ECONOMIC ANALYSIS OF RECOMMENDED TECHNOLOGIES ON PERFORMANCE AMONG SMALLHOLDER FARMERS IN EMBU COUNTY, KENYA

DANIEL MUSAU WAMBUA

A THESIS SUBMITTED IN PARTIAL FULFILLMENT FOR THE DEGREE OF MASTER OF SCIENCE IN AGRICULTURAL ECONOMICS OF THE UNIVERSITY OF EMBU

JANUARY, 2020

DECLARATION This thesis is my original work and has not been presented elsewhere for a degree or any other award.

Signature………………………………… Date……………….………………

Daniel Musau Wambua Department of Agricultural Economics & Extension A510/1115/2016

This thesis has been submitted for examination with our approval as University Supervisors.

Signature…………………………..…… Date………………………......

Dr. Samuel N. Ndirangu Department of Agricultural Economics & Extension University of Embu

Signature……………………………. Date………………………………

Dr. Lucy K. Njeru Department of Agricultural Economics University of Nairobi

Signature……………………………. Date………………………………

Dr. Bernard M. Gichimu Department of Agricultural Resource Management University of Embu

ii

DEDICATION

With profound appreciation, I dedicate this research thesis to my loving family for their unwavering and immeasurable support both morally and financially and good upbringing to realize the value of education.

iii

ACKNOWLEDGEMENT

I wish to thank the Almighty God for the life and an opportunity to pursue my studies at the University of Embu. I wish to record my sincere appreciation to my able supervisors led by Dr. Samuel N. Ndirangu, Dr. Lucy K. Njeru and Dr. Bernard M. Gichimu for their constructive criticism, valued contribution, unwavering support, tireless guidance, encouragement, availability and timely feedback which ensured completion of my research work. I would also register my profound appreciation to the University of Embu Management led by Professor Daniel Mugendi Njiru, for granting me a scholarship opportunity and also offering me a research grant which facilitated data collection for my research work. Special thanks go to the Chairman of the Agricultural Economics and Extension Department, my lecturers, staff and students of University of Embu for their continued support and their words of encouragement throughout my course work and research project. I would also like to thank the management of the six coffee cooperative societies led by Mr. Thomas Nyaga, and the farmers for their support and cooperation in enabling smooth data collection. My appreciation also goes to the team of enumerators led by Caroline who tirelessly assisted in data collection and cleaning in readiness for data entry and coding. Finally I would like to thank my family for their unending support in facilitating my studies and for the good upbringing to comprehend the value of education in my life.

iv

TABLE OF CONTENTS

DECLARATION...... ii DEDICATION...... iii ACKNOWLEDGEMENT ...... iv LIST OF TABLES ...... ix LIST OF APPENDICES ...... xi ABBREVIATIONS AND ACRONYMS ...... xii OPERATIONAL DEFINITION OF KEY TERMS ...... xiii ABSTRACT ...... xiv CHAPTER ONE ...... 1 INTRODUCTION...... 1 1.1 Background ...... 1 1.1.1 Status of Agriculture in the World and Kenya ...... 1 1.1.2 Status of Coffee Production and Marketing ...... 2 1.1.3 Coffee Research in Kenya ...... 3 1.1.4 Recommended Coffee Production Technologies and Management Practices ... 4 1.2 Statement of the Problem ...... 5 1.3 Justification ...... 6 1.4 Hypotheses ...... 7 1.5 Research Objectives ...... 7 1.5.1 General Objective of the Study ...... 7 1.5.2 Specific Objectives ...... 8 1.6 Assumptions ...... 8 CHAPTER TWO ...... 9 LITERATURE REVIEW ...... 9 2.1 Introduction ...... 9 2.2 Effect of Recommended Technologies on Productivity ...... 9 2.2.1 Coffee Varieties and Productivity ...... 9 2.2.2 Farm Inputs and Productivity ...... 10 2.2.3 Agronomic Practices and Productivity ...... 11

v

2.3 Effect of Recommended Technologies on Coffee Profitability ...... 12 2.3.1 Coffee Varieties and Profitability ...... 12 2.3.2 Farm Inputs and Coffee Profitability ...... 13 2.3.3 Agronomic Practices and Coffee Profitability ...... 13 2.4 Effect of Recommended Technologies on Coffee Quality ...... 14 2.4.1 Coffee Varieties and Coffee Quality ...... 14 2.4.2 Farm Inputs and Coffee Quality ...... 15 2.4.3 Agronomic Practices and Coffee Quality ...... 15 2.5 Research Gap...... 16 2.6 Theoretical Framework ...... 17 2.7 Conceptual Framework ...... 20 2.8 Operationalization of Variables ...... 22 CHAPTER THREE ...... 27 METHODOLOGY ...... 27 3.1 Introduction ...... 27 3.2 Study Area ...... 27 3.3 Research Design ...... 27 3.3.1 Target Population ...... 28 3.3.2 Sample Size ...... 28 3.3.3 Sampling Procedure ...... 29 3.3.4 Data Collection ...... 29 3.3.5 Accuracy and Reliability of Research Instruments ...... 29 3.4 Data Analysis ...... 30 3.4.1 Effect of Recommended Technologies on Coffee Productivity ...... 30 3.4.2 Effect of the Recommended Technologies on Coffee Profitability ...... 31 3.4.3 Effect of Recommended Technologies on Coffee Quality ...... 32 CHAPTER FOUR ...... 33 RESULTS ...... 33 4.1 Introduction ...... 33 4.2 Reliability and Validity Analysis ...... 33 4.2.1 Descriptive Statistics for Reliability Between the Two Halves ...... 34

vi

4.3 Descriptive Statistics of Socio-economic Characteristics of the Respondents ...... 34 4.3.1 Social Characteristics of the Respondents ...... 34 4.3.2 Economic Characteristics of the Respondents...... 35 4.3.3 Institutional Factors of the Respondents ...... 36 4.4 Coffee Input Use and Production ...... 37 4.4.1 Fertilizer and Manure Application ...... 37 4.4.2 Pest, Disease and Weed Control Chemicals ...... 38 4.4.3 Coffee Varieties ...... 39 4.4.4 Coffee Management Practices ...... 40 4.4.5 Coffee Production ...... 40 4.4.6 Effect of Recommended Technologies on Coffee Productivity ...... 41 4.5 Coffee Profitability Among Smallholder Farmers ...... 44 4.5.1 Estimation of Gross Returns (Margins) from Coffee Enterprise ...... 44 4.5.2 Estimation of Variable Costs for Coffee Production ...... 44 4.5.3 Effect of Recommended Technologies on Coffee Profitability (Gross Margin) ...... 45 4.6. Coffee Quality Among Smallholder Farmers ...... 48 4.6.1 Frequencies on Proportion of Mbuni to Cherry ...... 48 4.6.2 Descriptive Statistics on the Proportion of Mbuni to Cherry ...... 48 4.6.3 Effect of Recommended Technologies on Coffee Quality ...... 49 CHAPTER FIVE ...... 51 SUMMARY OF FINDINGS, DISCUSSIONS, CONCLUSIONS AND RECOMMENDATIONS ...... 51 5.0 Introduction ...... 51 5.1 Summary of the Key Findings ...... 51 5.1.1 Socio-economic Factors Influencing Coffee Performance ...... 51 5.1.2 Effect of Recommended Technologies on Coffee Productivity ...... 52 5.1.3 Effect of Recommended Technologies on Coffee Profitability ...... 53 5.1.4 Effect of Recommended Technologies on Coffee Quality ...... 53 5.2 Discussion ...... 53 5.2.1 Effect of Recommended Technologies on Coffee Productivity ...... 53

vii

5.2.2 Effect of Socioeconomic Factors on Coffee Productivity ...... 55 5.2.3 Effect of Recommended Technologies on Coffee Profitability ...... 57 5.2.4 Expenditure on Coffee Production ...... 59 5.2.5 Effect of Recommended Technologies on Coffee Quality ...... 60 5.2.6 Effect of Socioeconomic Factors on Coffee Quality ...... 61 5.3 Conclusions ...... 62 5.4 Recommendations ...... 63 REFERENCES ...... 66 APPENDICES ...... 74

viii

LIST OF TABLES

Table 4.1: Spearman-Brown Coefficient of Reliability Results ...... 33 Table 4.2: Descriptive Statistics for Reliability Between the Two Halves ...... 34 Table 4.3: Social Characteristics of the Respondents ...... 35 Table 4.4: Descriptive Statistics of Economic Factors of the Respondents ...... 36 Table 4.5 Descriptive Statistics on Institutional Factors ...... 37 Table 4.6: Descriptive Statistics for Fertilizer and Manure Application Rates ...... 38 Table 4.7: Frequencies for Pest, Disease and Weed Control Chemical Application Rates ...... 39 Table 4.8: Descriptive Statistics of the Grown Coffee Varieties ...... 39 Table 4.9: Descriptive Statistics of Coffee Management Practices ...... 40 Table 4.10: Descriptive Statistics for Coffee Production in the Study Area ...... 41 Table 4.11 Multiple Regression Results for Effect of Recommended Technologies on Coffee Productivity...... 42 Table 4.12: Multiple Regression Results for Effect of Socioeconomic Factors on Coffee Productivity...... 43 Table 4.13 Descriptive Statistics on Gross Margins (KES) from Coffee Production ...... 44 Table 4.14 Descriptive Statistics for Variable Costs per Acre ...... 45 Table 4.15 Multiple Regression Results for Effect of Recommended Technologies on Coffee Profitability ...... 46 Table 4.16 Multiple Regression Results on Effect of Socioeconomic Factors on Coffee Profitability ...... 48 Table 4.17 Frequencies on Proportion of Mbuni to Cherry ...... 48 Table 4.18 Descriptive Statistics for Total Loss and Proportion of Mbuni to Cherry49 Table 4.19 Regression Results for Effect of Recommended Technologies on Coffee Quality ...... 49 Table 4.20 Regression Results for Effect of Socioeconomic Factors on Coffee Quality ...... 50

ix

LIST OF FIGURES

Fig. 2.1 Conceptual Framework ...... 21

x

LIST OF APPENDICES Appendix 1: Household Questionnaire for Small-scale Coffee Farmers ...... 74

xi

ABBREVIATIONS AND ACRONYMS AFA Agriculture and Food Authority ASDS Agriculture Sector Development Strategy CBD Coffee Berry Disease CLR Coffee Leaf Rust CPU Coffee Pulpery Unit CRF Coffee Research Foundation CRI Coffee Research Institute DEA Data Envelopment Analysis DMU Decision Making Unit FAO Food and Agriculture Organization FYM Farm Yard Manure GDP Gross Domestic Product GoK Government of Kenya Ha Hectares KALR Kenya Agricultural and Livestock Research Act KALRO Kenya Agricultural and Livestock Research Organization Kg Kilogram KNBS Kenya National Bureau of Statistics KRDP Kenya Rural Development Programme M Metres MoA Ministry of Agriculture SDG Sustainable Development Goals SFA Stochastic Frontier Analysis SSA Sub Saharan Africa TFP Total Factor Productivity UM Upper Midland UN United Nations

xii

OPERATIONAL DEFINITION OF KEY TERMS Coffee performance: Production efficiency and economic viability of the crop to sustain itself measured in terms of output, quality and returns.

Coffee quality: The inherent and distinctive attributes or characteristics possessed by coffee cherry measured in terms of grade, fragrance, aroma etc.

Coffee variety: Commercial cultivars of Arabica coffee cultivated by farmers.

Crop productivity: Total crop yield per unit of input used.

Economic analysis: Study on optimization and allocation of the scarce production resources among various production entities through economic methods.

“Mbuni”: is a term commonly used to denote the coffee cherry that is dried before pulping, which is perceived as low quality coffee.

Profitability: A measure of financial performance of a farm on per unit or per output basis measured in terms of gross returns, production cost and net revenues.

Rate of adoption: Proportion of potential adopters of the recommended technologies.

Recommended technology: Any new innovation successfully integrated to make work easier in an economic or social process. In coffee production referring to new varieties, fertilizer, disease and pest control, pruning and weed control recommended by coffee research institutions.

Technology adoption: The use of a new technology in the long run equilibrium when the farmer has full information about the technology preceded by a period of trying and adaptation at the farm level.

Total factor productivity: The ratio of firm‟s total output relative to the total production inputs used in the production process.

xiii

ABSTRACT The coffee subsector in Kenya has been characterized by low and declining productivity at farm levels. Over the years, coffee research in Kenya has developed technologies that are aimed at increasing the productivity and improving the quality of coffee produced. Despite many agronomic recommendations, coffee productivity has not increased with increase in acreage. This may be attributed to the interaction between the recommendations and prevailing socioeconomic factors experienced by the farmers at the farm environment. There is limited research based information on the combined impact of the recommended coffee production technologies and prevailing socioeconomic environment on productivity, returns and quality of coffee at farm level. The purpose of this study was to analyze the combined effect of the recommended technologies and farmer socioeconomic characteristics on coffee productivity, profitability and quality among smallholder farmers in Embu County. Data was collected from a sample comprising of 376 farmers who were randomly selected from six cooperative societies using multistage stratified and probability proportional to size sampling techniques. Data was collected using semi-structured questionnaires, focus group discussion and oral interviews, and included farm demographic data, coffee management, production and quality data. The collected data on farm and farmer characteristics was analyzed using descriptive statistics such as means and frequencies. The combined effect of the recommended technologies and socioeconomic factors on productivity, profitability and quality was determined using Stochastic Cobb-Douglas Production Function, Profit Function and Binary Logit Models respectively. The regression results of estimated Stochastic Cobb-Douglas production function revealed that the recommended foliar feed, manure, herbicide and pesticide rates were significant and positive in affecting coffee productivity at 5% level. The results of the estimated Profit function revealed that recommended varieties, manure rate and capping were positive and significant in influencing coffee profitability at 5% level. Binary Logistic Regression results revealed that the recommended foliar feed rate, manure rate and pruning were significant in affecting the coffee quality. The results of the study also revealed that education, household size and off-farm income have significant impact on productivity and cherry quality. This indicates that there is possibility to increase productivity by using the recommended rate for foliar feed, manure, pesticide and herbicide and increase coffee returns by adoption of the improved coffee variety. Therefore, to increase coffee productivity, profitability and quality, more focus should be put on on-farm research, accompanied by farmer education, specifically on usage of manure, foliar feed, herbicide and pesticide and adoption of the improved coffee varieties. Policy makers need to develop a land use policy and also offer coffee farmers more incentives to access and use these inputs.

xiv

CHAPTER ONE INTRODUCTION 1.1 Background 1.1.1 Status of Agriculture in the World and Kenya The world population is projected to rise to about nine billion by 2050 (United Nations, 2015). To meet the food demand for the growing population, there is need to increase agricultural production by improving crop productivity. Agriculture plays a crucial role in the economies of developing countries. It is the main source of food, income and employment for rural populations. It has been established that the share of agricultural population is 67% of the world‟s total population (GoK, 2007). Agriculture accounted for about 39% of world gross domestic product (GDP) in 2015 and 28.6% of total exports in the world consist of agricultural goods (World Bank, 2016). In addition, agriculture plays a major role in enhancing food security and poverty alleviation in Sub- Saharan Africa (SSA), contributing on average 15% of total GDP. The contribution of agriculture to GDP in SSA ranges from below 3% in Botswana and South Africa to more than 50% in Chad, indicating the existence of different economic structures in SSA (FAO, 2015).

In Kenya, agriculture is the backbone of the economy, contributing 26% of the annual GDP directly and 25% indirectly, and accounting for 18% of the total formal employment. The sector accounts for 65% of Kenya‟s total exports, provides more than 60% of informal employment in rural areas and contributes over 75% of industrial raw materials (Nyamwamu, 2016). The development of the sector is anchored on improving productivity, commercialization and competitiveness of agricultural commodities, and efficient management of factors of production which is key to food security. To achieve this, the Kenya Vision 2030 sets the following targets for the agricultural sector by 2020; reduction of food insecurity by 30%, reduction of people living below absolute poverty lines by 25%, increase agriculture‟s contribution to GDP by Kshs. 80 billion per year (GoK, 2009).

1

1.1.2 Status of Coffee Production and Marketing Coffee is among the most traded tropical commodity worldwide accounting for nearly half of total exports of tropical products (FAO, 2008). Africa accounted for 12.8% of world coffee output, with Ethiopia and Uganda accounting for 62% of sub-Saharan Africa‟s coffee output in 2017/2018. Coffee is Kenya‟s fourth leading foreign exchange earner after tourism, tea, and horticulture (MoA, 2011). In 2015, total coffee production in Kenya was 42,037 metric tonnes, of which 44% came from coffee estates with the balance (56%) coming from smallholder coffee farms which are organized through producer cooperatives (AFA, 2016). The area under coffee production increased from 109,000 hectares in 2012/2013 crop year to 115,570 hectares in 2018/2019 crop year (ICO, 2019). In Embu County, coffee is one of the major industrial and export crop whereby nearly 70% of the crop is grown by smallholder farmers (GoK, 2013b).

Despite increase in area, gross returns have been on decline. Export licensing, growing inequality to value addition, minimum volumes for export and quality standards act as entry barriers to small scale coffee farmers to international markets leading to reduced economic incentives and low profit margins. Returns from coffee are majorly influenced by international market prices and therefore beyond the farmers‟ control (AFA, 2016). Increasing coffee productivity, which is largely within the farmers‟ control, would therefore mitigate the cost of production and improve incomes (AFA, 2016). High quality coffee produced by appropriate harvesting methods and good post-harvest practices will also guarantee high prices. This in turn would guarantee economic incentives and more competitiveness in the international markets, which would effectively maintain quality and productivity of coffee in Kenya.

Coffee growing in Kenya is done under the following agro ecological zones; Upper Midland (UM) 1 zone (coffee-tea zone), UM 2-3 (proper coffee zone) and UM 3-4 (marginal coffee zone) (Jaetzold et al., 2009). Coffee is largely grown in the UM 2 agro ecological zone which has a good yield potential (Gichimu & Omondi, 2010). Coffee is a highly environmentally dependent crop and increase in temperatures can substantially decrease the yields and quality of coffee (Gichimu & Omondi, 2010). In Kenya, climate change has rendered a significant proportion of traditional coffee growing zones less

2 suitable for coffee production (Cheserek & Gichimu, 2012) which has led to a shift in production suitability from optimal to sub optimal and marginal growing zones, resulting to changes in crop yields and quality.

1.1.3 Coffee Research in Kenya Stagnating coffee productivity has been a major policy concern in Kenya (GoK, 2009). This has led to increased investment in development and dissemination of yield- enhancing technologies. To deal with yield and quality variability, it is important to ensure that the yield-enhancing technologies developed are able to increase yields substantially and maintain high quality production. The national blue print Vision 2030 recognizes the role of research in improving coffee quality and also places a greater emphasis on value addition in coffee to boost household incomes (GoK, 2009). Agricultural Sector Development Strategy (ASDS) aims to streamline, rationalize and enhance coordination of agricultural research so that the sector can achieve the target of 10 percent annual economic growth envisioned in Kenya Vision 2030 (GoK, 2013a).

Coffee research in Kenya started in 1908 at Scott‟s Laboratories currently National Agricultural Laboratories (GoK, 2013a). In 1963, the Kenya government gave the research mandate directly to farmers. This led to formation of Coffee Research Foundation (CRF). The mandate of CRF was to promote research and also investigate all issues related to coffee and other agricultural and commercial systems associated with coffee (GoK, 2013a). The foundation established research stations across the country with substations in Kisii, Kitale, Mariene in Meru and Koru in Kipkelion. The Government of Kenya, under the Kenya Agricultural and Livestock Research Act of 2013, reformed the National Agricultural Research Systems through the creation of the Kenya Agricultural and Livestock Research Organization (KALRO). This was meant to restructure agricultural research into a dynamic, innovative and demand driven towards achieving Vision 2030. CRF was consequently transformed into Coffee Research Institute (CRI), one of the institutes under KALRO. CRI was mandated to expedite adequate access to research information, innovation and technology and promote application of research findings on coffee at the field level (Mati, 2016).

3

Coffee is one of the scheduled crops under the Crops Act 2013 (GoK, 2013c). There is therefore immense need to promote the crop by establishing linkages with government and private research institutions, disseminate market information, conduct farmers training programs, and develop varieties suitable for different agro-ecological zones (GoK, 2013c). These efforts are made to enhance productivity, product quality and competitiveness both in local and global markets. Increase in coffee productivity would reduce poverty, increase household incomes which would stabilize market prices hence increasing household consumption and saving. High quality coffee will compete globally, and guarantee high prices and market access to enhance profitability and incomes to rural populations. In undertaking its mandate of conducting research in all areas of production, processing and marketing of coffee, CRI has developed various technologies aimed at boosting productivity of coffee and reducing the production cost.

1.1.4 Recommended Coffee Production Technologies and Management Practices The research in coffee has developed and recommended technologies for coffee farmers in order to increase the productivity of the crop for increased household incomes and poverty alleviation among the small-scale farmers. The key recommended technologies (CRI, 2017) are discussed hereafter:

Varieties: The improved coffee varieties are Ruiru 11 and Batian which are resistant to coffee berry disease (CBD) and coffee leaf rust (CLR). Traditional varieties are K7, SL28 and SL34.

Spacing: For Ruiru 11 the recommended spacing is 2m × 2m (2500 trees per ha), Batian the spacing is 2.1m × 2.5m (1900 trees per ha), while for traditional varieties (K7, SL28 and SL34) is 2.75m × 2.75m (1300 trees/ha).

Fertilizer: NPK (17:17:17) is recommended at the rate of 250 g/tree six months before flowering and CAN is recommended at a rate of 300 g/tree per year and should be applied in two equal splits after the main flowering. Application of Zinc/Boron foliar fertilizer is recommended for application two months before flowering at rate of 2-3 kg of each per acre.

4

Disease control: For traditional varieties, copper based fungicides at 50 percent formulations are recommended for control of CBD and CLR at the rate of 7.7 kilograms per hectare or using tank mixtures of copper fungicides and organic fungicides as detailed in Coffee Research Institute Technical Circular No. 804 (CRI, 2018). Straight organic fungicides are also used for CBD control at the time when CLR control is not required.

Pest control: For control of insect pests such as berry borer, berry moth, thrips, antestia bugs and stem borers, different pesticides are recommended at different rates. It was important to find out whether farmers adhere to these recommendations and their effect on coffee productivity. Timely pruning, handling and de-suckering are also recommended for effective pest management.

Canopy management: The newly planted trees should be raised on a single head until the change of first cycle and then one or two heads in the subsequent cycles. Capping is recommended at a height not exceeding 6ft (1.83m) for easy harvesting and incase of uncapping, stems should be limited to 2-3 bearing heads.

Weeding: For control of annual weeds, application of foliar herbicides such as Sulfosate (Touch-down) is recommended at a rate of 0.5-1 litre per acre using low volume flood nozzle jets. Application of pre-emergence soil herbicides, such as Atrazine 900WDG or Simazine 900 WG (for broadleaf and grass suppression) are recommended at a rate of 5kg/ha before weeds emerge. These are just a few examples among others.

1.2 Statement of the Problem The coffee subsector in Kenya has been characterized by declining and low productivity at farm level which is of utmost concern to smallholder farmers and the government. Coffee productivity, like other crops, has been on the decline in the last five years, which means reduced household income, particularly in smallholder agriculture. Over the years, Coffee Research Institute (CRI) has been carrying out on-station research on coffee management, but has conducted limited on-farm research to assess the impact of research recommendations on coffee performance at farm level. It is at the farm level where the recommended technologies interact with the prevailing socioeconomic

5 environment. In addition, research centers carry out experimental studies on one recommendation at a time, at the station or demonstration plots.

Research centers have been carrying out on-station experimental research about the new technologies with little or no on-farm research to assess the economic impact of the released technologies given the prevailing socioeconomic factors at the farm level. On the other hand research centers have been conducting studies on one recommendation at a time over a certain period of time before releasing it to the farmers. CRI has been conducting research on field trials and experiments for specific input or recommendation without consideration of input variations and interaction of socioeconomic characteristics at the farm level. There is no quantitative information on site specific constraints on yield gaps attributed to factor-factor substitution of production factors used in the production process.

None of these studies have made emphasis on the economic impact of the recommended technologies on productivity variations across the households. These research centers give little consideration of the possible effects the technologies could have on one another under the influence of farm and farmer characteristics at the farm level. Therefore, there exists limited empirical information on the effect of these technologies on the performance (productivity, profitability and quality) of coffee at the farm environment. The current study fills this knowledge gap by analyzing the combined effect of the recommended technologies and farm socioeconomic factors on coffee productivity, profitability and quality in Kenya, using a case study of smallholder farmers in Embu County.

1.3 Justification In Kenya, agriculture plays a major role in poverty alleviation, rural development, economic growth and food security enhancement. In addition smallholder agriculture is seen as a vital development tool for achieving Sustainable Development Goals (SDG) number one and two which aim at ending poverty and hunger respectively, through promotion of sustainable agriculture (World Bank, 2015). The coffee industry has been one of key pillars of Kenya‟s economy. Currently, the industry contributes 8% of Kenya‟s total agricultural exports and benefits about 5 million people directly and

6 indirectly. Over the years, CRI and other research institutions have developed various technologies whose appropriate adoption at farm level is hypothesized to reverse the current declining trend in coffee production. Optimum input combination could be the solution to low productivity, quality and profitability.

The current study fills the knowledge gap that exists between research and technology adoption at farm level. This study endeavors to analyze the impact of already released coffee technologies on productivity at farm level in order to help in addressing the technology adoption challenges among small-scale farmers. It is expected that adoption of these recommendations will result in increased production of high quality coffee which will attract high prices in the international market thus increasing the farmer profit margins and improving farmers‟ livelihoods. The study will also assist in priority setting in developing demand driven technologies in future and developing policies that guide on technology implementation and best practices among small-scale coffee farmers. In depth analysis of the effect of adoption of recommended technologies on total factor productivity, quality and profitability of coffee will be a guide on input allocation and production potential to offset the production cost and protect farmers against international price shocks for increased profit margins.

1.4 Hypotheses The study tested the following hypotheses: 1. Adoption of recommended coffee production technologies have no significant effect on coffee productivity among smallholder coffee farmers in Embu County. 2. Adoption of recommended coffee production technologies have no significant effect on coffee profitability among smallholder coffee farmers in Embu County. 3. Adoption of recommended coffee production technologies have no significant effect on quality of coffee among smallholder coffee farmers in Embu County.

1.5 Research Objectives 1.5.1 General Objective of the Study The overall objective of the study was to analyze the economic impact of recommended technologies on coffee performance (productivity, profitability and quality) among small-scale farmers in Embu County.

7

1.5.2 Specific Objectives Specifically the study sought;

1. To analyze the effect of adoption of recommended coffee production technologies on productivity among the small-scale coffee farmers in Embu County. 2. To analyze the effect of adoption of recommended coffee production technologies on profitability among the small-scale coffee farmers in Embu County. 3. To evaluate the effect of adoption of recommended coffee production technologies on quality of coffee among the small-scale coffee farmers in Embu County.

1.6 Assumptions The study was based on the following assumptions. 1. That the main objective of smallholder farmers was profit maximization through increased productivity and enhanced quality as a result of adoption of the recommended technologies at farm level. 2. That some farmers had already adopted the recommended technologies whereby these farmers were perceived to be risk averse and they required certainty that these technologies and practices would guarantee them economic returns. 3. That the farmers‟ primary objective is cost minimization through least cost combination principle and optimum allocation of scarce production resources at the same time maintaining the level of production.

8

CHAPTER TWO LITERATURE REVIEW 2.1 Introduction The chapter reviews previous studies which have been conducted on the effects of recommended technologies on coffee productivity, profitability and quality among smallholder coffee farmers. The chapter identifies the gaps which exist in the reviewed literature, and also contains the theoretical framework, conceptual framework and operationalization of variables.

2.2 Effect of Recommended Technologies on Productivity Productivity can be defined as the ratio of output produced to inputs used in a production process (Muzari et al., 2012). There is a large gap between what the smallholder farmer gets and what is feasible, given the available technology (Muzari et al., 2012). Previous studies have found the use of modern agricultural technologies to cause an increase in agricultural productivity (Muzari et al., 2012). Available literature has shown that three main ways have explained growth in agricultural productivity. That is expanding the acreage under crop production, use of scientific research to generate high yielding and disease resistant varieties and finally increased efficiency in allocation of scarce production resources for output maximization (Bocher & Simtowe, 2017).

2.2.1 Coffee Varieties and Productivity A previous study by Gebeyehu (2016), reported positive and significant effect of improved seed on coffee productivity in Ethiopia as it was complimentary to inputs like fertilizers and other chemical inputs. He reported that improved coffee cultivars reduced chemical use due to resistance to major coffee diseases such as coffee berry disease (CBD) and coffee leaf rust (CLR). Kamau et al. (2016) found that Ruiru 11 and Batian coffee varieties were critical determinants of technical efficiency in coffee farming. This revealed a positive relationship between coffee variety and productivity. Van der Vossen et al. (2015) reported positive and significant effect of CBD and CLR resistant cultivars on coffee productivity. Meylan et al. (2017) found no significant effect of coffee varieties on coffee output variations across the coffee plantations in Costa Rica.

9

A study by Andrew & Philip (2014) on coffee production in Region, revealed that there were differences in coffee productivity resulting from differences in the varieties grown. The productivity of Arabica species, which was mainly grown in Kigoma District, was higher compared to Robusta species, mainly grown in Buhigwe District. This implied that the two cultivars of coffee grown in the two regions significantly explained coffee productivity variations. Haggar et al. (2017) found that the disease resistant cultivars for both CBD and CLR had significant effect on coffee productivity on sustainably certified coffee farms. These cultivars were resistant to common diseases with reduced usage of agrochemicals hence increasing the environmental economic trade off and sustainable agricultural production on the arable farms.

2.2.2 Farm Inputs and Productivity A study by Thuku (2013) on effects of reforms on productivity of coffee in Kenya found that labour, fertilizer, land and policy reforms implemented by the government explained 61.7% of the variations on coffee productivity in Kenya. The study found that the use of recommended fertilizer was positively related to productivity and significant in influencing coffee productivity. The effect of foliar fertilizer was found to positively and significantly affect the yields of tea grown in Kenyan highlands (Njogu et al., 2014). Dzung et al. (2013) in their study on the effect of compost application on growth and yield of coffee revealed that substituting part of chemical fertilizers with compost from coffee husk, which had rich organic matter, increased coffee yield by up to 14%. A study by McArthur & McCord (2017) found that the coefficient for the recommended fertilizer was strongly significant, implying increased coffee productivity in Mexico.

Inorganic fertilizers were found to increase coffee productivity under increased irrigation water, while organic manure performed better in water stress conditions (Chemura, 2014). The study reported positive interaction between irrigation water levels and organic fertilizer in affecting height and growth performance of coffee plants under low water supply level, pointing the key importance of organic manure in regulating soil water. A study by Worku and Astatkie (2015) revealed that increased water supply through irrigation is increasingly becoming important in coffee production given climate

10 change and unreliable rainfall. Organic manure plays a significant role in maintaining organic matter and moisture retention as a climate change adaption strategy and sustainable agriculture (Chemura, 2014).

Chemical inputs such as herbicides and pesticides are expected to protect coffee trees from pests and diseases and reduce weeding time and cost (Gebeyehu, 2016). Use of recommended pesticides and insecticides has been reported to have significant positive influence on coffee production (Ngeywo et al., 2015). They attributed the reduction in coffee production in Western Kenya as a result of low usage of insecticides and pesticides. Gebeyehu (2016) also reported positive effects of using recommended herbicides, pesticides, manure and inorganic fertilizers on coffee productivity in Ethiopia. Carvalho et al. (2014) found that herbicide use positively influenced coffee productivity and water use efficiency through controlled transpiration in case of water stress in Brazil. Adejumo (2005) found application of fungicides especially copper based formulations and organic formulations to be effective in control of CBD and CLR, hence increased coffee productivity.

2.2.3 Agronomic Practices and Productivity Appropriate pruning and shading have been shown to be important agronomic practices in control of coffee diseases for increased coffee output (Adejumo, 2005). Meylan et al. (2017) reported positive and significant effect of pruning and shade trees on coffee yields and argued that shade trees helped in nitrogen fixation which was taken up by adjacent coffee trees. Ameyu (2017) reported positive and significant effect of selective harvesting and pruning on coffee yields and overall quality. This allowed for air circulation between the coffee branches and reduced disease and pest incidences as dense canopy would create a conducive environment for coffee pests. On the other hand, tree capping has been shown to negatively affect coffee productivity by reducing coffee yields and quality of berries produced per unit area (Magha, 2013). Apparently, capping reduces apical growth of coffee trees leading to dense canopy which affects application of other agronomic practices such as disease and pest control and cherry harvesting.

Several other studies also revealed that systematic management practices and optimum spacing gave higher productivity scores (Läderach et al., 2011; Odeny, 2016; Belay et

11 al., 2016). Van Long et al. (2015) reported that optimum spacing and coffee pruning was significant in explaining variations in coffee output. Crop management practices have also been shown to have a significant effect on yields and quality in other crops. A study by Ghosh & Bera (2014) found that tree pruning significantly affected fruit yield in oranges. Worku and Astatkie (2015) found row spacing to have a significant effect on yield and increased fruit retention of soy bean. Similar findings were reported by Markos et al. (2012).

2.3 Effect of Recommended Technologies on Coffee Profitability Effective technology development must ultimately increase the farm‟s profits or decrease its losses (Afolami et al., 2015). Studies by Andrew & Philip (2014) and Mohammed et al. (2013) noted increased returns from technology usage although they did not consider all technologies and all factors of production.

2.3.1 Coffee Varieties and Profitability van der Vossen et al. (2015) found a positive impact of CBD and CLR resistant cultivars of Arabica coffee on profitability due to sustainable coffee production, reduced use of chemicals, and better adaptation to climate change. A study by Andrew & Philip (2014) on coffee production in , Tanzania revealed that coffee profitability varied between the two districts in the region. The profit gained from coffee production in Kigoma District was higher than in Buhigwe district. This was due to differences in productivity resulting from differences in the varieties grown. The productivity of Arabica species, which was mainly grown in Kigoma District, was 0.635 kg per tree compared to 0.52 kg for Robusta species mainly grown in Buhigwe District. van der Vossen et al. (2015) reported that Arabica coffee varieties fetched higher prices than Robusta coffee varieties due to higher beverage quality. They argued that disease resistant cultivars were more profitable and had more sustainable crop production. Studies by Minai (2014) and Muriithi (2016) concluded that adoption of improved varieties of coffee was positively related to household income. Mohammed et al. (2013) also reported a positive effect of improved varieties on coffee profitability in Nigeria. Studies on other crops also reported positive effects of improved varieties on profitability (Kolawole & Ojo, 2007; Afolami et al., 2015; Nguezet & Diagne, 2011).

12

2.3.2 Farm Inputs and Coffee Profitability Integrated fertility management, which entails the use of both organic and inorganic fertilizers as a climate change strategy, was found to increase profitability due to reduced cost of fertilizers and improved coffee growth and output (Chemura, 2014). Recycling of coffee wastes such as pulp as direct inputs or in combination with green manure as substitutes for inorganic fertilizers increased coffee profits (Chemura et al., 2010). A study by van der Vossen et al. (2015) found that reduced usage of fungicides for control of CBD and CLR, and herbicides for control of soil weeds, increased coffee profitability. Reduced agrochemical use through adoption of disease resistant cultivars of Arabica coffee is also a more sustainable crop production strategy considering the negative impacts of climate change resulting from the use of chemicals (van der Vossen et al., 2015). Haggar et al. (2017) reported similar findings on the effect of disease resistant cultivars on coffee profitability on sustainably certified coffee farms. This was as a result of reduced management and farm operations such as labour and costs on pest and disease control. Mohammed et al. (2013) reported a positive effect of organic manure on coffee returns. Organic manure increased the marketable value of coffee cherry compared to inorganic fertilizers and hence coffee profitability. Similar findings were reported by Chemura et al. (2010) and McArthur & McCord (2017).

2.3.3 Agronomic Practices and Coffee Profitability Pruning is an important management practice in coffee farming. Pruning of coffee trees maintains the tree canopy and facilitates field operations such as pest and disease control. This is turn improves coffee productivity and quality hence increased coffee returns at the farm level (Bigirimana et al., 2019). Bravo-Monroy et al. (2016) reported pruning as a conventional coffee management practice that is an economic driver of profitability. Pruning increases air flow and light penetration in the tree canopy resulting in high quality cherry which in turn fetches a premium price in the market (Bravo- Monroy et al., 2016). Appropriate harvesting methods also produce high quality coffee which provides higher economic incentives to farmers as well as increased coffee productivity (Bravo-Monroy et al., 2016).

13

Van Long et al. (2015) found a positive effect of pruning and optimum spacing of coffee trees on coffee returns. Proper spacing would ensure optimum plant population per unit area which would in turn increase productivity per tree hence higher net returns. Similar observations were made by Läderach et al. (2011) and Belay et al. (2015) who reported positive effects of pruning on coffee returns. However, tree capping was found to negatively influence coffee productivity and profitability (Magha, 2013; Odeny, 2016). This was as a result of dense canopy which interfered with farm operations such as pest and disease control and foliar feed application. This in turn reduced productivity per tree and also the quality of cherry, hence reducing the marketable value of the coffee cherry. There is need to quantify the relationship between these agronomic practices and coffee profitability.

2.4 Effect of Recommended Technologies on Coffee Quality Coffee quality can be defined as the inherent and distinctive attributes or characteristics possessed by coffee cherry measured in terms of grade, bean size and weight etc. at the farm level (Roba, 2017). High quality coffee with the best physical attributes will guarantee economic incentives with higher gross returns.

2.4.1 Coffee Varieties and Coffee Quality A study by Läderach et al. (2011) on systematic agronomic farm management for improved coffee, found that there was a significant quality difference in terms of beverage quality, fragrance, berry weight and size between the Red Caturra and the Yellow Caturra varieties in Mexico. Studies by Sualeh et al. (2014), Tirfe et al. (2015) and Ameyu (2017) found that coffee varieties were positively related to bean weight and size of coffee cherry produced. A study by Tsegaye et al. (2014) also reported a significant effect of coffee variety on the bean size and grade. They also reported significant variation between cherry layer thickness and coffee quality. These studies have concentrated on controlled conditions and experiments, hence the need to determine the effect of recommended coffee varieties on coffee quality at the farm level, where there is the influence of farm and farmer characteristics among other factors.

14

2.4.2 Farm Inputs and Coffee Quality Studies conducted on factors influencing production of quality coffee showed that pesticide and fertilizer applications have a significant influence on quality of coffee beans (Laderach et al. 2011; Bote, 2016). Application of coffee husk compost was reported to increase coffee quality by reducing the rate of fallen coffee berries from 25% to 18% (Dzung et al., 2013). A previous study by Abasanbi (2010) found that compost application was positively and significantly related with coffee quality as it increased soil aeration and microorganisms‟ activity. The study also revealed that use of CBD controlling fungicides had a positive influence on coffee quality. The fungicides reduced this disease that directly affects the cherries leading to production of immature and damaged berries of low quality.

Bote & Vos (2017) reported positive and significant effect of fertilizer application on bean size, weight and raw bean quality among Arabica coffee cultivars in Jimma State, Ethiopia. Castro-Tanzi et al. (2012) also reported that inorganic fertilizer and fungicide application had a positive effect on coffee quality. The study also revealed that nutrient application rates had a positive significant relationship with coffee yields. On another note, Andrew and Philip (2014) found that high cost of inputs, particularly for fertilizer and agrochemicals, was a major constraint to production of quality coffee. The study revealed that the price of home processed coffee was lower than the one processed at the Central Pulpery Unit (CPU). In Kenya, there exits limited research information on the effect of farm inputs on quality of coffee at farm level. This study will provide an understanding of the effect of these recommended farm inputs on coffee quality.

2.4.3 Agronomic Practices and Coffee Quality Läderach et al. (2011) found spacing and pruning to positively influence coffee quality with optimum spacing giving higher scores for quality. The study also found that coffee from denser canopy had higher quality scores than coffee from more open canopy. They further reported that berries from medium levels of canopy had the highest score when analyzed for beverage quality. This was supported by Bote & Struik (2011) and Odeny (2016). However, it contradicts the findings of Pinard et al. (2014) who found that canopy did not affect the final grade of green coffee. A study by Bote & Vos, (2017)

15 reported positive and significant effect of shade trees, pruning and fruit thinning on weight and bean size and also bean quality in Ethiopia.

Timely harvesting was also found to be positively related to coffee quality (Läderach et al. (2011). A study by Ameyu (2017) found that selective harvesting of coffee increased total raw coffee quality by 16% as compared to strip harvested coffee. Selective picking of only ripe red cherries improved total raw quality in terms of bean size, weight, and grade of coffee beans. However the study concluded that the effects of harvesting and post-harvest processing did not significantly affect the overall total coffee quality and grading. Kebati et al. (2016) found that shading and canopy management significantly affected the total loss due to effect of CBD and physiologic fall. The study also revealed that unpruned coffee had higher total loss of 74.9% compared to the pruned coffee at 64.42%. Despite these studies, there is inadequate knowledge on the impact of recommended agronomic practices on coffee quality at farm level.

2.5 Research Gap Evidence from the available literature has shown that many studies have considered factors that determine adoption of production technologies (Akudugu, 2012; Challa & Tillaum, 2014; Chepng‟etich et al. 2015; Belay et al. 2016; Kamau et al. 2016). These studies only focused on the determinants of technology adoption and factors that influence adoption decisions but failed to determine the effect on farm productivity, profitability and quality. Several studies have used data from field trials and expert opinions on induced effect of different farm inputs on per crop basis (Chemura et al. 2010; Dzung, 2013; Popp et al. 2013; Chemura, 2014). They construct alternative production scenarios for each crop to estimate changes in input use. One of the shortcomings of these studies is that field trials can hold constant all production factors while ignoring substitution possibilities. Therefore this study contributes to the existing literature by accounting for the effect of the recommended technologies, production factors and socioeconomic factors with consideration of substitution possibilities in explaining coffee output variations among the farms under study.

16

2.6 Theoretical Framework The study is based on the theory of the firm that the main objective of coffee farmers is profit maximization through cost reduction and increased production. The theory of the firm has two components; theory of production and theory of cost.

Theory of Production Production theory deals with the economic process in which production inputs are converted into output suitable for use (Hancock, 2012). Profit maximization is the ultimate goal for coffee producers. Production theory guides the producer in determining a profit-maximizing output-input combination. A farm is said to maximize profit at the level of output in which the marginal value product is equal to the unit factor price. Marginal value product is the value of additional output obtained from an additional unit of input. In the theory of production, the production firm is assumed to choose input- output combination which maximizes profit during the production process (Hancock, 2012). However, utility maximization is the primary goal of a household, which deviates from pure profit maximizing behavior assumed in theory of production (Gebeyehu, 2016). The farm-household therefore pursues both profit maximization as well as utility maximization.

Theory of Cost Cost theory constitutes allocation of factors of production in proportions that minimizes the total cost for each level of output derived from the production process (Uzawa, 1964). The farmer is assumed to choose input combination that minimizes the value of inputs given their market prices (Uzawa, 1964). The underlying assumption of the cost theory is that producers take input prices and wage rates as given and try to optimize production through cost minimization. Total cost depends on chosen output level, given the vector of input prices and the production function (Diewert, 1974). Profit maximization holds the assumption that any inefficiency in the production process is translated into reduced profit (Bocher & Simtowe, 2017). Choosing the optimal input combination to produce a given level of output is referred to as optimizing the factor- factor relationship (Hill, 2014). The optimum factor combination or the least cost combination entails the combination of production factors with which a farm can

17 produce a given amount of output at the least cost possible. For profit maximization, a rational farmer would combine various factors of production, given the production function, in such a way that maximum coffee output is achieved at the least cost.

Cobb-Douglas Production Function Stochastic Cobb-Douglas Production function model was used because it has the advantage of allowing for statistical inferences and the estimated coefficients are also easy to interpret. The model provides a convenient form of aggregation of the factors of production with other factors that affect production (Shepherd, 2015). The estimated coefficients of all parameters of the production function give the marginal effect of each variable used on output. Hence, it can be used to test the specification as well as different hypotheses on the error term and on all the other estimated parameters of the production frontier (Seyoum et al., 1998). The Cobb-Douglas production function was used because it is general and flexible and can allow for analysis of interactions among the predictor variables. The vector of inputs used in production of coffee output includes the variable production factors, recommended technologies and the farm socioeconomic factors. The general form of Stochastic Cobb-Douglas production function model as specified by Aigner et al. (1977) is as shown.

ln Y  ln    ln X   Z   D   o i i i j j j k k k ……………………… (1)

Where ln = Natural logarithm, Y = observed coffee output, 0 ,i , j and  k are vector parameters to be estimated, Xi = quantity of inputs, Zj = values of socio-economic factors, Dk = dummy variables for adoption of recommended technologies (1 = adopted, 0 = non adoption) and  = error term.

The Profit Function The profit function is a mathematical relationship that relates the production inputs, factor prices and socioeconomic factors to the maximum profit level attainable at those output prices and factor prices (Bocher & Simtowe, 2017). The profit function model was used since the function is able to derive indirect estimates that link the coefficients of the profit function with those of the production function (Adesina & Djato, 1996). The estimated profit function is characterized by increasing returns to scale holding

18 some factors constant (Hancock, 2012). Profit function was used to analyze the combined effect of recommended technologies, factors of production used and the farm socio-economic factors on coffee profitability. The general stochastic profit model is specified as applied by (Adesina & Djato, 1996).

w x ln  *  ln A   D   lnW   ln M   ln Y   Z   i i    k k  i i  j j  n n w *  ... (2)

Where  * = normalized profit, ln = Natural logarithm, ln A = Constant or intercept,  = parameter estimates, Dk = Dummy variables for recommended technologies, Wi = normalized factor prices, M = land area under coffee in acres, Y j = sum of costs of variable inputs, Z n = Dummy variables for socioeconomic factors, wi = wage rate normalized by price of coffee, xi = number of man days of labour used in production and  = Error term.

Binary Logit Model The study used Binary Logit model to analyze the combined effect of recommended technologies, inputs and farm socioeconomic factors on coffee quality. The model was selected due to the ease of interpretation of the coefficients in terms of the coefficients and odds-ratio. The model‟s mathematical convenience and the simplistic nature of interpreting the parameter coefficients, made it easy to apply in this study. The logistic model can be written in terms of the odds and the log of odds (odd-ratio). The odds ratio is the probability that a farm would produce high quality coffee ( Pi ) to the probability of producing low quality coffee (1 Pi ). The odd-ratio is specified as:

 P   i   e Zi 1 P   i  ………………………………………………..…………...... (3)

Given the natural logarithm of equation 3 yields

 P   i  ln   Zi    i X i   2 X 2  ....   n X n U i ………………...…..... (4) 1 Pi 

19

Where  and  are the parameter estimates, X i are the explanatory variables and U i is the disturbance term.

2.7 Conceptual Framework The main objective of this study was to analyze the economic impact of recommended technologies on coffee performance at the farm level. The study considered the impact of these recommendations after interactions with the prevailing socioeconomic factors at the farm level such as farm and farmer characteristics. The performance indicators considered were productivity, profitability and quality of coffee produced. The recommended technologies and other factors were expected to affect coffee performance as conceptualized in Figure 2.1.

20

Fig. 2.1 Conceptual Framework

21

2.8 Operationalization of Variables The nature of the dependent and independent variables in the conceptual framework and their apriori interactions are as described hereafter:

Productivity: Crop productivity can be defined as the total crop yield per unit of input used (Muzari et al., 2012). It was measured as a continuous variable in this study. Coffee productivity was measured using three ways; output per acre, output per tree and output per man-day. Output per acre was determined by calculating the ratio of quantity of coffee cherry in kilograms produced by the farmer and the total land size in acres under coffee. Output per tree was computed as the ratio of quantity of cherry in kilograms produced to the plant population. Output per man-day was determined as a ratio of total cherry produced in kilograms to total man-days of labour used in the production process.

Profitability: Crop profitability can be defined as a measure of firm‟s financial performance per unit of input or output (Uddin et al., 2016). In the current study, crop profitability was measured as a continuous variable. Profitability of coffee production was analyzed using two approaches; gross margin per acre and gross margin per tree. Gross margin per acre was determined as a ratio of the difference between the gross return in Kenya shillings and the total variable cost of production to land size under coffee. Gross margin per tree was determined as the ratio of the difference between gross return in shillings and total variable cost of production to plant population.

Quality: Quality of coffee may vary across the producer-consumer value chain. At farm level, coffee quality can be defined as physical attributes that determine the quality potential of coffee such as bean size, weight and grade which guarantee high prices (Roba, 2017). To determine the quality of coffee produced by the farmer, two measures were used in this study; proportion of “mbuni” to quantity of cherry produced by the farmer, and the total loss per tree (in kilograms).

Adoption of Recommended Varieties: The recommended coffee varieties for Embu County are Ruiru 11 and Batian varieties which are resistant to CBD and CLR (CRI, 2017). Adoption of the recommended varieties was measured as a dummy variable taking a value of 1 for adoption and 0 for non-adoption. Those farmers who planted any

22 of the two varieties were considered to have adopted while those who planted traditional varieties were considered not to have adopted.

Adoption of Recommended Fertilizer rate: Adoption of recommended fertilizer rate was measured as a dummy variable taking the value of 1 for adoption and 0 for otherwise. The current study considered two aspects of the recommendation associated with fertilizer application: quantity applied and the timeliness of application. Application of the recommended fertilizer rates was expected to be positively related to productivity, profitability and quality of coffee.

Adoption of Recommended Spacing: The adoption of this recommendation was measured as a dummy variable taking the value of 1 for application of recommended spacing and 0 for otherwise. The recommended spacing for each variety determines the plant population per acre for the variety. Spacing was measured in terms of distance (in metres) between the trees and that between the rows and compared to the recommended spacing as an indicator of adherence. Adoption of recommended spacing was expected to have a positive effect on coffee yield, profitability and quality.

Adoption of Recommended Chemical for disease, pest and weed control: Farmers with improved varieties were expected not to apply chemicals for disease control. The adoption of recommended rates of chemicals used in disease, pest and weed control was measured as a dummy variable taking a value of 1 for adoption and 0 for otherwise. Adoption of recommended rates of chemicals was expected to have a positive effect on coffee productivity and quality but negative or positive effect on profitability.

Adoption of Recommended Canopy Management: The canopy management practices considered in this study were capping height in metres and the number of bearing heads. Capping was measured as a categorical variable taking the value of one for the farms in which capping was done at the recommended height and value of zero for the farms that did not. The number of the bearing heads relative to the recommended 2-3 heads was used as a measure of adoption of the recommended bearing heads.

23

Intervening variables: These were farmer and farm characteristics that were hypothesized to influence the impact of the recommended technologies on coffee performance. The intervening variables were categorized into social, economic and institutional factors (Akudugu et al., 2012). Social factors can be defined as factors related to social welfare, culture and lifestyle of the farmer in a community which are likely to influence the farmer‟s capability to apply the recommended technologies. Economic factors are resource related factors of the farmer and household which include among others farm size and off farm income. Institutional factors are services offered to farmers by institutions such as extension and credit which are likely to influence the farmer‟s capability to adopt the recommended technologies. The following are the detailed descriptions of the farm and farmer characteristics that were hypothesized to influence the impact of the recommended technologies on coffee performance.

Age: Age of the farmer was measured as the number of years the farmer has lived and was measured as a continuous variable. Age was expected to have mixed effect either positive or negative as young farmers may have more knowledge about the technology while the older farmers may have more experience and resources to use the technologies.

Education: Education was measured as a categorical variable in terms of levels of formal schooling by the farmer. The levels of schooling considered were none, primary, secondary and tertiary levels of Kenyan education system. High level of education was associated with diverse knowledge base on different technologies. Level of education was expected to positively influence crop productivity and usage of recommended technologies as it increases farmers‟ ability to use resources efficiently.

Experience: It was measured as the number of years the farmer has been engaged in coffee farming and was a continuous variable. Increase in level of experience was hypothesized to positively influence coffee productivity, profitability and quality. Older farmers would have resources and more understanding of the knowledge about production technologies and possible strategies to mitigate risks associated with the new technologies.

24

Access to off-farm income: This was measured as a dummy variable taking the value of 1 for access to off farm income and 0 for otherwise. Off-farm income is defined as cash sourced from any other entity other than farming. It was hypothesized to positively influence usage of the recommended technologies as it was considered as a source of capital that can be used in production process. Off-farm income was expected to cushion against credit constraints. This is expected to have a mixed effect either positive or negative on coffee performance.

Farm size: This is the total land acreage owned by the farmer in acres and it was measured as a continuous variable. Farm size was hypothesized in this study to have a mixed influence on the usage and effect of recommended technologies on coffee performance. Large scale farmers were likely to spend more on efficiency improving technologies than small scale farmers who may use the technologies to increase productivity per unit area hence the expected effect on coffee performance was either positive or negative.

Land tenure: This refers to the ways in which property rights to land are allocated, transferred, used or managed in a society. It was measured as a dummy variable with 1 if households had a title deed and 0 if otherwise. With increased land ownership rights, farmers tend to use and adopt more recommended technologies to increase productivity. On the other hand if they do not have ownership rights they tend to be more risk averse. Security of land tenure was hypothesized to have a positive influence on coffee productivity, profitability and quality as the farmers could make long-term production decisions.

Access to extension services: This entailed technology transfer to the farmer through farm visits and farmer trainings. Access to extension was measured as a dummy variable taking the value of 1 for farmers who received extension services and 0 if otherwise. Access to extension was hypothesized to positively influence coffee productivity, quality and profitability as it would facilitate uptake of technologies by the farmers.

Credit access: It was measured as a categorical variable with 1 for households with credit access and 0 for otherwise. Access to credit would raise the purchasing power of

25 the farmer to finance production activities and help solve liquidity constraints. Credit access was therefore hypothesized to positively influence farm productivity as it would finance purchase of key farm inputs.

26

CHAPTER THREE METHODOLOGY 3.1 Introduction The chapter provides an overview of the research methodology which was employed in the study. The chapter gives a description of the study area, research design, target population, sample size, sampling procedure, data collection, accuracy and reliability of research instruments used and data analysis.

3.2 Study Area The study was conducted in Embu County in Eastern Kenya, where most of the marketed coffee is produced by small-scale farmers. The study covered Manyatta and Runyenjes sub-counties which are the main coffee growing areas in the County and are mainly located in Upper Midland (UM) 2-3 Agro-ecological zones (Ndirangu et al., 2017). The sub-counties have the majority of coffee farmers registered with various cooperative societies (GoK, 2013b). The rainfall pattern in the study area is bimodal with two distinct rainy seasons. Long rains occur between March and June while the short rains occur between October and December. Rainfall quantity ranges between 640- 1495mm annually with altitude ranging from 900m to 1400m above sea level. The rainfall pattern changes to bi-modal with increasing altitude (GoK, 2013b). Temperatures range from a minimum of 12˚C in July to a maximum of 30˚C in March (GoK, 2013b). Coffee being the main industrial crop in this area, agriculture is the main driver of the economy in the region with over 70% of the farmers being smallholders (MoA, 2011).

3.3 Research Design The study adopted cross sectional survey research design. The research design is used when gathering information about a population at a single point in time. This design helped the researcher yield both qualitative and quantitative data about the characteristics of coffee farmers in the study area. The cross sectional survey design was used to obtain information from a sample of individuals, which was a representative of the entire population, with adequate precision at a single point in time. The design was used because it allows use of a representative sample and is useful in identifying

27 associations or interactions of predictor variables on the outcome of interest (Sedgwick, 2014). The study collected data from the respondents using semi-structured questionnaires through farm visits.

3.3.1 Target Population The target population for the study comprised smallholder coffee farmers in Embu County. These farmers were the majority in the study area and contributed the largest share of marketed coffee and were also registered with cooperative societies. They were estimated to be 20,000 smallholder farmers in the study area (GoK, 2013b).

3.3.2 Sample Size The sample size for the study was 376 small-scale coffee farmers from the study area (Manyatta and Runyenjes sub-counties). The following formula was used to determine the sample size as recommended by Cochran (1963) and Muriithi (2016). Z 2 pq n0  2 e ………………………………………………...... (5)

Where n0 = required sample size Z = t value from normal table p = probability of success q = (1-p) probability of failure e = 5% level of significance (0.05) Using equation 5, the sample size was calculated as: (1.96)2 (0.5)(0.5) n   384……………………………..…………. (6) o (0.05)2 Given that the estimated target population of smallholder coffee farmers was less than 100,000 (GoK, 2013b), then the sample size was adjusted using the following equation for finite population correction (Cochran, 1963);

n n  o ……………………………………………………… (7) (n 1) 1 o N

28

384 n  = 376 farmers……………………………… (8) (384 1) 1 20000 3.3.3 Sampling Procedure The study applied multistage stratified random sampling to select the farmers to be interviewed. The study randomly selected six cooperative societies, three from each of the two sub-counties (Manyatta and Runyenjes). Out of the six cooperative societies selected, probability proportional to size sampling criteria was employed to randomly select 376 farmers from among the farmers who deliver coffee to the selected factories.

3.3.4 Data Collection Primary data was collected from the respondents by use of questionnaires. The primary variables were coffee varieties and proportions, tree spacing to determine plant population per land size, types and quantities of fertilizer applied, bearing heads and capping height, weeding methods, types and rates of pesticides and fungicides used. Data was also collected on socioeconomic factors of the respondent, coffee output and yield, variable costs of production, quantity of „mbuni‟ produced and input and output prices. Focus group discussions were used to collect qualitative data on farmers‟ perception of the crop and social attributes which may hinder production of coffee under the recommended technologies.

3.3.5 Accuracy and Reliability of Research Instruments The study conducted a pilot study of the respondents with a small sample of 20 respondents in the study area to assess the effectiveness of the research tool. Content analysis was done by taking a sample of the questions from each section of the questionnaire and comparing it with the desired outcome. To validate the effectiveness of the tool, inferences were drawn and compared to the recommendations. Reliability of the research tools was done using the split half method by dividing the questionnaires into two and administering them to two different groups of respondents to estimate the reliability. Correlation coefficient between the two halves was calculated using the Split half method as shown in the equation below.

29

…………………….. (9)

Where: = odd scores, = even scores, = sum of X scores, = sum of Y scores, = sum of squared X scores, = sum of squared Y scores, = sum of the product of paired X and Y scores, N = number of paired scores and r = coefficient correlation between halves. Since r represents one half of the instrument, Spearman- Brown coefficient was used to determine reliability of the full instrument as follows:

= 2 × reliability for ½ tests / 1 + reliability for ½ tests……...…… (10)

3.4 Data Analysis The data collected was processed, coded and analyzed using descriptive statistics. Descriptive statistics were used to analyze the demographic characteristics of the farmers hypothesized to influence their production capability. To determine the effect of recommended technologies on productivity, profitability and quality, multiple regression using stochastic Cobb-Douglas production function, profit function and binary logit models were applied.

3.4.1 Effect of Recommended Technologies on Coffee Productivity To model the combined effect of recommended technologies, input usage and farm socioeconomic factors on coffee output, a stochastic Cobb-Douglas production function was used. The empirical model was specified as proposed by (Aigner et al., 1977) and (Seyoum et al., 1998) shown below.

ln Y  ln    ln X   Z   D   o i i i j j j k k k …………………..…..……… (11)

Where Y = observed coffee output (in kgs), ln = Natural logarithm,  = parameter estimates, X = farm inputs used in production, Z = socioeconomic factors, D = dummy variables for recommended technologies, i = 1,2,…8, j = 1,2,…9, k = 1,2,…11, X 1 = land size under coffee, X 2 = fertilizer(kgs), X 3 = foliar feed (lts), X 4 = manure(debes),

X 5 = fungicides (lts), X 6 = herbicide (lts), X 7 = pesticides (lts), X 8 = labour (man-days),

Z1 = gender, Z 2 = age, Z 3 = education, Z 4 = experience, Z 5 = household size, Z 6 = off-

30 farm income, Z 7 = land ownership, Z 8 = extension, Z 9 = credit access, D1 = recommended fertilizer rate, D2 = recommended foliar feed, D3 = recommended manure rate, D4 = recommended fungicide rate, D5 = recommended herbicide rate, D6 = recommended pesticide rate, D7 = recommended pruning, D8 = recommended capping height, D9 = recommended heads per stem, D10= recommended coffee variety, D11= recommended spacing and = error term.

3.4.2 Effect of the Recommended Technologies on Coffee Profitability To model the combined effect of recommended technologies, input usage and farm socioeconomic factors on coffee profitability, a stochastic normalized restricted profit function was used. Using the coffee output price as the numeraire, the empirical profit model was specified as proposed by (Adesina & Djato, 1996). w x ln  *  ln A   D   lnW   ln M   ln Y   Z  i i   … (12)  k k  i i  j j  n n * 

Where ln = Natural logarithm,  * = normalized profit, ln A = constant or intercept,  = vector parameters to be estimated, D = dummy variables for recommended technologies (1 = adopted, 0 = non adoption),W = factor prices, M = land size under coffee, Y = cost of variable inputs, Z = socioeconomic factors, wi = wage rate per man day normalized by price of coffee, xi = number of man days of labour used in production, k = 1,2,…11,

i = 1,2,…6, j = 1,2,…6, n = 1,2,…9, D1 = variety dummy, D2 = spacing dummy, D3 = fertilizer dummy, D4 = foliar feed dummy, D5 = manure dummy, D6 = fungicide height dummy, D7 = herbicide dummy, D8 = pesticide dummy, D9 = pruning dummy, D10= capping dummy, D11= heads dummy, W1 = normalized fertilizer price per kg, W2 = normalized foliar price per litre, W3 = normalized manure price per debe, W4 = normalized fungicide price per kg, W5 = normalized herbicide price per litre, W6 = normalized pesticide price per litre, Y1 = fertilizer cost, Y2 = foliar cost, Y3 = manure cost,

Y4 = fungicide cost, Y5 = herbicide cost, Y6 = pesticide cost, Z1 = gender, Z 2 = age, Z 3 =

31 education, Z 4 = experience, Z 5 = household size, Z 6 = off farm income, Z 7 = land ownership, Z 8 = extension, Z 9 = credit access and  = error term.

3.4.3 Effect of Recommended Technologies on Coffee Quality Binary logit statistical model was used to evaluate the combined effect of recommended technologies, input usage and socioeconomic factors on coffee quality. In this study, the quality of coffee output was classified into two based on the proportion of ‘mbuni’ to cherry in the output sold; high quality coffee and low quality coffee. Coffee output comprising less than 15% of mbuni was classified as high quality coffee, while that comprising over 15% was taken as low quality coffee. In the binary logit model, high quality coffee took a value of 1 and low quality coffee took a value of 0. Coffee quality was explained by a set of explanatory variables including the recommended technologies (Aidoo et al., 2013). The empirical binary logit model was specified as:  P  ln i      X   Z   1 P  0  i i  n n  i  ………………...……….. (13) Where  are parameters to be estimated, X = dummy variables for recommended technologies, Z = socioeconomic factors, i = 1,2,…11, n = 1,2,…9, X 1 = Spacing dummy, X 2 = variety dummy, X 3 = fertilizer dummy, X 4 = foliar dummy, X 5 = manure dummy, X 6 = fungicide dummy, X 7 = herbicide dummy, X 8 = pesticide dummy, X 9 = pruning dummy, X 10= capping dummy, X 11= heads dummy, Z1 = gender, Z 2 = age, Z 3 = education, Z 4 = experience, Z 5 = household size, Z 6 = off farm income, Z 7 = land ownership, Z 8 = extension, Z n = credit access, = error term.

32

CHAPTER FOUR RESULTS 4.1 Introduction This chapter contains the results on reliability and validity of research instruments, and descriptive statistics on socioeconomic characteristics of the respondents. Descriptive statistics are also given for coffee input use, gross returns and variable costs. The chapter also contains the multiple regression results on the effect of recommended technologies on coffee production, profitability and quality.

4.2 Reliability and Validity Analysis The study determined the reliability and internal consistency of the research tool used using the Split half method of reliability analysis. The Spearman-Brown Coefficient results of this analysis are shown in Table 4.1. The results revealed a correlation coefficient value of 0.545 between the two halves. Spearman Brown coefficient of equal and the unequal lengths was found to be 0.706, which showed a strong correlation between the two lengths. Guttman Split-Half Coefficient value was found to be 0.705 which was almost equal to Spearman Brown coefficient (Table 4.1), which is an indicator of internal consistency between the scores. This implied high internal consistency of scores of one administration of the questionnaire from one set of respondents to another in the sample.

Table 4.1: Spearman-Brown Coefficient of Reliability Results

Cronbach's Alpha Part 1 Value 0.198 No. of Items 3 Part 2 Value 0.287 No. of Items 3 Total No. of Items 6 Correlation Between Forms 0.545 Spearman-Brown Coefficient Equal Length 0.706 Unequal Length 0.706 Guttman Split-Half Coefficient 0.705

33

4.2.1 Descriptive Statistics for Reliability Between the Two Halves Descriptive statistics for reliability analysis of the internal consistency of the research tool using the split half method is shown in Table 4.2. The two halves recorded very close mean values of 6.55 for part 1 and 6.10 for part 2 totaling 12.65. The standard deviation for Part 1 was 1.572 and that of Part 2 was 1.651, which showed internal consistency between scores or answers of the research tool from one group to another. This implied that the research tool used had internal consistency from one administration to another, hence reliable in giving consistent results.

Table 4.2: Descriptive Statistics for Reliability Between the Two Halves Items Mean Variance Std. Deviation No. of Items Part 1 6.55 2.471 1.572 3 Part 2 6.10 2.726 1.651 3 Both halves 12.65 8.029 2.834 6

4.3 Descriptive Statistics of Socio-economic Characteristics of the Respondents

This section provides background information of the farmers‟ demographic characteristics based on socioeconomic status of the respondents. The factors were grouped into three for ease of classification, that is, social, economic and institutional factors. The social factors included age, gender of the respondent, household size, education level and experience in coffee farming. Economic factors included farm size, land size, land tenure, off-farm income and occupation of the farmer. Institutional factors were access to extension services and credit access.

4.3.1 Social Characteristics of the Respondents Table 4.3 shows the descriptive statistics of selected social characteristics of the respondents that were hypothesized to affect coffee productivity, profitability and quality. The findings indicated that 74.7% of the respondents were males. Majority of coffee farms in the study area were found to be managed by males.

The study findings revealed that majority (84.6%) of the coffee farmers were aged between 41– 60 years with farmers below 30 years constituting only 1.1% (Table 4.3), implying that coffee farming is not popular among the youth which may be attributed to inadequate access to key production resources. Majority (75.8%) of respondents had

34 attained secondary education and above (Table 4.3). Majority of the respondents (90.1%) had 10 years and above experience in coffee farming (Table 4.3). Majority (89.4%) of the households had 6 members and below (Table 4.3).

Table 4.3: Social Characteristics of the Respondents Social Factors Frequency Percent Gender of the farmer Male 281 74.7 Female 95 25.3 Age of the farmer 18- 30 years 4 1.1 31- 40 years 26 6.9 41-50 years 158 42 51- 60 years 160 42.6 61 and above 28 7.4 Level of education Non formal 12 3.2 Primary 79 21 Secondary 245 65.2 Tertiary 38 10.1 Other 2 0.5 Experience in coffee farming (years) Less than 10 years 37 9.8 10-20 years 231 61.4 Above 20 years 108 28.7 Household size 1 - 3 74 19.7 4 - 6 262 69.7 7 - 9 40 10.6

4.3.2 Economic Characteristics of the Respondents Table 4.4 shows descriptive statistics of selected economic characteristics of the respondents hypothesized to affect coffee performance. The findings revealed that majority of farmers (77.6%) owned one acre of land and below (Table 4.4). Majority of the respondents (83%) had allocated half an acre and below to coffee production (Table 4.4). Majority of the households (85.9%) earned off-farm income which implied that these farmers were involved in other economic activities other than farming (Table 4.4). Less than half of the respondents (43.4%) had land ownership rights in form of land title

35 deed, implying that majority of the respondents did not have security of land tenure (Table 4.4). Table 4.4: Descriptive Statistics of Economic Factors of the Respondents Economic factors Frequency Percent Farm size (acres) 0 - 0.5 acres 111 29.5 0.51 - 1.0 acres 181 48.1 1.01 - 1.5 acres 39 10.4 > 1.5 acres 45 12 Land size under coffee (acres) 0 - 0.25 acres 153 40.7 0.26 - 0.5 acres 159 42.3 0.51 -1.0 acres 51 13.6 1.01 -1.5 acres 4 1.1 > 1.5 acres 9 2.4 Off farm income Yes 323 85.9 No 53 14.1 Monthly income (shs) < KES 10,000 109 29 KES 10,000 - 20,000 188 50 Above KES 20,000 28 7.4 No off farm income 51 13.6 Land ownership (title deed) Yes 163 43.4 No 213 56.6

4.3.3 Institutional Factors of the Respondents Table 4.5 shows the descriptive statistics of the selected institutional factors that were hypothesized to influence coffee production in the study area. The results of the study revealed that majority of the respondents (86.8%) received extension services (Table 4.5), which acted as a link between research organizations and the farmer through technology transfer for improved technical know-how and knowledge base. Extension service was hypothesized to positively impact on coffee productivity through increased technical know-how on efficient use of the recommended technologies.

36

Cooperative societies facilitated 83% of technology transfer through extension services in the study area (Table 4.5). This implied that cooperative societies were the major institutions acting as a link between research centers and the farmers at the farm level. Farmers‟ membership to a cooperative society was expected to positively impact on coffee productivity, profitability and quality.

Table 4.5 Descriptive Statistics on Institutional Factors Institutional factors Frequency Percent Extension services Yes 325 86.4 No 51 13.6 Number of visits (frequency) Once 85 22.6 Twice 186 49.5 Thrice 54 14.4 More than thrice 4 1.1 None 47 12.5 Type of institution Cooperative society 312 83 Research institutions 33 8.8 Any other 31 8.2 Credit access Yes 265 70.5 No 111 29.5

Majority of the respondents (70.5%) had access to credit (Table 4.5), implying that farmers had access to key farm inputs and other efficiency enhancing technologies. Access to credit provided a cushion against production risks and random shocks which motivated farmers to increase coffee production for repayment.

4.4 Coffee Input Use and Production The results of the study showed variations in input use among the farmers in the study area. Input use in the study area varied with the scale of production and farm size.

4.4.1 Fertilizer and Manure Application Table 4.6 shows the rates of fertilizer, foliar feed and manure used per crop year. Majority of the farmers (73.4%) used the recommended fertilizer application rates for

37 either NPK or CAN, as this was expected to have a significant impact on coffee productivity. Only (7.4%) of the respondents did not apply foliar feed while majority (92.6%) applied foliar feed as it was considered key in nutrient supplement to boost coffee productivity. Majority of the respondents (90%) applied manure at the recommended rate one debe per tree, implying high extent of adherence to the recommendation. This was guided by the aspect that organic manure was readily available and acted as a substitute for inorganic fertilizers.

Table 4.6: Descriptive Statistics for Fertilizer and Manure Application Rates

Farm Inputs Frequency Percent NPK rate (grams) per tree 0 - 99 61 16.2 100 - 199 37 9.8 200 – 299 143 38.0 300 & above 135 36 CAN (grams) per tree per split 0 - 99 51 13.6 100 - 199 36 9.8 200 – 299 134 35.4 300 & above 155 41.2 Foliar usage Yes 348 92.6 No 28 7.4 Amount of manure (debes) per tree Less than one 22 5.8 1 - 2 338 90.0 Above 2 16 4.2

4.4.2 Pest, Disease and Weed Control Chemicals Table 4.7 shows the rates of fungicides, pesticides and herbicides used across the farms. Only (33.5%) of the respondents applied fungicides at the recommended rate of 2-3 litres/acre. Majority (66.5%) of the respondents applied either more or lower than the recommended rates of fungicides (Table 4.7) for control of CBD and CLR as majority were growing traditional varieties (Table 4.8). Some of the respondents (47.6%) used less than one litre per acre, while only (3.2%) of the farmers applied herbicide mainly at the recommended rate (Table 4.7). The results revealed that majority of the farmers

38

(52.1%) used the recommended pesticide rate of one litre per acre. This was expected to impact on productivity and quality of coffee cherry produced. Table 4.7: Frequencies for Pest, Disease and Weed Control Chemical Application Rates

Farm inputs Frequency Percent Amount of fungicides used/acre ( lts) 0 - < 2 56 14.9 2 - < 3 126 33.5 3 & above 194 51.6 Amount of herbicides used/acre (lts) 0 - <1 179 47.6 1 - <2 12 3.2 2 & above 185 49.2 Amount of pesticide used/acre (lts) 0 - < 1 69 18.4 1 - < 2 196 52.1 2 & above 111 29.5

4.4.3 Coffee Varieties Table 4.8 shows the descriptive statistics of agronomic practices of coffee management. The recommended coffee varieties in the study area were Ruiru 11 and Batian as compared to traditional varieties such as SL 34, SL 28 and K 7, due to their productivity potentials and disease resistance. Majority of the respondents (67.2%) are still dependent on the traditional varieties. Only (38.2%) of the sampled farmers had planted the two improved varieties, implying that adoption of these improved varieties was quite low in the study area.

Table 4.8: Descriptive Statistics of the Grown Coffee Varieties Coffee variety Frequency Percent Ruiru 11 110 29.3 Batian 13 3.5 SL 34 220 58.5 SL 28 25 6.6 K7 8 2.1

39

4.4.4 Coffee Management Practices A significant proportion of farmers (59.3%) changed cycle of the coffee trees once while (40.7%) of the sampled farmers did not change cycle in the last ten years. Majority of the farmers (98.4%) pruned their coffee stems at least once while 1.6% of farmers didn‟t prune their coffee bushes. Pruning was done at least once per crop year and was seen a key management practice to reduce the unproductive stems and ensure efficient nutrient utilization and disease control for improved productivity.

Table 4.9: Descriptive Statistics of Coffee Management Practices

Categories Frequency Percent Change of cycle Yes 223 59.3 No 153 40.7 Pruning Yes 370 98.4 No 6 1.6 Capping Yes 58 15.4 No 318 84.6 Heads per stem 1 9 2.4 2 244 64.9 3 111 29.5 Above 3 12 3.2

Capping was not a common practice by the farmers as only (15.4%) capped their coffee stems while 84.6% did not practice capping. Results revealed that majority of the farmers (94.4%) maintained 2 or 3 heads per stem as recommended.

4.4.5 Coffee Production Table 4.10 shows descriptive statistics for coffee production across the farms in the study area. The results revealed that the minimum number of coffee trees was 30 per farm while the maximum number was 2000 trees. The mean number of coffee trees per farm was 262.43. The maximum quantity of cherry produced was 12,000 kgs per farm while the minimum was 100 kgs with an average of 1106.24 kgs. The minimum quantity of mbuni produced per farm ranged from 0 kgs to 430 kgs which averaged 27.85 kgs per

40 farm. The results indicated variations in productivity among the farms with minimum output per tree being 0.62 kgs while the maximum was 20 kgs. The mean production per tree was 4.66 kgs. Output per acre had a minimum of 82 kgs and a maximum output of 21,368 kgs with a mean of 2433.75 kgs. The minimum output per man day was 2.33 kgs and maximum output of 227.78 kgs which gave a mean of 34.3196 kgs. The minimum labour use per crop year was 7 man-days and a maximum of 99 with an average of 33.85.

Table 4.10: Descriptive Statistics for Coffee Production in the Study Area

Production Min Max Mean Std. Error Std. Deviation Trees owned (No) 30 2000 262.43 9.923 192.408 Cherry output (kgs/yr) 100 12000 1106.24 55.551 1077.18 Mbuni output (kgs/yr) 0 430 27.85 2.194 42.538 Output/tree (kgs) 0.62 20 4.6624 0.13291 2.57716 Output/acre (kgs) 82 21368 2433.75 104.320 2022..838 Labour use/year (days) 7 99 33.85 0.586 11.365 Output/man day (kgs) 2.33 227.78 34.3196 1.41797 27.49555

4.4.6 Effect of Recommended Technologies on Coffee Productivity A stochastic Cobb Douglas production function was used to show the combined effect of factors of production, recommended technologies and socioeconomic characteristics of the respondents on coffee productivity. The results of multiple regression (Table 4.11) shows that the fitted model gave an R-square value of 0.904, which implies that the predictor variables explained about 90% of variations in coffee productivity in the study area. The results also revealed an F value that was highly significant at 1% level (0.000), implying that the predictor variables explained significant variations in coffee output. The results of the multiple regression using Cobb-Douglas production function for the recommended inputs and crop management practices, and the socioeconomic characteristics of the respondents are given hereafter.

The t- value computed for each variable test the significance of its regression coefficient. Results revealed an increase in absolute t-values for significant variables implying an increase in the difference between the variables and the null hypothesis. VIF results

41 were below 5 for all predictor variables, implying that multicollinearity among independent variables used in the model was insignificant.

Table 4.11 Multiple Regression Results for Effect of Recommended Technologies on Coffee Productivity Dependent Var. (OUTPUT) Beta S. E t Sig VIF (Constant) 4.924 0.618 7.972 0.000 Fertilizer (kgs) 0.115 0.099 1.606 0.111 6.080 Recommended fertilizer rate 0.069 0.046 1.955 0.053 1.472 Foliar feed (ltrs) -0.009 0.045 .237 0.813 1.834 Recommended foliar rate 0.058 0.053 1.584 0.116 1.598 Manure (debes) 0.426 0.055 7.581 0.000*** 3.767 Recommended manure rate 0.196 0.068 5.580 0.000*** 1.474 Fungicide (ltrs) 0.166 0.060 3.511 0.001*** 2.656 Recommended fungicide rate 0.118 0.061 2.619 0.010** 2.409 Herbicide (ltrs) 0.074 0.071 1.342 0.182 3.645 Recommended herbicide rate 0.094 0.066 1.959 0.053 2.729 Pesticide (ltrs) 0.088 0.040 2.543 0.012** 1.432 Recommended pesticide rate 0.041 0.046 1.272 0.206 1.228 Recommended variety -0.017 0.044 -.527 0.600 1.290 Recommended spacing 0.001 0.044 .045 0.964 1.224 Recommended pruning 0.012 0.179 .386 0.700 1.185 Capping -0.015 0.064 -.453 0.651 1.267 Heads per stem -0.022 0.045 -.654 0.515 1.359 ***sig at 1%, **sig at 5%,

The coefficient for the recommended manure rate was positive 0.196 and significant at 1 percent level (t = 5.58, P = 0.000), implying that the coffee yield for adopters of recommended manure rate was 19.6% higher than that of non-adopters. The coefficient for the amount of manure currently applied by farmers was 0.426 and significant at 1 percent level (t = 7.58, P = 0.000). This implied that a 10% increase in amount of manure at the current rate increased coffee yields by 4.2%. The coefficient for recommended rate of fungicides was positive 0.118 and significant at 5 percent level (t = 2.619, P = 0.010), implying that the yield obtained by adopters of recommended fungicide rate was 11.4% higher than that obtained by non-adopters. The amount of fungicide used was positively related to coffee yield with a coefficient 0.166 and significant at 1 percent level (t = 3.511, P = 0.001), implying that at the current usage of

42 fungicides, a 10% increase in fungicide would increase coffee yield by 16.6%. The amount of pesticides used had a positive coefficient of 0.088 and was significant at 5 percent level (t = 2.543, P = 0.012), implying that at the current usage of pesticides, a 10 % increase in pesticide would increase coffee output by 8.8%.

The results on socioeconomic factors that were hypothesized to affect coffee productivity (Table 4.12) shows that the coefficient for respondent‟s engagement in off- farm income-earning activities was positive 0.084 and significant at 5 percent level (t = 2.223, P = 0.028), implying that engagement in off-farm activities increases coffee yield by about 8.4%. The coefficient for land size under coffee was positive 0.353 and was significant at 1 percent level (t = 7.521, P = 0.000), implying that increasing the land size under coffee by 10% increases coffee output by 3.5%. Increase in land size increases the scale of production and motivates the farmers to adopt new technologies, hence increasing coffee productivity. The coefficient for labour was positive 0.102 and significant at 5 percent level (t = 2.104, P = 0.038), implying that a 10% increase in labour increases coffee yield by 10%. Access to credit had a positive coefficient 0.074 and was significant at 5 percent level (t = 2.197, P = 0.030), implying that availability of credit increases coffee output by 7.4%.

Table 4.12: Multiple Regression Results for Effect of Socioeconomic Factors on Coffee Productivity Dependent Var. (OUTPUT) Beta S. E t Sig VIF Gender (male=1, female=0) 0.046 0.062 1.356 0.178 1.376 Age of the farmer (years) -0.031 0.037 -.768 0.444 1.883 Level of education -0.029 0.037 -.803 0.423 1.539 Experience (years) 0.016 0.044 .438 0.662 1.575 household size (Number) 0.012 0.042 .352 0.725 1.398 Off farm income (yes=1, no=0) 0.080 0.064 2.223 0.028** 1.539 Land ownership (Tenure) 0.091 0.047 2.579 0.011 1.477 Extension services (yes=1, no=0) -0.046 0.072 1.383 0.169 1.321 Labour (man-days) 0.102 0.085 2.104 0.038** 2.783 Land size (acres) 0.353 0.046 7.521 0.000*** 2.631 Credit access (yes=1, no=0) 0.074 0.054 2.197 0.030** 1.341 ***sig at 1%, **sig at 5%

43

4.5 Coffee Profitability Among Smallholder Farmers 4.5.1 Estimation of Gross Returns (Margins) from Coffee Enterprise The study hypothesized that variations in coffee profitability were due to differences in input use, adoption of recommended technologies with interaction of farm socioeconomic characteristics of the farmers. Adoption of these technologies was expected to increase economic efficiency among the farms. The study analyzed the gross returns across the farms in the sample and descriptive statistics results are given in Table 4.13.

Table 4.13 Descriptive Statistics on Gross Margins (KES) from Coffee Production Returns (Kshs) Min Max Mean Std. Error Std. Deviation GM -30,450 552,897.20 37384.01 3318.95 64356.80 GM / tree -268 3547 248.62 15.296 296.603 GM / man day -1155 8353 1052.47 75.42 1462.44 GM / acre -150,224 876755.60 80670.27 5973.77 115835.62 GM/shs -0.72 11.37 1.7612 0.09074 1.75942 Returns/ Kshs 0.28 12.37 2.7612 0.09074 1.75942

The results showed coffee production was profitable despite some farmers making losses due to high cost of production. Some farmers incurred losses maximum loss per farm being KES 30,450 but the gross margin averaged KES 37,384.01 per farm. Results of gross margin per tree had a mean of KES 248.62 and a standard deviation of 296.603. Gross margin per man day was estimated to be KES 1,052.47 with a standard deviation of 1462.44 due to variations in labour usage. Gross margin per acre averaged at KES 80,670.27 and a standard deviation of 115835.62. The study analyzed returns per shilling invested in coffee production and found that for each shilling invested, it generates a positive gross margin per shilling. This gave an indication of positive returns or earnings for every shilling invested making coffee production a profitable enterprise.

4.5.2 Estimation of Variable Costs for Coffee Production The costs of major variable inputs used in coffee production are given in Table 4.14. The average price of fertilizer per kilogram was KES 55.68 with average expenditure on fertilizer being KES 19,628.61 per acre. Expenditure on fertilizer constituted 19.8% of total variable cost implying a key factor of production in coffee. Expenditure on foliar

44 feed was 7.4% of total variable cost with an average of KES 956 per litre. Organic manure which acted as a substitute for inorganic fertilizer constituted 22.7% of total variable cost and an average expenditure of KES 22,427.12 per acre.

Table 4.14 Descriptive Statistics for Variable Costs per Acre Factors Unit Price/unit Av. cost %TVC S. E Std. Deviation Fertilizer Kilogram 55.68 19628.61 19.8 759.89 14734.80 Foliar feed Litre 956.20 7287.86 7.4 489.77 9497.06 Manure Debe 36.78 22427.12 22.7 1237.04 23987.14 Fungicide Kilogram 1057.60 10210.57 10.3 578.72 11221.74 Herbicide Litre 782.16 3464.23 3.5 321.52 6234.60 Pesticide Litre 1029.97 4670.30 4.7 281.83 5464.92 Labour Man-day 307.20 31313.10 31.6 1486.21 28818.62

Fungicides which were commonly used for control of coffee berry disease (CBD) and coffee leaf rust (CLR) constituted 10.3% of total variable cost with average cost of KES 10,210.57 per acre and an average price of KES 1,057.60 per kilogram. Expenditure on herbicides constituted 3.5% of total variable cost with an average expenditure of 3,464.23 per acre. Pesticides constituted 4.7% of total variable cost with an average price of KES 1,029.97 per litre and an average expenditure of 4670.30 per acre. Labour cost varied across the farms given the scale of production constituting majorly on total variable cost with 31.6% and average expenditure of KES 31,313.10 per acre. The average wage rate per man day was KES 307.20.

4.5.3 Effect of Recommended Technologies on Coffee Profitability (Gross Margin) A stochastic normalized restricted profit function was used in the study to show the responsiveness of the predictor variables on gross margins per acre across the farms. Adoption of recommended technologies was expected to have significant effect on gross margins. The fitted model gave a coefficient of determination value (R2) of 0.724, which implies that the predictor variables explained 72.4% of the observed variation in gross margins per acre.

Results revealed a standard error of the estimate of 0.69168 which was the difference between actual and the predicted scores in the null hypothesis. The results also revealed an F-value which was significant at 1% level (0.000), implying that the predictor

45 variables explained significant variation in the dependent variable. The predictor variables fitted in the model had a VIF value of less than 5 which implied no problem of multicollinearity between the variables. The results of the multiple regression using the estimated profit function for the recommended technologies and expenditure are shown in table 4.15.

Table 4.15 Multiple Regression Results for Effect of Recommended Technologies on Coffee Profitability Dependent Var.(Profit) Beta Std. Error t Sig. VIF (Constant) 2.54 2.594 0.979 0.329 Coffee varieties 0.114 0.125 2.19 0.030** 1.414 Tree spacing -0.067 0.117 -1.397 0.165 1.217 Recommended fertilizer rate 0.008 0.116 0.171 0.865 1.262 Recommended foliar feed rate -0.066 0.148 -1.051 0.295 2.066 Recommended manure rate 0.204 0.227 3.96 0.000*** 1.397 Recommended fungicide rate 0.034 0.148 0.574 0.567 1.797 Recommended herbicide rate -0.094 0.147 -1.547 0.124 1.937 Recommended pesticide rate -0.075 0.121 -1.572 0.118 1.179 Pruning -0.017 0.433 -0.372 0.710 1.156 Capping -0.128 0.177 -2.554 0.012** 1.318 Heads per stem -0.023 0.117 -0.474 0.636 1.267 Expenditure Fertilizer price 0.012 0.143 0.161 0.872 2.800 Fertilizer variable cost 0.118 0.159 1.474 0.143 3.381 Foliar feed price -0.182 0.15 -2.204 0.029** 3.564 Foliar feed variable cost -0.259 0.153 -3.291 0.001*** 3.239 Manure price -0.516 0.152 -7.026 0.000*** 2.829 Manure variable cost -0.398 0.145 -5.634 0.000*** 2.612 Fungicide price 0.032 0.129 0.565 0.573 1.657 Fungicide variable cost -0.121 0.138 -1.971 0.051 1.984 Herbicide price 0.061 0.123 0.878 0.381 2.533 Herbicide variable cost -0.076 0.139 -1.174 0.242 2.212 Pesticide price -0.038 0.118 -0.592 0.555 2.183 Pesticide variable cost -0.031 0.117 -0.503 0.616 1.986 Wage rate 0.321 0.018 6.743 0.000*** 1.192 ***sig at 1%, **sig at 5%

The coefficient for the recommended coffee varieties was positive 0.114 and significant at 5 percent level (t = 2.19, P = 0.030), implying that the gross margin for adopters of recommended coffee varieties was 11.4% higher than that of non-adopters. The

46 coefficient of recommended manure rate of one debe per tree was positive 0.204 and significant at 5 percent level (t = 3.96, P = 0.000). This implied that the gross margin for adopters of recommended manure rate was 20.4% higher than that of non-adopters. The coefficient of recommended capping was negative 0.128 and significant at 5 percent level (t = -2.55, P = 0.012), implying that gross margins for adopters was lower than that of non-adopters by 12.8%.

Price of foliar feed had a negative coefficient of 0.182 and significant at 5 percent level (t = -2.204, P = 0.029), implying that a 10% increase in unit price of foliar reduced coffee returns by 1.82%. The coefficient for expenditure on foliar feed had a negative coefficient of 0.259 and significant at 1 percent level (t = -3.29, P = 0.001), implying that a 10% increase in expenditure (amount used and the price) on foliar feed decreased coffee returns by 2.59%. Manure price had a negative coefficient of 0.516 and was significant at 1 percent level (t = -7.03, P = 0.000), implying that a 10% increase in unit price for manure reduced gross margins by 5.1%. Cost of manure (both price and amount used) had a negative coefficient of 0.398 and significant at 1 percent level (t = - 5.63, P = 0.000), which implied that a 10% increase in expenditure incurred on manure decreases coffee profit by 3.98%.

Wage rate was positively related to gross margin with a coefficient of 0.321 and significant at 1 percent level (t = 6.743, P = 0.000), implying that contrary to expectation a 10% increase in wage rate increases coffee profit by 3.2%. The prices of fertilizer, fungicide, herbicide and pesticide were not significant in explaining profit variations. Fungicide cost had a negative coefficient of 0.121 and significant at 5 percent level (t = - 1.91, P = 0.051), implying that a 10% increase in the cost incurred on fungicides decreases profit by 1.21%.

The results of the multiple regression using the estimated profit function for the socioeconomic factors that interact with adoption of technologies at farm level are shown in Table 4.16. The socioeconomic characteristics of the respondents hypothesized to influence coffee gross returns were not significant at 5 percent level (Table 4.16).

47

Table 4.16 Multiple Regression Results on Effect of Socioeconomic Factors on Coffee Profitability Dependent Var.(Profit) Beta Std. Error t Sig. VIF Socioeconomic factors Land size (acres) 0.081 0.124 1.110 0.269 2.788 Gender (male=1, female=0) 0.058 0.148 1.164 0.246 1.298 Age of the farmer (years) 0.037 0.094 0.672 0.502 1.619 Level of education 0.024 0.102 0.444 0.658 1.540 Experience (years) -0.021 0.117 -0.381 0.704 1.591 household size (number) -0.042 0.108 -0.828 0.409 1.368 Off farm income (yes=1, no=0) 0.005 0.178 0.084 0.933 1.518 Land ownership (tenure) -0.009 0.118 -0.19 0.850 1.305 Extension service (yes=1, no=0) 0.002 0.184 0.041 0.967 1.200 Credit access (yes=1, no=0) -0.045 0.136 -0.918 0.360 1.245 ***sig at 1%, **sig at 5%

4.6. Coffee Quality Among Smallholder Farmers 4.6.1 Frequencies on Proportion of Mbuni to Cherry Descriptive results statistics for proportion of mbuni to cherry is as shown in Table 4.17. Majority of the respondents (68.1%) had less than 20% proportion of mbuni to cherry, implying that they produced high quality cherry. A proportion of 28.7% produced 20 to 60% of mbuni while only 3.2% had more than 60 percent proportion of mbuni to cherry implying production of low quality cherry.

Table 4.17 Frequencies on Proportion of Mbuni to Cherry Proportion of mbuni to cherry Frequency Percent 0 - < 20 256 68.1 20 - < 40 76 20.2 40 - < 60 32 8.5 60 - < 80 10 2.7 80 and above 2 0.5

4.6.2 Descriptive Statistics on the Proportion of Mbuni to Cherry Table 4.18 shows the descriptive statistics on the proportion of mbuni to cherry and the total loss of coffee output per tree. The results revealed that the maximum proportion of mbuni to cherry was 80.05% with a minimum of zero and an average of 14.913% per farm. The total loss from physiological fall estimated by the farmer was divided by total number of trees and the maximum loss per tree was 5.71 kgs while the minimum was

48 zero with an average of 0.0530 kgs per tree which implied production of high quality cherry in the study area.

Table 4.18 Descriptive Statistics for Total Loss and Proportion of Mbuni to Cherry Factors Mean Std. Error Std. Deviation Prop of Mbuni to Cherry 14.913 0.83038 16.10167 Loss/tree (kgs) 0.053 0.01657 0.32137

4.6.3 Effect of Recommended Technologies on Coffee Quality Binary logistic regression was used to show the effect of recommended technologies (Table 4.19) with interaction of socioeconomic factors (Table 4.20) on coffee quality. In this analysis, the results revealed that the probability of the model Chi square (51.87) was 0.000, less than the level of significance at 5 percent (P = 0.05), implying that there was a statistically significant relationship between the predictor variables and the dependent variable (coffee quality). The model revealed a log likelihood of 61.147 almost equal to Chi square implying goodness of fit of the logistic regression model. The results for binary logistic regression are as shown thereafter.

Table 4.19 Regression Results for Effect of Recommended Technologies on Coffee Quality Dep. Var. (Coffee Quality) B S.E. Wald df Sig. Exp (B) Spacing -0.626 0.990 0.400 1 0.527 0.535 Variety -2.538 1.597 2.526 1 0.112 0.079 Fertilizer rate -0.347 0.915 0.144 1 0.704 0.707 Foliar feed rate -0.777 0.835 0.865 1 0.035** 2.175 Manure rate -0.901 1.141 0.623 1 0.043** 0.406 Fungicide rate 2.289 1.668 1.882 1 0.170 9.865 Herbicide rate 0.366 1.059 0.119 1 0.730 1.442 Pesticide rate -2.138 1.314 2.646 1 0.104 0.118 Pruning -4.364 1.717 6.461 1 0.011** 7.858 Capping -1.812 0.939 3.725 1 0.054 0.163 Heads per stem 0.368 0.746 0.244 1 0.621 1.445 ***sig at 1%, **sig at 5%

The coefficient of recommended foliar feed rate was negative 0.777 and significant at 5 percent level (P = 0.035), with an odd ratio of 2.175, implying that the probability of producing high quality coffee increases by a factor of 2.17 for adopters of the

49 recommended foliar feed rate. This implied that application of foliar feed at the recommended rate increases the odds of a farmer producing high quality coffee by a factor of 2.17. The coefficient for recommended manure rate was negative 0.901 and significant 5 percent level (P = 0.043), implying that adoption of recommended rate of manure increases the odds of a farmer producing high quality coffee by 0.901. Coffee pruning was significant at 5 percent level (P = 0.011) with a negative coefficient of 4.364 and an odds ratio of 7.858, implying that adoption of recommended coffee pruning increases the odds of a farmer producing high quality coffee by a factor of 7.86.

The coefficient for gender of the farmer was negative and significant at 1 percent (P = 0.002), with an odds ratio of 0.067, implying that being a male increases the log odds of producing high quality coffee by a factor of 0.067. The coefficient of household size was positive 2.594 and significant at 5 percent level (P = 0.007) with an odds ratio of 13.389, implying that an increase in household size decreased the odds of producing high quality coffee by a factor of 13.38. Age of the farmer, education, experience, land ownership, extension services and credit access had no significant effect on coffee quality.

Table 4.20 Regression Results for Effect of Socioeconomic Factors on Coffee Quality Dep. Var. (Coffee Quality) B S.E. Wald df Sig. Exp (B) Age (years) -0.793 0.602 1.736 1 0.188 0.453 Gender -2.700 0.851 10.063 1 0.002*** 0.067 Education -0.401 0.607 0.437 1 0.509 0.669 Experience 0.656 0.716 0.840 1 0.359 1.928 Household size 2.594 0.955 7.379 1 0.007*** 13.389 Off-farm income -1.227 1.011 1.473 1 0.225 0.293 Land ownership -1.021 0.928 1.209 1 0.272 0.360 Extension -0.578 1.037 0.310 1 0.578 0.561 Credit access 1.260 0.986 1.634 1 0.201 3.526 Farm size 0.201 0.470 0.183 1 0.668 1.223 ***sig at 1%, **sig at 5%

50

CHAPTER FIVE SUMMARY OF FINDINGS, DISCUSSIONS, CONCLUSIONS AND RECOMMENDATIONS 5.0 Introduction This chapter gives a summary of research findings, detailed discussion of research findings based on the study objectives, linking the findings with other previous studies and drawing conclusions based on the study findings objectives and hypotheses.

5.1 Summary of the Key Findings This section provides a precise summary of the research findings based on the objectives of the study. The key findings are used to infer and draw conclusions based on the research apriori expectations.

5.1.1 Socio-economic Factors Influencing Coffee Performance Coffee production was dominated by males (74.7%), implying that male household heads controlled the main resources used in coffee production such as land, capital and labour and also made the key production and marketing decisions. Results of the study revealed that (89.4%) of the households had six members and below. Increase in household size would increase household consumption and reduce household saving. Increase in household size would also increase family labour which in turn would reduce the cost of production given the scale of production.

The findings show that (75.8%) of the respondents had secondary school education and above which implied diverse knowledge base in the study area to adopt new production methodologies. Increase in level of education would play a key role in managing risks, taking mitigation strategies and long term production decisions. The results revealed that (50%) of the respondents were 51 years and above implying some generational gap in coffee production thus youth participation in coffee production is still a matter of concern. In terms of farming experience (number of years the farmer had been involved in coffee production), majority (90.1%) had 10 years and above. This implied they had diverse knowledge and skills to adjust to technological advancements, conduct long- term planning, and absorb or reduce production risks associated with new technologies.

51

Results revealed that (70.5%) of the respondents had access to credit which reduced the financial constraints faced by the smallholder farmers at the start of production process. Access to credit would offset liquidity constraints for purchase of key farm inputs, increase household risk bearing ability against production and market risks. The findings show that (85.9%) of the farmers had off farm income which played a significant role in cushioning farmers against crop failures, costs and risks associated with new technologies. Majority of the respondents (86.4%) had access to extension services which acted as a link between research institutions and the farmers at the farm level.

Farm sizes varied across the sampled respondents with majority (77.6%) having one acre of land and below. Land sizes under coffee were directly proportional to farm sizes. Farmers with large pieces of land were able to allocate more land to coffee, hence increased plant population and a reduction in average production costs due to economies of scale. Majority of the respondents (63%) allocated half an acre of land and below to coffee production. Almost half of the households (43.4%) had title deeds to the land entitlement. Ownership of a title deed provided security for land tenure which allowed the farmers to make long term production decisions and facilitate access to credit as collateral.

5.1.2 Effect of Recommended Technologies on Coffee Productivity Coffee productivity in the study area was still below the recommended output potential of 10 kg per tree (CRI, 2017) with mean yield per tree being 4.67 kgs. This can be explained by the low adoption level of the recommendations as majority of the respondents (67.2%) are still on traditional varieties (SL 34, SL 28 or K 7). Farmers in the study area (90%) have adopted the recommended rate for organic manure, but few (35%) are using inorganic fertilizers at the recommended rates. The estimated Cobb- Douglas production function revealed that adoption of the recommended rates for manure, fungicide and pesticide were positive and significant in affecting coffee productivity, implying that usage of these inputs at the recommended rates increased coffee yields significantly at 5 percent level. Other factors that were significant in affecting coffee productivity included amount of manure and fungicide used, off-farm

52 income, labour, credit access and land size. Farmers would be motivated to use high yielding technologies that guarantee high productivity per unit area.

5.1.3 Effect of Recommended Technologies on Coffee Profitability Coffee production in the study area was profitable despite high factor prices and variations in scale of production as results for Cost Benefit Ratio were more than one which implied that every shilling invested in coffee production was able to generate more than a shilling subject to cost of production. The estimated profit function revealed that recommended coffee varieties, manure rate and capping were significant in influencing coffee profitability at 5 percent level. Other factors which significantly influenced coffee profitability were prices for foliar feed, manure and labour. The variable cost for foliar feed and manure had also a significant effect on coffee profitability. Socioeconomic characteristics of the farmer were not significant in influencing coffee profitability.

5.1.4 Effect of Recommended Technologies on Coffee Quality Coffee quality was hypothesized to be influenced by many factors; resource related factors, recommended technologies and farm and farmer characteristics. The estimated binary logistic regression model revealed that adoption of the recommended rates for foliar feed, manure and pruning significantly affected coffee quality at 5 percent level. Socioeconomic characteristics of the respondents that were found to significantly affect coffee quality were gender and household size of the respondents.

5.2 Discussion 5.2.1 Effect of Recommended Technologies on Coffee Productivity Coffee production was hypothesized to be a function of factor inputs, recommended technologies and socioeconomic characteristics at the farm environment. R-square (R2), which is referred as the coefficient of determination, is the proportion of the variation in the dependent variable explained by the predictor variables in the model (Peng et al., 2002; Hanson, 2010). Test of the overall significance of the model (F-value) shows whether the entire set of predictor variables explain significant amount of variation in the dependent variable (Hanson, 2010).

53

The t-value is the regression coefficient of the variable divided by its standard error (Aiken et al., 1991). Variance Inflation Factor (VIF) is a statistic used to measure possible multicollinearity amongst the explanatory variables (Robinson & Schumacker, 2009). A general rule is that the VIF should not exceed 10 (Belsley & Kuh, 1980). The estimated Cobb Douglas production function revealed that adoption of the recommended rates of manure and fungicides was significant at 5% level in affecting coffee productivity. Factors of production that were found to significantly affect coffee productivity were amounts of manure, fungicides and pesticides. Socioeconomic factors that were significant include; off-farm income, labour, credit access and land size. The findings on each significant factor are discussed hereafter.

5.2.1.1 Manure Rate Use of manure per se and its application at the recommended rate had positive effect on coffee yield. These findings are supported by Chemura et al. (2010); Dzung et al. (2013); Chemura, (2014) and Gebeyehu, (2016), that organic manure had significant effect on coffee productivity. Organic manure would ensure moisture retention as a climate change strategy in water stressed areas and also increase microbial activity (Chemura, 2014). In addition, organic manure improves organic matter and availability of macro nutrients (Chemura, 2014). Organic soil fertility practice is among the important attributes of sustainable and climate smart agriculture through improvement of the physical, chemical and biological soil properties (Gebeyehu, 2016). Increased usage of organic manure increases plant height and stem thickening for increased production (Dzung et al., 2013). This was argued that farmers are driven by output maximization and would be motivated to use high yielding methodologies that guarantee high productivity with reduced cost. However, Mignouna et al., (2010) reported negative effect of manure use on productivity explained by inappropriate application rates.

5.2.1.2 Fungicide Rate This study established that majority of the farmers (62.7%) in the study area were still growing traditional varieties which are susceptible to fungal diseases such as coffee berry disease (CBD) and coffee leaf rust (CLR). This necessitated usage of fungicides for control of these diseases. Application of fungicides at the recommended rates had

54 positive and significant effect on coffee yield. These findings concur to that by Senkondo et al. (2014); Gebeyehu, (2016); Lechenet et al. (2017), who found positive interaction of fungicides and coffee yields. Copper based fungicides would promote leaf retention and therefore boost tree growth and yield. This in turn would increase plant growth and nutrient uptake for improved marketable value of cherry produced. However, Lenssen, (2013) found less interaction between fungicides and soybean productivity.

5.2.1.3 Pesticide Rate Amount of pesticide used was positively related to coffee yield in the study area, implying that increasing usage of pesticide had positive elasticity on yield. These findings are similar to that by Popp et al. (2013); Ngeywo et al. (2015); Gebeyehu (2016); Lechenet et al. (2017) who found pesticide use positively related to coffee yield. Coffee is susceptible to crop pests such as leaf miners, moths, stem borers, aphids etc. Pesticides for control and management of these pests are expected to increase coffee yield. Pests infest coffee trees negatively impacting on coffee production and quality of cherry produced. Insect pests act as pathogens in transmission of coffee diseases hence influencing coffee productivity. Control of these pests using recommended pesticides increases coffee yields, quality and also the marketable value of the coffee cherry produced. This in turn improves coffee quality and productivity per unit area.

5.2.2 Effect of Socioeconomic Factors on Coffee Productivity 5.2.2.1 Off-farm Income Engagement in off farm income generating activity had a positive and significant impact on coffee productivity. Studies by Akudugu et al. (2012); Minai et al. (2014); Challa & Tilahun, (2014) found similar results on effect of off-farm income on productivity of other crops. This was expected since availability of off farm income cushions farmers from liquidity constraints and facilitates purchase of key farm inputs and labour for increased coffee production considering the intensity of operations involved. Off-farm income would increase household risk bearing ability to mitigate or absorb production and market risks. Coffee production was expected to be dominated by economically

55 active people due to the intensity of farm operations. However, Chepng‟etich et al. (2015) found positive but insignificant relationship among sorghum farming households.

5.2.2.2 Labour Labour had a positive coefficient in explaining variation in coffee output. Ogada et al. (2010); Narayana (2016); Kamau et al. (2016) and Beck et al. (2016) also found labour to have the greatest and significant impact on yields. Labour availability facilitates farm operations such as weeding, fertilizer application, disease control and harvesting. Increase in labour supply accompanied by static labour demand, would decrease wage rate and subsequently increase coffee production per unit area of land given the scale of production. Labour availability would be key in coffee production given the intensity of farm operations. Mburu et al., (2014) found negative coefficient for family labour and wheat productivity.

5.2.2.3 Land Size Land size under coffee positively and significantly explained variations in coffee productivity at farm level. Studies by Gebeyehu, (2016) and Senkondo et al. (2014) reported similar findings on effect of land size and coffee productivity. Increase in land size under coffee would increase the number of coffee trees which would in turn increase coffee production per unit area. Increase in land size increases the scale of production and motivates the farmers to adopt new technologies, hence increasing coffee productivity. Increase in land size would also allow experimental trials on new technologies without affecting the main crop field hence increased coffee productivity. Chepng‟etich et al. (2015) and Chiona, (2011) support this argument on effect of land size on productivity of other crops. However studies by Minai et al. (2014) and Musaba & Bwacha, (2014) found land size negatively related to coffee and maize productivity respectively.

5.2.2.4 Credit Access Access to credit had a positive and significant elasticity in explaining variations in coffee yield. Studies by Mignouna et al. (2010); Akudugu et al. (2012); Musaba & Bwacha, (2014) reported similar findings on the effect of credit on farm productivity. Credit would enable coffee farmers to purchase key farm inputs for increased

56 productivity and also cushion them against random shocks and market failures. Access to credit would also finance investment in capital intensive technologies for increased production efficiency and productivity per unit area to avoid diseconomies of scale.

5.2.3 Effect of Recommended Technologies on Coffee Profitability Coffee profitability was hypothesized to be a function of factor prices, expenditure on inputs, recommended technologies and socioeconomic characteristics of the respondents. Multiple correlation coefficient or coefficient of determination (R2) is the proportion of variance explained by the regression model making it useful as a measure of success of predicting the dependent variable from the explanatory variables (Nagelkerke, 1991; Hanson, 2010). R square (R2) should lie between 0 and 1 which is invariant to units of measurement and becomes larger as the model fits better (Magee, 1990).

Variance Inflation Factor (VIF) measures the impact of multicollinearity among predictor variables in a regression (Robinson & Schumacker, 2009). The general rule is that the VIF should not exceed 10 (Belsley & Kuh, 1980). The estimated profit function revealed that recommended coffee varieties, manure rate and capping were significant in affecting coffee profitability. Other factors found to significantly influence coffee profitability were prices for foliar feed, manure and labour. Expenditure on foliar feed and manure was also significant in affecting coffee profitability. The findings on each significant factor are discussed hereafter.

5.2.3.1 Recommended Variety Adoption of improved coffee varieties (Ruiru 11 or Batian) was quite low (32.8%) but was found to be positive and significant in explaining variations in profit margins. These improved varieties were resistant to coffee berry disease (CBD) and coffee leaf rust (CLR) hence reduced production cost on fungicides. This in turn increased coffee productivity and high quality cherry which led to increased returns. Reduced cost of agrochemicals and labour through adoption of the improved varieties was found to increase the marketable value of cherry which guaranteed high returns. Andrew & Philip (2014) also reported that the use of disease resistant coffee varieties reduced the cost of agrochemicals which was positively related to coffee profitability in Kigoma, Tanzania.

57

Similar findings were also reported by Van der Vossen et al. (2015) and Haggar et al. (2017) on impact of improved coffee varieties on sustainable coffee production and profitability among arable farms. Nguezet et al. (2011) found adoption of improved rice varieties to significantly increase the income among rice farming households. However, Musaba & Bwacha, (2014) found impact of improved maize seed varieties insignificant in influencing farm returns.

5.2.3.2 Manure Rate Gross margins for adopters of recommended manure rate of one debe per tree was higher than that of non-adopters. Chemura et al. (2010); Mohammed et al. (2013); McArthur & McCord, (2017) found similar results for manure application and coffee profitability. Manure use acted as a substitute for inorganic fertilizers which were relatively expensive. The quality of the inorganic fertilizers, their costs and yield contribution varies from source to source due to variations in composition. Organic manure has been used as a strategy for climate change and sustainable production with reduced impact on the environment, which would in turn increase environmental- economic benefits and trade-offs for sustainable production and high returns on coffee farms. Contrary, Musaba & Bwacha, (2014) found manure use insignificant in explaining variations in returns from maize production.

5.2.3.3 Capping Gross margin for adopters of tree capping was lower than that of non-adopters. Tree capping encouraged lateral growth and bushy coffee trees at the expense of productivity, hence reduced net returns. Capping also increased the cost of hired labour due to increased demand for pruning. These findings concur to that of Magha (2013); Perdoná, & Soratto (2015); Odeny (2016) on effect of capping and coffee returns. Capping would lead to dense canopy and shade limiting light penetration. Dense canopy would also act as alternate host for pests which would lead to expenditure on agrochemicals, reducing the production potential of coffee trees hence negatively influencing coffee returns (Magha, 2013). However, Ghosh and Bera, (2014) found capping significant and positively related to profitability of sweet oranges.

58

5.2.4 Expenditure on Coffee Production 5.2.4.1 Foliar Feed Variable Cost This was a product of amount of foliar feed used and the price of foliar per unit. Increase in price for foliar per unit led to an increase in expenditure on foliar feed where both the price and expenditure on foliar negatively impacted on coffee returns in the study area, implying that marginal increase in expenditure on foliar was more than the marginal value product of coffee. Increase in cost of foliar implied that the marginal value product of coffee was more than the unit price of foliar fertilizer hence the negative returns. Foliar fertilizer is meant to correct micronutrient deficiency and maybe application of foliar was not important since no soil or leaf analysis had been conducted to ascertain the status, hence expenditure on foliar negatively influencing gross returns given the minimal marginal increase in coffee yields. This would reduce coffee productivity which translates to reduced profitability. These findings are similar to that by Castro-Tanzi et al. (2012); Alexander, (2012); Andrew & Philip, (2014); Komarek et al. (2017) who found cost of foliar negatively related to coffee returns.

5.2.4.2 Manure Variable Cost This was a product of amount of manure used and prices per unit. Increase in price of manure which was a substitute for inorganic fertilizers, increased expenditure on manure given the availability. Incurring cost to acquire manure negatively influenced coffee returns. This implied high marginal price elasticity for manure with respect to coffee returns with an increase in prices of manure whereby a unit increase in expenditure on manure led to less than a unit increase in marginal value product of coffee. Organic manure acted as a substitute for inorganic fertilizers and at higher prices households would apply less manure. Increased expenditure on manure means reduced application hence reduced coffee productivity ultimately impacting on net returns. Haggar et al. (2017); Oerke et al. (2012); Komarek et al. (2017) Bravo-Monroy et al. (2016); reported similar findings that cost of manure and agrochemicals negatively affected coffee profitability.

59

5.2.4.3 Wage Rate Cost of labour was positive and significant in influencing coffee gross returns. This was explained by the high marginal elasticity of hired labour with respect to farm profits. Hired labour was mainly engaged by business minded farmers whose farm activities could not be comfortably handled by family labour. The plausible explanation is that hired labour would increase efficiency and supervisory roles given that the farmer incurred cost compared to family labour. Increase in wage rate would act as an incentive and motivation for increased labour productivity hence increased coffee net returns. Increase in wage rate implied that the marginal value product for coffee was more than the price of labour per man-day. These findings concur to that by Thuku et al. (2013); Mohammed et al. (2013); Woodill et al. (2014) that expenditure on labour use increased coffee returns.

5.2.5 Effect of Recommended Technologies on Coffee Quality Coffee quality was hypothesized to be a function of recommended technologies and socioeconomic factors at the farm level. The maximum likelihood (-2log) ratio test evaluates the overall relationship between the explanatory variables and the dependent variable. The Wald test evaluates whether or not the predictor variables is statistically significant in differentiating between two categories embedded in binary logistic regression (Bayaga, 2010). Overall significance of the model is examined using three inferential statistical tests, the likelihood ratio, score and Wald tests. All these yield similar conclusions (Peng et al., 2002). The estimated Binary logistic regression model revealed that the recommended foliar feed, manure rates and pruning were significant in affecting coffee quality. Other factors found to significantly affect coffee quality were gender and household size. The findings on each significant factor are discussed hereafter.

5.2.5.1 Foliar Feed Rate Adoption of recommended foliar feed rate had positive and significant effect on coffee quality, which implied that adequate and timely supply of foliar feed increased the quality of coffee cherry produced. Foliar feed rich in Zinc and Boron will increase the growth of root system for nutrient uptake and increased photosynthetic activity (Belay et

60 al., 2016). This will improve the marketable value and quality of coffee berries produced. Castro-Tanzi et al. (2012); Njogu et al. (2014); Belay et al. (2016); also reported positive and significant relationship between foliar application rate and quality of coffee.

5.2.5.2 Pruning The study revealed that adopters of pruning had lower proportions of mbuni to cherry compared to non-adopters implying that farmers who pruned their coffee increased the quality of coffee produced. Pruning of unproductive stems allows light penetration and reduces pest and disease prevalence hence improved quality of coffee berries (Belay et al., 2016). These findings are in line with those of Läderach et al. (2011); Belay et al. (2016); Bote & Vos (2017) who found pruning to be positively related to quality as it modified air movement and reduced disease incidences. Long et al. (2015) found spacing and pruning of lateral branches of coffee trees to be positively related to quality of cherry produced.

5.2.5.3 Manure Rate Adoption of recommended manure rate was significant in explaining quality variations that is the reduction of proportion of mbuni to cherry. Tree pruning mulch used as green manure increases Carbon and Nitrogen availability which in turn will increase coffee quality and yield (Chemura et al., 2010). Organic manure increases the plant thickness and height as well as maintaining the physical, chemical and biological properties of the soil (Dzung et al., 2013). This promotes sustainable and climate smart agriculture without impact on the environment hence high quality coffee berries. Other studies conducted by Laderach et al. (2011); Bekeko, (2013); Tsegaye et al. (2014) reported similar findings of positive effect of organic manure on quality of coffee produced.

5.2.6 Effect of Socioeconomic Factors on Coffee Quality 5.2.6.1 Gender The study found gender of the farmer to be significant in influencing coffee quality (proportion of mbuni to cherry) in the study area due to high labour productivity associated with gender-yield differentials. Males would facilitate key production operations such as weed and disease control for improved quality of cherry and minimal

61 mbuni. Hill & Vigneri (2014) and Dzung et al. (2013) found gender-yield differential and specific constraints positive and significant in influencing the quality of coffee.

5.2.6.2 Household Size Increase in household size led to increase in family labour which increased productivity of coffee through farm operations hence improved quality and reduced cost of production. Increase in household size was associated with a reduction in proportion of mbuni to cherry. Family labour would facilitate the key farm operations such as weed control, disease control, pruning and harvesting hence improved coffee quality. Mignouna et al. (2010); Akudugu et al. (2012); Tirfe et al. (2015) found similar results as household was a source of family labour for farm operations which positively influenced quality coffee.

5.3 Conclusions Results of this study have revealed mixed effect of farm and farmer characteristics on coffee performance at farm level. Farmers in the study area applied the recommendations at different extents due to a combination of different factors. Farmers used the technologies if the tradeoff between perceived benefits and costs was greater than one, i.e. expected benefits outweigh the production costs.

Based on study findings for objective one, on the effect of recommended technologies on coffee productivity, it was concluded that these recommendations have significant effect on coffee productivity at the farm level. It is the interaction of these technologies with availability of off-farm income, labour use, credit access and land size that influenced coffee productivity positively at the farm level. The study tested the null hypothesis that the recommended technologies have no significant effect on coffee productivity. Results of the study revealed that application of manure, fungicide and pesticide at the recommended rates had significant and positive impact on coffee productivity. Therefore, the null hypothesis is not accepted while the alternative that recommended technologies have significant effect on coffee productivity is true.

Based on the study findings for objective two, results revealed that recommended variety, manure rate and capping were significant in explaining profit variations. Factor

62 prices and expenditure on key inputs had negative and significant effect on coffee gross returns. The study tested the hypothesis that the recommended technologies have no significant effect on coffee profitability. Results revealed that using the recommended coffee varieties (Ruiru 11 or Batian), manure rates and capping had significant effect on coffee profitability. Therefore, the null hypothesis is rejected and the alternative that the recommended technologies have significant effect on coffee profitability is true.

Based on the study findings on objective three, the study revealed that adoption of recommended foliar feed; manure and pruning were key determinants of quality variations across the study farms. The study tested the hypothesis that the recommended technologies have no significant effect on coffee quality. Results revealed that application of foliar feed, manure at the recommended rates and pruning positively and significantly influenced coffee quality. Therefore, the null hypothesis is rejected and alternative that the recommended technologies have significant effect on coffee quality is true.

5.4 Recommendations Based on the study findings on economic impact of recommended technologies on coffee performance among smallholder farmers, there was evidence of variations in output, gross returns and quality (proportion of mbuni to cherry) across the farms. Therefore, the study came up with the following recommendations to farmers for increased productivity, profitability and quality of coffee at the farm level; a. In order to improve coffee productivity at the farm level, the farmers should; i. Apply well decomposed manure or coffee pulp at the recommended rate of one debe per tree since it significantly increased coffee output per unit area. ii. Farmers who are still growing traditional varieties (SL 28, SL 34, K7), should apply fungicide for control of CBD and CLR at the recommended rates to improve on cherry productivity and the marketable value of cherry produced. iii. Apply pesticides for control of insect pests such as berry borer, thrips and stem borers etc. at the recommended rate for each pesticide to increase cherry productivity at the farm level.

63

iv. For maximum coffee productivity, farmers should adopt the improved variety, apply pesticides and manure at the recommended rate (profit maximizing input- output combination). v. Coffee farmers should diversify their income sources to finance key coffee farm inputs and also increase their household risk bearing ability to absorb production and market risks. vi. Government should formulate or review land use policies to regulate land fragmentation since land size under coffee significantly explained productivity variations across the farms. b. To increase coffee net returns and income at the farm level, the farmers should; i. Adopt the recommended and improved coffee varieties (Ruiru 11 or Batian) for they are disease resistant with high yield potential compared to traditional varieties (SL 34, SL 28 or K 7) to ensure a reduction in cost of agrochemicals hence high net returns. ii. Ensure application of organic manure at the recommended rate of one debe per tree given its availability as it is a substitute for inorganic fertilizer, readily available and cheap to off-set the negative effect of expenditure on inorganic fertilizers. Research institutions and technology transfer agents should train farmers on organic manure as it performs better than inorganic fertilizers under drought conditions hence better returns. iii. The Kenyan government should formulate an economic policy aimed at stabilizing factor prices of key inputs such as foliar feed and manure as expenditure on them significantly influenced coffee returns in order to improve competitiveness of the crop in the international markets and exchange rate volatility in the long run. c. To improve the quality of cherry produced, the farmers should; i. Apply foliar feed efficiently at the recommended rate in order to improve the quality of cherry produced and also promote bean color. This would boost coffee

64

quality through nutrient uptake in case of low irrigation levels hence reduced proportions of „mbuni‟. ii. Apply well decomposed organic manure at the recommended rate of one debe per tree as a strategy for sustainable agriculture and climate change for improved environmental-economic benefits. The government should formulate a policy on revitalizing coffee production and quality through green manure and decomposed coffee pulp for sustainable agriculture in moisture stress areas. iii. Timely prune their coffee trees and allow 2–3 bearing heads per stem as this increases air flow and also reduces pest and disease incidences for increased productivity and high quality cherry. iv. The government through research and technology transfer agents should develop training programs for farmers on key farm agronomic practices to boost coffee production and quality at the farm level.

65

REFERENCES Abasanbi, A. A. (2010). Assessment of coffee quality and its related problems in Jimma zone of Oromia Regional State. MSc thesis in Agriculture (Horticulture). 141p. Jimma (Ethiopia): Jimma University. Adejumo, T. O. (2005). Crop protection strategies for major diseases of cocoa, coffee and cashew in Nigeria. Adesina, A. A., & Djato, K. K. (1996). Farm size, relative efficiency and agrarian policy in Côte d'Ivoire: profit function analysis of rice farms. Agricultural Economics, 14(2), 93-102. AFA (2016). Agriculture and Food Authority (AFA) 2016 - 2021 Strategic Plan, 2016– 2021. Afolami, C. A., Obayelu, A. E., & Vaughan, I. I. (2015). Welfare impact of adoption of improved varieties by rural households in South Western Nigeria. Agricultural and Food Economics, 3(1), 18. Aidoo, R., Mensah, J. O., & Tuffour, T. (2013). Determinants of household food security in the Sekyere-Afram plains district of Ghana. In European Scientific Journal. Aigner, D., Lovell, C. K., & Schmidt, P. (1977). Formulation and estimation of stochastic frontier production function models. Journal of econometrics, 6(1), 21- 37. Aiken, L. S., West, S. G., & Reno, R. R. (1991). Multiple regression: Testing and interpreting interactions. Sage. Akudugu, M. A., Guo, E., & Dadzie, S. K. (2012). Adoption of modern agricultural production technologies by farm households in Ghana: What factors influence their decisions. Journal of biology, agriculture and healthcare, 2(3). Alexander, A. (Ed.). (2012). Foliar Fertilization: Proceedings of the First International Symposium on Foliar Fertilization, Organized by Schering Agrochemical Division, Special Fertilizer Group, Berlin (FRG) March 14–16, 1985 (Vol. 22). Springer Science & Business Media. Ameyu, M. A. (2017). Influence of harvesting and postharvest processing methods on the quality of Arabica coffee (Coffea arabica L.) in Eastern Ethiopia. ISABB Journal of Food and Agricultural Sciences, 7(1), 1-9. Andrew, R., & Philip, D. (2014). Coffee production in Kigoma Region, Tanzania: Profitability and Constraints. Tanzania. Journal of Agricultural Sciences, 13(2). Bayaga, A. (2010). Multinomial Logistic Regression: Usage and application in risk analysis. Journal of applied quantitative methods, 5(2). Beck, U., Singhal, S., & Tarp, F. (2016). Coffee price volatility and intra-household labour supply. No. UNU-WIDER Research Paper wp2016-0016.

66

Bekeko, Z. (2013). Effect of maize stover application as soil mulch on yield of Arabica coffee (Coffee arabica L., Rubiaceae) at Western Hararghe Zone, Eastern Ethiopia. Sustainable Agriculture Research, 2(526-2016-37774). Belay, S., Mideksa, D., Gebrezgiabher, S., & Seifu, W. (2016). Factors affecting coffee (coffea arabica l.) quality in Ethiopia: A. Journal of Multidisciplinary Scientific Research, 4(1), 22-28. Belsley, D. A., & Kuh, E. (1980). Welsch, RE-1980.“Regression Diagnostics. Identifying Influential Data and Sources of Collinearity”. Uiley Ser. Probab. Math. Stat. Bigirimana, J., Adams, C. G., Gatarayiha, C. M., Muhutu, J. C., & Gut, L. J. (2019). Occurrence of potato taste defect in coffee and its relations with management practices in Rwanda. Agriculture, Ecosystems & Environment, 269, 82-87. Bote, A. D., & Struik, P. C. (2011). Effects of shade on growth, production and quality of coffee (Coffea arabica) in Ethiopia. Journal of Horticulture and Forestry, 3(11), 336-341. Bote, A. D., & Vos, J. (2017). Tree management and environmental conditions affect coffee (Coffea arabica L.) bean quality. NJAS-Wageningen Journal of Life Sciences, 83, 39-46. Bocher, T. F., & Simtowe, F. (2017). Profit efficiency analysis among groundnut farmers from Malawi. Journal of Development and Agricultural Economics, 9(10), 278-288. Bravo-Monroy, L., Potts, S. G., & Tzanopoulos, J. (2016). Drivers influencing farmer decisions for adopting organic or conventional coffee management practices. Food policy, 58, 49-61. Carvalho, F. P. D., França, A. C., Souza, B. P. D., Fialho, C. M. T., Santos, J. B. D., & Silva, A. A. D. (2014). Water use efficiency by coffee arabica after glyphosate application. Acta Scientiarum. Agronomy, 36(3), 373-377. Castro-Tanzi, S., Dietsch, T., Urena, N., Vindas, L., & Chandler, M. (2012). Analysis of management and site factors to improve the sustainability of smallholder coffee production in Tarrazú, Costa Rica. Agriculture, ecosystems & environment, 155, 172-181. Challa, M., & Tilahun, U. (2014). Determinants and impacts of modern agricultural technology adoption in west Wollega: the case of Gulliso district. Journal of Biology, Agriculture and Healthcare, 4(20), 63-77. Chemura, A. (2014). The growth response of coffee (Coffea arabica L) plants to organic manure, inorganic fertilizers and integrated soil fertility management under different irrigation water supply levels. International Journal of Recycling of Organic Waste in Agriculture, 3(2), 59. Chemura, A., Mahoya, C., & Kutywayo, D. (2010). Effect of organic nursery media on germination and initial growth of coffee seedlings. In 23rd Colloquium of the

67

Association for Science and Information on Coffee (ASIC) (pp. 3-8). Chepng‟etich, E., Nyamwaro, S. O., Bett, E. K., & Kizito, K. (2015). Factors that influence technical efficiency of sorghum production: A case of small holder sorghum producers in Lower Eastern Kenya. Advances in Agriculture, 2015. Cheserek, J. J., & Gichimu, B. M. (2012). Drought and heat tolerance in coffee: a review. Chiona, S. (2011). Technical and allocative efficiency of smallholder maize farmers in Zambia. Unpublished MSc Thesis, University of Zambia, Lusaka, Zambia. Cochran, W. G. (1963). Sampling Techniques, New York, 1953. Statistical Surveys E. Grebenik and CA Moser. CRI (2017). Coffee Research Institute. Arabica Coffee Hybrid Resistant to Coffee Berry Disease and Leaf Rust. Revised May 2017, Ruiru. Diewert, W. E. (1974). Applications of duality theory. Dzung, N. A., Dzung, T. T., & Khanh, V. T. P. (2013). Evaluation of coffee husk compost for improving soil fertility and sustainable coffee production in rural central highland of Vietnam. Resources and Environment, 3(4), 77-82. FAO (2008). Food and agriculture organization of the United Nations. Retrieved on, 15 February 2018. FAO (2015). The state of food insecurity in the world 2014: strengthening the enabling environment for food security and nutrition. FAO, Rome. Gebeyehu, M. G. (2016). The impact of technology adoption on agricultural productivity and production risk in Ethiopia: Evidence from Rural Amhara Household Survey. Open Access Library Journal, 3(02), 1. Ghosh, S. N., & Bera, B. (2014). Effect of pruning on productivity in sweet orange. Journal of Horticultural Science, 9(2), 206-208. Gichimu, B. M., & Omondi, C. O. (2010). Early performance of five newly developed lines of Arabica coffee under varying environment and spacing in Kenya. Agric. Biol. J. N. Am., 1(1), 32–39. GoK (2007). Kenya Vision 2030. A Globally Competitive and Prosperous Kenya, 2007 (October) GoK (2009). Agricultural Sector Development Strategy (ASDS) 2009-2020. Nairobi. GoK (2013a). The Kenya Agricultural and Livestock Research Act, 2013. Nairobi, 25 January 2013. GoK (2013b). Embu County Government. County Integrated Development Plan 2013- 2017. GoK (2013c). The Crops Act, 2013. National Council for Law Reporting. Revised Edition 2013.

68

Haggar, J., Soto, G., Casanoves, F., & de Melo Virginio, E. (2017). Environmental- economic benefits and trade-offs on sustainably certified coffee farms. Ecological indicators, 79, 330-337. Hancock, D. (2012). A theory of production for the financial firm (Vol. 4). Springer Science & Business Media. Hanson, T. (2010). Multiple regression. Hill, B. (2014). An introduction to economics: Concepts for students of agriculture and the rural sector. Hill, R. V., & Vigneri, M. (2014). Mainstreaming gender sensitivity in cash crop market supply chains. In Gender in agriculture (pp. 315-341). Springer, Dordrecht. Jaetzold, R., Schmidt, H., Hornetz, B., & Shisanya, C. (2009). Farm Management Handbook of Kenya Vol.II: Nyanza Province, II, Sub-Part A. ICO,(2019). International Coffee Organization, International coffee council, 124th Session, (124-7), March 2019, Nairobi Kenya Kamau, V., Ateka, J., Mbeche, R., & Kavoi, M. M. (2016). Assessment of technical efficiency of smallholder coffee farming enterprises in Muranga, Kenya. In Scientific Conference Proceedings, November. Kebati, R. K., Nyangeri, J., Omondi, C. O., & Kubochi, J. M. (2016). Effect of artificial shading on severity of Coffee Berry Disease in Kiambu County, Kenya. Annual Research & Review in Biology, 9(2). Kolawole, O., & Ojo, S. O. (2007). Economic efficiency of small scale food crop production in Nigeria: A stochastic frontier approach. Journal of Social Sciences, 14(2), 128-130. Komarek, A. M., Drogue, S., Chenoune, R., Hawkins, J., Msangi, S., Belhouchette, H., & Flichman, G. (2017). Agricultural household effects of fertilizer price changes for smallholder farmers in central Malawi. Agricultural Systems, 154, 168-178. Läderach, P., Oberthür, T., Cook, S., Iza, M. E., Pohlan, J. A., Fisher, M., & Lechuga, R. R. (2011). Systematic agronomic farm management for improved coffee quality. Field Crops Research, 120(3), 321-329. Lechenet, M., Dessaint, F., Py, G., Makowski, D., & Munier-Jolain, N. (2017). Reducing pesticide use while preserving crop productivity and profitability on arable farms. Nature Plants, 3(3), 17008. Lenssen, A. W. (2013). Biofield and fungicide seed treatment influences on soybean productivity, seed quality and weed community. Agricultural Journal, (8), 138-143. Magee, L. (1990). R 2 measures based on Wald and likelihood ratio joint significance tests. The American Statistician, 44(3), 250-253. Magha, M. (2013). Factors contributing to unprofitable yield of coffee in the daughters of mary Ngorongoro farm in Karatu district (Doctoral dissertation, Mzumbe

69

University). Markos, D., Pal, U. R., & Uragie, E. (2012). Dry matter partitioning, nodulation and seed traits of medium and late maturing soybean varieties as affected by planting pattern and plant density. Journal of Agricultural Science and Technology.2012 Mati, B. M. (2016). Policy and institutional frameworks for sustainable land management in Kenya: Overview of the policy, legislative and institutional frameworks for sustainable land management in the public sector in Kenya. Kenya Country Report (Executive summary, February). Mburu, S., Ackello-Ogutu, C., & Mulwa, R. (2014). Analysis of economic efficiency and farm size: A case study of wheat farmers in Nakuru District, Kenya. Economics Research International, 2014. McArthur, J. W., & McCord, G. C. (2017). Fertilizing growth: Agricultural inputs and their effects in economic development. Journal of development economics, 127, 133-152. Mensah, K. K., Retief, J. V., & Barnardo-Viljoen, C. (2013). Reliability based application of Eurocode 2‟s Variable Strut Inclination Method for shear. In 11th International Conference on Structural Safety & Reliability (ICOSSAR) (pp. 16- 20). Meylan, L., Gary, C., Allinne, C., Ortiz, J., Jackson, L., & Rapidel, B. (2017). Evaluating the effect of shade trees on provision of ecosystem services in intensively managed coffee plantations. Agriculture, ecosystems & environment, 245, 32-42. Mignouna, D. B., Mutabazi, K., Senkondo, E. M., & Manyong, V. M. (2010). Adoption of a new maize and production efficiency in western Kenya. In Joint 3rd African Association of Agricultural Economists and 48th Agricultural Economists Association of South Africa Conference. Minai, J. M., Nyairo, N., & Mbataru, P. (2014). Analysis of socio-economic factors affecting the coffee yields of small holder farmers in Kirinyaga County, Kenya. Journal of Agricultural and Crop Research, 2(12), 228-235. MoA (2011). Ministry of Agriculture, Livestock and Fisheries Strategic Plan 2013-2017. Nairobi Mohammed, A. B., Ayanlere, A. F., & Ekenta, C. M. (2013). Profitability of coffee production in Kabba/Bunu local government area of Kogi State Nigeria. African Journal of Agricultural Research, 8(23), 2897-2902. Muriithi, L. W. (2016). Impact of coffee certification on small holder coffee farming in Embu County, Kenya. (Doctoral dissertation, Kenyatta University). Musaba, E., & Bwacha, I. (2014). Technical efficiency of small scale maize production in Masaiti district, Zambia: a stochastic frontier approach. Journal of Economics and Sustainable Development, 5(4), 104-111.

70

Muzari, W., Gatsi, W., & Muvhunzi, S. (2012). The impacts of technology adoption on smallholder agricultural productivity in sub-Saharan Africa: A review. Journal of Sustainable Development, 5(8), 69. Nagelkerke, N. J. (1991). A note on a general definition of the coefficient of determination. Biometrika, 78(3), 691-692. Narayana, M. R. (2016). Management of Leaf Rust Disease by Household Coffee Farmers in India: Evidence and Implications for Price and Income Effects on Chemical Control. Agricultural research, 5(1), 43-50. Ndirangu, S. N., Mbogoh, S. G., & Mbatia, O. L. E. (2017). Effects of land fragmentation on food security in three agro-ecological zones of Embu County in Kenya. Asian Journal of Agricultural Extension, Economics & Sociology, 18(4): 1- 9, 2017. Njogu, R. N. E., Kariuki, D. K., Kamau, D. M., & Wachira, F. N. (2014). Effects of foliar fertilizer application on quality of tea (Camellia sinensis) grown in the Kenyan highlands. American Journal of Plant Sciences, 5(18), 2707. Ngeywo, J. C., Shitandi, A. A., Basweti, E. A., & Agasa, L. O. (2015). Effect of Farm Succession and Farm Inputs use on Coffee Productivity in Kisii County, South Western Kenya. Asian Journal of Agricultural Extension, Economics & Sociology, 69-75. Nguezet, D., & Diagne, A. (2011). Impact of improved rice technology on income and poverty among rice farming household in Nigeria : A local average treatment effect (LATE) approach. Quarterly Journal of International Agriculture, 50(3), 267–291. Nguezet, P. M. D., Diagne, A., Okoruwa, V. O., Ojehomon, V., & Manyong, V. (2011). Impact of improved rice technology (NERICA varieties) on income and poverty among rice farming households in Nigeria: a local average treatment effect (LATE) approach. Quarterly Journal of International Agriculture, 50(3), 267-292. Nyamwamu, R. O. (2016). Implications of human-wildlife conflict on food security among small holder agropastoralists: a case of smallholder maize (Zea mays) farmers in Laikipia County, Kenya. World Journal of Agricultural Research, 4(2), 43-48. Odeny A. D. (2016). Effect of Cordia africana Lam shade on physiology, yield and quality of arabica coffee. (Doctoral dissertation, University of Nairobi). Ogada, M. J., Nyangena, W., & Yesuf, M. (2010). Production risk and farm technology adoption in the rain-fed semi-arid lands of Kenya. AfJARE, 4(1), 154-174. Oerke, E. C., Dehne, H. W., Schönbeck, F., & Weber, A. (2012). Crop production and crop protection: estimated losses in major food and cash crops. Elsevier. Perdoná, M. J., & Soratto, R. P. (2015). Irrigation and intercropping with macadamia increase initial Arabica coffee yield and profitability. Agronomy Journal, 107(2), 615-626.

71

Peng, C. Y. J., Lee, K. L., & Ingersoll, G. M. (2002). An introduction to logistic regression analysis and reporting. The journal of educational research, 96(1), 3-14. Pinard, F., Boffa, J. M., & Rwakagara, E. (2014). Scattered shade trees improve low- input smallholder Arabica coffee productivity in the Northern Lake Kivu region of Rwanda. Agroforestry systems, 88(4), 707-718. Popp, J., Pető, K., & Nagy, J. (2013). Pesticide productivity and food security. A review. Agronomy for sustainable development, 33(1), 243-255. Roba, (2017). Evaluation of coffee (Coffea arabica L.) on raw and cup quality aspect under the canopy of Cordia africana and Erythrina Abyssinica shade trees effect in Arsi Golelcha District, Ethiopia. Food Science and Quality Management. Robinson, C., & Schumacker, R. E. (2009). Interaction effects: centering, variance inflation factor, and interpretation issues. Multiple Linear Regression Viewpoints, 35(1), 6-11. Sedgwick, P. (2014). Cross sectional studies: advantages and disadvantages. Bmj, 348, g2276. Senkondo, Y. H., Tack, F. M. G., & Semu, E. (2014). Copper accumulations in soils, coffee, banana, and bean plants following copper-based fungicides in coffee farms in Arusha and Kilimanjaro regions, Tanzania. Communications in soil science and plant analysis, 45(15), 2032-2045. Seyoum, E. T., Battese, G. E., & Fleming, E. M. (1998). Technical efficiency and productivity of maize producers in eastern Ethiopia: a study of farmers within and outside the Sasakawa-Global 2000 project. Agricultural economics, 19(3), 341-348. Shepherd, R. W. (2015). Theory of cost and production functions. Princeton University Press. Sualeh, A., Endris, S., & Mohammed, A. (2014). Processing method, variety and roasting effect on cup quality of Arabica coffee (Coffea arabica L.). Disc. J. Agri. Food Sci, 2(2), 70-75. Thuku, G. K. (2013). Effects of reforms on productivity of coffee in Kenya. International Journal of Business and Social Science, 4(15), 196–213. Tirfe, E., Kufa, T., & Mohammed. A. (2015). Liquor quality performance of some early released coffee varieties at three locations in South-West Ethiopia, 10(2), 110–116. Tsegaye, B., Mohammed, A., & Getachew, E. (2014). Impact of Sun Drying Methods and Layer Thickness on the Quality of Highland Arabica Coffee Varieties at Limmu, Southwestern Ethiopia. Journal of Horticulture, 1-7. Uddin, M. T., Dhar, A. R., & Islam, M. M. (2016). Adoption of conservation agriculture practice in Bangladesh: Impact on crop profitability and productivity. Journal of the Bangladesh Agricultural University, 14(1), 101-112. United Nations (2015). Water Annual Report. http://www.unwater.org/publications/un- water-annual-report-2015. Accessed on 12th January 2018.

72

Uzawa, H. (1964). Duality principles in the theory of cost and production. International Economic Review, 5(2), 216-220. van der Vossen, H., Bertrand, B., & Charrier, A. (2015). Next generation variety development for sustainable production of arabica coffee (Coffea arabica L.): a review. Euphytica, 204(2), 243-256. Van Long, N., Ngoc, N. Q., Dung, N. N., Kristiansen, P., Yunusa, I., & Fyfe, C. (2015). The effects of shade tree types on light variation and Robusta Coffee production in Vietnam. Engineering, 7(11), 742. Woodill, A. J., Hemachandra, D., Nakamoto, S. T., & Leung, P. (2014). The economics of coffee production in Hawai „i. Econ. Issues, 25, 1-9. Worku, M., & Astatkie, T. (2015). Effects of row spacing on productivity and nodulation of two soybean varieties under hot sub-moist tropical conditions in south-western Ethiopia. Journal of Agriculture and Rural Development in the Tropics and Subtropics (JARTS), 116(2), 99-106. World Bank (2015). The World Bank Annual Report 2015.Washington DC World Bank (2016). World Development Indicators. Structure of Output, World Bank Group.

73

APPENDICES Appendix 1: Household Questionnaire for Small-scale Coffee Farmers Instructions Please tick appropriately in the box provided, and also fill in your responses in the blank spaces provided where detailed answers are required. Please fill the questionnaire honestly and objectively as the information will be treated with utmost confidentiality. Name of the farmer (optional)…………………...………………………………………………………..…… Date of response……………………………………………………………………………...…… Cooperative society ……………………………………………………………………… Demographic characteristics 1. Gender of the farmer male [ ] female [ ] 2. Age of the farmer 18-30 [ ] 31-40 [ ] 41-50 [ ] 51-60 [ ] 61 and above [ ] 3. Household head husband[ ] wife[ ] 4. Level of education non formal [ ] primary [ ] secondary [ ] tertiary [ ] other specify……………………………………………………… 5. How many years have you been growing coffee? Less than 10 years [ ] 10-20 years [ ] above 20 years [ ] 6. What is your household size? …………………………………………………….. 7. Do you have any other form of income other than farming? Yes [ ] no [ ] 8. If yes in (7) above, from what kind of occupation is the income from? Self-employed [ ] Formal employment [ ] Informal employment [ ] 9. What is the monthly income from the off-farm employment? Less than shs10, 000 [ ] shs10, 000-20,000 [ ] above shs20, 000 [ ] 10. How did you acquire the piece of land? Bought [ ] Ancestral land [ ] Family land [ ] 11. If bought, do you have a title deed for the land? Yes [ ] No [ ]

74

12. What is the size of your farm? …………………………………………………… 13. Do you receive extension services? Yes [ ] No [ ] If yes how often in a cropping season? Once [ ] Twice [ ] Thrice [ ] any other specify……………….. 14. What kind of institutions offers the extension services? Cooperative society [ ] Research institutions [ ] 15. Which coffee variety(s) have you planted in your farm? ………………………………………………………………………………..…… 16. Which is the source of the planting materials such as seedlings? Cooperative society [ ] Research centers [ ] others………………… 17. How did you come to know about the improved coffee varieties? Extension agents [ ] local media [ ] Fellow farmers [ ] cooperative [ ] 18. Have you grown any of the two varieties, Ruiru 11 or Batian? Yes [ ] No [ ] …………………………………………………………………………………….. 19. What other variety had you grown before the improved coffee variety? …………………………………………………………………………………….. 20. After first hearing about the varieties, how long did you take to make the decision to adopt? Less than a year [ ] one year [ ] two years [ ] more than two years [ ] 21. Do you have access to credit from lending institutions? Yes [ ] No [ ] If yes what is the repayment period for the loan? ………………………………………………………………………………..…… 22. Which marketing channel do you use to sell your coffee cherry after harvesting? Through cooperatives [ ] others (specify)………………………………….

Coffee Production 23. What average land size have you allocated to coffee? ……………………………………………………………………………………..

75

24. How many mature coffee trees do you have in your farm? ……………………………………………………………………………..……… 25. What is the age of your coffee trees? Less than 2 years [ ] 2 -5 years [ ] 5 – 7 years [ ] more than 7 years [ ] 26. Have you ever changed the cycle of coffee trees through top working? Yes [ ] No [ ] If yes which year …………………………………………………………………………………….. 27. Please indicate the spacing of the coffee trees ……………………………….……………………………………………..……... 28. Please indicate the amount of berries produced in the last crop year 2016/2017 in the table below. Type of berry Quantity in kilograms Average price per kilogram Coffee cherry Mbuni 29. Do you carry out sorting to remove the affected cherry at the farm level? Yes [ ] No [ ] 30. How many kilograms of coffee berry was lost after sorting at home before taking to the factory due to defects?…………………………………………………………..……………… 31. What was the grade of the coffee cherry you delivered to the factory? …………………………………………………………………………………… 32. What form of labour do you employ in harvesting? Family labour [ ] Hired labour [ ] both [ ] 33. If hired labour, what is the labour requirement for harvesting? Man days………………………… cost per man day……………………… Debes …………………………… Cost per debe ………………………… 34. What is the harvesting stage of your coffee cherry? Full maturity [ ] Just maturing [ ] 35. What method of harvesting do you use to harvest your coffee? Selective picking [ ] strip harvesting [ ]

76

36. After harvesting, how long does it take to dry the mbuni depending on weather conditions? Less than 1 week [ ] 1 week [ ] 1 month [ ] more than 1 month [ ] Inputs and farm operations 37. Number of coffee trees for each variety in case of more than one variety. Variety Number of trees Ruiru 11 Batian SL 28 SL 34 K 7 Other (specify)

38. Please indicate in the table the type and amounts of fertilizer used. Fertilizer Method of Amount Price per Time of Frequency of type application used/acre kg application application (splits)

39. In case of foliar fertilizer application, indicate in the table below. Type of Method of Amount Price per Time of Frequency of foliar application used/acre litre application application

40. How much of labour is required for fertilizer application? Man days…………………………….. Cost per man day in kshs …………… 41. In case of manure use, please indicate the type and amounts used.

77

Manure type Method of Amount used/ Price of Labour application acre manure required

42. Which is the common weed prevalent in your farm? …………………………………………………………………….………………. 43. Please indicate the method of weed control used in your farm? Manual control [ ] Chemical control [ ] other (specify) ……………… 44. If chemical control, indicate the type and cost of each herbicide. Type of Herbicide Amounts Price of herbicide Frequency of weed type used per acre per litre application

45. Please indicate the labour requirement for weeding indicated in (44) above? Man days …………………………….. Cost per man day …………….. 46. Indicate in the table below the fungicides used to control diseases if any. Disease Fungicide Amount Price per Cost of Time of type to used applied per litre/Kg spraying application control acre

47. What are the common pests that attack your coffee? …………………………………………………………………………………… 48. Do you use any pesticides to control the crop pests? Yes [ ] No [ ]

78

49. If yes in (48), indicate the type of pesticides and their prices used to control coffee pests. Type of Type of Amounts Price per litre Time of pest pesticide used per acre application

50. Please indicate the labour requirement for pest control. Man days …………………………….. Cost per man day…………………….. 51. Do you carry out pruning of your coffee trees? Yes [ ] No [ ] 52. Do you carry out capping of your coffee trees? Yes [ ] No [ ] 53. At what height have you capped your coffee trees? ………………………………………………………………………………….… 54. How many heads per stem? ……………………………………….……………… 55. What is the labour requirement for pruning and desuckering your coffee? Man days ……………………………… Cost per man day…………………….

79