access and governance, or livelihood problems th at cannot be solved at the level of the farm or household.
35 (German et al ., 2006, p. 2) Despite critics, the impe rative to look beyond the farm scale exclusively is also critical to the success of th is mode of decentralized resource management and requires identifying those aspects of the physi cal, cultural, and institutional setting that are likely to affect the determination of who is to be involved in a situation, th e actions they can take and the costs of those actions the outcomes that can be achieved (Ostrom, 1990, p.55) However, beyond the problem of actually material izing inclusive, multi-scale collective action for watershed management (Biggs, 2001), there remains a dearth of evidence that collective action in water management can actually support desired social outcomes in terms of poverty reduction and social inclusion or equity across a watershed (Swallow et al. 2006). The SCALES project, therefore, experiments with CPWF Theme 2s conceptual framework to test the hypothesis that collective action via improved water management is poverty-alleviating. The SCALES Framework The Sustainable Livelihoods framework is a well-known conceptual device which takes an integrated view of how livelihoods are made by connecting the microeconomy of the individual or household with the macro-level ins titutional, policy and vulnerability contexts in which it is situated (Ellis & Freeman, 2005). A ccordingly, the livelihood activities engaged in are determined by the assets a household has access to, alternatively referred to as resources or capabilities and clustered according to five capital categories (financial, human, natural, physical, and social capitals); the seasonality or other risk factors that make a households livelihood vulnerable; and the institutional and policy context, wh ich sets the wider, exogenous conditions that define peoples options for using their capital for genera ting a livelih ood (Barrett & Swallow, 2005; Swallow et al ., 2006). These three separate conditions the asset base, exposure to shocks and risks, and the socio-politi cal structure constitute the initial conditions shaping the context and opportuni ties for people to take action (Weingart & Kirk, 2008). The
36 emphasis placed on institutions of property right s and collective action as well as other legal structures in newer confi gurations of the livelihoods framework (Di Gregorio et al ., 2008; Weingart & Kirk, 2008; Swallow et al ., 2006) begins to address critics of the approach who claim that it has become overly economic in its focus on assets and capitals occluding other important factors of politic s and power (Scoones, 2009). The SCALES project builds upon this basic conceptual founda tion to incorporate issues of scale, both spatially and temporally, as a strategy to minimize th e likelihood of making a poverty-environment tradeoff in watershed manage ment. It endeavors to remedy the lack of integration between natural resources planning at large regional scales an d actions undertaken at local scales, as well as the problems caused by th e lack of linkages between natural resource managers and institutions operating at all sc ales. (Briggs, 2001, p.29) This new framework, displayed in Figure 2-1, is sh ared by every CPWF Theme 2 project and has been further developed under PN 20, the SCALES project. The models purpose is twofold. First, it seeks to unite the biophysical and instituti onal aspects of watersheds in one holistic, rather than separate, model as a first step in integrated management This integrated model thus recognizes that water management takes place at multiple and often overlapping scales, sometimes hydrologically defined, while at other times de fined according to social or administrative structures (Ravnborg, 2006, p. 395). Additionally, as a multi-scale and linked-system model, it captures such issues as coordination problems and spillover effects. Second, this combined model is viewed as a framework for understandi ng multi-scale collective action in watersheds by highlighting the key issues and pathways affecting collective act ion within and across scales (Swallow et al. 2006). The output produced from applying th is framework is a sort of poverty or livelihood map, a unified analysis of livelihoods, poverty and land (resource) use along different
37 points of a watershed (Johnson et al. 2006). The final stage in this an alysis is to interpret these results in the broader economic and environmental context of the watershed, and the implications for potential poverty-environment tr adeoffs are identified. (ibid, p.67) What this conceptual device essentially doe s is refashion the Su stainable Livelihoods framework in a way that is more representa tive of the dynamics, social and biophysical, found within a basin or watershed. The model is c onstructed around five hydrono mic zones: headwater ecosystems, uplands, midlands, lowlands, and lowland ecosystems (Molden et al. 2001). People dominate the upland, midland and lowland zones, labeled as primary nodes, where they live, interact and earn livelihoods. The two ecosystems zones affect and are impacted by water use in these primary nodes. Each primary node is define d as a locus of individual and collective actions that produce outcomes defined in terms of human welfare and modifications to water resources flowing to lower zones. (Swallow et al ., 2006, p.364) Essentially, a primary node can be conceived of as a context where individuals or households pursue a livelihood as per the dynamics of the Sustainable Livelihoods fram ework, adding in the special role of water resources and water technologies in producing that livelihood plus the possibility of spillover effects. The secondary and te rtiary institutional nodes repres ent meso and macro social, institutional and policy contexts that condition the availability and productivity of assets in generating a livelihood. In effect, they represent increasingly dist al sources and reflexives of micro-level poverty traps. The final component of the model is reverse flows, representing economic, social or political resources ( e.g ., commodities, money, regulation or influence) that flow back upstream from downstream residents in response to water transitions, changes in the quantity, quality or timing of water flows be tween hydronomic zones that may be negatively impacting downstream livelihood options (Swallow et al ., 2006). Given the variation in where
38 the poor are found in watersheds, these reverse flows may be mo re or less effective. For example, where the wealthy are found in upl and areas potentially providing a source of employment or goods to lowland re sidents as well as negative lateral flows such as, for instance, pollution or sediment, it is unlikel y that reverse flows can offset or compensate for the water transition. Indeed, in the wate r towers of East Africa, upland areas boasting the highest development and agricultural potential and the lowe st rates of livestock and human disease, this is a reasonable expectation ( ibid ). Such upstream-downstream interdependencies serve as strong justification for looking beyond the farm-scale to a multi-spatial scale approach. The unique contribution of the SCALES proj ect to the CPWFs emerging toolkit for assessing water-poverty linkages is the adaptation and appl ication of the poverty traps framework to watershed management. The basic idea of a poverty trap is any self-reinforcing mechanism which causes poverty to persist (Azariadis & Stachurski, 2004, p.33). Carter et al (2007, p.837) understand a poverty tr ap as a critical minimum asset threshold, below which families are unable to successfully educate their children, build up their productive assets, and move ahead economically over timeThose above th e threshold can be exp ected to productively invest, accumulate and advance. Further, poverty trap theorists argue this bifurcation of behavior reflects the bifurcati on of longer-term prospects and thus of economic wellbeing (Carter et al. 2007, p. 837), reflecting the importance of c onsidering the temporal scale and the potential for intergenerational transmission of poverty (Bird, 2007) implied when making resource use and management decisions. The pove rty traps framework represents something of an extension to the Sustainable Livelihoods fram ework as well by attempting to determine more precisely the necessary resources and asset thresholds that can distinguish a virtuous or selfreinforcing cycle out of povert y through endogenous asset accumula tion from a vicious cycle of
39 poverty through asset decumulation. It links th e choice of livelihood strategies, based on available assets and opportunities (including the vu lnerability and instituti onal contexts, level of risk, and entry barriers), to the equilibrium poi nt of wellbeing low or high that will be gravitated towards over time (Barrett & Swallow, 2006). The ability of collective action to be poverty-alleviating, part icularly sustainable poverty reduction, can be judged using th e poverty traps framework in the following way: The general evaluative question of collectiv e action would be does it act as a positive shock and lift households above the assets threshold such that investment, growth and further asset accumulation would happen endogenously. More parsimoniously, can collective action propel self-reinforcing growth among the poor away fr om a low-level equilibrium toward a high equilibrium level of wellbeing? (Barrett & Swa llow, 2006) In the more narrowly framed inquiry here, the question would be refi ned to whether the collectively improved management of water resources, reversing the trend toward resource degradation as sociated with the demands of poverty, can foster this upward mobility in a sust ained way by alleviating the chronic constraints facing poor households and promoting more remunerative live lihood activities. Alternatively, can water management via collective action act as a form of livelihood protection or safety net to cushion against short-term shocks, protect asse ts and help households recover essentially, increase their resilience while insulating them from asset-decumulating behavioral choices and risk-mitigating (low investment and low produc tivity) livelihood strate gies tantamount to lowering their vulnerability? Johnson et al. (2006; 2009) report results from the SCALES project in Colombia, South America, where the above framework was applied. The findings, converging with other CPWF studies, indicate only limited potential for pove rty alleviation via improved water management
40 and access. In the dynamic language of the povert y traps framework, only a minimum number of households in the subject Colombian basins are initially positioned such that improved water availability would make a struct ural change in their asset po rtfolio and mobilize them on a pathway of escape toward a higher equilibrium outcome. For poorer households, this new water supply would likely be insufficient to substantia lly change their conditions and they would remain in a poverty trap. The most successful livelihood strategies that is, those associated with the highest levels of wellbeing involve dive rsification into off-farm activities, which are recognized as causing spillover e ffects in the form of environmental degradation downstream. The authors argue that this form of multi-scale analysis allows them to identify alternative strategies that could alleviat e poverty and avoid future nega tive externalities. To enable participation in these win-win activities, however, means targeting interventions toward the various push and pull factors that otherwise determine the available livelihood options. Another report of findings from SCALES research in Colo mbia highlights that becau se of the weak direct and indirect linkages between wa ter and poverty, improvements in wellbeing are more likely to stem from developing labor and market linkage s than via better water management (SCALES, 2008). More about Poverty Traps and Their Links to a Systems Perspective Barrett (2005) argues that there are three reasons why the poverty traps framework is meaningful for agricultural research-for-developme nt, the principal activity of the CPWF and its parent, the CGIAR. First, poverty traps specify an asset-based approach and a focus on the accumulation of stocks of productive assets as the basis of livelihood strate gies. Critically, such an approach allows a distinction to be made betw een policies that promote or enable sustainable, structural transitions out of persistent or chr onic poverty versus the unsustainable, temporary or stochastic movements into or out of poverty th at do not transform the underlying state variables
41 of the system (Figure 2-2). This approach signals two practical challenges: (1) building the stock of productive assets under control of the chroni cally poor to lift their capabilities above the dynamic poverty line, and (2) prot ecting the assets of non-poor households to prevent them from being knocked down into a poverty trap (Barrett 2005). The research within the CGIAR that focuses on collective action, property rights and natural resource management is argued to address these challenges by speaking directly to the incentives people have for building up and protecting their assets. The s econd reason the poverty traps framework is meaningful for agricultural research-for-development is the empha sis it places on productivity growth, such as the WP paradigm more crop per drop that th e CPWF strongly pursues. Thirdly, the poverty traps framework promotes the idea of systems thinking, a concept well represented in the SCALES framework designed to capture the multi-scale spillover effects and coordination problems within a basin system. One of the most prolific authors on poverty traps, Barrett, identif ies six key features, which are summarized in Table 21. First is the focus on dynamics versus the historic, static approach to viewing and measuri ng poverty. The analytical concern with persistent, or chronic, poverty suggests the need to distinguish the pres ence and causes of structural from stochastic poverty, the latter referring to te mporary or transitory movements of individuals, households or communities, depending on the chosen level of aggregation, above or below the poverty line. Transitory or stochastic poverty occurs when people are better off one period than another without any significant or lasting change in their underlying circumst ances, particularly the stock of productive assets under their control (C arter & Barrett, 2006, p.180) Possible sources of these transitory changes in poverty status incl ude life cycle effects, random price and yield fluctuations, irregular earnings from remittances or gifts, or changes in luck (ibid). Conversely,
42 structural poverty is related to the decumulation of assets (e.g., c onsequential to shocks such as illness or natural disaster) or decreasing return s to assets already possessed (e.g., unemployment or declining terms of trade). Scholars argue that a new generation of poverty measures, for example, the Stages of Progress methodology ap plied by SCALES, must be used to properly identify the type of transition stochastic or st ructural that is being experienced in order to evaluate how well an economy is working for the poor, whether the poor are structurally positioned to take advantage of changes in th at economy, and whether those changes can obtain sustained poverty reduction by a lleviating those persistent lo ng-term structural conditions. (Convention is responsible for framing these assessments in the poverty traps literature relative to the economy; however, the subject here is improved water management.) Recent investigations into the patterns of household mobility, that is, escape or entry into poverty traps, have used quantitative or qualitative methods either individually or in combination ( e.g ., Adato et al ., 2006). Carter & Barre tt (2006, p.185) argue that the most effective measure for making this assessment is to determine econometrically the dynamic asset poverty line, defined as the asset threshold where a bifurcation in accumu lation dynamics occurs leading to the second feature of poverty traps: the existence of mu ltiple dynamic welfare equilibria. Alternatively, qualitative studies attempt a deeper, anthropological-type exam ination and identification of explanatory factors connected to the observed patterns of mobili ty including the specific role played by social relationships (Adato et al ., 2006). Multiple dynamic equilibria points are the consequence of different sets of initial conditions leading to qualitatively different equilibrium paths, which ultimately converge on different stea dy states, highor low-levels of wellbeing (Barrett, 2007). They represent critical thresholds or tipping points where the dynamics of accumulation and wellbeing tend to shift reinforc ing vicious or virtuous cycles of poverty.
43 The poverty traps framework has been used to link the state of natu ral resources to the prevalence of persistent poverty. Resource degradation, as a cons equence of being trapped in poverty, generates low-level stable dynamic equ ilibria for both economic and ecological state variables (Barrett, 2007, p. 11), with a good ex ample provided by Shepherd & Soule (1998) on soil fertility in western Kenya. With a very wide angle, Grey & Sadoff (2002) argue that, in part, the poverty and low-level economic growth of Africa is explained ge nerally by her water resources endowment. The natural water legacy of Africa can be characte rized by its hydrologic vulnerability that is, high vari ability of rainfall and river flow and landscape vulnerability to erosion and desertification. Worsening conditions of extreme vari ability are creating disincentives for investment in resource mana gement and encourage risk adverse behavior leading to low levels of productivity and bei ng trapped in poverty. Historically, people have adopted various livelihood strategies to cope wi th such vulnerability; however, the growth rates of poverty show the viability of these strategies is waning, leadi ng to increasingly unsustainable rural livelihoods, partic ularly those still largely agrarian ( ibid ). As suggested by the fractal qualities of poverty traps described below, the susceptibility of agricultural production to this natural vulnerability is pushing toward low-leve l equilibria points at the micro (subsistence farmer, farm level), meso (agribusinesses or co mmunity cooperatives) and macro scales (national economies). Consequentially, this means synergis tic solutions will have to be designed and implemented at all three scales. Ultimately, the puzzle for development practitioners is how to move or lift social systems (e.g ., households or communities) ont o growth paths that lead to breaking out of poverty traps, reflected by highe r levels of investment and productivity, and improved environmental quality via less resource-degrading behavior.
44 Barrett (2007) highlights the conceptual parallels between economists working with poverty traps and systems ecologists through terms such as multiple equilibria, thresholds or tipping points, and stable states ; going on, he argues that this overlap in language between poverty traps and resource dynamics opens a space fo r the development of integrated models and new understanding of the feedback between human-and-natural systems. For example, interpretations of the basic poverty traps framew ork to incorporate water yields the following enhanced conceptualization of the dynamic re lationship between wate r and poverty. Barrett & Swallow (2005; 2006) purport th at those operating above the dynamic asset threshold may stimulate a virtuous cycle of hi gher water storage, higher return s to water use, [and] lower environmental risk (Swallow et al. 2006, p.369) enabling higher investment levels in livelihood activities. Conversely, water poverty traps are characterized by low water storage, use, investment and productivity in a sel f-reinforcing vicious cycle. Swallow et al. s (2006) argument, which suggests that effective collect ive action in water management, can assist households and individuals to meet asset thre sholds constraining indi vidual production through such technologies as household wa ter taps or rainfall storage equi pment, indeed, is reflected in the hypothesis of the SCALES project. The existence of multiple equilibria points to the third factor about poverty traps: The structural conditions of a household and proxim ity to its dynamic asset poverty line determine the impact of risk and outcome of exogenous shocks. Drawing again on terms familiar to ecologists, the bifurcated growth model and th e possibility of multiple dynamic equilibria or steady states act as tools to understand the resilience and vulnerability of individuals or households when exposed to shocks and, in so doing, provides unique guidance to policymakers and development aid personnel who seek to mitigate risk and assist recove ry (Barrett & Carter,
45 2006). For conceptual clarity: Vulnerability is the likelihood that at a given time in the future, an individual will have a level of welfare below some norm or benchmark (Hoddinott & Quisumbing, 2003, p.9), with this benchmark often se lected as food insecurity. Resilience is the ability of a household (poor or other) to withstand and recover from hardships (for example, drought or illness), and to return to its pre-existing asset and welf are levels within a threeto five-year period. (Little et al ., 2006, p.201) Risk is one of th e hallmark characteristics of livelihood vulnerability that indu ces precautionary household behavior in terms of altering their livelihood strategies from those with high to those with low returns, commonly by opting to diversify in order to spread ri sk across activities and assets. Results from Dercon & Krishnan (1996, p.864) show diversification into off-farm ac tivities with low entry constraints, such as informal or casual labor, may offe r insurance against risk, but it is unlikely to cause substantial outcome differences compared to households not involved in these activities. Indeed, the SCALES project in Colombia obs erved this exact trend (Johnson et al ., 2006; see also Little et al ., 2006). Other empirical studies produce similar insights about vulnera bility and resilience based on the relationship between natural resource shoc ks and individual or ho usehold behavior in choosing livelihood strategies. The bifurcation represented in the poverty traps model implies that temporary shocks can have permanent a dverse effects on those dropped below the asset threshold, while those who are not decapitalized and do not transition between regimes would be expected to recover fu lly and quickly (Carter et al ., 2007; Barrett, 2005). Carter et al. (2007) comparing the environmental shocks, hurricanes and droughts2, to determine the long-term 2 These two types of events vary as follo ws. Hurricanes represent the direct destruc tion of assets while droughts are experience d more as income losses due to repeated crop failures, which for ce households to choose between sel ling assets to maintain curren t consumption and health or to preserve assets instead accepting lower levels of immediate consumption (Carter et al ., 2007). The latter can have significant long-term impacts on human capital through the malnutrition, underdevelopment or under-education of
46 effects on households and their livelihoods by evaluating the speed of recovery to rebuild up assets and the asset protection stra tegies adopted to self-insure agai nst risk discover that recovery time in terms of rebuilding of assets post-hurricane depends on in itial household wealth with poorer households experiencing the loss of assets longer and more acutely. During droughts in Ethiopia, the authors find that poorer households cope by adop ting patterns of asset smoothing and begin to accumulate assets faster than non-po or households post-drought (differentiated asset recovery experiences), however, never reaching a level of asset ownership that transcends the threshold to self-reinforcing growth (see also Little et al ., 2006). A fourth feature of poverty traps is that th ey depend on the existence of exclusionary mechanisms, which prevent or prohibit certai n individuals or groups from partaking in opportunities leading to sustained growth and a highlevel equilibrium state. Examples of factors that can lead to social and economic exclusion include gender, ethn icity, and governmental policies, which induce the various forms of res ource scarcity and other market imperfections. One of the most common forms of exclusion found pertains to economic exclusion manifest in lack of access to key financial services credit and insurance necessary to build and protect assets, to invest in improved technologies a nd to access higher-return markets (Barrett, 2005, p.7). Financial market failures are repeatedly id entified as the main structural reason driving chronic or persistent poverty and are essential to the emerge nce of a poverty trap (Carter & Barrett, 2006; Barrett & Swallow, 2006). The argument goes: if i ndividuals or households were free to borrow or insure themselves against ris k, they would do so moving above the critical asset threshold, adopt more remunerative strategi es and tend toward the high-level equilibrium. children, frequently referred to as the process of intergenerat ional transmission of poverty, a temporal scale issue of poverty (Bird, 2007).
47 Significant for the SCALES project and its hypo thesis that fosteri ng social capital and collective action for improved water management can be poverty alleviatio n, specifically, can lift households above the dynamic asset poverty th reshold, are the followi ng studies empirical findings. In light of the above di scussion, this hypothesis can be reposed as the question: Can increasing the stock of social capital change accumulation dynamics and shift individuals, households or other aggregate uni ts above the dynamic asset povert y line leading to higher-level equilibrium? Would increasing the stock of soci al capital affect social exclusion or other exclusionary mechanisms, improve the efficiency of meso-level institutions as mediators between the micro and macro scal es, and change investment in centives? How might investments in social capital address the tw o challenges posed by the asset-ba sed approach of poverty traps: building up stock of productive assets and protecting and increasing the productivity of those already possessed? Adato et al. (2006) investigated the ability of social capital to substitute or compensate for incomplete markets in the contex t of highly unequal societies. Their hypothesis was that social exclusion and ineffectual soci al capital explains South Africas patterns of economic mobility and poverty dynamics. Their conclusions are that social cap ital is more effective in procuring livelihoods for non-poor households, while in cases of the poor, social capital at best helps stabilize livelihoods at low levels a nd does little to promote upward mobility (Adato et al ., 2006, p.226). These findings lead the authors to proclaim that access to productive assets and the ability to accumulate thos e assets over time are much more critical to the escape from poverty traps than enhanced so ciability. Similarly, Cleaver (2005, p. 893) finds during research in a Tanzanian watershed that s ocial relationships, collective action, and local institutions may structurally re produce the exclusion of the poores t emphasizing that ways must be found to overcome their marginalizing effects. Barrett et al. (2006; see also Barrett &
48 Swallow, 2006) also report findings that show social capital only facilitates recovery to a lowlevel equilibrium point after a shock, not as a mean s to lift households to a high equilibrium level of wellbeing. Mirroring the Sustainable Liveli hoods framework, the fifth featur e of poverty traps is the now common recognition of the mu ltidimensionality of poverty, bot h in its causes and effects. This implies that targeted and coordinated mu lti-sectoral interventions will be necessary to contend with the various sources and drivers of poverty. Linked to this multidimensionality is the possibility of feedback across scales, the sixth feature of poverty traps. Barrett & Swallow (2006) interpret such feedback as a fractal process and argue that it can produce f ractal poverty traps, defined as one in which multiple dynamic equilibria exist simultaneously at multiple (micro, meso and/or macro) scales of analysis [or] aggr egation, with each scale re inforcing the pattern at other scales [meaning that] the low-level equilibrium prevaili ng at one scale makes a higherlevel equilibrium inaccessible at the ne xt lower scale, and vice versa. (ibid, p.3-6) Barrett & Swallow (2006) use the fractality of poverty trap s to explain low farm-level productivity as follows: They argue that farmers unwillingness to adopt or invest in improved technologies (or resource management practices) originates from coordination problems a nd social cleavages at the communityor meso-scale, while this higher-s cale failure is attribut able to the lack of individual incentive to adopt improved practices. Essentially, constraints that have spillover effects, which create a self-re inforcing feedback loop. As such Barrett & Swallow conclude, The challenge of reducing chronic poverty associated with povert y traps revolves around finding ways to remove or transcend the thre sholds and financial constraints that limit accumulation and access to remunerative strategies. (ibid p.6) But again affirming the need to look beyond the householdor farm-scale as productivity is not completely determined
49 endogenously, the key message from this fracta lity hypothesis for the design of watershed management which seeks to be poverty-alleviating is that interventions need to be simultaneously applied at all scales where lowproductivity strategies reinforce one another. ( ibid p.6) Synthesis In this chapter, I have disc ussed the objective of the CPWF to increase water productivity (WP) as a means to obtain improved environmental and social outcomes. During Phase 1 of the CPWF, studies found the concept lack ing in its ability to achieve the goal of poverty alleviation, suffering with the same handicap as IWRM generally to incorporate issues of social equity. One approach to remedy this is to view livelihoods as dependent on coupl ed human-and-natural systems to capture all three dimensions of water scarcity: physical, economic and social. Adopting this integrated systems perspective also reduces the risk of fostering a povertyenvironment tradeoff. The SCALES project de velops Theme 2s con ceptual model as an analytical tool to link unders tanding of the biophysical and social dynamics across watershed scales to assess the potentia l for collective action and introduces the fractal poverty traps framework to investigate the dynamic relationshi p between water, poverty and livelihoods. The project argues this approach to knowledge-building and expl anation of the poverty traps prevailing at multiple scales across a watershed can be used as the basis for crafting efficient and effective collective action for improved wate r management and poverty-alleviation. This research proceeds to apply the SCALES fram ework to understand the relationship between water, poverty and livelihoods within the broade r economic and environmental context of the Nyando River basin expressed in th e dynamic language of poverty traps.
50 Figure 2-1. The SCALES Framework (Source: Swallow et al ., 2006).
51 Figure 2-2. Graphic model of the poverty trap s framework. Livelihood function represents the particular livelihood strategies pursued a nd the associated utility. Vertical, doublebarbed arrows indicate possible stochastic transitions either temporarily moving above or below poverty line without any f undamental change in structural (assetownership) conditions. Horizont al, single-barbed arrows indi cate change in structural conditions. Solid arrows also indicate stru ctural changes that produce an immediate effect in terms of poverty (utility) without the intervening, transitory stage. The theory behind this model is to move households along the livelihood function and engage them in more remunerative strate gies. Linking the pove rty traps framework and SOP poverty assessment methodology, quadrants have b een labeled with letters to correspond with stages of progress pove rty dynamics: A = structurally poor; B = stochastically non-poor; C = stochastically poor; D = structurally non-poor. (Adapted from Carter & Barrett, 2006) Assets Utility Asset poverty line Income poverty line Livelihood function D C B A
52 Table 2-1. Key features of the poverty traps conceptual model. Poverty trap features Description Poverty is dynamic not static Structural versus stochastic poverty. Correlates of structural povert y produce changes in underlying conditions. Stochastic (temporary) poverty has its symptoms, which do not result in underlying changes. Multiple equilibria points High and low level steady states of wellbeing. Bifurcation of growth pathways Dynamic asset poverty line : Tipping point or threshold differentiating self-reinforc ing asset accumulation or decumulation. Initial asset conditions determin e resilience and vulnerability, and direction of mobility. Behavioral response to risk or shocks linked to initial asset conditions and final equilibrium outcome. Exclusionary mechanisms Factors prohibiting access to growth opportunities and highlevel equilibrium state. May be social or cultural; spatial; or economic and institutional. Multidimensionality of poverty Monetary and non-monetary aspects of poverty. Fractality Feedback across scales. Spillover effects. Coordination problems.
53 CHAPTER 3 STEPS ONE, TWO, THREE METHODOLOGY! The primary task of this research project is to develop an integrated understanding of water, poverty and livelihoods across the Nyan do River basin, Kenya, based on a unified analysis of the human and natural systems. Th e basic questions are: Who are the poor? Why are they poor? What livelihood strategi es or behaviors do the poor a dopt to contend with poverty as they understand and experience it? What are the dynamics of poverty? Answers to these questions describe the context as an important first step in how to make collective action inclusive of the poor and marginalized in the Nyando River basin. What type of policy or program interventions may be implicated by this knowledge of the initia l conditions associated with local poverty traps and substantiated means of escape and pathways of descent? So, how is it that I will get from asking these questions to providing some answers and drawing some conclusions? What data are relevant, what data am I to collect, and how will I analyze and interpret them? Step One: Picking a Research Design Chapter 2 established the argument that there exist multiple equilibria in livelihood outcomes, whether reflected in terms of WP or poverty. Su ch dynamics and the possibility of bifurcated outcomes are not captured in standard definitio ns and objective measures of poverty based on normative criteria (Peralta et al ., 2006). Instead, these measures paint an aggregate view of poverty that often belies worseni ng conditions of inequality or economic divergence and thus gives an inaccurate appraisal of policies or programs (Krishna et al ., 2006). Further, the poverty traps framework identifies the bifurcation in ho usehold or individual be havior in response to asset levels and vulnerability and institutional contexts, which, when combined, determine whether highor low-level equilibrium is reach ed. From the Sustainabl e Livelihoods framework,
54 it is well known that adopted livelihood strategies are not made solely based on factors endogenous to the householdor farm-level. Household behavior is also partially conditioned by contextual factors beyond the asset base, for ex ample, wider market im perfections, especially those which impose liquidity constraints; weak and exclusive institutions, legal structures and power relations; and exogenous risk factors (Weingart & Kirk, 2008). Basins as places where people live and a ttempt to procure a satisfactory level of wellbeing in linked social and environmental system s thus constitute the object of analysis that the CPWF argues must be understood and mana ged holistically to ach ieve its articulated, frequently incongruent goals. While a basin or wa tershed is admittedly a large unit to study with many potential variables involved, understanding of the multi-scale processes that condition behavior and, ultimately, determine livelihood outcomes can only be achieved by taking into account this wider context. It allows for a more complex and fuller explanation of the phenomenon of poverty within a basin by expand ing the scope to include the effects of organizations and institutions operating at various scales. From de Vaus (2001, p.234), by narrowing focus solely upon the household, farm or pl ot we risk wrenching traits (variables) out of the context in which they occur, we strip them of much of their meaning and consequently risk misreading their meaning and significance and th us misunderstanding their causes, which poses a serious threat to internal va lidity. By examining the basin as a unity, we attempt to draw a fuller and more rounded picture of the causa l processes surrounding a particular phenomenon (ibid, p.235), in this case the relationship betw een water, poverty and livelihoods. Accordingly, the following statement describing when to use a case study research design is persuasive: The study of context is important because behavior takes place within a context and its meaning stems largely from that context. (de Vaus, 2001, p.235)
55 A comparative case study design was selected by the principal investigators of the SCALES project. The primary objects of study, or units of analysis, are the Nyando River basin, Kenya, and two watersheds, Lake Fuquene and Coello River, in Colombia, South America. In the past, the case study method has been criticized for always being too at heoretical and unable to prosper theoretical generaliza tions. While this has been argued as being generally untrue (de Vaus, 2001; Yin, 2003), the SCALES proj ect at large and my research in particular have tried to establishe a firm theoretical and conceptual basi s to organize their invest igations. The endeavor here is to respond to Barretts (2005) observation that there remain s a need for further research into the emergence of poverty tr aps and best practices for escape to be used in the design of policy and development interventions. Accordin gly, this case study has ta ken a theory-building approach, selecting cases to help develop and re fine the propositions and develop a theory that fits the cases we study. (de Vaus, 2001, p. 223) Because this is the projects approach, I have only begun with some basic questions and one proposition rather than more precise hypotheses. Results from the case studies are then available to be used to build up the specifics of the theory and more detailed propositions. In addition to this basic multiple-case desi gn, SCALES pursued multiple units of analysis leading to a final Type 4 rese arch design, a multiple-case (embedded) design (Yin, 2003). Within the Nyando River basin a total of 20 villages (Figure 3-1) were se lected as sites to carry out a Stages of Poverty (SOP) particip atory poverty assessment, however, for most of the data analysis the SCALES project focused on the 15 sample villages in the Nyando and Kericho districts (Figure 3-2). This data collection method will be described subsequently. Embedded within this subunit, households were selected to conduct analys is at a further disaggregated subunit level to develop a deeper, subjective unde rstanding of household behavior and decision-making; a level
56 of analysis not replicated in the Colombian basins, but one that, it is argued, does not compromises the comparability of the three cases. Care is taken to be sure that the original phenomena of interest the relationship between water, livelihoods and poverty across basins remains the target of study instead of merely the context for analysis of villageor householdlevel poverty trends (Yin, 2003). The case st udy design adopted and revised for SCALES (Kenya) is displayed in Figure 3-3. Noteworthy are the parallels with the SCALES conceptual framework (Figure 2-1) with each larger, scaledup unit of analysis being characterized essentially corresponding to a higher node: primary:household; secondary:village; and tertiary:basin. In all, th ere are six elements that define the ap proach of this case study: It is a multiple-case (embedded) design used for theory-bu ilding. Further, it is meant to be descriptive rather than explanatory, and the cases have been studied in para llel and retrospectively (Yin, 2003). Step Two: Data Collection and Sampling The majority of data collected for the SCALES project was done using a Stages of Progress (SOP) poverty assessment (Krishna, 2004; Krishna et al ., 2005). Designed to counteract the failures of objective, static measures, this community-based, participatory poverty assessment derives a subjective and mostly qualitative loca l perception of poverty apposite to the poverty traps framework. This methodology was developed to capture the dynamic nature of poverty and seeks to differentiate the factors associated with escape from and descent into poverty. The methodology was inspired by empirical evidence showing economic growth is not homogenous across villages or households (Krishna, 2006). While economic growth is associated with households which manage to escape poverty, it ca nnot prevent the decline into poverty, therefore displaying the same type of bifurcated outco mes presented by the poverty traps framework. Questionnaires administered as part of the SOP methodology attempt to identify the household-
57 or local-level conditions associat ed with this difference in expe rience of economic growth, that is, why there is an economic divergence or increase in inequa lity versus a convergence as a consequence of growth. As part of the analysis these local-level conditio ns are interpreted in relationship to wider contextual factors based on data gathered from secondary sources. The existence of two different sets of factors e xplaining poverty escape or descent implicates the possible need for two sets of objectives for collective action: cargo net policies to facilitate escape from poverty (livelihood prom otion) and safety net policies to prevent the decline into poverty (livelihood protection) (Krishna et al ., 2006; Devereux, 2002). The SOP method provides data on a households progress vis--vis poverty in three times periods currently, 10 years ago, and 25 years ago to capture the dynamics and any intergenerational transmission of poverty. The reference years fo r the study are approximately 1977 to 2002 meaning readers are required to interpret these as fo llows: currently equals 2002; years ago equals 1992; and years ago equals 1977. Status in each time period is based on a locally-determined poverty line constructed by asking local representatives the sequence of expenditures, or stages of progress, a house hold would make as it moves away from acute poverty toward prosperity (Krishna, 2004). Agre ement is reached on the threshold or tipping point stage that equates to the local poverty line. Households are then id entified by their stage of progress and classified into one of four categories rela tive to their position to the poverty line: A: Poor then and poor now; chronically and structurally poor B: Poor then and not poor now; escaped poverty, transitory/stochastic C: Not poor then but poor now; declined into poverty, transitory/stochastic D: Not poor then and not poor now ; persistently and structurally non-poor Information is also gathered on what livelihood ac tivities a household is e ngaged in to link their stage of progress (alt. level of wellbeing) to what they do to ach ieve it. Krishna (2006; see also Krishna et al. 2004) argues that the livelihood strategies households adopt are di rectly related to
58 how they understand and perceive poverty. Ther efore, establishing a local understanding of poverty helps to identify successful versus unsuccessf ul livelihood strategies in terms of attained wellbeing outcomes. The attained stage of progre ss then is the basis of the poverty-livelihoods link, and policies or development interventions are purported to be most successful when they recognize this link and support more participation in proven re munerative activities versus scuttling them (Krishna, 2006). Th e final type of data collected as part of the SOP methodology is a list of reasons cited during household interviews to explain eith er stability or change in the stages of progress of a sample, usually 40%, of village households. These data are interpreted as explaining why the subject household has or has not progressed out of or has even declined into poverty. One uniqueness of the case study design is that it does not pursue statistical generalization hence is not subject to the same probability sampling logic of other research designs. In a case study design, cas es are specifically chosen based on their known outcome to evaluate a proposition, to disprove it or establ ish the particular conditions under which it holds (de Vaus, 2001). However, this does not preclude the case study design fro m satisfying the tests of generalizability and validity through selecting cases based on one of two types of replication logic: literal (conditions under which phenomenon can be found; case matches what theory predicts) or theoretical (conditions under which phenomena of intere st are not likely to be found) (Yin, 2003). The size of the sample, consequentially, is also of less concer n with the appropriate number varying in step with wh ether literal or theoretical replication is being pursued and the number of external variables that may be im plicated in the phenom enon under investigation. Accordingly, each of the units of analysis unde r investigation has been chosen strategically, known as purposive sampling, driven by replica tion logic as follows: Choosing the Nyando
59 River basin as part of the SCA LES project provides researchers with the opportunity to compare and contrast the ability of the SCALES conceptu al framework to evaluate the poverty-alleviating potential for collective action via water manage ment in a South American and an African context. The two locations differ in terms of the content and influence of reverse flows with the water towers of east Africa expecting to pr oduce flows of power traveling in the same downstream direction as water. This rival situation to Colombia se rves as an opportunity to test for theoretical replic ation, although the actual comparison of results between Colombia and Kenya does not fall under the sc ope of this thesis. SOP data on poverty and livelihoods for 15 of the 20 villages selected in the Nyando River basin to be part of the SCALES project we re actually collected as part of an earlier study, Safeguard, conducted by the World Agroforestry Ce ntre (ICRAF). It was a cost-savings decision on behalf of SCALES to incorporate these same villages into the new study meaning the SOP poverty assessment only had to be executed in five additional vill ages. The justifications for selecting the particular village s is unknown to me, but given th eir distribution across the basin (Figure 3.2) and their differences based on certa in variables I see no grounds for finding them unacceptable choices. Krishna et al. (2004; 2005) argue that sample villages are to be selected based on their differences in geographical, econo mic and other exogenous or contextual factors (for example, variables such as remoteness; de gree of commercialization; population size; ethnic composition; agroecological c onditions; watershed location, el evation and sub-catchment) to focus attention on the potential similarity of lo calor household-level reasons for poverty escape or descent. Table 3-1 presents the descriptive data on the vill age subunits including some of these variables of difference. Se lecting 15 villages that do disp lay such differences or rival
60 conditions, it is submitted, is a necessary number of cases in order to achie ve literal replication within these subgroups as well as theoreti cal replication across these subgroups. The SOP data were used in selecting the household subunit cases for more detailed examination. Selection was confined to two of the 15 villages only Chepkemel (village no. 4) and Miolo (village no. 12) due to logistical reasons and the necessi ty to piggyback a concurrent SCALES experiment. As these two villages differ on many factors as seen in Table 3-1, the same confidence is warranted with respec t to literal replication as for the village level of analysis. The purpose of these household-level analyses is to identify with more precision why there are differences in livelihood outcomes following th e asset-based argument of the poverty traps framework. As such, this effort is a response to Krishna et al .s (2005) call for further refinements to the SOP methodology by collecting more information from households on asset ownership, income sources and expenditures, an d intra-household differences regarding gender. In keeping with the case study approach, househol ds were chosen that differ on their livelihood outcome represented by their SOP category (depende nt variable), seeking to compare households that are structurally non-poor (c ategory D) and households that are transitioning toward poverty (category C) to determine what accounts for such stability or change. Female-headed households also are to be amongst the sample to probe th e impacts of this diffe rence. Approvals were granted by the village elders to interview the hous eholds listed in Table 3-2. While eight out of ten households interviewed do not show change in terms of their attained SOP category (columns 3 and 6), the fluctuations in their actual stages of progress (columns 2 and 5) do not inclusively suggest stability. Some of the cases may be exhibiting some normal transitory or stochastic type change associated with bad luck or moderate life cycle eff ects (Baulch & McCulloch, 1998), yet others may be showing signs of the beginnings of structural cha nge leading to a descent into
61 poverty. The relative lack of households that are or have escaped poverty that is, category B households made this a difficult group to sample plus, as the results presented in the next chapter show, the volume of households descendi ng into poverty appears to be a much more significant phenomenon to address. Initially, a new survey instrument was prepared to collect additional data on the household subunits. This survey was designed to measure the levels of the five capital assets accessible to the household; the resulting farm pr oductivity; on-farm, offand non-farm income; and consumption of food and other consumption e xpenses. The initial intention was to use these data to construct a similar socioeconomi c linear programming m odel to Magombeyi & Taigbenus (2008) reported in Chapter 2. While in th e field the decision was made to revise this intended household data analysis method aw ay from modeling in favor of a deeper, anthropological-type (ethnographic) analysis to better capture th e wider, contex tually-driven causes of behavior regarding resource use and livelihood activities pursued. This change was justified on the following grounds: First, while not exhaustive, the SOP data provided a reasonable level of understanding about househ old microeconomic conditions. Second, this avoided the mistake of neglecti ng wider issues of context obsc ured by an exclusive focus on microeconomic conditions, a common critique of the livelihoods approach (Scoones, 2009). Validating this second point, initial discussions with the selected cases revealed them to be highly integrated and reactionary to off-farm conditions implyi ng that they are much more determining of livelihood outcomes and wellbeing than on-farm conditions. Accordingly, the data collection method changed from a survey inst rument to a participatory rural appraisal or PRA process (Chambers, 1997) to develop a more thorough understandin g of the subjective meaning of observed behavior.
62 Step Three: Data Analysis and Interpretation Because of the essential need for consistency in a comparative case study, this study follows for the most part the data analysis procedures used by the project team working on SCALES in Colombia (Johnson et al ., 2006, 2009; Peralta et al ., 2006) as well as the or iginator of the SOP method, Krishna (2004; Krishna et al ., 2004). This was done to ensu re reliability and validity, and to avoid circumstantial evidence that could threaten theoretical or literal replication. To synthesize results and interpret findings from SCALES (Kenya), I empirically construct from in-depth household interviews archetypical schematics of livelihood systems in the Nyando River basin reflecting the different Stages of Progress dynamic poverty categories (de Vaus, 2001). These diagrams show the differe nces between household categories in terms of wealth; market integration or exclusion; livelihood diversification; dependency on the natural system; and vulnerability and re silience. This descriptive under standing is useful for assessing the potential and opportunities for collective action around water management to be povertyalleviating.
63 Figure 3-1. Map of sample villages across the Nyando River basin where the SOP poverty assessment was applied. The SCALES proj ect subsequently narrowed focus to 15 villages. (Source: SCALES project) Figure 3-2. Map of political districts and sampled village locations for SCALES project. (Source: SCALES project)
64 Figure 3-3. Multiple-case (embedded) re search design for SCALES (Kenya). N y ando River basin unit ( macro-scale ) Village subunits across basin (meso-scale) Household subunits, Chepkemel (local-scale) Household subunits, Miolo (local-scale) Proposition: Improved water management via collective action across scales can be poverty-alleviating.
65 Table 3-1. SCALES project sample v illages in Nyando River basin, Kenya. a Village number corresponds to location on map. b Lower (1100-1250 masl), middle (1250-1750 masl) a nd upper (1750-2500 masl) correspond to elev ation in meters above sea level (masl), while Awach and Kapchorean are two of three sub-catchments within the basin. Village numbera Village name Location & subcatchmentb Political district Population density (persons/sq.km.) Land tenure Ethnic composition 1 Kaminjeiwa/Kedowa Upper Kapchorean Kericho 13-87 Sub-divided leasehold Kalenjin 2 Nyaribari Upper Kapchorean Kericho 150-223 Sub-divided leasehold Mixed 3 Kiptegan/Ketitui Upper Awach Nyando 528-751 Adjudicated Kalenjin 4 Chepkemel Upper Awach Kericho 528-751 Adjudicated Kalenjin 9 Nakuru/Ahero Irrigation Lower Kapchorean Nyando 752-1322 Adjudicated Luo 11 Karabok/Gem Nam Lower Awach Nyando 400-557 Adjudicated Luo 12 Miolo/Agoro East Middle Awach Nyando 304-399 Adjudicated Luo 13 Kasiwindhi/Jimo Middle Lower Awach Nyando 150-222 Adjudicated Luo 14 Achego/Awach Scheme Lower Awach Nyando 400-557 Adjudicated Luo 15 Ochoria Middle Kapchorean Nyando 13-87 Settlement scheme Mixed 16 Kakmie/Chesinende Upper Kapchorean Kericho 304-399 Settlement scheme Mixed 17 Soliat Middle Awach Kericho 400-527 Adjudicated Kalenjin 18 Kakmie Lower Kapchorean Nyando 224-303 Adjudicated Luo 19 Rawana/Border 2 Middle Kapchorean Nyando 224-303 Settlement scheme Mixed 20 Owaga Middle Kapchorean Nyando 224-303 Settlement scheme Mixed Table 3-2. Sampled household cases in two village subunits of analysis. a Female-headed households. b Local poverty line at stages c Numbers refer to in order: male adults, female adults, male adolescent children, female adol escent children, children under 10 years. Village 4 (Chepkemel) Village 12 (Miolo) Household number Stage of progressb (1977 1992 2002) SOP category (1977 1992 2002) Household compositionc Household number Stage of progress (1977 1992 2002) SOP category (1977 1992 2002) Household composition 26 11-10-8 D-D 1-1-0-0-3 4 13-9-11 D-D 1-1-1-2-4 31a 12-8-12 D-D 0-1-4-0-0 5 13-8-11 D-D 1-1-1-14 32a 13-13-10 D-D 0-1-1-5-1 8 13-11-7 D-D 1-1-6-0-0 51 12-8-9 D-D 1-2-0-0-2 10a7-12-6 D-C 0-1-1-2-0 52 12-8-2 D-C 1-1-1-0-0 47 8-10-7 D-D 1-1-1-1-1
66 CHAPTER 4 POVERTY DYNAMICS AND DRIVERS OF CHANGE IN LIVELIHOODS AND WELLBEING ACROSS THE NYAN DO RIVER BASI N (1977-2002) As is to be anticipated from a multi-scale, integrated analysis, there are several layers of results to report and discuss which are divided into two parts. Chap ter 4 considers the larger units of analysis, specifically the story of the Nya ndo River basin as built up and reflected by the sampled case study villages. The goal of this chapte r is to weave together the wider institutional and geographical contextual factor s, including water, into unders tanding of the observed changes in poverty and livelihoods in the basin followi ng the SCALES framework. The cornerstone of this analysis is poverty dynamics as an indi cator of welfare stabili ty over time (Muyanga et al ., 2007). The value of focusing on inter-temporal cha nges in poverty of speci fic households is to identify the correlates of poverty, that is its causes rather than symptoms ( ibid ; see also Baulch & McCulloch, 1998). In particular, what are the di fferent determinants of stochastic movements or transitions in or out of poverty versus those structural factors which render households trapped in poverty? This topic is further addressed in Chapter 5, following the poverty traps framework, where I seek to identify the hous ehold-level microeconomic factor s associated with poverty and wellbeing, contextualized and interpreted in view of this wider, integr ated discussion of the historic conditions of the Nyando River basin. The presentation of results in this chapter ar e organized as follows: First, for purposes of comparability, are the results on village-level poverty dynamics from across the Nyando River basin according to the standard procedures established by Kr ishna (2004) and replicated by SCALES in Colombia (Johnson et al. 2006). Second, processing more of the raw SOP data, results on the livelihood stra tegies associated with stability or change in attaining stages of progress in relation to the local poverty line are identified. This an alysis zeros in on two of the case villages Chepkemel (village no. 4) and Miol o (village no. 12) as representatives of the
67 divergent experiences of povert y and wellbeing between the uppe r and lower regions of the watershed and, thereby, satisfies the maximum difference criterion of case study methodology as discussed in Chapter 3. This sets up for drawing conclusions about what constitute successful versus unsuccessful or more ve rsus less remunerative strategies and related needs for action. Throughout these findings are inte rpreted in reference to the institutional and geographical context of the Nyando River basin over the subject 25 year period. At the end of this chapter, the reader should have a grasp of the historical a nd spatial dimensions of poverty across the Nyando River basin and the differences in livelihood strategies that are consequential to them. Based on the analysis, some conclusions will be offered as to the role of water in poverty, livelihoods and wellbeing, and the potential opportu nities and challenges for collective action and water management generally in the Nyando River basin. Trends in Poverty Dynamics across the Nyando River Basin Using SOP data economic transition matrices that present the trends in poverty dynamics over time are constructed. The tables below should be interpreted as follows: The percentage of households that have remained poor in both ti me periods are defined as chronically poor (category A); households that have shown upward mobility or progress, changing from being poor in the past to not poor now, have escaped poverty (category B); house holds that have shown downward mobility, changing from not being poor in the past to poor now, have descended into or entered poverty (Category C); and, finally, those non-poor households th at were not poor in either time period (category D). Table 4-1 presents these trends for the peri od between 25 years ago to the present (19772002) for 1247 rural households in all 20 sample villages from across the Nyando River basin. Overall, a total of 16.9% (A + B) of households were poor 25 years ago, while 28.5% (A + C) are reported as poor at the pres ent time indicating a rise of 1 1.6% in the total number of poor
68 households during this period. Tabl e 4-2 presents these trends fo r the more recent period of between 10 years ago to the present (1992 -2002) for a sample si ze of 1372 rural households. In the shorter time period, 15.1% of households were poor initially, in 1992, while 27.2% of households were poor by 2002. This equates to a 12.1% rise in poverty during the decade. Viewed together these partic ipatory poverty assessment fi gures convey the respondents perception of relative stability in aggregate poverty levels between 25 and 10 years ago (1977 to 1992) with most of the change in poverty and declines in wellbeing experienced during the decade from 1992 to 2002. Further to the concern to identify worsening conditions of economic inequality or divergence, these poverty dynamics data also show that, while the percentages of structurally chronic poor (categor y A) and non-poor (category D) are relatively similar over both time periods, between 1992 and 2002 there is a growing gap betwee n households that manage to escape poverty (category B) versus those that des cend into it (category C). The observation that almost four times as many households fell into poverty during the 1990s suggests that actions, programs and safety net policies which mitigate the relevant push factors to prevent or arrest the descent into poverty stop what may be stocha stic transitions into poverty of category C households from becoming permanent, structur ally poor category A households may be of greater importance than fostering pull factors to faci litate the ascent out or alleviation of poverty. Tables 4-3 and 4-4 are economic transition matrices disaggregating the Nyando River basin data and clustering into smaller subunits the sampled villages by their district location, Nyando or Kericho. From Figure 3-2 both political districts can be seen to fall almost entirely within the watershed. These districts also fall within different larger political provincial boundaries, with Nyando district in the Nyanza provi nce and Kericho district in the Rift Valley province. Significantly, the districts can be used as representatives to analytically divide the
69 Nyando River basin into two regions according to their differences in maize agroecological and elevation zones: Nyando district represents the lower basin area described as a lowland tropical (LT) agroecological zone and Keri cho district represents the u pper basin described as a moist transitional (MT) agroecological zone (De Groote et al. 2006). The average annual rainfall in the LT zone ranges between 800 1200 millimeters per year (mm/yr), while the average range in the MT zone is 1200 1600 mm/yr (WRI et al. 2007). These geographical factors will be discussed in-depth below for their significance in determining the livelihood strategies adopted and poverty outcomes realized. For now what should be noted is the relative affluence of the upper basin versus the lower basin indicating cond itions of economic inequality across spatial and political scales. This pattern of inequality also prevails within the political districts themselves, the gap between non-poor and chro nically poor househol ds indicating more inequality within the upper basin and Kericho di strict than in the lower basin. Both of these findings mirror results found by WRI et al. (2007) in their report on spatial patterns of poverty and human wellbeing in Kenya. Significant for the SCALES project, inequali ty is a known threat to a communitys social fabric, including qualities of social ca pital and social cohesion (WRI et al. 2007), hence poses a risk to the possible emerge nce and durability of co llective action. In the terminology of the SCALES framework, such upstream-downstream inequalities could undermine the reverse flows from Nyando to Kericho district residents and limit the opportunities for collective action across spatial scales. Adding in consideration of movements in and out of poverty, both district s exhibit the same low rates of escape from poverty, while the lowland Nyando district shows a comparatively la rger percentage of hous eholds entering poverty meaning, should these trends pers ist, inequality between the tw o regions of the basin can be expected to continue to widen.
70 The variability in poverty rates between di stricts suggests the possibility that poverty might be heterogeneous within th e districts themselves. To inves tigate, Table 4-5 presents the dynamic results for all 20 sampled villages ordere d by the degree of change in poverty from the largest percentage drop in poor households to the largest increas e in the percentage of poor households. This rank ordering is based on the change observed for the more impoverishing period, 1992 -2002, although data for other time periods are also included in the table. Villages are labeled only by their assigned number rath er than by name, which corresponds to their location on Figure 3-1, for simplicity in referenc ing. Only two villages, nos. 3 and 20, managed to decrease the percentage of poor households in all three time periods. In the time period 1977 1992, seven villages managed to exhibit a net decrease in the number of poor households; however, in three out of five cases these gains were more than eliminated during 1992 2002. Plotting the percentage of poor households at each of the three time points (Figure 4-1 Nyando district villages; Figure 4-2 Kericho district villages) visu ally reinforces the mean and aggregated basin results which show, for the mo st part, poverty levels were remaining fairly steady or declining between 1977 and 1992, but that they appear generally to have jumped dramatically from 1992 to 2002. Before leaving such trends analysis, an economic transition matrix for the two case villages Chepkemel (no. 4) and Miolo (no. 12) selected for more in-depth household interviews is included. Table 4-6 shows the very different dynamics between the two villages in 1992 2002. The observed values on poverty in the lowland village of Miolo, ranked 20th and worst in the sample in terms of percentage change in poverty over the last ten years, vary substantially from both those of the basin as a w hole and the Nyando district in which the village lies. Miolo has 83% fewer non-poor households than the basin average, but most striking is the
71 differential between households which were able to escape poverty compared to those who descended into poverty, a ratio of 6:1 in th e direction of entering poverty. Conversely, Chepkemel (village no. 4) has fewer chronically poor and more non-poor hous eholds than either the basin as a whole or its home Kericho distri ct. However, the number of households able to escape poverty was less than either the basin or Kericho averages. Th is is most likely a reflection of the overall fewer number of households that were poor to begin with. The most significant trend, though, is the ratio between entry into and escape from poverty at 10:1. So, while the absolute numbers of households currently or chr onically below the poverty line in Chepkemel is relatively small, there appears to be a growi ng and not offset trend of households entering poverty in this village. In both village cases, these ratios of poverty entry versus escape reinforce the objectives of this research to identify the most remunera tive livelihood strategies, presumably resistant or resilient against the push factors driving househol ds into poverty, and factors which account for households being able to adopt th em at a sufficient level of productivity. Indeed, mobilizing and maximizing efficiency of these factors to attain such productivity levels are the drivers behind CPWF technology development generally. The firs t of these objectives is undertaken over the remaining balance of this chapter. Having esta blished a basic understand ing of the spatial and temporal distribution of poverty across the basin, the next section will at tempt to connect these dynamics to livelihood strategies also across s cales. Later, explanati ons are sought for the marked increase in poverty in the period 1992 2002 by analyzing significant changes in the institutional environment and their impacts on liv elihoods. And further to the larger goal of linking water, poverty and livelihoods, information and interpretation of the role water (as the
72 primary environmental or geographical factor of interest) plays in hous ehold behavior, decisionmaking and success regarding these li velihood activities is discussed. Summarizing key results presented thus fa r, the SOP poverty dynamics data converge with results from other poverty mapping studies (WRI et al. 2007) and Government of Kenya statistics (KNBS, 2007) in term s of distributional patterns. Po verty levels according to the subjective, participatory assessment were fa irly stable for the period between 1977 and 1992 with conditions becoming much worse betw een 1992 and 2002. Generally, the incidence and density of poverty are worse in the lower basin with its drier agroecological conditions than the upper basin, although there remains some heteroge neity across villages within the districts. Conditions of inequality are vi sible between the upper and lowe r regions and within the upper region itself threatening the like lihood of collective action based on a weakened social fabric. Livelihood Strategies across the Nyando River Basin This section presents findings on the associa tion between specific li velihood activities and poverty across temporal and spatia l scales in the Nyando River ba sin. In all, there are eight primary identifiable, significan t livelihood activities or groups of activities found within the basin: formal employment; informal employment or casual labor; livestock; maize production; operating a small business; horticulture production (e.g. fresh fruits and tomatoes); other subsistence crop production ( e.g ., sorghum, beans, kale and sweet potato); cash cropping or commercial agriculture ( e.g ., sugarcane, tea, coffee, and irrigated rice). Appendix B (Table B-1) presents data on the percentage of househol ds participating in these livelihood activit ies in years 1992 and 2002 by sample v illage. (Due to incompleteness of the available dataset not all the sample villages are included.) From this information some answers to questions elicited by th e poverty dynamics data can be inferred. For instance, village nos. 3 and 20 appear to continue reducing thei r poverty levels during the 1990s because of
73 expanding participation in cash cropping/comm ercial agriculture of tea and sugarcane respectively. These villages are located geographical ly very near key tradi ng centers, Kericho for tea and Muhoroni for sugarcane, undoubtedly help ing in their trajectory. Tea production began in village 3 in the 1980s with strong support from the Ministry of Agriculture, so by the 1990s the plants were mature and yields high. The high-value of tea was pivotal in maintaining the low poverty levels of the village wh ere worsening conditions of land scarcity in 1970 when title deeds were first issued, the larg est farm owned 30 acres ; at the time of inte rviews in 2003, the largest landholding was 3 acres af ter 30 years of customary subdivision have meant reductions in the number of other high-value crops ( e.g. tomatoes, bananas and pineapples) that can be grown. Respondents also discusse d land scarcity as a reason so me households were reducing livestock numbers and switching from cattle to chickens because of the lack of space for grazing. Were et al. (2006), in an analysis of collective action for water mana gement in the village, found a high degree of social organi zation with one of the primary impacts being the freeing up of womens time from the burden of water coll ection. Tea production offers a lot of casual employment opportunities that are available to women, which can enhance the wellbeing of households as evidence presented later will affirm. This does pr esume the ability of women to turn time savings into money income (Roy et al., 2005, p. 7), which cultural reasons may prohibit. On the whole, other onfarm activities such as livestoc k, maize, and other subsistence crops (the latter in village 20 only) also have high participation ra tes contributing to the livelihood security of these vi llages through diversification. Of the remaining five villages that reversed 1977-92 poverty reduction trends during 1992-2002 village nos. 5, 12, 13, 14, and 19 relevant sh ifts in participati on appear to be the exit from cash cropping, if it was practiced in the village initially, as we ll as significant exits
74 from livestock and maize production. In particular the findings about livestock reflect the wellestablished correlation between livestock and lack of poverty in Kenya (Burke et al. 2007). Onfarm maize production for household consumption is also important be yond food security; it reduces the exposure households ha ve to the potentially impoverish ing effects, in terms of cash and nutrition, of price fluctuations on the maize market (De Groote et al. 2006). Looking at the histories of these villages points to some reasons why there has been an increase in poverty. The people of upland village 5 are described as land less forest dwellers, who, in 1987, were granted some non-statutory rights by the government to fa rm border areas of the forest. Their insecure land tenure accounts for the observed lack of conservation farming practices leading to local land slides articulating an import ant connection between the legal system and the incentive for collective action. In 1991, widespre ad food insecurity in the v illage due to the restricted livelihood system precipitated a general divestit ure of livestock to enable maize purchases. Lowland villages 13 and 14 are areas that were settled, deforested and converted to rice farming, making use of the fertility provided by the Awach Rivers silt deposits during episodes of flooding. In 1980, in village 14 small-scale irrigated rice producti on was started by the Ministry of Agriculture and supported by the cereals marketing board. In 1991, both agencies stopped their support of local farmers who were th en on their own to procure seed and market their produce. By 1994, irrigation canals that we re dug in village 13 w ith the support of the European Union a decade earlier had become silted and were not re-dug due to local liquidity constraints. For both villages, droughts betw een 1993 and 1996 caused widespread death of cattle and the local rivers to become seasonal, destroying many rice farms. Flood events in 1997 and 1998 worsened the siltation problem causing the collapse of the rice schemes by 1999. The case of village 12, Miolo, is taken up in-depth below but it, too, has seen its farmers exit
75 remunerative crops such as groundnuts, greengram s and cotton due to a combination of change in climate and closure of agri cultural marketing boards. Formal employment rates showed expansion in 8 out of 17 case villages and contraction in 7 out of 17 case villages over the decade with the highest participation reco rded in village 6 at 35.2% of households. Regardless, 15 of the 17 villag es showed an increase in the percentage of households in poverty during the same peri od, including village 6, suggesting formal employment may not guarantee or be sufficient for avoiding poverty. Conve rsely, in 14 out of 17 villages, participation rates in low barrier-to-entry informal employment or casual labor increased in 1992-2002 with 12 of the 14 villages e xperiencing an increase in the percentage of households beneath the poverty line. These ini tial observations can be investigated more thoroughly and linked to poverty by narrowing focus to district representatives, case study villages, Chepkemel and Miolo, with the added support of ethnographic data collected in June 2007. Appendix Table B-2 presents livelihood activity participati on rates of these villages households disaggregated by their relevant SO P category, A through D, during the period 1992 2002. Water, poverty and livelihoods in Chepkemel Beginning with upland village, Chepkemel, households classified as non-poor are generally well-diversified, but si gnificantly, indicate intensive participation in livestock, maize and cash crop production. This intensity of land use is reflected in Figure 4-3 and a typical planting calendar (Table 4-7) documented in consultation with smallholder farmers through participatory rural appraisal (PRA) methodologies (Chambers, 1997). The planting calendar shows crop production in the village to be a ye ar-round activity possible because of a fairly reliable secondary, short-rains season. Intercropping and narro w crop-spacing practices also indicate the worsening conditions of land scarcity in the village. Land pressure was evident back
76 in 1990 when there was reported to have been a cl earing of more village land to allow for more settlement and cultivation. The seasonality of ha rvesting and marketing of crops means there are periods when households are both cash-poor and food-insecure (shaded boxes) leading to the adoption of short-term crops such as sweet pot atoes. Households descending toward poverty are exiting livestock and maize production in significant numbers all the while adopting seasonal casual labor. Land scarcity is also pushing households out of cattle because of the lack of land for grazing, a situation made worse by the fenci ng of property after la nd titles were issued. Although it is essentially endogenous casual labor is perhaps more accurately viewed as a response to rather than a cause of poverty and th e associated decapitalization of livestock and other resources because of its low barriers-to-e ntry. Formal employment is reported for small numbers of households in both categories hence is ambiguous in its effect on poverty in this village context. While there are too few cases repr esented in the sample to make generalizations possible, what data are available reinforce the id ea that households in penury do not participate in any livelihood activities except for informal employment. Obvious questions emerge asking why households are pushed out of maize and/or cash crop production and decapitalize their livestock. From the opposite perspe ctive, the question can be posed as: Why are households able to continue to successfully practic e these livelihood activities? Both perspectives are useful to try to pinpoint significan t factors. As was described for neighboring village 3 above, cash crop agriculture in Chepkemel is predominantly tea, which is a widely reporte d activity in non-poor households. Tea production was introduced to the village in 1980 with strong extension support from the Ministry of Agriculture. Due to its regular harvesting throughout the year, it plays an important role in seasonally smoothing out household income and c onsumption thereby reducing the chances of
77 having to divest livestock savings to meet immediate needs. Tea production is also attractive to villagers because of the close proximity of buying centers and the sure ty and regularity of payments. It provides a key source of income for women as tea pickers, who, in turn, spend their cash income on supporting household wellbeing through purchases of food, clothing, domestic supplies and payment of educationrelated expenses (see also Were et al ., 2006). This off-farm income for women, however, does come at the expense of subsistence crop productivity on their own farms, neglected in particular regarding weeding, another traditionally female activity. The Kenyan Tea Development Authority (K TDA), a now privatized parastatal organization, serves a critical function by facilitating input purc hases, technology use (e.g. fertilizer) and credit loans, a ll critical factors in helping to enhance crop productivity and overcoming liquidity constraints, offsetting the push down toward poverty in tea-growing households. Discussions with tea farmers in the v illage, however, revealed that some smallholder producers disregard extension recommendations provided by the KTDA for example, by planting tea on less-than-recommended crop spacing or by intercropping with subsistence crops, often maize, both behaviors jeopardizing maximu m tea yield based on their need to satisfy short-term food security requirements, their pers istent livelihood vulnerab ility thus threatening tea productivity as well. These behaviors were no t observed on more established tea farms whose yields are sufficiently high to financially-secure households and will likely be abandoned when the young tea plants are large enough to produce highe r yields. Similar discrepancies, also with productivity implications, are evident between other crop re commendations made by Kenyan Ministry of Agriculture and Rural Development (MOARD) extension agents and on-plot practices, where farmer decision-making and beha vior are attributed to limitations posed by input prices (the rising cost of fertilizer over the year be fore this fieldwork was conducted was
78 repeatedly mentioned during farmer interviews), lack of credit access an d lack of land. Longer term, these constraints are going to outweigh the compensatory demands placed on the local soils through year-round cultivation indeed, area soil s are already claimed to be highly nutrient depleted (Place et al ., 2007) that will decrease the leve ls of productivity and remuneration attainable from these activities. What these an ecdotes highlight is the interven ing role poverty plays in restricting the cooperati on households have with governme ntal agencies over land (crop) management practices. Combining quantitative result s with ethnographic information gathered via PRA, these described differences in livelihood strategies between SOP categorical poverty outcomes are schematically presented as representative types in Figures 4-4 4-6. The arrows show pathways of cash, goods or materials flowing in and out of households. What is communicated by these livelihood system diagrams is the progression by which these pathways representing livelihood strategies go missing as househol ds transition into poverty and di sconnect from the market, with two important implications. First, poverty ha s pushed households in creasingly into the nonmarket sector the widespread adoption of info rmal employment/casual labor a strong indicator of this hence they are unlikely to benefit from future market reforms a nd thus require specially targeted policies (Kabubo-Mariara & Kiriti, 2002 ). Second, these diagrams show the dependence of rural smallhold farmers on the natural sy stem for their livelihoods and wellbeing. For example, in Chepkemel 96.8% and 48.4% of househol ds interviewed (n=31) rely on either their own or other peoples land as the primary s ources of fuelwood and construction materials respectively. This recognition is critical to avoid i nducing a poverty environment trade-off as discussed in Chapter 2 in the absence of compensatory policies.
79 Figure 4-4 shows the on-farm livelihood st rategies of a typical non-poor household reflecting the following initial conditions or c ontextual factors. In terms of assets and capabilities, these households have a minimum la nd size (two acres in crop production appeared to be the critical threshold in households selected for the PRA) to support a combination of subsistence and cash crop production with surp lus sales made to mark et. Occasional use of fertilizer is reported with positive impact on yiel ds. A variety of livestock strategies are used, mostly involving cattle and dair y sales. Fodder production was in troduced into the village by MOARD and for households able to adopt it, ca ttle ownership remains possible. Chickens and egg sales are also common. Offfarm or non-farm employment are not common strategies for this group as on-farm strategies are sufficiently re munerative. What is comm on is participation in womens or other self-help groups, the most commonly referred to being merry-go-rounds, which is a form of group savings in a context where formal credit services are minimal. Households are dependent on the natural system for the supply of water for production and consumption and local forest produc ts for fuelwood and construction. Figure 4-5 shows a downwardly-mobile household that is transitioning toward poverty. The solid pathways of a non-poor ho usehold are beginning to break re vealing a shift in strategies away from solely on-farm produc tion. Subsistence crops may remain but there are no marketable surpluses and an increasing re liance on markets for food purchas es. Crop input purchases are reduced perhaps to zero except for seed. Ca sh crop production is abandoned. There are no normal livestock trading activities, only the forced divestiture of this fo rm of savings to meet immediate cash needs. Secondary school education for children becomes cost-prohibitive. Some non-farm activities such as ope rating a small trading business may be adopted, but more
80 commonly casual labor is taken up. Some gr oup membership continues although they are reported to become more exclusive because of the cost to join or participate. Lastly, Figure 4-6 presents the remaini ng livelihood strategies of a poor household. Minimal subsistence production re mains although the household is very dependent on the market for food purchases. The lack of cash means food insecurity and reports of hunger. Only a few chickens may remain as livestock. Casual la bor is the main source of cash income. The dependency on the natural system fo r water and wood resources remain s. The relationship between water, poverty and livelihoods in Chepkemel can be summarized as follows. The village is described by farmers as receiving ample annual rainfall and all of its crops are sufficiently rainfed producing relatively good yields (Table 4-8 compares maize productivity and input usage between Chepkemel and Miolo). Of 31 households interviewed by original SCALES project personnel 29 produced mai ze, retaining either all or a portion of the harvest to provi sion the household and marketing any remainder thus generally characterizing them as autarkic or net sellers, depending on land size. However yields and, thus, water productivity could be substantially improve d through the use of fertilizers reported as being sporadically used and in amounts r oughly half of what is recommended by MOARD extension agents for reasons de tailed above or alleviating ot her conditions of poverty that minimize spacing or weeding of crops. Poverty and factor constrai nts on livelihood strategies can be seen in the following examples of farmer behaviors th reatening negative externalitie s involving water. Steep slope farming is evident (Figure 4-7) due to the shorta ge of land in the village, a behavior contributing to erosion and other lateral flows and water trans itions. Livestock, a critical livelihood activity, is watered daily in the local springs often requiring a s ubstantial labor time investment, except on
81 rare farms with private boreholes (6 out of 31 households interviewed). This practice also contributes to shoreline erosion and poses potential human health risks. Some discussion is warranted on the other dimensions of water scarcity and water-related poverty and livelihood impacts within Chepkemel. While there is seasonal variability in rainfall, relative to the majority of Kenya, Chepkemel would not be characterized as a water scarce village in terms of physical scarcity. Over th e period 1992-2002, water qual ity was considered to have stayed the same or improved by 90.3% of survey respondents (n=31) while water access was reported to have remained unchanged or improved by 80.6% of respondents. Quality improvements were the result of community actio n against an upstream, polluting coffee factory, a positive example which may inspire future collective action over water management. Despite these perceptions about water qual ity and access, there remain conditions of economic water scarcity in terms of a lack of water infrastructure in Chepkemel. In the sample of 31 households interviewed, only 23 % reported ha ving an on-farm water source, perhaps a borehole, well or roof catchment system. Women, as the primary water collector in 87.1% households interviewed (n=31), tell of spending c onsiderable amounts of time per day collecting water for a variety of domestic uses: cooki ng, drinking, bathing, and washing clothes. Commonly, water collection involves walking to the local rivers or springs an average of three times per day, traveling distances ranging from 0.1 to 2.5 kilometers each direction (in sample of 31 households interviewed, 68% tr aveled between 0-500 meters each way), and returning with about 30 liters of water in jerry cans perche d on their heads when donkeys are unavailable. (Were et al. (2006) report some skepticism over the accuracy of similar figures regarding distance and the same caution is advised here.) Aside from the onerousness of water collection, there is the associated impact on the availability of womens la bor for livelihood activities. As
82 becomes apparent in the household-scale analysis coming in Chapter 5, womens cash-earning activities are critical to household wellbeing and can mean the difference between livelihood security and insecurity. Indeed, freeing up wo mens time for production-related activities is argued as a key motivation for local colle ctive action on water development (Were et al ., 2006; Roy et al. 2005). From a different dimension of poverty, reduced water consumption with its associated health implications is affirmed by data involving women from non-poor households indicating they spend more time and collect more water than women from households descending into poverty. Water, poverty and livelihoods in Miolo Turning attention to lowland village, Miol o, from the data in Table Appendix B-2 the livelihood strategies of households disaggregated by SOP category vary substantially from their upland neighbors. Diversification into multiple livelihood activities is the dominant livelihood strategy adopted by all households regardless of category. No particul ar livelihood activity assumes the kind of dominance as maize and liv estock do in Chepkemel. From interviews, because of adverse climate conditions, exits from all available activities except casual labor predominate in the period between 1992 and 2002. Non-farm strategies specifically two activities, formal employment or working as a skilled laborer, seem to be critical in differentiating households which remain nonpoor from those who are in poverty, a less ambiguous finding in this village context. Neverthe less, these strategies are limited because of a lack of education and formal job market opportunities. While not pr acticed as widely in Miolo as in other lowland villages, again purportedly b ecause of unfavorable climate conditions and the closure of the cotton marketing boa rd, some forms of cash crop/com mercial agriculture is carried on in households that are nonpoor or exiting poverty.
83 Figures 4-8 and 4-9 are photographs of farms and the environs of Miolo, which indicate why a diversified strategy is essential for inhabita nts of this village to try to secure a livelihood and lower risk. Miolo is already located in the drier LT agroecological zone, where annual rainfall is only marginally sufficient to suppor t non-drought resistant crops including subsistence crop, maize. The rainfall pattern is reflected in the typical planting ca lendar (Table 4-9) documented in collaboration with local farm ers showing only one cropping season, indicating the necessity of livelihood st rategies either beyond crop farming or involving new crop technologies. Some examples of local coope ration, innovation and technology use to contend with problems of physical water scarcity include a flexible sharing scheme of a local, privatelyowned generator for pumping river water to irrigate fields; collective gardening on rented lands nearer river; and river-fed aquaculture ponds. Additionally, the bimodal rainy seasons are too short to produce high yields (Table 4-8) insp iring the Kenyan government to support work on the development of new, drought-tolerant crop st rains for use in the lower basin although no MOARD extension officers ha ve ever visited Miolo. Significant for understanding the relationshi p between water, livelihoods and poverty, this marginal productivity in yields and crop freq uency reinforcing the poverty trap means Miolo residents are much more dependent on markets for the purchase of food and vulnerable to price fluctuations as net buyers of staple grains even in households that are classified as not poor. For instance, household interv iews in Chepkemel determined that maize purchases are made, if at all given the predominantly surplus or autarkic pr oduction, usually between April and July, stores having lasted from the previous August, while in Miolo the crop harves ted in August more or less has been consumed by January. From a timing perspective, this means households face competing cash demands between daily food purcha ses, the need to purch ase crop inputs for the
84 sole planting season, January to Au gust, and the payment of school f ees or related expenses all at the same point. Thus market dependency has im plications on every dimension of poverty but specifically so in terms of food security and nutr ition of the poor. Aside from these initial geographical conditi ons of physical water scarcity, the lower Nyando River basin experienced El Nino floods in 1997-98 and a La Nina drought, officially recorded as occurring between 1998 and 2000 alt hough respondents in Miolo referred to drought conditions lasting until 2002. Beyond the local live lihood insecurity and declines in wellbeing, these events had significant impacts on the overall Kenyan economy with the El Nino floods estimated to have cost 11% to GDP and the La Nina drought estimated to have cost 16% to GDP in 2000-02 (WRI et al. 2007). In May 2004, a flood event in Miolo affected 4500 people; 1500 households were lost; 2700 acres of crops were destroyed; and 150 h ead of livestock were killed. Miolo and the lower Nyando basin, therefore, represent the sort of natural hydrologic vulnerability which compounds th e vulnerability of the land to processes of erosion and desertification that Grey & Sadoff (2002) claim ar e innate to Africa and responsible for water poverty traps found across the continent. Wate r poverty traps qualify as a special type of geographical poverty trap defined as the insu fficiency of natural resource endowments to support economic growth and argued most famously by Sachs (Gallup et al ., 1999) as explanatory of the lack of economic growth acr oss Africa. Farmland with in Miolo, indeed, has been subjected to extensive erosion (Figure 48), attributed to over 100 years of extensive upstream deforestation, the effects of which are made worse based by the regions innate geographical characteristics for example, an average slope of 5% and sediment transport capacity index score of 0.30, both values the highes t in the Lake Victoria basin and cited as reasons why the Nyando River basin has had a pa rticularly significant impact on the Lakes
85 eutrophication (Douglas et al. 2006; Odada et al. 2004) and exogenous shocks, compounding its already low agricultural productivity. Farmers reported needing to replant crops two or three times during the 2007 growing season due to r ecurring flood events putti ng strain on already constrained household resources for the purchase of replacement seed only, fertilizer neither being affordable nor incentivized based on pers onal histories with fr equency of flooding. Water-related poverty is also expressed in Miolo in economic terms. For instance, the bridge, which crosses the Awach River (tributary of the Nyando River) linking the village to the main access road, was washed out during flooding in December 2006 and remained unrepaired during field visits in June 2007. In spatial terms, this situation deepened Miolos isolation and exclusion from markets and health care services During fieldwork, local villagers were working cooperatively to shore up the bri dge and stabilize the make-shift repair allowing foot traffic to cross. Another example of local collective action to mitigate the effects of flooding in the village includes the planting of sisal to stabilize plot edges from encro aching water courses with varying degrees of success (Figure 4-10). For households th at can afford to do so, the preferred strategy is to rent plots on higher lands away from streams prone to overflowing. The immediate pressures of poverty, however, in cite contradictory conservation behaviors; for example, women have adopted as a livelihood activity the mining of earth from one of the local gullies, the consequence of flooding and the vulnerability of th e soils to erosion, to sell to road-building contractors (Figure 4-11). The Awach and Sare Rivers are the villag es principle drinking water sources and locations for bathing and for livestock watering. Water quality and access are reported to have remained the same or improved in the last ten years according to 96.6% of respondents (n=29) suggesting they are not positively correlated with th e substantial growth of poverty in the village
86 over the same period. Travel distance to water sources for 65% of the 25 households interviewed is 0-500 meters, roughly the same as Chepkemel. Roof catchment wate r collection systems are more prevalent in Miolo than Chepkemel with 40% of households possessing one, the depth of poverty in the village suggesti ng these technologies were f acilitated via non-governmental organizations working in the area. The potential opportunities for women to use their time freed up from water collection in productio n activities is more limited in Miolo, restricted to casual labor on farms away from the village. Statistical results on the relationship between livelihoods and poverty Logistical regression analysis modeling the correlations between available livelihood strategies and poverty confirm the above inferences about which activities are the most remunerative in Chepkemel and Miolo. Tables B3 6 displaying regressi on results are to be found in Appendix B. From those tables, the co efficients have been exponentiated (Pampel, 2000) to determine the percentage change in odds of poverty based on the adoption of a particular livelihood activity (Table 4-10). The significan ce of maize and livestock to reduce the odds of being poor and of descent into poverty resp ectively in Chepkemel is confirmed, as is the significance of formal employment for reducing the odds of both be ing poor and pov erty entry in Miolo. In Miolo, working as a skilled laborer an d producing cash crops are also significant in reducing odds of being poor but not for preventi ng the transition into poverty. In sum, these findings reflect the livelihood systems as define d by extant agroecologica l conditions, the risk and vulnerability associated with climate conditions and lateral flows, and changes in the economy and governments involvement in agriculture. Re-aggregating the data, Table 4-11 presents the percentage changes these livelihood activities have on the probability of being poor across the entire Nyando River basin in each of the three time periods. Between 1992 and 2002, live stock, specifically ca ttle, remained most
87 associated with not being poor despite decreasing in importance relative to historic times, a drop more likely the result of less participation and decapitalization due to pove rty than a change in the remuneration attainable from it. Formal employment remains a critical activity; however, its success over the same time period has waned, a di rect consequence of slowed growth in the Kenyan economy and structural ad justment policies applied over the 1990s, a topic yet to be considered. According to results, cash cropping/c ommercial agriculture gene rally and of sugar as one example became a more significant activity as sociated with decreasing the probability of poverty in the 1990s. However, most livelihood o pportunities relating to sugar production are as informal laborers, an activity correlated with an increasing probability of poverty. Commercial, irrigated rice production indi cates the opposite tre nd to sugar, a direct consequence of institutional changes associated with structural adjustment polic ies and liberalization of Kenyan parastatals, a history that will be discussed in a later section of this chapter. Subsistence cropping, for instance in the exam ple of sorghum, has become increasingly correlated with poverty reflecting the stagnation of yields, reductions in farm si ze, and, consequentially, lower labor productivity all linked to high growth in population of households participating in agriculture (Jayne et al ., 2001). This impoverishing effect of staple grain production, however, has not been evenly experienced with households which are net sellers or autarkic less vulnerable than those which are net buyers ( ibid ). Livelihood summary Synthesizing this section, understandi ng of the relationship between the poverty dynamics and livelihood strategi es across spatial and tempor al scales was developed. Conditionally, on-farm income-earni ng strategies continue to be sufficiently remunerative for households to remain above the poverty line in the upper basin. Factors limiting agricultural productivity include the lack of applied inputs associated with th e lack of cash availability or
88 access to credit on terms that reflect seasonal fluctu ations in cash availabi lity, and lack of land associated with population growth and density. Lo wer yields and lower returns are helping to push households to decapita lize, especially of livestock, and to adopt coping behaviors, such as increased farming intensity via intercropping, year-round production and slope farming, which can produce water transitions downstream exacerbated by extant geographical conditions and exogenous events. In the lower basin, on-farm cropping activities produce very low yields attributable to less-favorable agroecological conditions regular exogenous shock events such as flood and droughts and concomitant land degradation including extensive erosion, and the unaffordability of inputs. Irrespective of wealth levels, all households are forced to adopt diversified strategies to gene rate a livelihood. Formal employmen t and operating a small trading business are most correlated with non-poor households in the lower ba sin, essential to offset low and seasonal agricultural income High barrier-to-entry commerc ial cash cropping historically was successful, but economic reform po licies have reduced opportunities. Of the relationship between water, poverty a nd livelihoods, from this point in the analysis the Nyando River basin varies in terms of phys ical water scarcity between upper and lower reaches with varying implications for CPWF technology development. Water productivity could be increased in the upper basin, where water is not physically scar ce, by seeking economic solutions such as enabling smallholders to a ccess input markets. An interview with local Ministry of Agriculture (MOARD) officials exposed the governments strategy to promote to smallhold farmers alternative high-value crops ( e.g., onions, tomatoes, Irish potatoes, and fruits including mangoes and avocados plus honey and da iry), another means of increasing water (and labor) productivity to increase on-farm incomes, and, hopefully, alleviate problems associated with land scarcity and environmentally-detri mental coping behaviors. The lower basin,
89 conversely, is a drier and hydrologically-vul nerable area requiring a combination of environmental and social solutions (Chapter 1). Thusly, successful on-farm strategies will require on-going development of drought-re sistant crops and, likely, irrigation. The extensiveness of land degradati on and soil erosion observed in th e Miolo area suggests strong need for accessible and affordable land rehabili tation technologies. Strategies to mitigate the risks of flooding are also essentia l. All of these needs will re quire forms of collective action across scales, the scope and cost of solutions being well beyond the cap acity of local Miolo residents. In consideration of the disparities or ine qualities in wellbeing livelihoods and poverty dynamics across the basin and between the two focal case villages, some initial conclusions are proposed about the likelihood of cooperation for resource management across scales. Cultural questions pertaining to ethnicity are beyond the scope of this anal ysis, but should be teased out of the economic experimental games to be conduc ted at a later stage of the SCALES research. Historical clashes between the Luo and Kipsig is, including violent c onflicts in 1994 in the vicinity of Miolo suggest ethnic tensions will be a si gnificant hurdle to the development of trust, a key factor in collective ac tion. Miolo village shows compar atively less inequality than Chepkemel with implications for their willingness to work collec tively. At present, there are two significant interactions between these two neigh boring villages (located approximately 10 aerial kilometers apart) which are likely to im pact opportunities for co llective action on water management. First is the informal employment, casual off-farm day labor, the lowland village residents, particularly women, re gularly obtain in the upland vill age. The critical importance of this casual labor income to lowland villagers reco rded during interviews i ndicate a flow of power and influence upland residents have over thos e in the lowland regi on. Second, along with cash
90 income, agricultural products flow downstream from the upland village to markets nearby to Miolo linking lowland food security to upland productivity. Combined, these two conditions suggest a minimum of reverse flows to influen ce upstream residents to change their resource management behavior or internalize the costs of negative environmental externalities. Further, even if the relatively impoverished lower ba sin could find government or donor financing to institute changes in upper basin management and land use to mitigate negative downstream externalities through such fashionable, compensa tory schemes as payments for environmental services programs, there remain larger questi ons about food security and economic growth for the Kenyan state as will be discussed shortly. Geographical Factors and the Spatial Di stribution of Poverty and Livelihoods One of the narratives being induced from this integrated, multi-scale analysis of the Nyando River basin is the importance of geographical and spatial factors in determining livelihoods and poverty in rural Kenya. Okwi et al (2007) model a combination of exogenous ( e.g ., rainfall, elevation, slope, type of land cover, soil type, isolation from markets, and length of growing season) and endogenous ( e.g ., population, market and road de nsity, and transportation time) variables finding that they explain more than 53% of the variation in rural poverty across Kenya. The exogenous spatial variables alone explain 36 % of variability in poverty. Extracting results relevant to the Nyando River basin, exogenous variables eleva tion, slope, exposure to flooding, and length of growing period all return as statistically significant, converging with these analytical results. Kenya is a state recognized for its lack of natural resource endowments. This is particularly true regarding wa ter. The countrys total annual renewable water resource is estimated at 30.7 billion cubic meters per annum one third of this (approximately 20.2 billion cubic meters) coming from rainfall as internal renewable surface water and the balance supplied
91 by groundwater and transboundary water fl ows from neighboring countries (WRI et al. 2007). When divided by Kenyas estimated 2008 popula tion of 37.95 million people (CIA, 2008), the total renewable water resource av ailable per year is 532 cubic meters per person. This is a substantial drop in four years from the 936 cubic meters per person calculated to be available based on the 2004 population of 32.8 million people (WRI et al. 2007) and it is predicted to drop further, to 190 cubic meters per person by 2025 based on population growth estimates (Gleick, 1993). A country is defined as water sc arce when annual per capita water availability falls below 1,000 cubic meters per person, consider ed the minimum per capita water requirement for an efficient, industrialized state (Gleick, 1993) This degree of water scarcity is predicted to limit economic development in Kenya for the next few decades. Part of this scarcity is that Kenyas rainfall pattern is variable seasonally with a bimodal rainfall pattern occurring acro ss the country. With 98% of Ke nyas crops rainfed, there is widespread exposure to rainfall variability (WRI et al. 2007). In the Nyando River basin, the long rains occur between March and September and the short rains between November and January. The general challenge this rainfall pattern poses is that neither of the two rainy seasons is quite long enough to allow very high crop yi elds. The minimum required amount of rainfall per year to support maize and other non-drought re sistant crops is 800 millimeters per annum. As a result only 15% of Kenyas total land area, including the upper Nyando basin, receives enough rain to support its primary stap le food crop, maize, while only an additional 13% of the country, including the lower Nyando basin, is classified as receiving marginal rainfall sufficient for growing drought-resistant crops (WRI et al ., 2007). This means 72% of Kenya has no agronomical growing season. Therefore with on ly 28% of Kenyas land area suitable for agriculture, its intensification and extensification w ithin the Nyando River basin is to be
92 anticipated yet challeng ed by the continued growth and incr easing density of population settling marginal, vulnerable lands. Expanded irrigation infrastructure is a frequently mentioned solution with only 19% of Kenyas irrigation potentia l argued to be currently activated (WRI et al. 2007). Significant for the Nile Basin at large and the CPWFs intention to assist in the development of a basin action plan is that all bu t one (Uganda) of the upper riparian countries of the White Nile basin Kenya, Tanzania, Rwanda and Burundi are projected to experience extreme water stress and scarcity with implications for transnational cooperation on watershed management. Within Kenya, the Lake Victoria dr ainage area, which includes the Nyando River basin, captures 65% of the countrys internal re newable surface water s upply per annum. This means that as Kenya becomes increasingly water st ressed, there will be more demands placed on the Lake Victoria region to tran sfer water out domestically physic ally and virtually in the form of other goods or services to areas unable to meet demand based on local supplies. However, based on anticipated population growth alone, the upper Nyando River basin is predicted to experience its own water deficit by 2010. This worsening condition makes the pursuit of improved water productivity by the CPWF even more imperative, but rev eals the wider political economic challenges that lie ahead. The argument here is that, in view of wors ening national conditions of water scarcity, agricultural production within th e Nyando River basin, particular ly in the more naturallyendowed Kericho district and other areas of the upper basin, will b ecome even more strategically important for national food security and economic growth, suggesting that the power of this east African water tower is potentially unassailable. However, this will place even more pressure on already substantially degraded en vironmental resources and threaten even more lateral flows and
93 water transitions downstream. Assessments on the environmental state of the lower Nyando River basin concur on the changes in water qua lity and quantity with measurable impacts on human health, visible issues of sedimentati on and eutrophication, and lost soil and water productivity (Mungai et al ., 2002). These environmental outcomes are argued to be the result of the well-established cause and effect relationship between upstream land use behaviors and downstream environmental impacts magnified by the geographical characte ristics of the basin ( ibid ). In view of the poverty-environment relationship Okwi et al. (2007) affirm, the risk and severity of water poverty traps within the basin is reasonably expected to grow if these upstreamdownstream interdependencies are not integrated in to future basin policy and rural development. The imperative for social equity also must be observed meaning how new agricultural opportunities are to be instituted to be inclusive of the poor is a critical question. This segues into the role institutions have played in the pove rty dynamics of the Nyando River basin. Institutional Factors behind Poverty and Livelihoods in the Nyando River Basin The poverty dynamics of the Nyando River ba sin can be contextualized and better explained in reference to the hi story of structural adjustment programs (SAPs) enacted in Kenya beginning in the 1980s but which did not reac h full force until the 1990s. (For a detailed presentation of neoliberal economic reform in Kenya, see Legovini (2002) and Were et al (2005).) The macroeconomic change s introduced during this time under the principal direction of the International Monetary Fund (IMF) involved reform of sector s including social services (health and education) and agriculture, which are now recognized as having had a disproportionate and detrimental effect on the poor. Assessments of structural adjustment in Kenya conclude that it has faile d to create the cond itions for sustainable recovery of gross domestic product (GDP) growth to levels atta ined in the 1960s and 1970s. With this slow growth, poverty has been increasing while social indicators (life expectancy, child mortality and
94 primary school enrolment) have shown negativ e trends in the adjustment period (KabuboMariara & Kiriti, 2002, p. 2). The basic objective of SAPs is to restor e developing countries to macroeconomic stability and to revive economic growth thr ough increased resource mobilization and more efficient use of resources. (Kabubo-Mariara & Kiriti, 2002, p. 4) From independence in 1963, to support its economic growth the Kenyan government had sought actively to mobilize and efficiently use its limited resources with special focus on rural smallholder farmers. This included property rights reform and resettle ment onto productive lands where commercial agriculture was actively state-supported via parastatals depl oying market controls and abovemarket pricing ultimately incentivizing over-produc tion (Legovini, 2002). As a result, agriculture output grew at rates greater than 5% during th e 1970s and, as a sector, was buoyed by the coffee and tea booms of the late 1970s and 1980s in time s of decline. With parastatals, notably the Kenyan Cereals and Produce Board (KCPB), cont rolling interregional movements of maize, wheat, rice, sugar, cotton and milk, producer pri ces remained above market pricing, enriching net sellers while impoverishing net buyers, and drivin g the KCPB among others into unsustainable deficits (ibid). De Groote et al. (2006) present findings on the impact of market liberalization involving the privatization of parastatals on agriculture, specifically maize production, in Kenya for the period from 1992 to 2002 when macroeconomic re forms were more rigorously applied. Their results are disaggregated by agroecological zones he nce it is possible to look at the difference of impacts in the upper basins moist transitional (MT) zone versus the lower basins lowland tropical (LT) zone. Generally, th e authors reach two conclusions : Market liberalization had a positive impact on the price of maize that is, lowered it to near parity with world markets
95 making maize theoretically more affordable for the poor after bei ng disadvantaged by high producer prices while decreasing the ratio betw een input and output pri ces thus lowering its profitability to producers. From the producers pers pective, the social bene fit of these outcomes, however, does not overcome the uncertainty ar ound maize production cons equential to price fluctuations in the early 1990s after relative stability throughout th e 1970s and 80s, and the reduced institutional support in th e form of research and extension services, agricultural credit and marketing boards. As a result, after the gr owth periods of the late 1970s and early 1980s, maize yields began to stagnate in the 1990s wh ich, in the context of population growth, means a decrease in maize production per capita. When the grain-basket of Kenya was hit with the exogenous shocks of flooding and dr ought in the late 1990s, it was no longer self-sufficient in maize and remains a net importer to this day linki ng food security of the poor to fluctuations on the world maize market. Similar histories describe the impacts of economic reform on producers of cotton, wheat and sugar during the 1990s lowe ring remuneration and yields a nd eventually causing exit from the activities by marginal producers. Looking more in-depth at the case of irrigated rice production in the lower basin, the eventual clos ing of operations of the parastatal, National Irrigation Board (NIB) in 1997 has been determined to be directly accountable for the explosion of poverty in Village 9 of the SCALES study (Swallow et al ., 2008). Table 4-5 shows poverty grew in this village from 35.4% to 62.5% of households be tween 1992 and 2002 distinguishing it with the highest percentage of poor households in the Nyando River basin sample. The NIB served similar functions as the Kenyan T ea Development Authority (KTDA) in providing farming inputs, marketing and timely payment for crops. When the NIB ceased to function, smallholders were unable to cont inue producing rice in part due to the property rights conditions
96 of the village and lack of resour ces. This story of the success and failure of irrigation schemes in the lower basin is instructive, linking irrigation and stat utory property rights to poverty as well as to the potential for cooperation over water mana gement (ibid). The Provincial Irrigation Unit (PIU), another organization facilitating irrigation schemes in village nos. 13 and 14 discussed earlier, faired better in maintaining livelihoods and offsetting poverty although the scheme in village 13 was finally forced to shut down due to siltation and the inabilit y of local residents to pay for its removal. One of the fundamental social sector reforms introduced by adjustment was cost sharing in health and education with user fees taking full effect in 1991-92 after a preliminary attempt to institute them in 1989 (Were et al. 2005). These cost-sharing programs also had a disproportionate impact on the poor where the effect of education and health sector reforms has been a decline in household savings and increas e in poverty since a larger proportion of the consumers income is devoted to education and health care financ ing. (Kabubo-Mariara & Kiriti, 2002, p. 2) The accuracy of this quotation is made flesh by the rural inhabitants of the Nyando River basin as is reflected in the shar p increases in poverty in the sample villages (Figures 4-1 and 4-2). The linkage between thes e institutional changes and livelihood strategies appears visible when, from 1992 to 2002, households were pushed to decapitalize of livestock, fewer were able to afford to participate in crop farming, and many adopted casual labor (Table B-1). At a macro-scale, these institutional changes contributed to a decline in investment, lower productivity, and, by increasing dependence on natural resources, both common property and private, worsened the degradation of an innate ly vulnerable environment, the three essential characteristics of a poverty trap.
97 There are important insights to be draw n from this experience for SCALES as it considers the opportunities for collective action via improved water management to alleviate poverty in the Nyando River basin. Decentralized resource management based on collective action essentially represents anot her form of cost-sharing for th e provision of public goods, in this case, improved water quality and quantity, an d environmental security While user fees, to my knowledge, are not implicated in proposed project interventions in this case, investments of labor are essential. Household constraints and poverty may prohi bit the possibili ty of such investments without first increa sing current labor productivity le vels. However, there is an implicit tautology in this if it is collective action that is suppo sed to be what improves labor productivity via water management as a first step toward improved wellbeing. This discussion concludes by referring back to the trends observed in Table 4-11 displaying the percentage change of various livelihood activities on being poor, which can be more thoroughly explained when vi ewed in context with the changes in agricultural and social policies instituted as part of economic liberaliz ation. A significant retrench ment of public sector wage employment occurred during the 1990s, which accounts, in part, for the drop by formal employment. As per the intention to shift the ec onomy to export-oriented agriculture, the rise by cash crops, particularly tea, is expected given the support sma llholders receive from private market organizations. With the collapse of th e National Cereals and Produce Board (NCPB) and other marketing boards and parastatals, the declines in importance of crops such as rice and sorghum are also expected. The dramatic increase in importance of small business reinforces the pattern of a shift from formal to informal em ployment common to liberal ization programs. These statements confirm the importance of developing an integrated, cross-scale and coupled systems
98 understanding of poverty and liv elihood dynamics. The relationshi p between water, livelihoods and poverty is shown to be heavily influenced by factors exogenous to the farm or plot scale.
99 Table 4-1. Economic transition matrix of tre nds in household poverty in 20 villages across Nyando River basin over a 25 year period (approx. 1977 2002). 25 years ago At present Poor Not Poor Poor 9.0% (A) chronic poor 19.5% (C) entered poverty Not Poor 7.9% (B) escaped poverty 63.6% (D) non-poor Table 4-2. Economic transition matrix of tre nds in household poverty in 20 villages across Nyando River basin over a more re cent ten year period (1992 2002). 10 years ago At present Poor Not Poor Poor 10.9% (A) chronic poor 16.3% (C) entered poverty Not Poor 4.2% (B) escaped poverty 68.7% (D) non-poor Table 4-3. Economic transition matrix of tr ends (1992 2002) in household poverty for Nyando district (N = 627), lowland areas of the Nyando River basin and a lowland tropical agroecological zone. 10 years ago At present Poor Not Poor Poor 15.3% (A) chronic poor 18.7% (C) entered poverty Not Poor 4.6% (B) escaped poverty 61.4% (D) non-poor Table 4-4. Economic transition matrix of tre nds (1992 2002) in household poverty for Kericho district (N = 411), upper elevation and mois t transitional agroecological zone of the Nyando River basin. 10 years ago At present Poor Not Poor Poor 7.3% (A) chronic poor 11.9% (C) entered poverty Not Poor 3.9% (B) escaped poverty 76.9% (D) non-poor
100 Table 4-5. Trends in the percentage of poor households disaggregated by village. Villages have been listed in rank order from the largest percentage decrease to the largest percentage increase of poor house holds during the period 10 years ago to the present (1992 2002). Data reported fo r all 20 villages where SOP poverty a ssessment was conducted on behalf of SCALES and Safeguard projects. Village no. District location Number of households presently Percentage of poor households Change in percentage of poor households 25 years ago 10 years ago present 25 to 10 years ago 10 years ago to present 25 years ago to present 3 Kericho 129 16.3% 8.5% 4.7% -7.8% -3.9% -11.6% 20 Nyando 52 20.8% 14.0% 13.0% -6.8% -1.0% -7.8% 8 Nandi 58 3.7% 4.7% 7.1% 1.0% 2.4% 3.4% 14 Nyando 94 45.2% 33.2% 37.4% -12.0% 4.3% -7.7% 6 Nyando 71 7.4% 20.9% 25.2% 13.5% 4.3% 17.8% 15 Nyando 58 0% 0% 4.5% 0% 4.5% 4.5% 1 Kericho 63 3.0% 3.1% 10.4% 0.1% 7.3% 7.4% 19 Nyando 47 4.7% 2.2% 11.1% -2.4% 8.9% 6.5% 11 Nyando 101 9.1% 11.0% 21.0% 1.9% 10.0% 11.9% 17 Kericho 60 15.3% 21.8% 32.8% 6.6% 10.9% 17.5% 13 Nyando 52 38.5% 25.0% 36.5% -13.5% 11.5% -1.9% 4 Kericho 73 0% 4.2% 16.9% 4.2% 12.7% 16.9% 7 Nandi 92 1.2% 7.3% 21.9% 6.1% 14.6% 20.7% 2 Kericho 28 0% 0% 20.0% 0% 20.0% 20.0% 18 Nyando 49 14.1% 20.4% 40.4% 6.3% 20.0% 26.3% 5 Nandi 50 18.4% 14.1% 36.4% -4.2% 22.2% 18.0% 16 Kericho 58 20.9% 28.4% 51.3% 7.6% 22.9% 30.4% 10 Nyando 63 3.3% 4.8% 29.0% 1.6% 24.2% 25.8% 9 Nyando 73 35.2% 35.4% 62.5% 0.2% 27.1% 27.3% 12 Nyando 101 34.7% 24.8% 56.4% -9.9% 31.7% 21.8% Mean 1372 16.9% 15.4% 27.8% -1.5% 12.4% 10.8%
101 Table 4-6. Poverty dynamics (1992 2002) for case study villages, Chepkemel in Kericho district and Miolo in Nyando district. 10 years ago Case village At present Poor Not Poor Miolo (no. 12) (N = 101) Poor 18.81% 37.62% Not Poor 5.94% 37.62% Chepkemel (no. 4) (N = 73) Poor 2.74% 13.7% Not poor 1.37% 82.19% Table 4-7. Typical planning cale ndar for Chepkemel smallholder farm. Shaded months indicate common periods of food insecurity and liquidity constraints. Crop Maize Sweet Potato Beans Bananas Homegarden Expected yield 10 sacs/acre (900 kgs) 25 sacs /0.5 acre 4 sacs/ac (360 kgs) 1 bunch per month approx. produced, sold Veggies for home use January Prepare land; plant Plant intercropped with maize February March April Harvest/sell May plant Harvest June July Plant pure stand August Harvest and sell September Plant quick harvesting maize October November Harvest, sell Harvest December Harvest and sell Plant
102 Table 4-8. Comparison of physi cal and human capital inputs into crop farming between Chepkemel and Miolo. Data reported are ranges based on household interviews and are separated for households at low-level versus high-level equilibrium states. (*= amount spent on casual labour net of fulltime live-in farm boy). Chepkemel Miolo Poor household Non poor household Poor household Non poor household Seed purchases (KSH/ac) 1153-1560 956-1780 500-3610 2010-6480 Fertilizer expenses (KSH/ac) 2333-3000 1878-4600 0 0-400 Hired supplemental labour (KSH/ac) 0 0-4733 0-800 1600-8750* Maize yield (kgs/ac) 540-900 630-900 90-324 270-720 Total annual farm expenses (KSH/ac in production) 5220-5280 3007-13955 1000-7390 11220-15720* Table 4-9. Typical planting cale ndar for Miolo smallholder farm. Crop Maize + Beans Sorghum Sukuma Wiki (local kale) Tomatoes Expected yield 3 sac maize/acre (270 kgs); 2 sacs beans/acre (180 kgs) 3 sacs/acre (270 kgs) January Maize sacs empty Sorghum sacs empty February Plant Plant End harvest/ sales March April May June Harvest beans July Harvest August Harvest maize September Bean sacs empty plant planting October November harvest/sell harvest/sell December
103 Table 4-10. Percentage changes in odds of be ing poor and odds of entering poverty. *. These independent variables are determined to be significant in the logi stic regression model equations. The unmarked values were not deemed to be significant independent variables in the regression hence their actual influence on changing percentage odds is unlikely. Odds of being poor Odds of entering poverty Chepkemel Miolo Chepkemel Miolo Maize -95.2%* 131.4% -49.2% -77.6%* Livestock -35.4% -62.8% -83.0%* 187.9% Casual Labor 32.9% 2.3% -64.2% 64.5% Formal Employment 38.4% -96.6%* -84.4% -85.7%* Trading Business 0 -75.3% 0 -44.4% Skilled Labor 0 -86.8%* 0 -36.7% Cash Crops -37.2% -84.7%* -45.5% 54.4% Table 4-11. Percentage impact identified liveliho od strategies have on th e probability of being poor. Strategies 25-10 years ago (1977-1992) 10 years ago to present (1992-2002) 25 years ago to present (1977-2002) Average Cattle (Livestock) 18.0% 15.1% 20.9% 18.0% Formal employment 10.5% 8.6% 16.0% 11.7% Cash crops 6.7% 9.1% 17.9% 11.3% Sugarcane 9.0% 9.0% 11.0% 9.7% Irrigated Rice -5.6% -5.8% -5.7% -5.7% Small business 6.0% 14.5% 5.0% 8.5% Informal employment (Casual labor) -6.0% -5.7% -10.6% -7.4% Sorghum Production (Subsistence farming) -5.8% -9.7% -12.8% -9.4%
104 Figure 4-1. Poverty dynamics, Kericho district, for sample villages. Figure 4-2. Poverty dynamics, Nyando district, for sample villages.
105 Figure 4-3. Intensive farming of maize and sweet potatoes on smallholder farm, Chepkemel. Crops are maize and sweet potato.
106 Figure 4-4. Livelihood st rategies of a typical high-level equilibrium state household in Chepkemel.
107 Figure 4-5. Livelihood st rategies of a downwardly-mobile household on a pathway toward a lower equilibrium state in Chepkemel. Da shed lines represent pathways that are beginning to break.
108 Figure 4-6. Livelihood strategi es of a low-level equilibrium household in Chepkemel.
109 Figure 4-7. Slope farming in Chepkemel. Spring vi sible in bottom right co rner is one of main sources of drinking water in the village. Figure 4-8. Miolo environs showing the soil er osion and loss of land of the local vulnerability context.
110 Figure 4-9. Miolo farm plot after previous nights heavy rain s and flash flooding. Elevated land in upper right of picture shows an attempt to berm the land to protect from flooding. This crop should be almost ready for harvesting. Figure 4-10. Miolo farmers plan t sisal on plot boundaries to pr otect from flooding and soil loss. Leaves are cut to make rope and baskets.
111 Figure 4-11. Earth harvested from local gully for purchase by local construction companies.
112 CHAPTER 5 POVERTY DYNAMICS AND DRIVERS OF CHANGE IN LIVELIHOODS AND WELLBEING: EVIDENCE FROM HOUSEHOLD LEVEL ANALYSIS (1977-2002) This chapter continues to construct an unde rstanding of the correlates of the poverty dynamics within the Nyando River basin by prob ing household-level microeconomic conditions across scales. If Chapter 4s analysis of the geographical and institutional history of the basin developed, in an integrated way, comprehension of the extant macro-level poverty traps, Chapter 5 endeavors to provide the reader with knowledge of what constitute micro-level poverty traps at the household scale. This analysis goes deeper into the focal villages of Chepkemel and Miolo and into select households to investigate the various features of micro-level poverty traps (Table 2.1). Ethnographic and quantitative data are analyzed to query w hy households are resilient or vulnerable to stochastic transitions or structur al changes in poverty for its ultimate value in crafting targeted policy solutions Given the planned action of th e SCALES project to foster social capital as a resource for collection acti on around water management, the role of social relationships and other potentially exclusionary mechanisms, including gender, in household resilience or vulnerability vis-vis poverty are highlighted. This chapter is organized as follows: It begi ns with presenting the local definitions of poverty and prosperity as constructed using th e SOP methodology. It is in reference to the poverty line in particular that household conditions, behaviors and livelihood activities are evaluated. The most substantial part of the ch apter comes next involvi ng a discussion of the reasons or extant conditions randomly sampled hous eholds have been identified as being either poor or not poor, speaking to the whys behind the patterns of mobility or poverty dynamics at the household level. This analysis is first done base d on basin-wide data colle cted during the initial stages of the SCALES project, converting thes e reasons to capital asse t categories for their ability to reflect structural c onditions. Evidence collected during 2007 fieldwork is used to probe
113 the behavior and decision-making of certain representative households in Chepkemel and Miolo in response to their own conditions and experi ences, and how these get reflected in their livelihood strategies and poverty dynamics. Conc erns over the reliability and accuracy of quantitative data provided by households as well as the small number of cases means that construction of a quantitative dynam ic asset poverty line has not been attempted in this research. Nevertheless, some factors signif icant to its position and discussi on as to which equilibrium state household conditions and responses have caused them to gravitate toward is possible. Household-level details are then extrapolated to construct an understanding of the correlates of micro-level poverty traps in the Nyando River basin, the significance of water to their emergence, and the role enhanced water productivity could play in helpin g the poor to break out. Some detail as to the role of social organizati on in the livelihoods of the villagers is also discussed. At the end of the chapters presentation the reader should be left with an appreciation of how the larger human-cum-natural system impacts livelihoods, the resulting behaviors adopted, the household characteristics that lead to bifurcated equilibrium outcomes, and, ultimately, an integrated understanding of household poverty dynamics. Stages of Progress in the Nyando River Basin The SOP methodology establishes locally m eaningful poverty lines by asking village representatives what a household would buy as it upwardly moves from abject poverty to prosperity. The community then agrees at whic h stage poverty and prosperity lines would be drawn essentially separa ting the list into three types of households: poor; in an intermediate stage; or prosperous. In all, the 20 villages from across the Nya ndo River basin participating in the SOP poverty assessment (combined SCALES a nd Safeguard project da tasets) identified a total of 25 distinct stages of progress relating generally to basic needs for food (1.0), shelter (0.61), and clothing (0.90), to childrens prim ary education (0.78), livestock purchases (0.50-
114 0.59), the acquisition of farming inputs or ne w land (0.38), and investments in livelihood activities such as cas h cropping or the purchase of a bicy cle or an ox-plough (< 0.10). The figures in brackets are scores for each stage on a scale of importance ranging from 0 (not important) to 1.0 (extremely important). Each individual village identif ied between 11 and 14 stages and, while the ordering of these stages va ries across villages, f ood purchases are always ranked first as reflected by its score. Figure 5-1 contains a prototypi cal stages of progress list including the average poverty and prosperity lines for the Nyando River basin constructed from a review of all the sampled villages. The fact that a prototyp e could be made indicates that th ere are not gross differences in the stages of progress nor in how poverty is defined across the Nyando basin regardless of the agroecological conditions, elevation, land tenure type, ethnicity or degree of marketization and commercialization in the village. From this list it is possible to construe the livelihood strategies a household passes through as it moves into or out of poverty. Further, this list is interpretable as a view of household behavior or decision-making regarding asset accumulation and decumulation and investment in livelihood activities relative to their basic needs and proximity to the poverty line. This connects the stages of progress to the poverty traps framework by showing associations between specific assets and low versus high equilibrium states and growth paths (Krishna, 2004). Table 5-1 presents the stages of progress lists publically derived for Chepkemel and Miolo, which, upon reviewing, reflect the establis hed differences between upland and lowland livelihood systems in Chapter 4. Livestock owners hip is clearly linked to upward mobility, the lack of ownership registering t oo on separate lists of the attri butes of poverty for both villages. The preference for local cattle reflected in the data for Chepkemel has shifted toward hybrid
115 cattle for dairy production, a government (MOARD ) supported activity. Interviews with Miolo residents alluded to some lingering preference for local breeds for cultural reasons. Since primary school education fees were lifted in Kenya in 2002, a partial reversal of cost-sharing liberalization policies, combined with the strict enforcement of attendanc e by the state (Personal Communication, E. Nyaoro, Area Educa tion Officer, Lower Nyakach Division1), not being able to send young children to school is associated with the depths of poverty. Secondary or tertiary education, however, remains expensive and is only available to the most prosperous households, linking initial wealth conditions to the ability to obtain non-farm empl oyment. The relative importance of formal employment and its relati onship to being not poor in Miolo and not in Chepkemel is reflected somewhat by its stage ear lier ranking. The need for diversifying into nonfarm income sources through such activities as rental housing is seen in Miolo as well, contrasting with diversification and investment in on-farm activ ities such as cash cropping in Chepkemel where it remains associated with pros perous levels of wellbeing. The lack of land available in the upper basin due to population growth is evident in the small land sizes targeted for rent or purchase beyond any questions of affo rdability. Other unique attributes of poverty to Chepkemel that are not cited for Miolo include small land size (0.5 acres), having a large family, and lack of (financial) capital, the last a key defining feature of poverty traps. Household-level Conditions and Behaviors Correlated with Poverty Dynamics Part of the SOP methodology involves aski ng respondents in a focus group setting the reasons why a sample of village households has a ttained their particular stage of progress. For comparability with the SCALES project sites in Colombia once again, these data are presented 1 Statistics on drop-out rates for 2007 provided by Education Officer for three zones in the Lower Nyakach division within the lower basin indicate only 83 primary school drop-outs for the year, 36 boys and 47 girls, with the highest drop-out totals associated with Level (or grade) 8, the final year of primary school.
116 following standardized procedures applied to the ba sin-level unit of analysis. In total, 35 separate reasons, clustered into 19 groups, were identifi ed from across the Nyando River basin to account for a households level of wellbeing (See Appendix C, Table C-1 for the comp lete list including the number of times each reason was cited.) Vi a statistical regression these reasons were modeled against the binary, whether the househol d was poor or not poor, to identify the effect each has on the probability of being poor. Regres sion results for the pivotal period between 1992 and 2002 are presented in Table 5-2 with values on the probability of major reasons correlated with poverty detailed in Table 5-3. From Table 5.2, drugs and alcohol use (P-v alue = 0.0010) and death of major income earner (P-value = 0.0000) were the two most significant reasons for a household to be poor based on basin-wide data. Factors pertai ning to household demographics or life cycle effects were not significant in a households poverty level; for example, old age (P-value = 0.1990); child-related expenses (P-value = 0.1520); and large family size (P-value = 0.0950). Converging with the descriptive analysis in Chapte r 4, non-farm livelihood strategies involving formal employment (P-value = 0.0000) or operating a small business (P-value = 0.0060) were also significant for their effects on not being poor wh en aggregating basin data. From Table 5-3, drug and alcohol usage is f ound to have the larges t association with poverty, increasing its probability by 38%, a finding which contra sts against other nearby SOP assessments done in Vihiga and Siaya districts in Wester n Kenya (Krishna et al. 2004) and neighboring country, Uganda (Krishna et al ., 2006) where they were found to have no or only minimal impact on poverty. Death of major inco me earner has a predictable impact on being poor (25.8% increase in probability) also beca use of the reduction in a households human capital (labor supply and productivity). In the next two sections retu rning to the two focal
117 villages, upper basin repr esentative, Chepkemel, and lower ba sin representative, Miolo, the task is to probe for deeper insight into the local reasons cited for poverty. The stories of households selected for in-depth interviews are analyzed to determine the impact on livelihood strategies and what accounts for their resilience or vulnerabil ity. The roles of water and collective action are discussed for their impact on these reasons and th e opportunities for future poverty-alleviation. Household-level correlates of poverty dynamics: Chepkemel In Table C-1, Appendix C, the frequencie s of cited reasons for poverty in a random sample of 19 households are listed. While the sample sizes here are all too small to claim generalizability, the convergence of results with other studies, acro ss cases and levels of analysis strengthens the validity of findings. The most common household problem cited is the high cost of health care (quantity 8 or 42% of households sampled); however seven out of eight (88%) of those identified with this problem are downwardly mobile toward a low equilibrium state, with three households having actually moved below the lo cal poverty line and three others sitting at stage six just perched above it. In most cases these households also are characterized as possessing either a small parcel of land or unpro ductive land, indicating an initially high asset poverty line, enhancing their vulner ability to stochastic shocks. In response to medical expenses, these households sold land or liv estock, such asset decumulation symptomatic of a structural transition to poverty. This problem and its association with local poverty dynamics are highlighted to introduce a discussion about hum an illness conspicuously absent from this analysis thus far. Medical poverty traps are an increasing phenomenon linked to the introduction of user fees for public health care se rvices and rising costs of priv ate care driving households into poverty and increasing the depth of povert y for those poor already (Whitehead et al. 2001). The economic reforms in Kenya in the 1990s included re form of the health care sector, indeed setting
118 up these conditions for medical poverty traps to em erge. The Nyando River basin is plagued with several forms of illnesses including malaria, HIV/AIDS, diarrheal diseases, and respiratory and skin conditions. HIV/AIDS prevalence is high and variable across the basin estimated inflection rates of 28% in Nyando distri ct and 12% in Kericho district (Mungai et al. 2002) reflecting the same patterns as poverty distribution. Contesting state statistics claiming only a 3% morbidity rate from malaria, Cohen (2004), in a study of case villages also within the lower Nyando River basin, calculated the rate as a pproximately 49%. In Chepkemel, out of 31 households interviewed 24 (77%) re port an incidence of malaria, often recurring more than once during the year and caught by more than one person in the household. By contrast, Miolo respondents report malaria incidence in only 4 out of 26 households (15%), which seems strikingly too low. However duri ng households visits, sleeping nets to protect against malaria were frequently seen with one report of them being provided for free by a local organization perhaps accounting somewhat for this low figure. Illness reduces labor productiv ity in relation to all livelihood activities whether on-, offor non-farm, frequently placing households on a pa thway toward low-level equilibrium such as those witnessed in Chepkemel. (For a detailed discussion of the specific impacts illness has on livelihood activities in the Nyando River basin, see Cohen (2004).) Aside from the demands illness places on household cash for treatment expe nses, the timing of these needed purchases also can exacerbate its effects by competing with agriculture input purch ases, payment of schoolrelated expenses or between-h arvest food purchases. This pushes households toward underinvestment, one of the defining features of pove rty traps, along with the lower productivity and future transmission of poverty that will result In the Nyando River basin, illnesses such as malaria often overlap with labor-intensive peri ods in the farming calendar with a significant
119 impact on agricultural productivity, food security and the possibility of having agricultural surpluses to sell. In Chepkemel, one househol d interviewed commented on his wifes malaria actually during this field visit (June 2007) and coinciding with when the household was most dependent on the market for food purchases, farm-produced supplies having being exhausted and the new crop not yet ready for harvest. This coincidence placed addi tional pressure on the household to sell livestock, which would be difficu lt to repurchase given the farms small plot size and topography, features th at had already been compensated for by a mixed intercropping strategy spaced so close together it was difficult to enter the field. It also reduced her productivity as an off-farm casual laborer, a common female activity that makes substantial contribution to household wellbeing (livelihood security) as is reinforced in many of the cases subsequently described. Putting another face on a medical poverty trap is Silvanuss2 household, which has made a structural transition from being above the pros perity line to below the poverty line, located on the steep slopes of Chepkemel, very near the pr imary spring used by villagers, on land that is very rocky and unproductive. The lo cation of the household within th e village thus helps to raise its asset poverty line and has increased the poten tial for resource-degrading effects of poverty given the households proximity to the spri ng and the shadow of compensating land use behaviors. The house is a mud, one room hut with old thatch and a leaking roof. The household plants maize and little else, and was divested of its livestock more than ten years ago (just after user fees were introduced) to pay medical expens es for the now-deceased parents and sister of the current household head, age 40. Silvanus no longer works and is reported to be a drunkard, while his wife works as casual labor off-farm earning the households cash. 2 All names of individuals have been changed to protect their privacy.
120 The differences between these two households, which appear to account for the currently stochastic drop toward poverty in the first case and the structural decline of the second, are land size and quality, livestock ownership, and th e husbands alleged drunkenness. By contrast, households that are structurally not-poor manage to avoid the push from health care costs toward decapitalization by having larger resources of productive land, livestock ownership, being engaged in cash-crop production and ample labor supply, either domestic or hired. Still in Chepkemel, correlation analysis of the village data shows there to be no relationship between poverty and assorted household demographic va riables including the sex of the household head; the age of the household head; or the size of the hous ehold, the latter despite claims by some respondents that too large a fa mily can induce poverty. For instance, out of 79 resident households 18 can be described as on the pathway to low-level equilibrium and structural transition, yet the ages of househol d heads range from 25 to 80 years old suggesting the economy is not working for a broad cross-section of rural smallholders. Their declining welfare, instead, is attributed to asset divestiture due to medical expenses, drunkenness, or unproductive land. Households which remain at a stable high-eq uilibrium point (an alternative way to view category D, never poor households) share in common ownership or access through rental to large amounts of land (more than two acres appeared to be critical). So endowed, these households are able to carry on a diversified crop production including a combina tion of staple and cash crops, and own livestock. This was found to be true ev en in widowed, female-headed households where these women, Benta and Pauline, engage the mark et in various ways fr om selling to local crop brokers (Benta) or acting as a broker herself (P auline). Neither household receives remittances; instead, they are self-supporting. Both women have adult children who as sist in the farming
121 enterprise, the farms are geographically well-situated, and either large enough to still be supportive once subdivided among married sons or not needing subdivision for daughters who have relocated to their husbands households. Pauline, with the larger of the two farms, has provided secondary education to her three sons with the last one still in attendance, while the older two work on the farm earning prosperous livelihoods having not been pulled to migrate based on their education level. Paulines povert y dynamics record indicates an upward mobility in her stages of progress now that the secondary education expenses have been reduced to one child. She did experience a slight, temporary drop in wellbeing this spring before my visit due to ceremonial expenses for her younge st son, but the steady, diversif ied income of the household soon returned it to its usual high-equilibrium state. Another household that maintains itself at high equilibrium (stage 9) is that of Charles (age 28) and his young wife, Peris. Charles took over part of hi s fathers farm, which he operates with great savvy. Maize is the pr imary crop grown and productivity is high average of 20 90 kilogram sacs per acre, the proper yield accordin g to MOARD, but twice that reported by other farmers interviewed due to fertilizer use and no intercropping on the main plot. Just prior to this fieldwork, a maize buyer from the capital, Na irobi, had purchased the entire crop as green maize, an indication of the degree of agricult ure market inclusion of some households in Chepkemel. This early harvest means Charles can possibly plant a third crop of maize this year two crops per year is a normal practice maki ng use of new, quick-growing maize varieties available on the market, but this early off-take of the first crop puts three within reach altogether an indication of the intensity of farming in Chepkemel. Other crops grown include beans, sweet potatoes and bananas. A stand of napier grass is also produced as fodder for the households cattle, an activity introduced by MOARD which helps households to retain livestock
122 in areas where land and grazing area is scarce. The farm is so productive Charles is able to support his parents, his wife a nd his younger brother, and also manages to put some cash into savings. The household factors that account for th is positive outcome are the size of the farm approximately, a minimum of 2 acres of planti ng area I found to be th e tipping point between stability in wellbeing versus vul nerability to decline its to pography, and Charles hard-work and capacity. There are no reports of illness or other shocks to the household draining financial resources or other conditions lowering labor productivity. While Charles and Peris have no children of their own yet, the fact that the fa rm supports extended family means its success cannot be attributed to a small family size. Inquir es as to possible waterrelated constraints on the success or adoption of livelihood activities elicit ed the response that it was only difficult, in terms of quantity, with respect to livestock spra ying (vaccinations). Households in the village do not support the group operating a local cattle dip because, they argue, the service should be provided free of charge. This household is the best example found of se lf-reinforcing growth based on favorable initial conditions, enabling it to remain thus far structurally secure in the face of exogenous changes. In contrast, I turn to the case of Rich ards household, Charles brother and neighbor. Initial conditions are roughly the same in te rms of land size and location, but their poverty dynamics are opposite with Richards household ha ving dropped from a reported stage 8 in 1992 to stage 2 in 2002. What accounts for this bifurcat ion, this difference in experience between the two households? The village focus group attributes the households decline to the subdivision of land for Charles, lower farm productivity because of Richards increased drinking and family medical expenses necessitating the selling off of livestock. In this household Leah, Richards wife, was interviewed. Their living quarters is a two room mud and that ch hut in very poor
123 disrepair, markedly different from the polis hed, multi-building, tin-roofed homestead of the younger Charles. They have two small children who a ttend pre-school, essentially daycare, which frees Leahs time for onand off-farm wo rk. There are different plots of maize, millet, beans and some vegetables, but the spacing is very close and weeds abound. No fertilizers are used because of a lack of cash thus maize yields average 10 sacs per acre, half of that produced by Charles next door. There is also a plot of tea plants, nearing five years old, and beginning to produce. The household has a cow a nd three calves meaning there is an on-farm supply of milk but no dairy sales. Leah works when she can on th e farm, but she is mainly working off-farm as a tea picker to smooth income and consumption l eaving her time constraine d to tend to activities such as weeding. Her time is also partly cons trained due to water collection from the nearby spring, making three trips daily taking roughly a total of one hour. Her cash income from tea picking is a large proportion of the farm income (estimated at 24% per an num) and is vital to wellbeing, the household being deeply cash constraine d in months when this seasonal income is low, forcing the household to reduce food consump tion. Tea yields are lowe st in December and January hence tea income is also lowest in the same months when input purchases are highest for planting for the long rains growing season. The different outcomes in terms of agricultu ral productivity and wellbeing between these two households appear to stem in part from the differences between the two male household heads, Charles being hands-on and more hard-working versus Richard, who is reported by the villagers to be a drunkard. Whet her or not his drinking is a response to increasing poverty induced by asset loss (land and lives tock), it has contri buted to the decreased productivity of the farm in the same fashion as can illness or deat h of a major income earner. Such low productivity means the household is not self-supporting and Leah compensates for this situation by working
124 as casual labor. Casual labor is interpretable as a response to a lack of credit access and the overall lack of liquidity combin ed with the immediate need to smooth income and consumption, linking it to larger, structural economic problems. Two of the most common mechanisms described which exclude households from credit markets include the need to produce collateral, usually in the form of a title deed and difficult for sons starting up new farms on lands subdivided from their fathers, and terms of repayment that do not align with the planting calendar meaning the principle is due before the crop is harvested and sold. There is also the perception that there is isolation from credit ma rkets here in Chepkemel despite the high degree of market integration in other ways thus representing a failure in this sector of the Kenyan economy the Agricultural Finan ce Company (AFC), a credit provid er to the smallhold farmer, was another parastatal agency of the government divested duri ng adjustment that has yet to recover as a private enterprise. The women of Chepkemel, in particular, have responded to this lack of credit availability and seasonality of incomes with merry-go-rounds a popular and apparent ly ubiquitous revolving credit-type system where members make regular c ontributions and take tu rns receiving the kitty. The women interviewed reported using their income from this sy stem in important ways to smooth household consumption, purchasing everyt hing from household goods to clothing and paying of school-related expenses. Some men spoke n with have started their own versions of these as has the village elder, his with a lo w weekly contribution to make it accessible to everyone in the village. Mary used her cash receip t to purchase a cow, an investment which has reversed the stochastic, downward mobility of her househol d. Mary and her husband, Joseph (age 30), have three boys in prim ary school and live on a small farm (1.5 acres) with a mud and tin-roofed house. They plant small stands of mai ze, beans, bananas and sweet potatoes, selling
125 surpluses of the last three but retaining all the ma ize produced for household consumption, which does not last the full year between harvests. They have also planted a small area of tea, which is beginning to produce. The low maize harvests of the previous two years forced the household to sell every one of its 25 chickens to avoid what they described as starvation. The topography of the farm, its small size and minimal input usage means cropping activities only earn the household a small income, making it vulnerable to exogenous shocks such as these recent weather-related drops in producti vity. Mary works with Leah as a tea picker, this income accounting for an estimated 32% of household cash income per annum. The cow recently had a calf meaning the household has an on-farm surplus supply of milk and dairy sales now add to the households cash income. The stock of chickens is also being built back up. Despite their temporary drop into poverty that required asse t decumulation, this household has reestablished itself at a stage 8 through Marys actions. Altogether these case histories help to identify features of microlevel poverty traps in Chepkemel. Correlates of poverty dynamics that stand out are things that reduce labor productivity, especially illness. A household may experience this as either a stochastic or structural shift into poverty de pending on initia l asset conditio ns; higher wealth levels can promote faster recovery. One example was di scussed of a household that experienced a stochastic fall into poverty because of weather-relat ed declines in yield over the past two years. A small and/or unproductive farm (rocky or sloped) are two features of vulnerable households that can cause downward mobility and induce behavi oral responses such as the divestiture of livestock and zero or minimal input usage. A tr agic dimension of poverty that appears to be exploding in Chepkemel is alcoholism, accelerating the descent into poverty by reducing labor productivity and liquidity even more. Exclusi onary mechanisms that limit opportunities for
126 growth in the village are spatial and economic The local differences in topography and land availability and quality suggest a local geographical (spatial) povert y trap exists in Chepkemel, which reduces a households resilience to exogenous shocks such as the introduction of health care fees and variability in w eather. The most significant ec onomic exclusion prevailing in Chepkemel is the inaccessibility of credit markets, which means the initial asset conditions of a household have to be very high in order to avoid a structural tr ansition into poverty. Local collection action groups have served important functions in Chepkemel. There are four recorded womens self-help groups in the villa ge; one in particular ha s played a pivotal role in introducing a new livelihood activity, the pr oduction of tea seedli ngs, which is being increasingly adopted by male and female farmers. Success of these seedling nurseries depends on an available water supply hence is one opportu nity for improved water development to be poverty-alleviating. These womens groups also operate merry-go-round credit schemes, which help households contend with liquidity issues for the purchase of farming inputs or other cash needs. These womens groups have also engage d in cooperative farming of maize and sweet potatoes. Other groups have helped to construct local schools and churches and rehabilitate local roads, although no collective act ion on water infrastructure development was reported. These groups suggest how women might use their freedup time in future should the burden of water collection be lifted. Cooperation from government agencies, especially the Ministries of Education, Social Services and Health, on challe nges facing local households is also considered extremely important supporting residents with servic es and supplies. The Ministry of Agriculture is not viewed as important to the villagers as th e other agencies because of the lack of extension services it offers. Instead, extension and cr op education has been assumed by non-governmental organizations, but they have reportedly only bene fited a few households in the village. The local
127 tea factory is viewed as extremely important, pa rticularly for the school fee loans it extends to tea-growing households. Household-level analysis of poverty: Miolo From Table D-1, the reasons provided to e xplain poverty dynamics in Miolo diverge in many ways from Chepkemel. In Chepkemel severa l reasons are identified that involve factor endowments and labor productivity as compared with Miolo where the emphasis is on the quality and availability of labor. These reasons converg e with the livelihood anal ysis results regarding the importance of formal employment as a m eans to avoid poverty: 8 out of 10 (80%) of upwardly mobile households re port formal employment income. In terms of household characteristics, this importance is expressed in the emphasis on education 80% of cases on an upwardly mobile path are described as educated linked to the ability to gain a non-farm job versus households described by a lack of edu cation where 100% of cases are fixed below the poverty line. Constraints due to land quality or size are not cited as reasons for poverty in Miolo reinforcing the results that show on-farm liveliho od activities rarely enable a household to not be poor in this context of hydrol ogic vulnerability; from the sa mple only 1 out of 8 upwardlymobile households is principally a farmer. In Miolo, correlation analysis of the village data shows there to be no relationship between poverty and assorted household demogr aphic variables including the sex of the household head and the age of the household he ad, which ranges for households either at or moving toward low-level equilibrium from between 23 to 80 years. As with Chepkemel, in part this indiscrimination reflects the wider, struct ural failures in the Kenyan economy rather than endogenous life cycle effects. In this village, however, size of house hold is statistically correlated with poverty although in only one case is it (having too small a family) explicitly mentioned as a reason why a particular household is poor. Household size is also locally
128 impacted by the taking in of AIDS orphans. Povert y due to high health care costs and illness are found in 4 out of 10 households. Death of a majo r income-earner has oc curred in 6 out of 10 (60%) of households now below the poverty line, 5 out of 6 (83%) of these formerly identified with intermediate stages of wellbeing befo re this occurrence indi cating a strong, direct relationship with structural tr ansformation. The greater importa nce of death over illness and illness-related expenses in Miolo versus Chepkemel reflects the difference between HIV/AIDS and malaria effects. Whereas the critical role women play in household wellbeing in Chepkemel is evident from the amount of time spent with them and th eir willingness to disc uss their lives openly, womens voices were less heard in Miolo providing only a mo re secondhand understanding of their role and agency. Interviews were mos tly with male household heads in Miolo, not by design or omission but simply because of access granted. Interviews were conducted with two widowed women now managing their own households and whose stories illuminate the meaning of death of a major income ear ner. Since being widowed, Esthe rs (age 38) stage of progress dropped in the last ten years from above the prosperity line to ben eath the poverty line and continues to gravitate toward a low-level equilib rium. She has three children in primary school under the age of 15 years. She ow ns three connected plots of land, located away from the Awach River and the threat of being inundated by river su rges during heavy rain events. She plants one crop per year of maize and sorghum with the assistance of male neighbors who plow her fields, but she cannot afford to use any fertilizers or other inputs. Esther owns a couple of sheep and goats plus the usual chickens. Her depth of poverty was indicated not only by the visual cues around her dwelling a brick, iron-roof house speak ing to her former level of wellbeing but also through her use of water in stead of milk to make tea for her guests, an embarrassment she
129 repeatedly apologized for emphasizing the psychological impacts of poverty. Esther reported earning cash income via some small market trad ing of woven mats, th e revenue from which could not support the level of hous ehold expenses she detailed. Ruth (age 65) is also a widow with only a small piece of land where she cultivates a mixed homegarden to provide the household with food, which she never sells. Her main source of cash income is the weaving of sisal baskets, the raw material being harvested from plants located around the village plante d to protect against soil eros ion. Ruth, like Esther, is now classified as structurally-poor with a stage of progress at le vel 2 down from 8 ten years ago. During that time all the livestock, excluding a few chickens, has been sold. Both women are threatened by extreme vulnerability to their livelihoods. The agency these two women have in the face of their poverty is limited, excluded from the formal job market by their lack of education and, for Ruth, age. Their poverty has effectively pushed both women outside of the market economy meaning they are excluded from any reforms enacted or market-based forms of collective action that might otherwise benefit th em. Narrowing attention to the central question the SCALES project asks can improved water management help to pull these women back above the poverty line by enabli ng them to participate in more remunerative liv elihood activities would mean focusing on developing their on-farm potential. Esther has land assets for crop production but substantially lacks the human and fina ncial capital to invest in it, household-scale constraints beyond the prevailing ecological vulnera bility. Irrigation could help to overcome the variability and insufficiency of rainfall, but only through gender-neutral programs that also mobilize labor and credit. Improving the crop yiel ds for both womens subsistence crops (maize and sorghum as staple grains would be most im portant) through the use of inputs would increase their water productivity and lower their food in security with positive effect on nutrition.
130 Facilitating access to inputs specifically fertilizer markets, and promoting high-value crops are two possible on-farm solutions. Esther and Ruths situations contrast with the four households interviewed as representatives of a high equilibrium steady state. These households were deliberately chosen based on the fact that their livelihood strategies do not include formal employment because of the CPWFs overall interest in the relationship between increa sed agricultural water productivity and improved livelihoods. While ove r the last ten years all of th ese households have remained between poverty and prosperity in the interme diate stages of wellbeing, their pathways are moving in opposite directions, trends for which e xplanations are now sou ght beginning with the two households that indicated upward mobility for the period 1992-2002. Benjamin (age 34) and Richard (age 45) ar e brothers, each with their own farms of approximately the same land size (2.5 acres). Benj amins farm is located very near the Awach River and his plots are being eroded away, a pr ocess he is attempting to slow by maintaining sisal hedges along field perimeters. He owns anothe r plot further away from the river and rents a third on much higher ground out of the village. Staple crops, including maize and sorghum, are produced on the two owned fields, with minimal a pplications of fertilizer. The household owns a lot of livestock, a combination of oxen, cows sheep, goats, donkeys and chickens. The normal selling of livestock (that is, not a divestiture of livestock induced by shocks and the need to raise cash) and a chicken brokering business are ma in livelihood activities. On the main farm Benjamin occasionally rents the local generator to irrigate his tomato crop, a new activity which is becoming a steady flow of weekly cash during th e off-season months. The rented field is used to grow other cash crops including groundnuts an d greengrams, which in 2006 contributed an estimated 44% to annual cash income. Due to re peated flood events in early 2007 that inundated
131 this rented field, Benjamin was unable to plant these cash crops, which will push the household into a vulnerable position as the year closes out. This livelihood vulnerability is being compounded by a current outbreak in livestock diseases, which has forced the closure by MOARD officials of livestock ma rkets. Constance, Benjamins wife, contributes to household cash income with a small trading business of food stuffs operating out of her home and by working as a casual off-farm laborer on upper basin farms during weeding season. Constance belongs to a local, women-only, merry-go-round al so prevalent in this village using her income to buy childrens clothes and household goods. Given the absence of financial credit markets and overall low agricultural returns, Benj amin also will take on casual labor to make-up cash shortfalls in the household, pr eferring this activity to the forced decumulation of livestock. In summary, Benjamins household possesses a high level of assets, human and physical capitals in particular, that have enabled it to obtain a moderately upwardly mobile growth path in the direction of prosperity based on diversified farming. Events in the time leading up to this fieldwork, however, have substantially increased the vulnerability of th e household, undoubtedly reversing this direction. Push f actors are exogenous to the hous ehold, specifically, the hydrologic vulnerability of the households location lim iting investment and productivity compounded by a lack of credit access. It is antic ipated that asset leve ls are high enough that Benjamins family should remain above the poverty line in the near term despite recent events, but some decumulation of assets is likely to occur. Richards livelihood activities are broadly similar to Benjamins, centered on diversified crop farming. His three fields are not as expos ed to flooding, yet productivity remains low because of poor soil fertility and zero fertilizer use given the lack of available cash to purchase. As is a common response in Chepkemel and Mi olo, Richard tries to maximize his productivity
132 per unit area by planting mixed crops on very close spacing (Figure 52). All staple crop production is retained for household consumption, ex cept in cases of emergency when it will be sold to meet urgent cash needs. Annual seed purchases are made by selling young livestock. In early 2007, Richard was forced to sell some of hi s livestock (goats and sheep) because of dire food insecurity and hunger. He has also harves ted his small cassava plot, viewed often like livestock as a form of savings, because maize yields have been so low in the last five years due to persistent drought conditions. Most of the cu rrent cash income is coming through dairy sales from his three lactating cows. During this visit, Richards wife, Ruth, was in ill-health and could no longer continue her trading bus iness. While it is unknown to what extent she contributed to household cash income, given the wide-sprea d importance womens work has for livelihood security her poor-health is likely pushing the household towa rd poverty. Should Richard be forced to divest more livestock to pay medical bills, the already visible insecurity of the household is likely to worsen. Shifting the spotlight to the two househol d cases on a downward path of wellbeing, Ericks (age 66) household has abund ant physical assets in land a nd livestock. He farms a mix of maize, sorghum and groundnuts, retaining the enti re yield of the first two staple crops for household consumption and selling the groundnut harv est. While not located near the river, his fields are commonly inundated due to heavy rain s and flooding. The late and heavier than normal rains during the 2007 planting season necessitate d repurchasing maize seed and replanting the field three times, putting a strain on cash resources with no access to cred it. Fertilizers are not applied to fields because of their cost, but some livestock spraying is done to control for diseases albeit not on a regular schedule again because of the lack of cash. To reduce the impact of flooding, part of one field was bermed-up allowing a small area of the crop to grow, while the
133 remainder is a complete loss with implications for liv elihood vulnerability (Figure 4-9). On-farm income mostly attains from chicke n and sheep sales. More livest ock are sold in years of hunger or to pay for emergencies. In previous years, Erick had earned a small income from plowing fields for farmers without the capacity to do it th emselves, but this activity is less in demand now as households are increasingly pushed out of cr op farming. The largest co ntributor to household income is earned by the Ericks wife, Matilda, as an off-farm casual laborer three days per week for sugarcane and maize growers in the Rift Va lley (Kericho district). Matilda walks to these farms, an average of two hours in each direction. On her return trip, she stops to collect fuelwood from the hill slopes above Miolo, in effect contributing to the de forestation and, consequentially, worsen the water transitions from the upper basin. Both Erick and Mathilda participate in merrygo-round groups using the income to purchase h ousehold goods. Three of their six sons are living away from home with rela tives who are paying for them to be educated. Otherwise, the household receives no remittance income. Of the rema ining sons, the eldest one is disabled and, after being in a state institution, has returned to live at home. The last household case to present belongs to Moses (age 60) and his family, which migrated to the village from a nearby city where Moses had formal employment. Since relocating, in the last ten years the household has dropped in its wellbeing and now sits just above the poverty line. The household farms a combin ation of staple crops, entirely retained for household consumption, and a small amount of cash crops, greengrams and cowpeas. Livestock assets include oxen, cows, goats, sheep and chicke ns, with one or two of the smaller ruminants sold per year. Pamela, Moses wife, is the la rgest contributor to household income through a combination of off-farm casual labor and fire wood sales (Figure 5-3) Moses is a trained carpenter and finds the occasional bit of work locally. He does not travel outside of the village to
134 practice his trade. Pamela belongs to a lo cal merry-go-round group, which helps to build up household assets by investing the money in young livestock. Two adult sons have migrated away but do not send remittance income, only household goods, often bringing clothing when they come to visit. There appears to be no identifiable factors contributing to the households recent decline in wellbeing (from stage 10 to stage 7 on village SOP list) othe r than wider economic and geographical conditions. Ther e are no reports of major illne sses in the household; family size, with three adults and an infant in th e household, appears not to be an overburden on resources; and there are consid erable livestock assets. Limited crop productivity means the household has a high dependency on the market fo r food purchases balanced against a low level of income. If hunger becomes a pressing problem, the household does have more livestock than some to sell to alleviate immediate suffe ring with longer-term consequences. Assimilating these case histories to construc t an understanding of micro poverty traps in Miolo reinforces the significan ce of water, in terms of agroecological conditions and hydrologic vulnerability, in explaining local poverty dynamics both stochastic and structural. Floods and droughts push households toward lo w-level equilibrium states, where the critical assets to prevent this push from becoming permanent involve the ability to diversif y into multiple income sources and self-insure against risk, primarily through human capital. Importantly though, the poverty dynamics data analyzed here show that diversification, while nece ssary, is not sufficient to attain upward mobility. The prevalence of disease in the area, specifically HIV/AIDS causing the death of major income earners, has re duced household human capital resources and productivity. Due to the low productivity of the land and, hence, low remuneration levels of onfarm activities, physical assets provide littl e protection. Livestock pr ovides more livelihood
135 protection but not as much as in Chepkemel given the overall, more pronounced vulnerability context forcing its divestiture to pay for food and health care. Tweaking the term a bit, Miolo can be descri bed as experiencing a high degree of social scarcity related to water. Soci al scarcity is defined by the CPWF (2005, p. 11) as a lack of social adaptive capacity to changes in water supply and demand. Some local behavioral adaptations have been reported to try to contend with variability in water supply including land management or conservation farming practices (b erming), land rental, small-scale irrigation, and barrier planting. However, for all the households where these practices have been adopted, they have not facilitated upward mobility or growth nor protected households from being pushed further down. Concomitant to these are behavi ors induced by poverty for example, earth mining, firewood collection, and livestock watering destabilizing shorelines which magnify the vulnerability of the landscape and cause more degradation. The challenge in Miolo is to increase th e adaptive capacity of households to their hydrological vulnerability in ways that foster resilience and improve wellbeing. The SCALES project proposes to do this thr ough the fostering of soci al capital and collec tive action to enable participation in more remunerative livelihood ac tivities. Examples of local collective action include three womens self-help groups, which operate as merry-go-r ounds, facilitating the payment of school and medical expenses, and ha ve purchased goats and chickens for member households. The elders self-help group is considered the most important of these groups, involved in an on-going cooperative garden project, initially launched to pay funeral expenses, and the construction of the local primary school Churches are of central importance to the villagers for the support and social safety net they provide in times of distress. None of the local groups were reported to have done any organized collective action around water in the village.
136 The main motivation for joining groups was stated to be that it is the on ly way to benefit from external assistance. Indeed, there are five non-governmental organization reported to have worked in Miolo, providing a variety of assist ance ranging from credit and microfinance to education in tree planting. The orga nization that was to assume the task of agricultural extension from the Ministry of Agriculture, which was repo rted to have never visited the village, also failed to provide services. The most important gove rnment agency to villagers was the Provincial Administration, which has acted as a conflict mediator and security provider during ethnic clashes and will provide support during times of crisis, includ ing illness. The Nyando County Council was also recognized as an important institution, however, the village felt isolated from it and its services. Opportunities for local colle ctive action to be po verty-alleviating in Miolo include the replication of the community-garden project, im prove its performance, and place an emphasis on homegarden-style vegetable production as an appr oach to address the nutrition and food security non-monetary dimensions of poverty. This woul d also lower dependency on market purchases for food, which would help to lower cash needs. For those members of the current community garden project, however, this collective action has not resulted in upw ard mobility or growth, echoing the findings of Adato et al. (2006). The groups facilitating the accumulation of livestock seem better poised to do so, assuming the livestock market once again opens up. If crop production is to be promoted as a povert y-alleviating strategy then, as Barrett et al. (2001) highlight, structural exclusions from agricultu ral extension and support must be overcome. For example, the SCALES project might try mob ilizing additional credit access and marketing support as ways to help promote higher-value crops, developing local initiatives already being experimented with including aquaculture and tomato production, and thus potentially materialize
137 growth (Speelman et al ., 2006). Because initial w ealth levels and motivations for joining groups commonly vary across gender lines ( e.g. Davis & Negash, 2005; Kari uki & Place, 2005), market access policies will have to be so crafted to target the poor, women, and other marginalized groups. The ultimate viability of this sort onor off-farm approach to improving wellbeing is questionable, however, based on results on the im portance of higher education linked to formal employment and high-level equilib rium states within the lower Nyando River basin, converging with those of Mathenge & Tschirley (2007) who show the necessity of higher education in low potential agriculture zones of Kenya to break ou t of poverty traps by shif ting into the non-farm sector.
138 Table 5-1. Stages of progress for Chepkemel and Miolo villages. Darkened lines represent location of poverty line (after stages 5 and 6 respectively) and prosperity lines (after stages 10 and 11 respectively). Stage # Chepkemel (village no. 4) Miolo (village no. 12) 1 Food Food 2 Clothing Local chicken 3 Repair shelter, rethatch roof (1 room hut) Clothes 4 Purchase local chicken Rethatchi ng roof or hut repair (max 2 rooms) 5 Primary education Purchase goats or sheep 6 Purchase goats or sheep Primary education 7 Rent land (0.5-1 acre maximum) Farming by hired oxen (1-2 acres) 8 Purchase local cattle/cow Purchase local cattle 9 Construct a semi-permanent house with iron roof, mud walls, 2-3 rooms Secondary education for children 10 Secondary education Build semi-permanent house, 2-3 rooms, mud walls, iron roof 11 Purchase land (0.5-1 acre maximum) Purchase ox plough, oxen 12 Plant cash crops (tea, coffee, sugar), 0.51 acre maximum Purchase land 13 Construct permanent house (stone brick, iron roof) Build rental houses 14 END OF LIST Build permanent residential house Table 5-2. Regression results on reasons identified for poverty and whether household was poor or not poor. (Source: SCALES project.) Reason Coef. Std. Err. z P> z 95% Conf. Interval Mean Farm income, diversified farming -0.3254 0.2501-1.30.1930-0.8156 0.16480.2791 Child-related indebtedness -0.3433 0.2394-1.430.1520-0.8125 0.12600.1744 Drugs, alcohol 1.0213 0.29623.450.00100.4409 1.60180.1434 Old age or disabled 0.3150 0.24521.280.1990-0.1655 0.79560.1473 Large family size -0.4672 0.2797-1.670.0950-1.0155 0.08100.1202 Formal employment -1.7397 0.4455-3.900.0000-2.6128 -0.86650.3372 Death & funeral expenses 0.7369 0.20803.540.00000.3293 1.14450.3178 Polygamy, domestic violence -0.7161 0.3266-2.190.0280-1.3561 -0.07600.1163 Operating small business -1.5090 0.5527-2.730.0060-2.5922 -0.42580.1085 Constant 0.0001 0.18940.000.9990-0.3712 0.3714 LR chi square (9) 144.49, l og likelihood = -102.46489, Pseudo R2 = 0.4135
139 Table 5-3. Percentage influen ce of specified reason on probab ility of being poor. (Source: SCALES project.) Reason Percentage change in the probability of being poor minus these reasons Formal employment 45.1 Operating small business 30.2 Polygamy, domestic violence 19.2 Large family size 13.6 Child-related indebtedness and loans 10.5 Farm income from selling surpluses, diversified farming 10.2 Old age or disabled -11.1 Death or major income earn er & funeral expenses -25.8 Drugs, alcohol -38.0 Figure 5-1. Prototypical Stages of Progress. 1. Food 2. Clothing 3. House repairs 4. Purchase chickens 5. Primary school education 6. Purchase furniture 7. Purchase sheep or goats 8. Build a semi-permanent house 9. Purchase cattle sp. cross-breed 10. Secondary or tertiary education for children 11. Expand cash crop farming or horticulture 12. Purchase or rent more land 13. Build a permanent house. Poverty line Prosperity line
140 Figure 5-2. Intercropping of mai ze, sorghum and greengrams on Miolo farm. Figure 5-3. Firewood collection.
141 CHAPTER 6 SUMMARY, CONCLUSIONS AND TRAGE DY: THE WATERSHED BECOMES BLOODSHED Innately within the SCALES project and its approach to in tegrating watershed management is a multidimensional conceptualization of scale: temporal, spatial, and social. To be effective and equitable, the argument goes, watershed management will have to reflect and coordinate across these realities. This necessitates understanding the historical and social as well as natural (environmental) factors that define the landscape to be managed; otherwise, assuming an ahistorical and acontextual stance is to eclipse the human or social side of the system. The implication of this is the real ization of poverty environment tr adeoffs. This research project was an attempt to operationalize the CPWFs Theme 2 conceptual framework to uncover the holistic relationship between wate r, poverty and livelihoods and, in so doing, discovered that endogenous (plotor farm-level) factors alone do not explain wellbeing outcomes in the Nyando River basin. Exogenous, contextual factors involving the institutiona l and vulnerability contexts also play a determining role. This multi-scale ap proach helps to build knowledge of poverty traps at various degrees of resolution thusly can be used to develop efficient and targeted policy. Drawing on this research, it is time to consider: In the Nyando River basin can improved water management help to lift households above the applicable SOP poverty line and stimulate asset accumulation and growth away from low equilibrium points, the e ssential question of the project yet an outcome that was found to be untru e in SCALES research locations in Columbia (Johnson et al. 2006)? Would improved water manageme nt place households on an upwardlymobile path ( i.e. water management as livelihood promotion) or, at least, establish a safety net to prevent a decline in wellbeing ( i.e., water management as live lihood protection)? The chapter begins with a synthesis of information from which to draw conclusions about this question. It then makes a brief foray into a discussion of the opportunities for colle ctive action around water
142 management to be poverty-alleviating in the Ny ando River basin based on the content of this analysis. It ends with some reflection sadly i nduced by the post-election violence that struck western Kenya in winter 2007-08. Synthesis and Conclusions The case analysis of Miolo village in the lo wer basin shows a str ong relationship between water, poverty and livelihoods due to geographica l conditions and landscape vulnerability. Land use in the upper Nyando River ba sin, after 100 years of intensive and extensive agriculture, is argued to be causal in the observed changes in wa ter quality and quantity in the lower basin and for the environmental degradation of th e lowlands and Lake Victoria (Mungai et al. 2002). Lateral flows from the upper to the lower basin have magnified already marginal agronomic, lowland conditions such that non -irrigated crop farming as a singular livelihood activity is not sufficiently remunerative for the smallhold farm er. Even irrigated crop production has become challenged through issues related to siltation, insecure property rights and changes in market conditions; yet, opportunities exist for multi-sc ale collective action to resolve governance and maintenance issues and to return these system s to operation. Households are more asset-poor since the economic reforms of the 1990s, effect ively further lowering their already meager resilience and increasing their vulnerability to exogenous shocks such as climate, which coincidently got worse. These shocks came in the form of floods then dr oughts covering the last five years of the study period. Low asset levels, expressed primarily in human and financial capital terms, mean low investment in liveli hood activities as well as environmentally-degrading behaviors while geographical conditions mean low productivity. In sum, self-reinforcing water poverty traps have emerged. It is important to recognize, however, that these poverty traps are not solely the product of the na tural system; changes in legal and economic institutions the
143 human system contributed signif icantly to this process. More over, this is true beyond the borders of Miolo and applies to th e entire lower basin. The relationship between water, poverty a nd livelihoods for smallholders in the upper Nyando River basin is less clear cut. The area receives ample ra infall to produce high yields, which could be more productive if inputs were mo re widely and/or effi ciently used. Credit market failures, cost-sharing and other economic reforms, and low crop pr ices constrain water productivity and force coping behaviors that lower overall productivity. Population growth and density, attracted in part by the comparativel y rich resource (water ) endowments, and the resulting land scarcity mean farming activities are pushed into clearing marginal and vulnerable lands, increasing the potential for negative ex ternalities downstream. Once again, the three hallmark signs of a poverty trap are visible low investment, low productivity and resource degradation however, in this context, they do no t pivot around water or water scarcity, at least from a local perspective. However, the national problem of water scarcity requires increasing the water productivity of an area wh ich receives the majority of Ke nyas renewable supply. Policies and actions facilitating increasi ng returns to onand off-farm labor and thereby water would have a positive effect on livelihoods mostly serv ing as a safety net, but careful, conservation farming techniques would be required to arrest future water transitions. Environmental protection, if not rehabilitation of the lowlands and Lake Victoria would need to remain state priorities along with agriculture sector growth, such that the tertiary node, hearkening back to the conceptual framework, acts as an effective force supporting reverse flows on behalf of lowland inhabitants to mitigate the power of the Rift Valley water tower. Indeed, the multifaceted challenges of poverty alleviation, system rehab ilitation and national wellb eing converging in the Nyando River basin are well articulated by Swal low (2002, p. 29) as how to maximize the off-
144 take of above ground carbon (harve st of vegetation as crops, forage and wood) while maintaining sufficient vegetation cover to minimize adverse consequences of soil eros ion, nutrient depletion and eutrophication of Lake Victoria. Despite these very different geographical contexts, the pus h and pull factors identified vis--vis micro-level poverty were similar (Barrett et al ., 2001). Pushing households toward lowlevel equilibrium states were issues pertaining to human capital, including illness or death and the costs associated with health care, and alc oholism; the lack of credit and low returns to agriculture combined as general liquidity constrai nts; the lack of higher education (lower basin); and lack of factor endowments es pecially land (upper basin). Social organization into local selfhelp groups, particularly by wo men and manifesting most comm only as merry-go-round credit schemes to address the lack of liquidity, are playing a key role in lowering household vulnerability, effectively shifti ng the asset poverty line left (Figure 2.2) with the potential to enable participation in new livelihood activities and asset accumulation. These groups took on important roles as leaders in de veloping new activities such as t ea nurseries in Chepkemel. This fieldwork was not long enough to de velop a full sense of any social exclusion from groups such as these, but it is a reasonable hypothesis that, depending on the re quired investment level, the poorest households can be excluded potentially replicating Adato et al .s (2006) finding that social capital works better for non-poor hous eholds. Cultural questions raised by Roy et al. (2005) on the fear men have about the empowe rment of women via these groups, which may hinder their formation and, hence, thwart possible improved livelihoods, is a topic that too merits further investigation. The much smaller number of cases of households (classified as SOP category B) which have managed to exit poverty in the last decade makes it harder to identify factors pulling them
145 out of poverty. Further, it makes it harder to de termine the forms of colle ctive action that would facilitate mimicking them; yet, from the househol ds worked with intensively on the PRA, it was found that stochastic transitions were reversed by the acquisition of liv estock facilitated by womens cash income and group membership. Form al employment and diversification emerged as necessary but often not sufficient livelihood stra tegies that also enable asset accumulation and improved wellbeing. However, there would appear to be various economic and perhaps social exclusionary mechanisms that k eep the poor and marginalized out of these strategies, which are likely not entirely with in the scope of local collectiv e action to effectively remedy. Through this type of analytical, also describable as an ecolo gical approach, the situation of the Nyando River basin comes to be viewed as a sort of perfect storm. It was the layering of history, geography, globalized macroeconomy, polic y, culture and population concentrated on an inherently vulnerable landscape, which combined to produce some of the most extreme poverty and environmental conditions within the entire Nile River basin. Speaking personally for a moment of reflexivity, I confess th at when I started this researc h, my attention was fixed solely on conditions at the farm scale as if I believed that households had complete autonomy and agency to change their own trajectory. I was ye t to engage with the concept and meaning of integrated watershed management. Fortunately, I was given the opportunity to visit the Nyando River basin to develop familiarity with the context as I embarked on analyzing a dizzying dataset and collecting my own. What happened was that I be gan to listen, to hear the histories of the people who reside there, and to realize that to understand their circumstances meant I must understand the subjectivity of the Nyando River basin itself, as something beyond a natural landscape rather also as a social product. Globalization, a process that has unequivocally impacted the lives within and history of the Nyando River basin not to mention the lingering
146 effects of colonization and oppression, which are s ubjects left for another day is noted for its ability to deterritorialize space, meaning to em pty it of its history and culture, to suppress difference, and to reform relations of produc tion (Gregory, 2007). While this task is not complete, what I hope I have done here is to lay out a rough sketch of the Nyando River basin as just such a coupled human-and-natural system to explain its poverty dynamics over the period 1977 2002. If you, the reader, are convinced, I will have kept my pledge to the people I met to shine a light on the whole trut h of their experience. No one ever asks for the truth, is the tr anslation of a pithy comment made by Rose, a blind and generous, elderly woman, who, in the mont h before my arrival, had buried the last of her 12 children, two adult daughters deceased unspokenly due to AIDS. She and her husband, John, live on a small, unproductive farm in Miol o. John had lost his formal government job as part of economic restructuring. Their depth of poverty was the most severe I witnessed and was particularly notable because of the sense of powerlessness Rose and John displayed. They scrapped by on their meager harvest, their few chickens, and the ge nerosity of neighbors. It is difficult to imagine how collective action around improved water management might alleviate this degree of poverty, it probably can t, but I turn now to consider briefly the opportunities that might exist more generally. Opportunities for Collective Action via Water Management to Alleviate Poverty A full empirical assessment of the local oppor tunities for collective action for improved water management based on institutional analysis (Ostrom, 1990) falls outside the ambit of this research project. Instead, it is the explicit purpose of the next stage of the SCALES project in Kenya, the conduct of experimental economic games (Cardenas & Carpenter, 2005). That methodology will be testing local understanding of upstream-downstream interdependencies (specifically, in biophysical terms of hydrology and water transitions) and will try to show
147 participants the rationality of decisions to cooperate on resource management based on the presumption of improved livelihood outcomes. Howeve r, it is argued that it is necessary also to ask whether the appeal for cooperation and cost-s haring across scales, espe cially socially, is legitimate in this tribal context. In the African context generally, where moral obligations weaken as concentric social circles widen, would the Luo, Ki psigis and Kalenjin peoples occupying different regions of the basin accept c ooperation as a legitimate obligation particularly after a history of conflict? (K elsall, 2008) This is just one question that needs further investigation as part of a study of the cultural politics (Alvarez et al ., 1998) in the Nyando River basin to make a full assessment of the opportuniti es for collective action and equitable watershed management. Nevertheless, throughout the text experiences with and pot ential opportunities for collective action have been described. Extrapol ated from these findings, Table D-1, Appendix D, is an attempt to synthesize the correlates of poverty dynamics, materialized poverty impacts and household responses, bifurcated based on wealth level, to speculate on some opportunities for collective action via water management to be po verty-alleviating within the Nyando River basin. These are focused mainly on endogenous factors in accordance with the CPWFs statements. Final conclusions about the SCALES projects hypothesis regarding the ability of improved water management, or increased water productivity in the argot of the CPWF, via collective action to alle viate poverty by opening up more remune rative livelihood activities in the Nyando River basin are as fo llows: First, the hypot hesis appears to refl ect and desires to replicate the successful Kenyan experience with decentralized resource management through community organizing, arguably st arted in Kenya in 1977 with Wangari Maathais Green Belt movement. Local, village-based collective action opportunities exist to imp rove water access and supply that would help to offset the unreliabi lity of rainfall, lift the burden on womens time
148 devoted to water collection, and im prove health. People express inte rest in farming a variety of high-value crops based on the resulting secure wa ter supply, strategies, however, which require market reforms. Indeed, villages surveyed by the SCALES project from across the basin equally express their willingness to pa rticipate in collective action groups to develop local water resources through the construction of, for inst ance, boreholes or wa ter tanks by providing unskilled labor, cash, and other re sources, although it is not clear benefits from these activities would be inclusive of the poor. These groups are presently hindere d in their emergence because of a lack of leadership and or ganization skills, technical knowledge and the unaffordability of all the materials required, necessita ting external partnerships. Conversely, the projects attempts to develop social capital thusly cr eate opportunities for inter-village cooperation on water management s eems more difficult for cultural and political reasons. Issues of trust, recipr ocity, legitimacy and the same un ity of purpose felt toward local collective actions are real impediments. A second challenge, therefore, is how to link collective action across scales to manage an entire watershe d in an integrated and inclusive way, balancing the variety of water users and uses within. Instit utional diagrams revealed there to be a mix of governmental and non-governmental organizations working in the villages indicating a degree of cross-scale action. Concerns were raised, however, as to the exclusivity of some, isolating the village entirely or some social groups only. Improved performance of these organizations is critical to the success of local self-help groups. As a safety net or manner of livelihood pr otection, improved water productivity has the potential to slow or arrest the descent into poverty by incr easing returns to land and labor. Whether these increases can outpa ce population growth or the incr easing prices of consumables and inputs will determine if, in fact, rural well being actually improves. It may only stave off
149 vulnerability rather than increase resilience or promote growth. The estimated 33% price increase in fertilizer between the 2006 and 2007 planting seasons (Chepkemel farmer, personal communication) forced even the wealthiest of households to reduce or eliminate its use indicating how severely the ratio between input and output prices has narrowed (crop revenue had not changed over the same period, in fact it ha d declined due to lower yields). This example is cited to illuminate anecdotally why, once again, water management is a necessary but not sufficient means to improve wellbeing. Policies th at target and extend market access for various marginalized groups must accompany water if gain s, measurable in terms of asset accumulation, are to materialize. Collective action for water management is no t an assurance of changing labor relations nor of restoring citizenship in te rms of social protection, therefore, may be limited in its ability to mobilize households out of poverty. While comm unal activities that im prove water supply and quality are meaningful for addr essing food security, nutrition and human health and, thusly, are unquestionably, unassailably valuable they seek to remedy only the defining features of abject poverty. Unsupported with wider systemic change, these activities cannot promote livelihoods or move households away from vulnerable, low-leve l equilibrium states. Issues of population size and density and concomitant land sc arcity are the critic al issues facing the upper basin that feel well-beyond the scope of local action alone. The degree of environmenta l degradation and the hydrologic vulnerability of the lo wer basin, including worsening climate conditions, pose vast challenges to inhabitants and the Kenyan governme nt. Balancing social and environmental needs will require substantial cooperation a nd investment across scales. Epilogue: Violence Erupts in Western Kenya In December 2007 national elections were he ld in Kenya. The challenger, Raila Odinga, to the incumbent, Mwai Kibaki, is Luo, fr om western Kenya. Talk of the election was
150 everywhere during my time in Kenya that summ er and I was impressed with how politically engaged everybody was. When the election was d eclared won by Kibaki despite evidence of voter fraud and election-rigging, et hnic violence ensued in areas ju st to the north of the Nyando River basin and spread to urban centers. A field report from SANA (2008), the SCALES projects local NGO partner tasked with the actual fostering of social capital between four different tribal communities, assessed the damage thusly: The post election violence which divided the country into tribal divide made th e three communities in this location turn on each otherthe tension is still ve ry highand 1 person was shotAll the areas have severe economic effects andthe trust and co-existence has been interfered w ith and rebuilding will have to start afresh [ sic ]. This tragic outcome and the frailty of diffuse social cap ital fostered as a resource for environmental management rein forces comments above about the need to understand the cultural politics of the Nyando River basin. What is significant about this tragic turn of events is the sense of depth of political and economic exclusion felt within areas within the Kenyan Lake Victoria dr ainage basin, of which the Nyando River basin constitutes a part. The fa ct that this sparked a peasant uprising is noteworthy and merits detailed investigation for its significance upon watershed management. As water becomes increasingly scarce and, conseque ntially, this water-rich area assumes even more strategic importance, is peasant uprising likely to become a more frequent occurrence? How would IWRM as a governance t ool need to be amended to c ontend with these possibilities? These are questions proposed for future research.
151 APPENDIX A LOGISTICAL FRAMEWORK
152Table A-1. Logistical framework for CPWF, the SCALES project, and this research project. Narrative Summary Challenge Program on Water and Food (CPWF) Theme 2: Water and People in Catchments SCALES Project (CPWF PN20) SCALES (Kenya): Understanding the link between water and poverty in the Nyando Basin Goal To produce more food using less water. To improve sustainable livelihoods for people who live both in upper catchments and downstream, through improvements in water productivity. To contribute to poverty alleviation and sustainability in the upper watersheds of the tropics through improved collective action for watershed management within and across social and spatial scales. 1. To better understand the relationship between water and poverty in the Nyando Basin; 2. To identify opportunities for reducing poverty via collective action around improved water management. Purpose or Objective To create and disseminate international public goods that improve the productivity of water in river basins in ways that are pro-poor, gender equitable and environmentally sustainable. Address upstream-downstream interdependencies by: 1. Enhancing the capacity of basin communities to improve their collective use of water resources in support of rural livelihoods; 2. Shifting focus of management from farm to basin scale; 3. Seeking equity in distribution of costs and benefits from improved management. 1. To understand the role of water and other natural resources in the livelihoods of the poor in upper watersheds; 2. To understand factors that affect the emergence, performance and equity of collective action; 3. To use knowledge gathered to strengthen the poors access to resources and participation in decisions about how those resources will be managed to enable the escape from poverty traps. 1. To test empirically the ability of the conceptual framework, developed in Output #1 of SCALES, to explain the link between water and poverty in the Nyando Basin, Kenya; 2. To fulfill Output #2 of SCALES by describing the poverty traps of the Nyando Basin, Kenya; 3. To assess the potential for collective action around water management to facilitate escape from these poverty traps; 4. To prioritize sites, identify the poor, the needed activities and factors for development practitioners. Research Outputs Innovations to address the needs of rural poor through increasing water productivity: 1. Technologies and management practices; 2. Institutional mechanisms and partnership arrangements; 3. Policy recommendations. 1. A deeper understanding of the relationship between water and poverty; 2. Improved knowledge of how upstream land management affects downstream water availability; 3. Institutional and policy innovations to improve land and water management at di fferent scales; 4. Developing, adapting and scaling up technical and institutional innovations for sustainable and equitable catchm ent management. 1. Conceptual framework for analyzing water, poverty and scale; 2. Identification of poverty traps at different scales across a catchment; 3. Understanding of the determinants of collective action at different catchment scales; 4. Watershed action plans; 5. Recommendations and tools for policymakers and development practitioners; 6. Capacity building. 1. Identification of the livelihood strategies of the poor and critical constraints for breaking out of poverty traps at different scales; 2. Characterization of the livelihood systems across the basin; 3. Existing water productivity levels from a comparative case study of two villages; 4. Fulfill in part the scoping part of the CPWF basic research process for SCALES (Kenya). Principle Research Question Using less water, how can more food be produced, and rural livelihoods be improved, in a manner that is socially acceptable and environmentally sustainable? What is the nature of water-related poverty in primary catchments? To what extent can it be alleviated through better water management? Where are the poor? What is their relationship with environmental degradation? Can improved watershed management rooted in collective action help alleviate poverty by increasing the availability or productivity of water, opening up new livelihood strategies and leading poor households and communities out of poverty traps? What is the relationship between water and poverty in the Nyando Basin? What opportunities exist for collective action to help the poor escape from poverty traps in the Nyando Basin?
153 APPENDIX B STATISTICAL RESULTS ON LIVELI HOOD ACTIVITY PARTICIPATION
154Table B-1. Percentages of households participating in livelih ood activities (or activity groups) in years 1992 and 2002. Village no. % in poverty (1992) % in poverty (2002) Cash crop/ commercial agriculture Formal employment Horticulture Informal employment Livestock Maize Other subsistence agriculture Small business 1992 2002 1992 2002 1992 2002 1992 2002 1992 2002 1992 2002 1992 2002 1992 2002 3 8.5% 4.7% 47.3 52.7 3.1 8.5 3.1 5.4 20.2 26.4 79.8 74.4 100 98.4 1.6 3.9 9.3 10.1 20 14.0% 13.0% 3.8 82.7 0.0 7.7 0.0 100 0.0 42.3 0.0 96.2 0.0 63.5 1.9 90.4 7.7 75.0 8 4.7% 7.1% 53.4 35.0 10.3 13.8 0.0 0.0 96.6 100 0.0 0.0 6.9 0.0 0.0 0.0 1.7 6.3 14 33.2% 37.4% 80.6 78.5 22.6 14.0 16.1 15.1 32.3 33.3 104.3 123.7 80.6 76.3 82.8 76.3 8.6 8.6 6 20.9% 25.2% 56.3 33.8 35.2 35.2 0.0 5.6 23.9 47.9 16.9 18.3 54.9 47.9 8.5 2.8 5.6 9.9 15 0% 4.5% 3.8 6.3 1.3 2.5 36.7 36.7 68.4 83.5 38.0 88.6 72.2 74.7 135.4 135.4 17.7 26.6 1 3.1% 10.4% 4.8 2.8 22.2 21.1 0.0 0.0 6.3 11.3 31.7 43.7 77.8 100 0.0 0.0 7.9 8.5 11 11.0% 21.0% 28.4 11.9 26.5 19.8 24.5 34.7 28.4 40.6 125.5 95.0 77.5 60.4 83.3 58.4 23.5 31.7 13 25.0% 36.5% 78.8 32.7 19.2 21.2 3.8 7.7 23.1 21.2 61.5 50.0 76.9 63.5 80.8 50.0 30.8 34.6 4 4.2% 16.9% 31.5 26.0 9.6 13.7 0.0 0.0 21.9 28.8 95.9 72.6 100 89.0 2.7 2.7 20.5 16.4 7 7.3% 21.9% 79.3 75.0 9.8 6.5 28.3 23.9 58.7 68.5 25.0 20.7 47.8 35.9 21.7 18.5 41.3 31.5 2 0% 20.0% 82.1 74.3 7.1 8.6 10.7 2.9 28.6 25.7 71.4 65.7 85.7 100 85.7 100 10.7 11.4 18 20.4% 40.4% 61.7 43.8 8.3 10.9 5.0 6.3 18.3 32.8 100 100 63.3 60.9 76.7 59.4 6.7 15.6 5 14.1% 36.4% 0.0 2.0 0.0 0.0 2.0 6.0 50.0 42.0 56.0 50.0 98.0 94.0 2.0 2.0 12.0 14.0 10 4.8% 29.0% 93.7 68.3 20.6 9.5 146.0 152.4 25.4 38.1 160.3 122.2 74.6 73.0 60.3 31.7 52.4 36.5 9 35.4% 62.5% 75.3 4.1 8.2 5.5 2.7 5.5 61.6 64.4 31.5 30.1 9.6 20.5 5.5 5.5 42.5 38.4 12 24.8% 56.4% 0.0 1.0 11.7 9.9 3.9 3.0 58.3 69.3 49.5 31.7 53.4 37.6 137.9 91.1 31.1 18.8
155Table B-2. Percentages of households part icipating in each livelihood activity or su ite of activities disaggregated by the app licable SOP category over the period of 10 years to now (1992 2002). Chepkemel data are represented in the top table and Miolo in the lower. Percentages are presented for two points in time, 10 years ago and now, to show the dynamics in livelihood activity. Chepkemel (N=73) A: chronically poor (N=2) B: escaping poverty (N=1) C: entering poverty (N=11) D: never poor (N=59) 10 years ago now 10 years ago now 10 years ago now 10 years ago now Casual labour 100% 100% 9.09% 45.45% 20.34% 16.95% Formal employment 9.09% 9.09% 10.17% 11.86% Livestock 50% 100% 100% 90.91% 45.45% 96.61% 77.97% Maize 100% 100% 100% 100% 54.55% 100% 94.92% Other subsistence crops 3.39% 3.39% Cash crops 9.09% 38.98% 30.51% Trading business 9.09% 25.42% 18.64% Skilled labour 9.09% 1.69% 5.08% Miolo (N=102) A: chronically poor (N=19) B: escaping poverty (N=6) C: entering poverty (N=38) D: never poor (N=39) 10 years ago now 10 years ago now 10 years ago now 10 years ago now Casual labour 78.95% 52.63% 83.33% 66.67% 28.95% 73.68% 23.08% 33.33% Formal employment 16.67% 15.79% 2.63% 15.38% 20.51% Livestock 16.67% 31.58% 18.42% 20.51% 20.51% Maize 57.89% 31.58% 66.67% 50.0% 55.26% 36.84% 48.72% 38.46% Sorghum 52.63% 26.32% 66.67% 50.0% 52.63% 31.58% 46.15% 38.46% Other subsistence crops 36.84% 26.32% 50% 50.0% 42.11% 18.42% 33.33% 23.08% Cash crops 10.53% 10.53% 16.67% 33.33% 39.47% 15.79% 30.77% 30.77% Trading business 26.32% 10.53% 16.67% 10.53% 5.26% 15.38% 7.69% Skilled labour 15.79% 15.79% 16.67% 16.67% 26.32% 13.16% 35.9% 38.46%
156 Table B-3. Chepkemel logistic regression results modeling th e relationship between livelihood activities and poverty status (poor versus not poor) as the dependent variable. Independent Variable Unstandardized Logistic Regression Coefficient (b) Standard Error of b Statistical significance of b Exponentiated coefficient eb Standardized Logistic Regression Coefficient Maize -3.047 1.109 .006 .048 -1.042 Livestock -.437 1.013 .666 .646 -.203 Casual Labor .285 1.045 .785 1.329 .123 Formal Employment .325 1.240 .793 1.384 .111 Trading Business -18.745 1.265E4 .999 .000 -6.411 Skilled Labor -21.286 2.004E4 .999 .000 -4.193 Cash Crops -.466 1.192 .696 .628 -.204 Intercept 42.277 2.370E4 .999 2.294E18 --Table B-4. Miolo logistic re gression results modeling the relationship between livelihood activities and poverty status (poor versus not poor) as the dependent variable. Independent Variable Unstandardized Logistic Regression Coefficient (b) Standard Error of b Statistical significance of b Exponentiated coefficient eb Standardized Logistic Regression Coefficient Maize .839 .763 .272 2.314 .408 Livestock -.989 .837 .237 .372 -.352 Casual Labor .023 .562 .968 1.023 .012 Formal Employment -3.390 1.171 .004 .034 -1.014 Trading Business -1.398 .892 .117 .247 -.377 Skilled Labor -2.025 .641 .002 .132 -.863 Cash Crops -1.877 .860 .029 .153 -.762 Intercept 7.396 2.093 .000 1.630E3 --
157 Table B-5. Chepkemel logistic regression results modeling th e relationship between livelihood activities and entering into poverty as the dependent variable Independent Variable Unstandardized Logistic Regression Coefficient (b) Standard Error of b Statistical significance of b Exponentiated coefficient eb Standardized Logistic Regression Coefficient Maize -.678 1.018 .505 .508 -.232 Livestock -1.771 .814 .030 .170 -.822 Casual Labor -1.026 .832 .217 .358 -.441 Formal Employment -1.855 1.226 .130 .156 -.634 Trading Business -21.277 1.226E4 .999 .000 -7.277 Skilled Labor -22.024 2.220E4 .999 .000 -4.339 Cash Crops -.606 .715 .396 .545 -.265 Intercept 46.942 2.536E4 .999 2.436E20 -Table B-6. Miolo logistic re gression results modeling the relationship between livelihood activities and entering into povert y as the dependent variable. Independent Variable Unstandardized Logistic Regression Coefficient (b) Standard Error of b Statistical significance of b Exponentiated coefficient eb Standardized Logistic Regression Coefficient Maize -1.497 .724 .039 .224 -.728 Livestock 1.058 .796 .184 2.879 .377 Casual Labor .498 .514 .332 1.645 .249 Formal employment -1.948 1.136 .086 .143 -.582 Trading Business -.587 .943 .534 .556 -.158 Skilled Labor -.458 .572 .423 .633 -.195 Cash Crops .434 .799 .587 1.544 .176 Intercept 2.593 1.895 .171 13.372 -
158 APPENDIX C REASONS FOR POVERTY Table C-1. Reasons identified for povert y status based on household interviews. Group Reason Definition Basin Frequency (n=258) Chepkemel Frequency (n=19) Miolo Frequency (n=20) 1 FARM On-farm income (crop sales) 90 5 2 IDF Diversification into cash crops 25 3 2 FEE School fees 53 1 DEBT Indebtedness 1 HDEP Taking in orphans 28 2 LOAN Loans 7 3 IJW Takes drugs or is a drunkard 56 7 4 ILL Bad health, accident 38 1 4 ILLH High health care expenses 51 8 2 5 AGED Elderly 38 1 2 DISABLE Disabled 11 6 IBFS Large family size 53 4 7 JOB Has formal employment 78 1 8 IEC Is a hard worker 59 2 8 IDM Death of major income-earner 65 3 6 ICTD High funeral expenses 40 2 2 9 DOME Marital separation, domestic violence 22 3 ICTN Polygamy 23 ILG Litigation 2 10 LNSD Small land size 42 6 ICTM Wedding expenses 5 1 LNBG Land subdivision 3 11 ISFS Small family size 26 2 1 12 LJB Lost job 38 LJEMP Unemployment 21 1 13 LOWPAY Low paying job (casual labor) 42 4 3 14 NOEDU Illiteracy, lack of education 25 2 15 IED Has education 17 8 16 LAZY Is lazy 21 1 17 PB Family is rich; intergenerational wealth transmission 14 HRH Help from family 16 1 18 NAL Accidental loss from exogenous shock 28 LDCL Land use conflicts 8 19 BUSS Operates a small business 37 1 20 UNPLAND Unproductive land 4
159 APPENDIX D OPPORTUNITIES FOR COLLECTIVE ACTION: SYNTHESIS
160 Table D-1. Opportunities and c onstraints for collective action. Specific impacts attributable to water on each asset are ident ified. Capital asset Sources of limitations Poverty impacts Household Responses Opportunities and constraints for collective action Non-poor Poor Human Death of main income earner Education level Health (illness and disease) Personal behaviours ( e.g. excessive drinking) All these limitations impact every one of the livelihood outcomes measuring poverty-income, food security, well-being, sustainability and vulnerabilityby the restrictions they place on the type of livelihood strategies available and the quantity of time available for participation. Can invest in human capital (supplemental labour) to offset any household constraints. Combined with baseline levels of natural capital, households can achieve surplus production. Have the education level to pursue formal employment if chosen. Age and health effects are minimal. Cannot offset their household labour constraints with the hiring of supplemental labour. Lower education implies casual labour. Age and health effects more considerable. Usually more of the personal behaviours limiting human capital present. Cooperative farming activities or laboursharing. Requires human capital be available. Water collection time Well-being impacts given variations in volume collected Constraints on womens available human capital resources Women not compelled into so much off-farm labour that they spend more time collecting water with more volume being collected. Possible additional time spent laundering clothes and socializing positive reflection in well-being Less time spent collecting water; smaller amounts collected Less time spent at water sources to launder clothes or socialize Lift some of the constraints on womens time due to water collection. Associated gains in volume of water collected with implications for wellbeing. Reduce exposure to health-related impacts of diseases associated with water. Sanitation services may reduce health exposures. Financial No or restricted credit access Education expenses Sufficient cash income from onor off-farm sources Food insecurity: reduce, sometimes substantially, food purchases due to lack of cash especially in season when there is no surplus farm production to sell. Some participation in formal credit schemes run by parastatals (particular to households engaged in cashcrop (tea) production) Cash income close or sufficient enough to narrow the gap to No credit access. Wide deficits between cash income and household expenses resulting in material deprivation. Cannot invest or take risks in Merry-go-round groups to mobilize cash to satisfy immediate material needs or for investment in livestock already common. Cash threshold for
161 General material deprivation. Future opportunities and well-being: children may be pulled out of school or unable to get secondary education due to lack of funds household expenses. Fewer forced cuts to expenses. Cope with shocks by selling available livestock. Can invest and take risks in new livelihood strategies. Wide range of strategies available if household chooses to diversify. alternative livelihood strategies. Fewer strategies are available. Little if any livestock remains to be sold to cope with shocks. participation may be a limiting factor. Cooperative farming groups of cash crops also in operation to increase income. None related to water since it is a free good Boreholes, wells Should water fees be imposed would place added demands on cash resources. Some can afford to dig on-farm boreholes freeing up more time. No on-farm boreholes or wells Private boreholes not equitable in access. Natural Vulnerability context Assets controlled by household (e.g. land size) Farm location Vulnerability context conditions whether or not farming is a viable livelihood strategy Stock of natural assets directly influences livelihood outcomes: income from farming, food security (food grown for household consumption), demand on financial resources because of forced participation in the market economy for food Have financial resources to relocate farm production when surplus land is available Farm size large enough that household can produce a surplus or participate in cash-cropping Intercropping patterns to even out seasonality in when crops are available for consumption and sale Higher yields reduce need to go to market for food Household on most productive lands No response to vulnerability context Farm size too small for surplus production, cashcropping or to meet household consumption needs Intercropping patterns to improve fertility (no financial resources to invest in physical assets) and to maximize production; subsistence crops only Lower yields mean more market purchases or greater food insecurity Household on more marginal lands Collective farming outside of vulnerability context (again requires certain levels of human and financial capital to participate) Water quality and quantity available Flooding and erosion (part of the vulnerability context) Poor quality can result in health effects; lower well-being and constraints on human capital availability Water-borne or associated diseases decrease health and Equal exposure to quality of water and related diseases. Financial assets may help evade some of the worst effects. Equal exposure to quality of water and related diseases. Lack of financial resources makes evading effects harder. Some community efforts to reduce impacts from flooding and erosion were tried in lower Village 12 with limited effect. They broke down due to lack of participation. Incentive to participate
162 well-being; also impact human capital availability Loss of income and decreased food security probably lessened by view that livelihood security in the future is less likely to come from the land (i.e. farming) than formal employment. Physical Labour-saving means of production Storage capacity Crop inputs (e.g. seeds and fertilizers) Livelihood outcomes including income levels, food security and wellbeing. Also sustainability of use of natural resource base. More ownership of means of production. Can store surplus production (depending on crop) for sale at times of higher market value Can purchase crop inputs although perhaps: not on time for optimal planting; not insufficient quantities reinforcing the need for intercropping for fertility; may need to be re-purchased due to vulnerability related losses; livestock spraying varies with cash availability May be available on a reciprocal basis assuming household has the labour to provide in return Storage not available Seeds must be purchased as the practice of retaining seeds from previous harvest is not pursued No fertilizer use due to lacking financial resources Decreased frequency or reduced need (due to lack of livestock) for livestock spraying Reciprocal use of resources although depends on human capital availability. Location of water source, seasonal flows and quality Impacts on well-being and livelihood strategies Have physical assets to assist in water collection. More rainwater collection capabilities Fewer physical assets to assist in water collection. Less rainwater collection capacity Move water source closer to village. Sanitation services to decrease exposure to health-related problems in water quality.
163 Social (group membership) Time availability (shortages in human capital) Financial resources for cash contributions Community cohesiveness Increased availability of financial resources. When in form of cash used to invest in human capital of children (pay education expenses); purchase of physical inputs to farming (seeds, fertilizers); increase household well-being (purchase of household goods). Financial assets may also be in convertible form (livestock investments) Able to participate in groups although does depend on specific household characteristics and endowment levels as groups vary in their demands Unable to participate in groups due to time and cash poverty Lower time and cash thresholds for participation
164 LIST OF REFERENCES Adato, M., Carter, M.R., & May, J. (2006). Expl oring poverty traps and social exclusion in South Africa using qualitative and quantitative data. Journal of Development Studies, 42 (2), 226 247. Alvarez, S.E., Dagnino, E., & Escobar, A. Eds. (1998). Cultures of Politics Politics of Culture: Re-visioning Latin Am erican Social Movements. Boulder, CO: Westview Press. Azariadis, C., & Stachurski, J. (2005) Povert y traps. In Aghion, P., & S. Durlaub, Eds. Handbook of Economic Growth. Retrieved March, 2009, from http://emlab.berkeley.edu/users/chad/azstach.pdf Baulch, B., & McCulloch, N. (1998). Being poor and becoming poor: Poverty status and poverty transitions in rural Pakistan IDS Working Paper 79. Sussex: Institute of Development Studies. Barrett, C.B. (2005). Poverty traps and agricultural resear ch: Improving policies, institutions and technologies to support su stainable poverty reduction Paper presented at the Consultative Group on International Agricultur al Research Science Forum. December 6, 2005. __________. (2007). Poverty traps and resource dynamics in smallholder agrarian systems. Retrieved March, 2009, from http://www.aem.cornell.edu/faculty_s ites/cbb2/Papers/WageningenFeb2007.pdf Barrett, C.B., Bezuneh, M., & Aboud, A. (2001). Income diversification, poverty traps and policy shocks in Cote dIvoire and Kenya. Food Policy, 26, 367-384. Barrett, C.B., Carter, M.R., & Little, P.D. (2006). Understanding and redu cing persistent poverty in Africa: Introduction to a special issue. Journal of Development Studies, 42 (2), 167 177. Barrett, C.B., Reardon, T., & Webb, P. (2001). Nonf arm income diversification and household livelihood strategies in rura l Africa: concepts, dynamics and policy implications. Food Policy, 26, 315-331. Barrett. C.B., & Swallow, B.M. (2006). Fractal poverty traps. World Development, 34 (1), 1 15. _________________________. (2005). Dynamic poverty traps and rural livelihoods. In F. Ellis, & H.A. Freeman (Eds.), Rural Livelihoods and Poverty Reduction Policies (pp. 16 27). New York: Routledge. Bird, K. (2007). The intergenerational transmission of poverty: An overview. ODI Working Paper 286. London: Overseas Development Institute.
165 Briggs, S.V. (2001). Linking ecological scales and institutional frameworks for landscape rehabilitation. Ecological Management & Restoration, 2 (1), 28 35. Burke, W.J., Jayne, T.S., Freeman, H.A., & Kristjanson, P. (2007). Factors associated with farm households movement into and out of poverty in Kenya: The rising importance of livestock. MSU International Development Working Paper No. 90. Michigan: Michigan State University. Cardenas, J.C., & Carpenter, J. (2005). Experiments and Economic Development: Lessons from Field Labs in the developing world Retrieved October, 2008, from http://www.middlebury.edu/NR/rdonlyres/64DC78BC-D459-48D1-A160ECDE5B370EA2/0/0505.pdf Carter, M.R., & Barrett, C.B. (2006). The economics of poverty traps and persistent poverty: An asset-based approach. Journal of Development Studies, 42 (2), 178 199. Carter, M.R., Little, P.D., Mogues, T., & Negat u, W. (2007). Poverty trap s and natural disasters in Ethiopia and Honduras. World Development 35 (5), 835 856. CIA (Central Intelligence Agency) (2008). The World Factbook. Retrieved March 9, 2009, from https://www.cia.gov/library/publicati ons/the-world-factbook/geos/ke.html Chambers, R. (1997). Whose Reality Counts? Putting the First Last. London: IDTG Publishing. Chambers, R., & Conway, G.R. (1991). Sustainable rural livelihoods: Practical concepts for the 21st century. IDS Discussion Paper 296 Brighton, England: IDS. Cleaver, F. (2005). The inequality of social capital and the reproduction of chronic poverty. World Development 33 (6), 893-906. Cohen, L. (2004). The impact of illness on livel ihoods in rural western Kenya: The influence of livelihood type, gender, and seasonality Unpublished masters thesis, University of Florida, Gainesville. CAWMA (Comprehensive Assessment of Wate r Management in Ag riculture) (2007). Water for Food, Water for Life: A Comprehensive Assess ment of Water Management in Agriculture. London: Earthscan, and Colombo: International Water Management Institute. CPWF (Challenge Program on Water and Food) (2005). CGIAR CPWF: Research Strategy 2005-2008. Colombo, Sri Lanka: CPWF. _____. (2006). Challenge Program for Water and Food Report 2005, Theme 2: Water and People in Catchments (February 2006). Colombia: CIAT. Retrieved from http://gisweb.ciat.cgiar.org Davis, K.E., & Negash, M. (2005). Gender, wealth and participation in community groups in Meru Central District, Kenya. Paper presented for Intern ational Research Workshop on Gender and Collective Action, Oc tober 17-21, 2005, Chiang Mai, Thailand.
166 Dayal, B. (2008, November). Lessons learnt from the Practi cal Experience of Scaling up the Water Related Technologies at the Community level. Poster session presented at the CGIAR Challenge Program on Water and Food, 2nd International Fo rum on Water and Food, Addis Ababa, Ethiopia. De Groote, H., Kimenju, S., Owuor, G., & Wanyama, J. (2006, August). Market liberalization and agricultural intensification in Kenya (1992-2002). Paper presented at the 26th Conference of the Internati onal Association of Agricultural Economics, Gold Coast, Australia. de Haan, L., & Zoomers, A. (2005). Exploring the frontier of Livelihoods Research. Development and Change, 36 (1), 27 47. De Vaus, D.A. (2001). Research Design in Social Research. California: Sage Publications, Inc. Devereux, S. (2002). Can social safe ty nets reduce chronic poverty? Development Policy Review, 20 (5), 657-675. Dercon, S., & Krishnan, P. (1996). Income portfo lios in rural Ethiopia and Tanzania: Choices and constraints. Journal of Development Studies, 32 (6), 850 875. Di Gregorio, M., Hagedorn, K., Kirk, M., Ko rf, B., McCarthy, N., Meinzen-Dick, R., & Swallow, B. (2008). Property rights, collective action, and poverty. The role of institutions for poverty reduction. CAPRi Working Paper No. 81. Wa shington, D.C.: International Food Policy Research Institute. Din Ahmad, M., Islam, M.A., Mashi, I., Mut huwatta, L., Karimi, P., & Turral, H. (2008, November). Water productivity mapping to identify opportunities to improve agricultural water management in the Karkheh River Basin, Iran. Paper presented at the CGIAR Challenge Program on Water and Food 2nd International Forum on Water and Food, Addis Ababa, Ethiopia. Ellis, E., & Freeman, H.A. (2005). Conceptual framework and overview of themes. In F. Ellis, H.A. Freeman (Eds.), Rural Livelihoods and Pover ty Reduction Policies (pp. 3 -15). New York: Routledge. Gallup, J.L., Sachs, J.D., & Mellinger, A.D. (1999). Geography and economic development. International Regional Science Review, 22 (2), 179-232. German, L.A., Kidane, B., & Mekonnen, K. (2005). Watershed Management to Counter Farming Systems Decline: Toward a DemandDriven, Systems-Oriented Research Agenda African Highlands Initia tive (AHI) Working Paper # 16. Uganda: ICRAF. Gibbon, P., Havnevik, K.J., & Hermele, K. (1993). A Blighted Harvest: The World Bank and African Agriculture in the 1980s. African World Press. Gleick, P.H. (1993). Water and conflict: Fresh water resources and in ternational security. International Security, 18 (1), 79 -112.
167 GWP (Global Water Partnership) (2003). Poverty reduction and IWRM. TEC Background Paper no. 8, GWP, Stockholm. Retrieved February 2, 2008, from http://www.gwpforum.org /gwp/library/TEC%208.pdf Gregory, S. (2007). The Devil behind the Mirror: Globaliz ation and Politics in the Dominican Republic. Berkeley: University of California Press. Grey, D., & Sadoff, C. (2002, August). Water resources and poverty in Africa: Breaking the vicious circle. Paper presented at the Inaugural Meeti ng of AMCOW, Abuja, Nigeria. Retrieved from www.africanwater.org/Documents/amcow_wb_speech.pdf Gurung, H., Bhandari, H., Velasco, M.L., Keona khone, T., Pandey, S., Velarde, O., & Samson, B. (2008, November). Poverty, water, and livelihoods: A case of two upper-catchment villages in the northern Lao PDR. Paper presented at the CGIAR Challenge Program on Water and Food 2nd International Forum on Water a nd Food, Addis Ababa, Ethiopia. Haileslassie, A., Gebreselassie, S., Peden, D., Amede, T., & Descheemaeker, K. (2008, November). Impacts of access to resources on wate r productivity: the case of the Blue Nile. Paper presented at the CGIAR Ch allenge Program on Water and Food 2nd International Forum on Water and Food, Addis Ababa, Ethiopia. Hoddinott, J., & Quisumbing, A.R. (2003). Methods for microeconomic risk and vulnerability assessments: a review with empirical examples. Social Protection Discussion Paper No. 0324. Washington, D.C.: World Bank. IFAD (International Fund for Agricultural Development) (2003). Report and recommendation of the President to the Executive Board on a proposed loan to the Republic of Kenya for the Southern Nyanza Community Development Project. IFAD: Rome. Jayne, T.S., Yamano, T., Nyoro, J., & Awuor, T. (2001). Do farmers really benefit from high food prices? Balancing rural interests in Kenyas maize pricing and marketing policy? Tegemeo working paper 2B. Kenya: Egerton University. Johnson, N., Garcia, J., Moreno, A., Granados, S ., Rodriquez, H., Rubiano, J.A., Peralta, A., Rubiano, J.E., Quintero, M., Estrada, R., & Mwangi, E. (2006). Watershed management and poverty alleviation in the Colombian Andes. In Decision Support Project BP-2: Spatial and economic analysis for policy and decision s upport in agriculture and environment. Annual Report 2006. Colombia: CIAT. Kabubo-Mariara, J., & Kiriti, T.W. (2002). Structural adjustment, poverty and economic growth: An analysis for Kenya. AERC Research Paper 124. Na irobi, Kenya: African Economic Research Consortium. Kariuki, G., & Place, F. (2005). Initiatives for rural developmen t through collective action: The case of household participation in group activ ities in the highlands of central Kenya. CAPRI Working Paper 43. Washington, D.C.: CAPRI
168 Kemp-Benedict, E., Chadwick, M., & Krittasudthacheewa, C. (2008, November). Impact of water-for-food research on poverty alleviation: Bayesian methods for livelihood, water, and poverty analysis. Paper presented at the CGIAR Challenge Program on Water and Food 2nd International Forum on Water and Food, Addis Ababa, Ethiopia. Kelsall, T. (2008). Going with the grain in African development? APPP Discussion Paper No. 1. London: Overseas Development Institute. Kerr, J. (2002). Watershed development, enviro nmental services, and poverty alleviation in India. World Development, 30 (8), 1387-1400. KNBS (Kenya National Bureau of Statistics) (2007). Basic Report on Well-being in Kenya. Nairobi: Ministry of Planning and National Development. Kitchaicharoen, J., Ekasingh, B., Dithapra yoon, S., & Chaiwinit, W. (2008, November) Impact of access to water on livelihood strategies in northern Thailand. Paper presented at the CGIAR Challenge Program on Water and Food 2nd International Forum on Water and Food, Addis Ababa, Ethiopia. Krishna, A. (2001). Moving from the stock of social capital to the flow of benefits: The role of agency. World Development, 29 (6), 925-943. _________. (2004). Escaping poverty and becoming poor : Who gains, who loses, and why? World Development, 32 (1), 121-136. _________. (2006). Pathways out of and into poverty in 36 villages of Andhra Pradesh, India. World Development, 34 (2), 271-288. Krishna, A., Kapila, M., Porwal, M., & Singh, V. (2005). Why growth is not enough: Household poverty dynamics in northeast Gujarat, India. Journal of Development Studies, 41 (7), 1163-1192. Krishna, A., Kristjanson, P., Radeny, M., & Ni ndo, W. (2004). Escaping poverty and becoming poor in 20 Kenyan villages. Journal of Human Development, 5 (2), 211-226. Krishna, A., Lumonya, D., Markiewicz, M., Mu gumya, F., Kafuko, A., & Wegoye, J. (2006). Escaping poverty and becoming poor in 36 vi llages of central and western Uganda. Journal of Development Studies, 42 (2), 346-370. Legovini, A. (2002). Kenya: Macroeconomic evolution since independence. Retrieved March, 2009, from http://www.ke.undp.org Little, P.D., Stone, M.P, Mogues, T., Castro, A.P., & Negatu, W. (2006). Moving in place: Drought and poverty dynamics in south Wollo, Ethiopia. Journal of Development Studies, 42 (2), 200 -225.
169 Magombey, M.S., & Taigbenu, A.E. (2008, November) Simulation of smallholder farming systems in the Olifants River Basin, South Africa. Paper presented at the CGIAR Challenge Program on Water and Food 2nd International Forum on Water and Food, Addis Ababa, Ethiopia. Maneta, M., Torres, M., Wallender, W., Howitt, R ., Vosti, S., Rodrigues, L.N., Bassoi, L.H., Allen, S., Bennett, L., & Young, J. (2008, November). Water management across scales in the Sao Francisco River Basin, Brazil: policy options and poverty consequences. Paper presented at the CGIAR Challe nge Program on Water and Food 2nd International Forum on Water and Food, Addis Ababa, Ethiopia. Mathenge, M.K., & Tschirley, D.L. (2007, July). Income growth and mobility of rural households in Kenya: Role of education and historical patterns in poverty reduction. Paper presented at the American Agricultural Economics Association meeting, Portland, OR. Molden, D.J., Sakthivadivel, R., & Keller, J. (2001). Hydronomic zones for developing basin water conservations strategies. Research Report 56. Colombo, Sri Lanka: IWMI. Moriarty, P., Butterworth, J., & van Koppen, B. Eds. (2004). Beyond domestic: Case studies on poverty and productive uses of wa ter at the household level. Technical Paper Series 41. IRC (International Water and Sanitation Centre), Delft, the Netherlands. Mulligan, M., Rubiano, J., White, D., Hyman, G., & Saravia, M. (2008, November). Participatory modeling and knowledge integration Paper presented at the CGIAR Challenge Program on Water and Food 2nd International Forum on Water and Food, Addis Ababa, Ethiopia. Mungai, D., Swallow, B., Mburu, J., O nyango, L., & Njui, A. (Eds.) (2004). Proceedings of a workshop on reversing environmental and agricu ltural decline in the Nyando River Basin. Nairobi: ICRAF. Muyanga, M., Ayieko, M., & Bundi, M. (2007). Transient and Chronic Rural Household Poverty: Evidence from Kenya. PMMA Working Paper 2007-20 Available at http://ssrn.com/abstract=985167 Odada, E.O., Olago, D.O., Kulindwa, K., Ntiba, M., & Wandiga, S. (2004). Mitigation of environmental problems in Lake Victoria, Ea st Africa: Causal ch ain and policy options analyses. Ambio, 33 (1-2), 13-23. Okwi, P.O., Ndenge, G., Kristjanson, P., Ar unga, M., Notenbaert, A., Omolo, A., Henninger, N., Benson, T., Kariuki, P., & Owuor, J. (2007). Spatial determinants of poverty in rural Kenya. PNAS, 104 (43), 16769-16774. Ongile, G.A. (1999). Gender and Agricultural supply res ponses to structural adjustment programmes: A case study of smallholder tea producers in Kericho, Kenya. Research Report No. 109 Upsala, Norway: Nordiska Afrikainstitutet.
170 Ostrom, E. (1990). Governing the Commons: The evolution of institutions for collective action. New York: Cambridge University Press. Owoyesigire, B., Mpairwe, D., Mutetika, D., Bashasa, B., Kiwuwa, G., & Peden, D. (2008, November). Socioeconomic factors affe cting livestock water productivity in rainfed pastoral production systems. Paper presented at the CGIA R Challenge Program on Water and Food 2nd International Forum on Water a nd Food, Addis Ababa, Ethiopia. Pampel, F.C. (2000). Logistic Regression: A Primer. Thousand Oaks, CA: Sage Publications, Inc. Peralta, A., Garcia, J.A., & Johnson, N. (2006) Dynamics and definition of poverty in the Colombian Andes: Participativ e vs. objective approaches. In Decision Support Project BP2: Spatial and economic analys is for policy and decision su pport in agriculture and environment. Annual Report 2006. Colombia: CIAT. Perez, C., & Tschinkel, H. (2003). Improving watershed management in developing countries: a framework for prioritiz ing sites and practices. AgRen Network Paper No. 129. London: Overseas Development Institute. Place, F., Adato, M., & Hebinck, P. (2007). Unde rstanding rural poverty and investment in agriculture: An assessment of integrated quan titative and qualitative research in western Kenya. World Development, 35 (2), 312-325. Rahaman, M.M., & Varis, O. (2005). Integrat ed water resources management: evolution, prospects and future challenges. Sustainability: Science, Practice, & Policy, 1 (1), 15-21. Ravnborg, H. M. (2006). Water management a nd the poor: Issues a nd scales of action. Water International, 31 (3), 387 397. Reardon, T., & Vosti, S. (1995). Links between ru ral poverty and the environment in developing countries: Asset categories and investment poverty. World Development, 23 (9), 14951506. Rocheleau, D., Thomas-Slayter, B ., & Wangari, E.(Eds.) (1996). Feminist Political Ecology: Global issues and local experiences. London: Routledge. Roy, J.L., Crow, B., & Swallow, B. (2005, January). Getting access to adequate water: community organizing, women, and soci al change in Western Kenya. Paper presented at the African Water Laws: Plural Legislative Frameworks for Rural Water Management in Africa Conference, Gauteng, South Africa. SCALES. (2008). Sustaining inclusive collective acti on that links across economic and ecological scales. Poster presented at the CGIAR Ch allenge Program on Water and Food 2nd International Forum on Water and Food, Addis Ababa, Ethiopia. Scoones, I. (1998). Sustainable rural livelihoods. A framework for analysis. IDS Working Paper 72. Brighton: IDS.
171 _________. (2009). Livelihoods perspectiv es and rural development. Journal of Peasant Studies 36 (1). Speelman, S., DHaese, L., Ochieng, C., & Vandermeulen, V. (2006, August). Productive water uses at household level in rural Kenya: case study of the Ukambani district. Poster session at the International Association of Agricultural Econom ists Conference, Gold Coast, Australia. Swallow, B. (2005). Poverty reduction strategi es for addressing multi-level poverty traps: reflections from Kenya. In F. Ellis, & H.A. Freeman (Eds.), Rural Livelihoods and Poverty Reduction Policies (pp. 311 331). New York: Routledge. Swallow, B., Johnson, N., Meinzen-Dick, R., & K nox, A. (2006). The challenges of inclusive cross-scale collective action in watersheds. Water International, 31 (3), 361 375. Swallow, B., Onyango, L., & Meinzen-Dick, R. (2003). Catchment Property Rights and the case of Kenyas Nyando Basin. In Proceedings of the Afri can Regional Workshop on Watershed Management. Available at http://www.iwmi.cgiar.org/Assessment/files/pdf/publications/WorkshopPapers/Catchment PropertyRights.pdf _____________________________________. (2007). Irrigation management and poverty dynamics: Case study of the Nyando basin in western Kenya. Available at: http://www.iwmi.cgiar.org/Publications/CAB I_Publications/CA_CABI_Series/Community _Law/protected/Ch%2012.pdf Swallow, B., Onyango, L., Meinzen-Dick, R., & Holl, N. (2005, January). Dynamics of poverty, livelihoods and property rights in the Lower Nyando basin of Kenya. Paper presented at the African Water Laws: Plural Legislative Frameworks for Rural Water Management in Africa Conference, Gauteng, South Africa. Teyie P., Swallow B., & Onyango L. (2006, August). Opportunities for fost ering inclusive multiscale collective acti on for watershed management in the Nyando River Basin, Kenya. Paper presented at the Stockholm World Water Week, Stockholm. Todaro, M.P. (1977). Economic development in the Third Wo rld: An introduction to problems and policies in a global perspective. New York: Longman Inc. van Koppen, B., Smits, S., Moriarty, P., & Penning de Vries, F. (2008, November). Communitylevel multiple-use water services: MUS to climb the water ladder. Paper presented at the CGIAR Challenge Program on Water and Food 2nd International Forum on Water and Food, Addis Ababa, Ethiopia. Weingart, A., & Kirk, M. (2008). Escaping poverty traps? Collect ive action and property rights in post-war rural Cambodia. CAPRi Working Paper No. 89. Washington, D.C.: International Food Policy Research Institute.
172 Were, M., Ngugi, R.W., Makau, P., Wambua, J., & Oyugi, L. (2005). Kenyas Reform Experience: What have we learnt? KIPPRA Working Paper No. 12. Nairobi, Kenya: Kenya Institute for Public Policy Research and Analysis, Macroeconomics Division. Were, E., Swallow, B., & Roy, J. (2006). Water, women, and local social organization in the Western Kenya highlands. CAPRi Working Paper No. 51. Wash ington, D.C.: International Food Policy Research Institute. Whitehead, M., Dahlgren, G., & Evans, T. (2001). Equity and health sector reforms: can lowincome countries escape the medical poverty trap? The Lancet, 358, 833-836. World Agroforesty Centre. (2006). Improved land management in the Lake Victoria Basin: Final report on the TransVic project. ICRAF Occasional Paper No. 7. Nairobi: World Agroforesty Centre. World Resources Institute; Depa rtment of Resource Surveys and Remote Sensing, Ministry of Environment and natural Resources, Kenya; Cent ral Bureau of Statistics, Ministry of Planning and National Developmen t, Kenya; & International Livestock Research Institute. (2007). Natures Benefits in Kenya, An Atla s of Ecosystems and Human Well-Being Washington, DC and Nairobi: World Resources Institute. Yin, R.K. (2003). Case study research: design and methods, Third Edition. California: Sage Publications, Inc.
173 BIOGRAPHICAL SKETCH Jillian Jensen completed her Bachelor of Sc ience degree at the University of Toronto, Canada, in 1989, majoring in environmental sciences with minors in economics and botany. From 1990 to 1997, Jillian was the Sales and Market ing Director of one of Canadas leading interior landscaping firms, winning multiple in ternational awards. In 1997, she founded Jill Jensen Botanical Specialties, a wholesale greenhouse operation producing exo tic tropical foliage plants, distributed across eastern Canada, and continued her award-winning ways with custom design landscape projects in Canada and the United States. Jillian, inspired by her certification as a yoga instructor after ten years of study, returned to the Universi ty of Toronto in 2002 to add a minor in Religion and the Envi ronment to her Bachelors de gree. In 2006, Jillian entered graduate school at the Univers ity of Florida to develop her un derstanding of issues of poverty and the role agricultu ral development should play in so lving the plight of the poor. Jillian is an avid traveler having been around the world thre e times, touring agriculture operations and engaging with small farmers on six continents.