ORGANIC ECONOMIC ANALYSIS OF SMALLHOLDER VEGETABLE PRODUCTION SYSTEM IN KIAMBU AND KAJIADO COUNTIES OF
SAMUEL KIRUKU NDUNGU (BSC)
A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN
AGRIBUSINESS MANAGEMENT, KENYATTA UNIVERSITY
I declare that this thesis is my original work and has not been submitted or published for any award of a degree in this or any other university.
Signature……… Date: ………
Samuel Kiruku Ndungu
Recommendation by supervisors
We confirm that the work reported in this thesis has been submitted for review with our approval as university supervisors.
Dr. Ibrahim Macharia
Signature……… Date: ………
Department of Agribusiness Management and Trade
Dr. Ruth Kahuthia-Gathu
Signature……… Date: ………
Department of Agricultural Science and Technology
The successful completion of this work is highly indebted to my supervisors; Dr. Ibrahim Macharia and Dr. Ruth Kahuthia-Gathu for guidance throughout the whole process of proposal development, data collection, data analysis and thesis writing.
I am also indebted to Productivity for Growth of Organic Value Chains (ProGRoV) project under the stewardship of Prof. Raphael Wahome of the University of Nairobi for financial support and guidance through data collection, data analysis and thesis writing.
Table of contents
Declaration ... i
Dedication ... ii
Acknowledgement ... iii
Table of contents ... iv
List of tables ... vii
List of figures ... viii
Abbreviations and acronyms ... x
Abstract ... xii
CHAPTER ONE ... 1
1.0: INTRODUCTION ... 1
1.1 Background... 1
1.2 Statement of the problem and justification ... 2
1.3 General objective ... 4
1.4 Specific objectives of the study ... 4
1.5 Hypotheses of the study ... 4
1.6 Significance of the study ... 5
1.7 Conceptual framework ... 5
1.8 Scope of the study ... 9
1.9 Assumptions of the study ... 10
1.10 Operational definition of terms and concepts ... 10
CHAPTER TWO ... 12
2.0: LITERATURE REVIEW ... 12
2.1 Introduction ... 12
2.3 Organic agriculture development in Kenya... 14
2.4 Current status of organic production in Kenya ... 14
2.5 Economic analysis ... 15
2.6 Determinants of farm economic performance ... 16
2.7 Variables affecting profitability of organic production systems ... 17
2.8 Methodologies and techniques of comparing organic and conventional production systems 18 2.9 Impact evaluation of organic production system ... 25
2.9.1 Impact evaluation using propensity score matching ... 26
2.9.2 Matching algorithms ... 28
2.9.3 Gaps in literature ... 29
CHAPTER THREE ... 30
3.0 MATERIALS AND METHODS ... 30
3.1. Introduction ... 30
3.2 Study area ... 30
3.3. Research design ... 33
3.4 Target population ... 33
3.6 Pretesting survey tools ... 36
3.7 Data management and analysis ... 37
CHAPTER FOUR ... 48
4.0 RESULTS AND DISCUSSION ... 48
4.1 Introduction ... 48
4.2 Descriptive statistics comparing organic and conventional farmers. ... 48
4.3 Factors associated with adoption of organic production system ... 51
4.4 Factors associated with profitability of investment in organic production system ... 55
4.6 Impact of organic farming on smallholder vegetable farm profitability ... 59
4.6.1 Estimation of propensity scores ... 59
4.6.2 Impact evaluation ... 62
4.6.3 Evaluation of psm quality indicators ... 67
4.6.4 Sensitivity analysis ... 68
CHAPTER FIVE ... 70
5.0 CONCLUSIONS AND RECOMMENDATIONS ... 70
5.1 Conclusions ... 70
5.2 Recommendations ... 71
5.3 Areas for further study ... 73
List of tables
Table 2.1: Summary of reviewed methodologies and techniques ... 19
Table 3.2: Summary of sampling design ... 36
Table 3.3: Variable definitions ... 37
Table 3.4: Summary data analysis ... 39
Table 4.2: Factors associated with adoption of organic vegetable production system ... 52
Table 4.3: Factors associated with profitability of investment in organic vegetable production system ... 55
Table 4.4: Gross margin analysis of organic and conventional production vegetables ... 58
Table 4.5: Maximum likelihood estimators for factors impacting on profitability of organic vegetable production system ... 61
Table 4.6: Impact of organic vegetable production system on profits ... 63
Table 4.7: Quality indicators across matching logarithms ... 67
List of figures
Figure 1.1 Conceptual Framework Adapted from Nemes (2010) and Offermann and Nieberg (2000) ... 8 Figure 3.1 Map of study area ... 31 Figure 4.1: Distribution of Propensity scores on region of common support using KBM, Nearest Neighbour and Radius matching ... 60 Figure 4.2: Cost analysis for organic and conventional production system per unit acre ... 64
List of appendices
Appendix 1: Correlation matrix of coefficients for testing multicollinearity for logit
regression estimators ... 80
Appendix 2: Goodness of fit diagnostic for logit regression estimators ... 80
Appendix 3: Summary characteristics of logistic regression estimators ... 81
Appendix 4: Normality and multicollinearity test for OLS estimators ... 81
Appendix 5: Correlation matrix test for OLS estimators ... 82
Appendix 6: Swilk test for normality of OLS explanatory variables ... 82
Appendix 7: Shapiro test for normality of explanatory variables ... 83
Appendix 8: Kernel density graph for normality of OLS residuals ... 84
Appendix 9: Linktest for model specification ... 84
Appendix 10: White's test for homoskedasticity ... 84
Appendix 11: Cameron and Trivedi's decomposition of IM test ... 85
Appendix 12: Instrumental IV regression variables ... 85
Abbreviations and acronyms
ATE Average Treatment Effects CBO Community Based Organizations
CBTF Capacity Building Taskforce on Trade and Development CV Coefficient of variation
CSHEP Community Self Help Empowerment Program EAC East African Community
ETE Estimated Treatment Effect FADN
Farm Accountancy Data Network
FAO Food and Agriculture Organization
IFOAM International Federation of Organic Agriculture Movements KIOF Kenya Institute of Organic Farming
KNBS Kenya National Bureau of Statistics KOAN Kenya Organic Agriculture Network MOA Ministry of Agriculture
NGO Non Governmental Organization OLS Ordinary Least Squares
ProGrOV Productivity and Growth in Organic Value Chains PSM Propensity Score Matching
ROA Return on Assets ROE Return on Equity
xi UNON University of Nairobi
1.0: INTRODUCTION 1.1 Background
Agriculture has historically been the source of livelihood for a majority of population in Africa where more than 75% of the population practice subsistence and traditional agriculture which rarely relies on purchased inputs (Lyons and Burch, 2007; Rundgren and Lustig, 2007). Despite the need for increased food production to feed increasing population in Africa, agricultural productivity has declined over the years and is currently 2-3 times lower than the world average (FAO, 2006). The increasing costs of inputs coupled with low agricultural products market prices have made rural populations be in a poverty trap caused by the farming system, its commercial pressures and marketing obligations (UNEP-UNCTAD CBTF, 2010). The widely practiced conventional production system has also contributed to increased soil erosion, environmental pollution and degradation as well as loss of indigenous crop diversity and continued poor health among the people (FAO, 2002; UNEP-UNCTAD-CBTF, 2010).
waste, compost manure, agroforestry and ecological pest management (IFOAM, 2002; FAO, 2006). It offers particularly developing countries a wide range of economic, environmental, social and cultural benefits (UNEP-UNCTAD-CBTF, 2008). In Kenya organic farming has been promoted by more than 30 NGOs which have trained more than 200,000 smallholder farmers with less than 2 ha land holding (Mwaura, 2007; KOAN, 2010). However, the number of certified organic farmers is 12,647 representing less than 0.01% of the farming population (KNBS, 2010; Willer and Lukas, 2010; UNEP-UNCTAD CBTF, 2011). In addition land under organic management in Kenya is partly 1.0% of the total organic certified arable land in Africa (Willer and Lukas, 2010) which is 0.36% of the total arable land area in Kenya (KOAN and MOA, 2010). The low adoption of organic production system may be attributed to lack of adequate skills and knowledge in organic farming technologies, high cost of certification, low market development, lack of organic agriculture policy, poor post harvesting handling and processing (Mwaura, 2007; KOAN and MOA, 2010). The problem is exacerbated by lack of strong institutional support in organic agriculture education, research and development (Mwaura, 2007; UNEP-UNCTAD CBTF, 2008). There is therefore a gap in research, policy, and technology to be filled for increased adoption and growth of organic agriculture in Kenya.
1.2 Statement of the problem and justification
Fowler, 2002). This limits information available to inform pro-organic policies, organic technical advisory services and increasing a body of knowledge available in organic agriculture (UNEP-UNCTAD CBTF, 2008). For instance, only few studies evaluate comparative benefits of organic versus conventional production systems in Africa (Bolwig, Gibbon & Jones, 2009; Nermes, 2009). In general, many studies which evaluate economic benefits of organic production system recommend its adoption as a more profitable production system (Offermann and Nieberg, 2000; Canavari, et al., 2004; Bolwig et al., 2009; Argiles and Brown, 2010).
Economic benefits are one of the key motivating factors of converting to certified commercial organic production (Cobb et al., 1999; Pimentel et al., 2005; Kerselaers et al., 2007; Bolwig et al., 2009). With availability of information on benefits of converting to organic farming, farmers will adopt the technology due to the various benefits accrued (Parrott and Elzakker, 2003).
there is only 30 and 41 smallholder farmers certified to produce organic vegetables in Kajiado and Kiambu counties of Kenya (KOAN, 2010; ENCERT 2010). This is despite the proximity of large and increasing market for organic products in Nairobi (Ndungu, 2013). The study therefore was set to contribute to the knowledge gap of the economic benefits of organic vegetable production as a motivation for its adoption.
1.3 General objective
The study aimed at analyzing organic production system in order to establish its contribution to profitability of smallholder vegetable farms in Kajiado and Kiambu counties of Kenya.
1.4 Specific objectives of the study
The study had the following specific objectives:
1. Analyze the factors associated with adoption of organic vegetable production system in Kiambu and Kajiado counties of Kenya.
2. Analyze factors associated with profitability of investment in organic vegetable production in Kiambu and Kajiado counties of Kenya.
3. Estimate the impact of organic production system on profitability of smallholder vegetable farms in Kiambu and Kajiado counties of Kenya.
1.5 Hypotheses of the study
1. Social economic, farm and market characteristics are associated with adoption of smallholder organic vegetable production system
2. Social economic, farm and market characteristics influence profitability of organic smallholder production system.
3. Organic production system has an impact on the profitability of smallholder vegetable production systems in Kiambu and Kajiado Counties of Kenya.
1.6 Significance of the study
The study provides information to extension officers, field officers and value chain actors on impact of organic farming system on profitability of smallholder vegetable production system. This contributes to information available on benefits of adopting this system among smallholder farmers in the counties of Kajiado and Kiambu. It also provides smallholder farmers with the necessary information on profitability of organic vegetable production system to facilitate making informed decisions when converting to organic farming. By determining the effect of social economic, farm and market characteristics on profitability of smallholder vegetable production system, the study assists providing additional way of increasing profit margins in smallholder vegetable production systems and value chains.
1.7 Conceptual framework
profitability of smallholder organic vegetable farmers and social economic and farm charactersitics on one hand and costs and reveue factors on the other hand.
Impact evaluation indicates whether changes in the wellbeing in a community after an intervention is indeed due to programme interventions and not to other factors (Khandker et al., 2010). The study followed the reflexive comparisons where through ex post evaluation a programme impact is examined across the difference in outcome of participants and counterfactual of no particiaption. Due to lack of counterfactual in this case, evaluation therefore involved data for participants (Organic Farmers) and non partipants (Non organic farmers) as well as other social and economic factors that may have determined the course of intervention (Khandker et al., 2010).
factors also referred to us market environment include transaction costs, market channel and market prices.
Furthermore, the variables that affect organic farm profitability after conversion as identified by various scholars were analyzed as 1) Potential yield loss (Pimentel et al., 2005; Zanoli et al., 2007; Nermes, 2009), 2) Change in land use due to rotational requirements (McCrory, 2001; Shadbolt et al., 2009), 3) Price variation for organic products (Pimentel et al., 2005), 4) Managerial difficulties and cost variation for purchased inputs (Cobb et al., 1999), and 5) Variation in labour cost (Pimentel et al., 2005) and Marketing Channel (Offermann and Nieberg, 2000).
of selection bias from unseen differences between control and treatment groups since the treatment is not randomized as observed by other authors (Nicolette, 2011; Khandker etal., 2010; Lu, 2005). Following the work of Rosenbaum and Rubin (1983), propensity score is therefore used for constructing matched sets which are comparable in terms of pretreatment characteristics as advanced by Zhao (2008) and Frolich (2004) enabling us to estimate the impact of investing in organic production system. This relationship can be illustrated as follows:
Figure 1.1 Conceptual Framework Adapted from Nemes (2010) and Offermann and Nieberg (2000)
gender, occupation) and farm characteristics (number of parcels, irrigation, ownership and location) and market channel. The study therefore investigated these relationships and demonstrated the impact of organic production system on profitability of smallholder vegetable production system.
1.8 Scope of the study
The study collected data from all organic smallholder farmers producing vegetables for domestic market in Kiambu and Kajiado Counties of Kenya. Though organic farmers are required to adhere to organic principles they are also certified for compliance with organic standards. For this study, certification was considered as an indicator of compliance with organic standards and adherence to principles. Farmers who were under conversion1 were included in the survey. Conventional farms were selected based on comparability (Offermann and Nieberg, 2000; Zanoli et al., 2007).
There are many activities that are carried out in a farm. For determination of costs associated with production, the focus of the survey was those that are directly related to organic/ conventional vegetable production. Transaction costs were taken as costs involved in facilitating exchange along the vegetable value chain (Brouthers, 2002).
1.9 Assumptions of the study
Analysis of profitability of a farming system requires availability of costs and income information accrued from the system. Small holder farmers sometimes don’t see the need or have no skills to keep records necessary for such an evaluation. In carrying out the study, it was assumed that the smallholder farmers interviewed were able to reconstruct the required data on farm costs and incomes for appropriate determination of farm profitability.
1.10 Operational definition of terms and concepts
Impact evaluation involves a systematic and objective assessment of effects of a programme on individuals, households or communities to establish whether they are due to specific interventions of the programme undertaken.
Organic farming system: is a system of production that uses biological cycles and processes and avoids the use synthetic pesticides, chemical fertilizers, genetically modified organisms and synthetic feed additives such as hormones in livestock (IFOAM, 2004). Organic farming follows set standards of production and processing.
Conventional production system: Means non organic farming system or the system that farmers normally practice where they use synthetic pesticides and chemical fertilizers.
Economic analysis is a systematic approach to determine the optimal use of scarce resources involving comparison of two or more alternatives in achieving a specific objective under given assumptions and constraints.
Smallholder farmers: Smallholder farmers are defined by scale of their operation. They are farmers whose farm is mainly managed through farm labour and predominantly practices farming on an area less than 3 ha. (Taylor, 2006)
2.0: LITERATURE REVIEW
This section contains a review of information and work done by other researchers in the area of profitability of organic production systems and impact evaluation in farming systems. Background information relating to the topic is initially discussed i.e. organic agriculture and its importance, development and status in Kenya. This is followed by description of methods of economic evaluation and factors that affect farm profitability. Finally, the methodology of impact evaluation including matching logarithms is discussed.
2.2 Importance of organic agriculture
An organic production system complies with set organic standards which define requirements for production, conversion, handling, storage, processing and packaging (EAC, 2007). To confirm compliance with standards, farmers undergo a verification system, where they are monitored annually by an independent certification body (Elzakker and Eyhon, 2010). Conventional production system on the other hand is the production system that utilizes materials, production or processing practice that is not organic (EAC, 2007).
Organic farming system has been found to have environmental, social and economic benefits (Pimentel et al., 2005). For instance, it reduces run off by allowing more water percolation, uses less energy in terms of fossil fuels for farm machinery, fertilizers, seeds and herbicides (Pimentel et al., 2005). It also contributes to building more soil carbon, increasing water holding capacity, retaining more nitrogen in the soil and increasing biodiversity (Cobb et al., 1999; Pimentel et al., 2005; Scialabba, 2007). Organic produce have also been found to contain more vitamins, minerals, enzymes and micronutrients compared to conventional produce (Reganold, 2006). Studies demonstrate organic products as safe with no risk of containing chemical residues (UNEP-UNCTAD CBTF, 2008). Organic farming system is shown to be more resilient and less risky thereby providing the base for immense potential towards food security (Pimentel et al., 2005; Kerselaers et al., 2007; UNEP-UNCTAD CBTF, 2008).
immunity, resilience and regeneration. It also stands for clean environment, protected from destruction through deforestation, soil erosion, extinct plant and animal species (IFOAM, 2002).
2.3 Organic agriculture development in Kenya
Organic agriculture has been practiced in Kenya for many years since most farmers traditionally used manures without application of synthetic pesticides and fertilizers (UNEP-UNCTAD CBTF, 2011). Formal organic agriculture in East Africa was pioneered in Kenya in the eighties when Kenya Institute of Organic Farming (KIOF) was established (Parrott and Elzakker, 2003; Taylor, 2006; Mwaura, 2007). The initial efforts to promote organic agriculture in Kenya were made by Non Governmental Organizations, Faith Based Organizations, and Community Based Organizations through diversification of food production at household level and use of intensive ecological methods to ensure sustainable production and increased household incomes. This has changed over time to integrate commercial approaches and adoption by large scale farmers (KOAN and MOA, 2010).
2.4 Current status of organic production in Kenya
grew by 6.2%. Currently, 1.8 million producers are involved in production of different crops globally on 89.1 million hectares (Willer and Lukas, 2010).
The Kenya organic sector is relatively small but growing fast especially in fruits and vegetables. About 12,647 farmers are involved in production of vegetables, fruits, chillies, coffee, tea, nuts, herbs and spices cultivating on 104,211 ha (Willer and Lukas, 2010). Most of the production is for export market with only 45 producers certified for selling in the domestic market (UNEP-UNCTAD CBTF, 2011). The Domestic market price premiums range between 15 to 150%, while the value of domestic organic market is estimated to be Ksh. 0.25 billion (UNEP-UNCTAD CBTF, 2008).
2.5 Economic analysis
Economic analysis is a systematic approach to determine the optimal use of scarce natural resources involving comparison of two or more alternatives in achieving a specific objective under given assumptions and constraints (USDA NRCS, 1998) It can be used to determine if a farmer will receive a greater economic and more stable return from adopting a technology.
Bolwig et al., 2009; Nermes, 2009). This study analyzed the stated variables to arrive at a sound economic analysis comparing organic and conventional production systems.
2.6 Determinants of farm economic performance
The determination of the farm performance can be made through net returns (Nermes, 2009; Vongpaphane, 2009) or gross margins (Cobb et al., 1999). It can also be made using Return on Assets (ROA) and Return on Equity (ROE) (McCrory, 2001 ; Canavari et al., 2004; Shadbolt et al., 2009). Other methods of measuring farm perfomance include Estimated Treatment Effects (ETE) relative to the counterfactual of no treatment (Bolwig et al., 2009). The use of ROA and ROE as methodology for evaluation is limiting and is not appropriate where land valuation is difficult (Shadbolt et al., 2009).
2.7 Variables affecting profitability of organic production systems
Numerous studies showing organic production system as being more profitable compared to conventional production system attribute the difference to various variables. The variables include; premium (Cobb et al., 1999; Pimentel et al., 2005; Zanoli et al., 2007; Nermes, 2009), support payments in the EU (Cobb et al., 1999; Offermann and Nieberg, 2000), and production costs (Cobb et al., 1999; Offermann and Nieberg, 2000; McCrory, 2001; Zanoli et al., 2007; Demiryurek and Ceyhan, 2008; Nermes, 2009). Other factors include yield (Offermann and Nieberg, 2000; Zanoli et al., 2007; Nermes, 2009), marketing channels (Offermann and Nieberg, 2000), and social economic characteristics (Demiryurek and Ceyhan, 2008; Bolwig et al., 2009; Oxouzi and Papanagiotou, 2010).
Few studies have reported absence of a relationship between variables in an organic production system and profitability (Low Pseudo R) (Canavari et al., 2004). Premiums have a positive relationship to farm profitability (Cobb et al., 1999; Pimentel et al., 2005; Nermes, 2009). These fndings are however disputed by Delate et al. (2002) who found no relationship between the two variables. Support payments is documented as having influence on profitability positively through contributing to 15-24% of farm incomes in EU countries (Cobb et al., 1999; Offermann and Nieberg, 2000).
and Nieberg, 2000; Demiryurek and Ceyhan, 2008; Oxouzi and Papanagiotou, 2010) and others show it as relating negatively to farm profits (McCrory, 2001). Yield has been shown to have both positive and negative effect to profits since it is largely affected by other variables outside the system (Nermes, 2009). Marketing system has a positive relationship to profitability (Offermann and Nieberg, 2000) while social economic characteristics are shown to have both positive relationship with profitability (Bolwig et al., 2009) and no relationship (Oxouzi and Papanagiotou, 2010). The study therefore evaluates the influence of social economic factors (age, gender, occupation) and farm characteristics (land ownership, number of parcels, land size, irrigation and location) as the main factors that affect profitability of an organic production system.
2.8 Methodologies and techniques of comparing organic and conventional
Table 2.1: Summary of reviewed methodologies and techniques
Authors and year
Study title Objectives Analytical method
Key findings Gaps
Cavigelli M.A., Hima B.L., Hanson J.C., Teasdale J.R., Conklin A.E., and Lu Y, 2009.
Long-term economic performance of organic and conventional field crops in the mid-Atlantic region To Evaluate the sustainability of organic and conventional grain crop production Present value for returns, Net profit analysis ANOVA using Proc mixed When organic price premiums are included, net returns were at least 2.4 times greater and risk was at least 1.7 times
lower for organic systems than for conventional systems Disparities between the rotations and lack of universal application.
Shadbolt N., Kelly T., Horne D., Harrington K., Kemp P., Palmer A., Thatcher A., 2009 Comparisons between organic and conventional pastoral dairy farming systems: cost of production and profitability To compare the cost of production and profitability of certified organic and conventional dairy farms over five years of a Massey University system comparison trial Grossmargin analysis and Return on Assets (ROA)
Authors and year
Study title Objectives Analytical method
Key findings Gaps
Cisilino F. and Madau F.A. 2007 Organic and Conventional Farming: a Comparison Analysis through the Italian FADN Characterize organic and conventional groups of farms to better address differences (if any) in production technology, costs and revenues Data Envelopment Analysis (DEA) using Constant Return of Scale (CRS) and Variable Return of Scale (VRS) The productivity and gross margin for conventional farms are higher than organic farms. Does not consider the whole farm as a system. Ignores conversion status as a variable influencing profitability in his economic analysis
Charyulu D. K and Biswas S. 2010 Economics and Efficiency of Organic Farming vis-à-vis Conventional Farming in India Establish the economics and efficiency of organic and conventional farming for cotton, sugarcane, paddy and wheat in India. Nonparametri c Data Envelopment Analysis (DEA) The gross margins for organic wheat and cotton per acre were 16% and 72% per cent higher respectively when compared to conventional fields. Did not take care of self selection bias
Authors and year
Study title Objectives Analytical method
Key findings Gaps
Bolwig S., and Gibbon P., 2009
The Economics of Smallholder Organic Contract Farming in Tropical Africa Examining the revenue effects of certified organic contract farming for smallholders and of adoption of organic agricultural farming methods in a tropical African context. OLS and FIML estimation of the Heckman model. The average effect of participating in organic coffee production is a revenue increase of USH 170,430 per household, equivalent to a gain of 75% in net coffee revenue relative to the
Authors and year
Study title Objectives Analytical method
Key findings Gaps
Oxouzi E. and Papanagiotou E., 2010 Comparative analysis of organic and conventional Farmers and their farming systems. Where does The difference lie? Comparative analysis of viticulture farms economic performance under organic and conventional management practices. Gross Margin Analysis Organic viticulture farms’ net profit was estimated to be lower by 55.0% than that of conventional viticulture farms. Did not take care of self selection bias
Delate K. Duffy M. Chase C. Holste A. Friedrich H. and Wantate N. (2002) An economic comparison of organic and conventional grain crops in a long-term agroecological research (LTAR) site in Iowa To examine the agronomic and economic performance of conventional and organic systems, using required practices for certified organic production. (Proc ANOVA). No significant differences between organic and conventional corn yields for 7 years of study. Organic soybean yields were consistently equal to conventional yields in all rotations in all years Did not consider experiential learning effect Reganold, J. 2006 Sustainability of Organic, Conventional, and Integrated Apple Orchards To evaluate the number of indicators of environmenta l and economic sustainability with these three apple production systems ANOVA Gross margins and net returns The organic system was more profitable than either the conventional or integrated system; The organic fruit was sweeter and as firm or firmer than conventional and
Authors and year
Study title Objectives Analytical method
Key findings Gaps
Pimentel D., Hepperly P., Hanson J., Douds D., And Seidel R. 2005
Environmenta l, energetic, and economic comparisons of organic and conventional farming systems Economic comparison of conventiona l and organic rotation systems with and without rotation period. General Linear Model Univariate Analysis of Variance. Organic corn (without price premiums) was 25% more profitable than conventional corn ($221 per ha versus $178 per ha).
Did not analyse social economic factors and role in profitability . Also lacked experiential effect
Jans S. and Cornejo J.F., 2001
Economics of organic farming in the US: The case of Tomato Production
To Estimate the effect of adopting organic organic tomatoes on yields, revenue and profits. Probit analysis
There was a negative effect (elasticity of probability of -0.04) of organic farming on yield and variable profits which was reduced by time factor The number of farmers involved (33 organic) were few and had a wide disparity of acreage with conventiona l farmers which was not controlled. Khanal K., 2004 Organic and
conventional vegetable production in Oklahoma To calculate cost and returns of selected vegetables for both conventional and organic systems in Southeastern Oklahoma Linear Programming Model and Target MOTAD Model Organic production system provides less return over the conventional production system when the same prices are considered. Use of simulation assumed optimal conditions in modeling
organic farm should be if it were conventional. The first system ignores the changes that will occur outside the farm and their potential impact on the two systems on a time scale (Offermann and Nieberg, 2000; Kerselaers et al., 2007; Shadbolt et al., 2009). Though a robust comparison method, modelling does not have a universal application and differs according to level of application. It also does not represent the actual situation and its probability of occurrence depends on factors outside the model (Offermann and Nieberg, 2000). This is in addition to need for calibration, possibility of bias towards optimizing solutions and insufficient accounting for time and risk aspects. The comparison of conventional farms as an approximation or how an organic farm should be if it were conventional has been commonly used in FADN data analysis. The methodology is however criticized since it makes comparison across countries difficult (Lee and Fowler, 2002).
studies comprise largely of qualitative assessments with inclusion of quantitative measurements. They are not popular and require a long-term study. This study utilized farm survey technique where organic farming group is compared to conventional farming group at the same point in time.
2.9 Impact evaluation of organic production system
The number of studies evaluating the impact of organic production system on profitability are high. However, only few studies consider long term economic impact and most of them have been undertaken in developed countries (mainly USA) and on certain crops (corn, soy and wheat) (Nemes, 2009). The comparison between the two systems however faces several challenges (Offermann and Nieberg, 2000; Canavari et al., 2004; Cisilino and Madau, 2007; Zanoli et al., 2007). The challenges can be categorized as; a) high differences as far as the productive techniques are concerned; b) different technical- productive paradigm which is difficult to define a peculiar one for each group; c) heterogeneity mostly because conventional farming is a mix of agronomic techniques, some of which are similar to the organic ones.
al., 2004; Manivong, 2009). Some studies show organic production system having no impact on farm profitability during conversion but show profitability increasing with achievement of full organic status (Cobb et al., 1999; Pimentel et al., 2005). The impact of organic system on profitability is shown to have disparities depending on crops, regions and technologies employed in the study (Pimentel et al., 2005). This study focused on establishing the impact of organic farming on profitability of vegetable production system among smallholder producers.
2.9.1 Impact evaluation using propensity score matching
There is need to control factors that may have an impact on farm profitability for the two groups and secondly take care of potential large differences in covariates between treatment and control groups which may lead to confounding when estimating treatment effects (Lu Bu, 2005). Propensity score matching (PSM) has been widely used in biometrics, econometrics and other social sciences to estimate average treatment effects of labour markets, training prorgammes and medical treatment among others (Frolich, 2004; Diaz and Handa, 2004).
Following the work of Rosenbaum and Rubin (1983), propensity score is a commonly used device for constructing matched sets which are comparable in terms of pretreatment characteristics (Frolich, 2004; Lu Bu, 2005). It reduces the dimension of estimation problem by controlling only one-dimensional propensity score (the conditional probability of treatment receipt) to remove selection bias instead of controlling all the confounding factors (Rosenbaum and Rubin, 1983).
2.9.2 Matching algorithms
Matching is a technique used to assign control subjects to be matched with treated subjects based on the propensity scores (Khandker et. al, 2010). The choice of matching technique used is important since the estimators differ in the way that the neighborhood surrounding each treatment is defined (Caliendo and Kopeing, 2008).
There are several matching algorithms that can be used and the most popular ones include; nearest neighbor, caliper or radius, stratification or interval and Kernel and local linear. In the nearest neigbour matching, each treatment unit is matched to the control unit with the nearest propensity score (Khandker et al, 2010; Caliendo and Kopeing, 2008). Caliper method on the other hand imposes a threshold on the maximum propensity score distance (Caliper) (Khandker et al., 2010). It involves matching with replacement only among propensity scores within a certain range thereby enforcing a common support area thereby reducing the number of possible matches (Nicolette, 2011).
neighbour, kernel matching and radius matching was used. In addition, a calipher of 0.3 was used to provide a reasonable range within the region of common support where suitable matches were obtained.
2.9.3 Gaps in literature
Conversion to organic production system has been shown to have varied benefits to a farmer. The literature indicates a big gap in studies comparing economic performance of organic system and conventional system in Africa. This is because most of the studies are conducted in Europe and US. Due to the realization of the importance of such studies and their role in stimulating interest and adoption of organic production system in Kenya, this study seeks to add to the few studies conducted in this field.
3.0 MATERIALS AND METHODS
This section first describes the two study areas that were sampled and the characteristics of the target population. Secondly, it explains the research design that was used, data collected and sampling procedures used. It concludes by showing how data analysis was conducted for the three objectives so as to answer the set research questions.
3.2 Study area
31 Figure 3.1 Map of study area
on volcanic foot ridges and plutonic soils. These Nitisols and Andosols are generally developed, well drained, deep, fertile and suitable for cultivation of various crops (FAO, 2007). The repeated and high intensive cultivation over time however has led to depletion of soil nutrients.
Kiambu is predominantly agricultural with nearly 70% of the population engaged in farming (KNBS, 2007). Most of the cultivation is by smallholder farmers (90%) while the remaining large scale farmers concentrate mainly on tea and coffee. The food crop patterns are dominated by production of maize, beans, potatoes in a mixed intensive cultivation. Tea, coffee and horticultural products (flowers, cabbages and potatoes) are the main income earners. Due to its proximity to Nairobi, the County provides a significant percentage of the food consumed in Nairobi.
Kajiado County is located at the former Rift Valley Province. As shown in Figure 3.1, it is bordered by Tanzania to the southwest and Taita Taveta County to the SE, Machakos County to the E, Nairobi County to the NE, Kiambu County to the N and Narok County to the W (Appendix 1). The County has population of 687,312 (KNBS, 2010), and covers an area of 19,600 km2. Its numerous rocky scarps and slopes have shallow, reddish-brown, stony clay-loams (FAO, 2007). The area has varied soils, including alluvial deposits.
forests. The annual rainfall varies between 500 to 1,250mm. There are two wet seasons, the ‘short rains’ between October and December and the ‘long rains’ between March and May. The main economic activities include livestock herding, tourism, agriculture and urban-life activities like cattle trading. Of the whole households’ population, 44% derive their income from agriculture while 45% rely on urban self-employment (FAO, 2007).
3.3. Research design
To realize the set objectives and be able to test the set hypotheses data were collected using farm survey design where organic farms were compared to conventional farms at the same point in time (Lee and Fowler, 2002). Data on production, income, expenditure, social economic and farm characteristics was collected from the two sample sets of farmers. The survey was done during the period when farmers are engaged in farm preparation, planting and management activities and therefore could easily be located. Interviews were done with household heads and where not available members of the houshold who had full knowledge or were involved in farming activities. Where household head was not involved in farming activities, the spouse was interviewed instead.
3.4 Target population
grocers, supermarkets, farmers’ markets, restaurants and hotels in Nairobi were involved in the survey. The study focused on farmers who had cultivated for two seasons between 2010 and 2011 the vegetables kales, spinach and cabbages. For comparison purpose, smallholder conventional vegetable producers with similar farm characteristics in the target area were surveyed.
3.4 Data collection
Secondary data were collected between January and February 2012 while primary data were collected between March and June 2012. Primary data were collected using structured questionnaire administered through household interviews for both smallholder organic and non organic farmers. Data collected included acreage of the farm, yield, prices, costs, target market for the previous two seasons for the year 2010/2011 among others (Appendix 6).
3.5 Sampling procedure
The conventional farmers were sampled using stratified sampling method where K-means clustering approach as postulated by Cisilino and Madau (2007) and Zanoli et al. (2007) was applied. Stratified sampling technique is a probability sampling method in which the entire population is divided into subgroups or strata from where the final subjects are sampled proportionately from the different strata (Kothari, 1990). Lists of farmers growing identified vegetables as generated by District Agriculture Officers Annual crop reports was used to develop sampling frame for conventional vegetable farmers (District Agricultural Crops Reports, 2010 a,b,c). The farm acreage was used to determine strata for the conventional farmer’s clusters as follows: (0.25-2), (2-5), (5-10) and more than 10 acres. To develop the clusters, each organic farm was used as a centroid while comparable conventional farms were clustered around each organic farm according to the closeness of their characteristics as postulated by Offermann and Nieberg (2000) and further developed by Zanoli et al. (2007). The characteristics considered were farm type, location, production system, topography, soil type and market distance. A sample of 144 smallholder conventional farmers was used to achieve a significance level of 0.05, power of 0.8 to detect the postulated effect of 10% change in means between the two cohorts and a 30% coefficient of variation (CV) (Belle and Martin, 1993) using Lehr’s equation
where n is the sample size, CV is the coefficient of variation and
conventional farmers was distributed in proportion to organic population as shown in the Table 3.2 below.
Table 3.2: Summary of sampling design
Method of Sampling
Sample sites Total
Organic organic Organic Non-organic Stratified
30 61 41 83 215
Population 30 12,155 41 289,470 301,625
As shown in Table 3.3 above, the distribution of the sample between the two Counties was based on the population of organic vegetable farmers. Kiambu having a population of 41 organic vegetable farmers had a sample of 83 non organic farmers while Kajiado with a population of 30 organic farmers had a sample of 61 non organic farmers.
3.6 Pretesting survey tools
evaluated. The exercise enabled the questionnaire to be reviewed so as to incorporate changes that enhanced the capturing of required data.
3.7 Data management and analysis
The filled in questionnaires were first numbered and then coded for ease of referencing. Data were then entered into an excel sheet and cleaned. To facilitate analysis in order to make empirical conclusions, variables were defined as shown in the Table 3.3 below;
Table 3.3: Variable definitions
Profitability This is the financial gain earned when revenue is higher than the expenses incurred in operating a business activity. This was measured using gross margin analysis i.e. total revenue minus total costs
Impact Effect of adopting organic on gross margins of vegetable production system.
Acceptance of new product or innovation in this case organic farming technology.
Age Age in years of the person managing the farm.
Sex Sexual orientation household head either male (1) or female (0).
Occupation (OCC) Refer to job, role or regular activity done for payment, income or livelihood. The variable was used to determine the main activity of the farmer and was a dummy where main engagement in farming activity was taken as (1) and non farming activities (0).
Land size (SIZ) Total land size in acres being managed by the farmer. Land owner ship
Nature of formal ownership which was a dummy with farmer owned being (1) and (0) otherwise.
Irrigation (IRR) Presence or absence of irrigation facility on the farm which was a dummy with availability (1) and non availability being (0)
County location Where the farm is located among the two Counties which was dummy with Kiambu (1) and Kajiado (0).
Target market (MKT)
Where the farmer sells or intend to sell his/her produce which was dummy variable with retail being 1 and 0 wholesale markets.
Transaction costs (TC)
The costs incurred in making an economic exchange for example facilitating product flow from or to the farm expressed in Kenya shillings. This included sourcing information and transportation of produce.
Position in the house hold
Role of the respondent in the family; whether head of the household (HH), spouse or sibling (son or daughter).
Marital status Whether the respondent is married or single.
Level of education Formal education level attained whether primary, secondary or college.
Trainings related to farming attended since the farmer started farming.
Topography Surface features of a place described for respondents whether; flat, undulating, steep or very steep. This was taken as a dummy where flat surface was represented by 1 and zero otherwise.
Source of labour Where the farmer gets labour for undertaking farm operations. This was represented by a dummy where own labour source was represented by one and zero otherwise.
Comparisons for social economic, farm and market characteristics were done using cross tabulation.
Data analysis was done using different methods as per stated objectives (Table 3.4 below).
Table 3.4: Summary data analysis
Objective Independent Variables
Statistical test Analyze the factors
associated with adoption of organic vegetable
production system in Kiambu and Kajiado Counties of Kenya.
measured as participation as an organic farmer
occupation of the household head, experience in farming, farm size, ownership, irrigation number of parcels, target
Analyze factors associated with profitability of investment in organic vegetable production in
Kajiado counties of Kenya.
Profitability measured as gross margin
occupation of the household head, experience in farming, farm size, ownership, irrigation number of parcels, target
Ordinary least Square
Estimate the impact of organic production system on profitability of smallholder
vegetable farms in
Kajiado counties of
Impact measured as a percentage change in average treatment effect of organic farmers.
occupation of the household head, experience in farming, farm size, ownership, irrigation number of parcels, target
Kenya. market and
Objective one and three were analyzed using logit model while objective two was analyzed using OLS regression. Logistic regression was preferred to linear probability models in analyzing objective one and three due to the well-known shortcomings of the linear probability model especially the unlikeliness of the functional form when the response variable is highly skewed and suitability for predictions that are outside the [0, 1] bounds of probabilities (Caliendo and Kopeing, 2008). The logit regression model was also preferred since its distribution has more density mass in the bounds than probit model and that it is commonly used in adoption studies. In addition, logit model is documented to be consistent in parameter estimation associated with the assumption that the error term has a logistic distribution (Zhao, 2008).
To achieve the first objective of the study, the effect of Social economic, farm and market characteristics on adoption of smallholder organic vegetable production system was evaluated using a logistic regression model which was analyzed using Stata version 11.0 of 2009.
counterfactual can be expressed as Pr (Yi=0). The model below therefore will suffice the analysis required thus:
……….. ……… (1)
Where Pr (Yi=1) is the probability that a farmer is practicing organic and β is the coefficient while x represents adoption factors (social economic, farm and market characteristics) and Ԑ is the error term. By taking the natural logarithm of equation 1 above we get a simplified form of adoption logistic probability model as follows:
The equation 1 above can be expanded to a full logistic regression equation with explanatory variables included as follows:
... (3) By taking the log odds the equation can be simplified thus;
Where the default value β0 is the constant and Β1, β2, β3, β4, β5, β6. and Β7 are
parameters for explanatory variables; sex, age, occupation of the household head, experience in farming, farm size, ownership, irrigation facility on the farm, number of parcels and target market respectively eirepresents the error term.
Y =β0 +β1 Age + β2SIZE + OWN β3 + β4PAR+ β5MKT+ ei ……….. (5)
Where the default value β0 is the constant and Β1, β2, β3, β4, β5, β6. are parameters for
explanatory variables; age, experience in farming, farm size, ownership, number of parcels and target market respectively eirepresents the error term.
The third objective was achieved by evaluating the impact of organic production system on profitability using propensity score matching where the observable estimated treatment effects were compared to counterfactual of no treatment (Rosenbaum and Rubin, 1983). The propensity score matching was used to evaluate the impact of organic production system on smallholder vegetable farm production. Propensity Score Matching (PSM) was used as an impact estimator to get unbiased estimates of average treatment effects. The choice of PSM as an impact estimator was informed by its reliability and comparability with experimental impact estimators especially when similar survey instruments are used (Diaz and Handa, 2004). This therefore infers that the method can be used as a robust impact evaluation method where timeframe is short. This was done first by establishing the estimators for logit regression used in estimating propensity scores. Nine variables representing social economic and farm characteristics were used in matching. They included land size, location, gender, age, occupation of household head, years of experience, number of farm parcels owned, availability of irrigation, and land ownership.
practicing organic farming. For PSM to hold, the following matching assumptions were made;
Common support assumption
Where is the outcome for non organic smallholder vegetable farmers, the outcome for practicing organic farmers, D as the treatment indicator where D=1 signifies a farmer practicing organic is the notation for statistical independence and is the propensity score. When the matching assumptions are met, the unbiased impact of organic production system on organic producers through matching by propensity score can therefore be estimated. In the logit regression model µ was assumed to follow a logistic distribution. The error terms in the outcome equations of both the organic smallholder farmers and non organic farmers were allowed to be correlated with cov( )=0 and cov( )=0 so that the unconfoundedness assumption can be satisfied (Zhao, 2008).
The average causal impact of practicing organic was therefore measured by average treatment effect as follows:
And also by average treatment effect of the treated
Where Di indicated whether the smallholder vegetable farmer was practicing organic farming (D=1) or not practicing organic farming (D=0). The symbol measured the impact of organic production system to the whole population in this case referred to as the treatment while represented the impact for the sub population. The mean difference between observables can therefore be written as:
Where is the bias and can also be given by
Where ATT is the average treatment of the treated, representing the profitability outcome of practicing organic smallholder vegetable farmers,
represented profitability outcome of non organic farmers, the profitability outcome of nonorganic farmers if they practice organic farming and the error term.
Correspondingly, the true parameter of ATT is identified if the outcome of the treatment and control on condition of no practicing organic farming is the same;
By putting the propensity scores, unbiased estimate of the average treatment effect was got thus:
is the expected profitability of smallholder, i practicing organic vegetable production with propensity score (x) and is the expected profitability of non organic smallholder vegetable farmer i with propensity score (x).
Since the dependent variable in the above model takes a binary form; where it takes the value of 1 for organic farmers and 0 for non-organic farmers, logit regression model was therefore used. The cumulative treatment effect or impact for organic production system on the participants was therefore deduced from the summative difference between the participants and matched non participants.
For such models sensitivity analysis is usually undertaken to check the conditional independence assumption and the influence of unmeasurable variables on the matching procedure (Khandker et al., 2010). Balancing tests were also conducted to check whether within each quartile of the propensity score distribution, the average propensity score and mean (X) were the same.
4.0 RESULTS AND DISCUSSION
This chapter presents the results of data analysis. First, the factors influencing adoption of organic production system are presented where mean differences of adopters and non adopters were compared. Secondly, the logit regression estimators for adoption of organic vegetable production system relating to social economic, market and farm characteristics are presented. Thirdly, the ordinary least squares estimators relating profitability of organic vegetable production system to social economic, farm and market characteristics are presented. The chapter concludes by presenting the impact of converting to organic vegetable production system on farm profits. Quality of data was established using covariate balancing tests, PSM quality indicators and sensitivity analysis.
4.2 Descriptive statistics comparing organic and conventional farmers.
Table 4.1 Difference In Means Of Characteristics of Adopters and Non Adopters
Table 4.1: Difference in means of characteristics of adopters and non adopters
Variables Unit Conventional
χ2 /Ttest Sample characteristic (proportion)
% female headed 54.20 69.0 1.78** Education level % with
45.0 42.30 16.56***
Occupation % fully occupied with farming
51.70% 59.2% 1.01
Source of capital
% farmers using own capital
93.0% 88.0% 1.67
Source of Labour
% using family labour
81.0% 54.9% 16.85***
% with individual owned farms
46.2% 84.5% 7.28***
Availability of irrigation
% with irrigation
49.6% 62.9% 3.12*
Marital status % married 88.30 80.30 3.42
Sample Characteristic (Means)
Age No. of years 37.73 46.68 -8.95***
Experience No. of years 9.35 6.37 2.99***
(1.02) Knowledge Number of
1.75 2.94 -1.19**
Total land size Ha 0.57 3.04 -2.47***
(0.59) Number of
1.17 1.43 -0.27**
Age, level of education, farming experience, number of training, land size, number of parcels of land owned by the farmer, and source of labour were significantly different for the two cohorts (Table 4.1). However position in the household, marital status, topography, occupation, source of financing and type of irrigation for the two cohorts was the same. As observed by Demiryurek and Ceyhan (2008) and Jans and Cornejo (2001), the organic vegetable farming group was older compared to conventional farmers group and had bigger land sizes and more parcels of land compared to non organic farmers. The adoption of organic vegetable production system by aged population is expected as the general trend of farming in Kenya is by aging population while most of the youth go to towns to seek employment (Mwaura, 2007). The preference of organic production by older generation can also be said to relate to their preference for health benefits associated with consuming organic foods as observed by Ndungu, (2013).
inform of many parcels contrary to expectation that organic farmers have small farms as observed by Demiryurek & Ceyhan, (2008).
The existence of significant difference between the two groups for selected variables suggests that they may have an influence on farmer’s decision whether to adopt organic vegetable production system. It is therefore important to use econometric analysis to understand motivation for adoption.
4.3 Factors associated with adoption of organic production system
Table 4.2: Factors associated with adoption of organic vegetable production system
Variable definition Coefficient Standard error
Sex of household head Occupation
Land size Irrigation Land ownership County
Target market Constant
0.056 -0.152 1.621*** 1.123*** -1.917***
0.023 0.411 0.608 0.146 0.592 0.436 0.731 0.521 0.950 Wald Chi2=42.51, Prob> Chi2=0.001, Pseudo R2=0.63, Log likelihood=-79.01 F(9,191) =4.7; Significance level of regression estimators: *0.1, **0.05, ***0.01
The Pseudo R2 indicated goodness of fit for regression estimators meaning that they were able to explain the adoption probability. This was also confirmed by pearson goodness of fit test that yielded large P value showing that the model fitted well with explanatory variables (Appendix 2). The model was also shown to have well and correctly specified predictor values with high percentages. In addition, correlation matrix for the coefficients as shown in Appendix 2 reported weak relationships which can be interpreted to mean low or absence of multicollinearity.
retail markets positively influence adoption of organic production system while County of residence has an inverse relationship to adoption of organic vegetable production system.
The findings presented demonstrate that social economic factors, farm and market characteristics can significantly explain the adoption of smallholder organic vegetable production system. The study established that older farmers were responsive to adoption of organic agriculture compared to the youth. This finding can be explained by the labour intensity associated with organic production system as observed by Cisilino and Madau (2007). Young people are also generally less interested in agriculture production enterprises due to perceived length of time to earn returns (Republic of Kenya, 2010). Other studies by Bolwig et al. (2009) on organic coffee in Uganda and Oxouzi and Papanagiotou (2010) on organic grapes in Greece indicated that organic farming population was more aged compared to population of conventional producers.
The target market was found to influence the decision to convert to organic production system. Majority of farmers expressed the desire to sell their produce to retail markets as compared to whole sale markets. Retail markets are associated with higher prices due to elimination of middle men and hence the preference compared to whole sale markets Elzakker and Eyhon (2010).
4.4 Factors associated with profitability of investment in organic production
Profitability of smallholder organic vegetable production system is influenced by several factors. In the study social economic variables (age, gender, level of education and occupation), farming characteristics (farming experience, number of trainings attended, land size, number of parcels, irrigation, production per acre, average price cost of production and land ownership) and target market were evaluated to determine their relationship with gross margin of the organic production system. Table 4.3 shows the coefficients for the variables.
Table 4.3: Factors associated with profitability of investment in organic vegetable production system
Variable Coefficient Robust Standard Error
County of residence 1.35 7.57
Gender -6.66 2.12
Age 2.77** 1.21
Level of Education 1.40 1.01
Occupation -3.61 2.37
Number of trainings attended 8.70* 4.77
Land ownership 3.93 1.87
Total farm size 7.28 3.42
Irrigation 0.60* 3.28
Target market -0.661*** 0.235
Constant -90.38*** 9.25
Number of observations=66; F (7,59)=2.44; Prob> F=0.029; R2=0.79 and Root