BAHIR DAR UNIVERSITY ADOPTION OF IMPROVED POTATO VARIETIES AND ITS IMPACT ON HOUSEHOLD NUTRITION: EVIDENCE FROM EMBA ALAJE WOREDA, NORTHERN ETHIOPIA. M.Sc Thesis By MOHAMMED EBRAHIM JUNE, 2019 COLLEGE OF AGRICULTURE AND ENVIRNOMENTAL SCIENCE BAHIR DAR UNIVERSITY ADOPTION OF IMPROVED POTATO VARIETIES AND ITS IMPACT ON HOUSEHOLD NUTRITION: EVIDENCE FROM EMBA ALAJE WOREDA, NORTHERN ETHIOPIA. PRINCIPAL ADVISOR: ZEWDU BERHANIE (Ph.D.) CO- ADVISORS: ERMIAS TESFAYE (Ph.D.) AND FITSUM HAGOS (Ph.D.) ATHESIS SUBMITTED TO THE COLLEGE OF AGRICULTURE AND ENVIRONMENTAL SCIENCES BAHIR DAR UNIVERSITY IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR DEGREE OF MASTER OF SCIENCE IN AGRICULTURAL ECONOMICS BY MOHAMMED EBRAHIM JUNE, 2019 BAHIR DAR UNIVERSITY THESIS APPROVAL SHEET As member of the Board of Examiners of the Master of Sciences (M.Sc.) thesis open defense examination, we have read and evaluated this thesis prepared by Mr Mohammed Ebrahim entitled “Adoption of Improved Potato Varieties and Its Impact on Household Nutrition: Evidence from Emba Alaje woreda, Northern Ethiopia”. We hereby certify that; the thesis is accepted for fulfilling the requirements for the award of the degree of Master of Sciences (M.Sc.) in Agricultural Economics. Board of Examiners ____________________ ______________ ______________ Name of External Examiner Signature Date ____________________ ________________ ______________ Name of Internal Examiner Signature Date _____________________ _____________ _______________ Name of chairman signature DECLARATION This is to certify that this thesis entitled with “Adoption of Improved Potato Varieties and Its Impact on Household Nutrition: Evidence from Emba Alaje woreda, Northern Ethiopia”. Submitted in partial fulfillment of the requirements for the award of Master of Science in agricultural economics to the Graduate Program College of Agriculture and Environmental Sciences, Bahir Dar University by Mr. Mohammed Ebrahim who is an authentic work carried out by himself under our guidance. The matter embodied in this project work has not been submitted earlier for award of any degree or diploma to the best of our knowledge and belief. _______________________ _________________ _________________ Name of the Student Signature Date Name of the Supervisors: 1._____________________ __________________ __________________ Main Supervisor Signature Date 2. _____________________ __________________ __________________ Co-Supervisor Signature Date 3. _____________________ __________________ __________________ Co-Supervisor Signature Date AKNOWLEDGMENT First, I would like to express my deepest gratitude to my advisors Dr. Zewdu Berhanie, Dr. Fitsum Hagos and Dr. Ermias Tesfaye for their highly valuable comments, timely response, open face and continuous intellectual guidance for successful accomplishment of this paper. Secondly, I would like to express my greatest thanks to the International Livestock Research Institute (ILRI). This research was undertaken with support from Africa RISING, a program financed by the United States Agency for International Development (USAID) as part of the United States Government’s Feed the Future Initiative. The content is solely the responsibility of the author/s and does not necessarily represent the official views of USAID or the U.S. Government or that of the Africa RISING program. Africa RISING is aligned with research programs of the CGIAR. Lastly, my special thanks goes to Kagnew Kassahun (Africa RISING driver), for his kind assistance by implementing different field assignment and errands while I was attending the class, transporting enumerators and facilitating the data collection work. Above all, it is a gift of Allah, the cherisher and the sustainer of the world, who bestowed on me the divine guidance, enough courage, and patience to complete my thesis work in difficult conditions and workload. Had it not been for His gracious assistance, it would have been impossible to overcome the challenges of the MSc rigorous work, and without His willingness and mercy I would have not reached this stage, great thanks be to Him forever. i ABSTRACT High population pressure and continuous decrease of the land holding size results in increase of food insecurity. To meet the increasing food demand of the growing population there is a need to intensify production practices. In this regard, improved potato varieties production plays a great role in improving the household’s food security, food consumption, and food diversity, and there-by contributing to nutrition security. This study analyzes the probability and use intensity of improved potato varieties adoption and, the effect of adoption on households’ nutrition security. The data was collected in 2018 at Emba Alaje woreda from a survey of 370 households (185 improved potato variety growers and 185 non-growers). Sampling weights were used to account the proportion of the sample compared to the whole population. Tobit model was used to analyze the factors affecting the probability and use intensity of improved potato varieties adoption. Both propensity score matching (PSM) and endogenous switching regression model (ESRM) were used to analyze the impacts of improved potato varieties on households’ nutrition using Food security scale, Food consumption score and Household dietary diversity score proxy variables. To control the possible selection problem invers mill’s ratio was included in the second stage equations. Size of own land, distance of the nearest plot, access to extension services, the existence of neighbor adopter, perception on the improved potato varieties’ maturity period and tuber yield potential were found as the main factors of adoption probability and use intensity of improved potato varieties. The PSM result indicated that, adoption of improved potato varieties increases the average food security scale, food consumption score and the dietary diversity score by 1.79, 6.6 and 0.8 points, respectively. Similarly, the ESRM result confirmed that, improved potato varieties adoption increases the average food security scale, food consumption and dietary diversity score by 2, 6.1 and 1.4 points, respectively. Thus, to improve the nutritional status of the farming households, government should give due emphasis for potato production and the extension service need to be strengthen. Keywords: Improved potato variety, Adoption, Nutrition security, Endogenous switching regression model, Propensity score matching, Tobit model, Northern Ethiopia. ii Table of Contents ABSTRACT ................................................................................................................ i LIST OF TABLES ..................................................................................................... v LIST OF FIGURES .................................................................................................. vi APPENDIXS ........................................................................................................... vii LIST OF ABBREVIATION ................................................................................... viii 1. INTRODUCTION ............................................................................................. 1 1.1. Background and Justification ............................................................................................................. 1 1.2. Statement of the problem ................................................................................................................. 4 1.3. Objectives of the Study ................................................................................................................. 5 1.3.1 General objective ............................................................................................................................. 5 1.3.2 Specific objectives ............................................................................................................................ 6 1.4 Research Questions ............................................................................................................................. 6 1.5. Significance of the Study .................................................................................................................... 6 1.6. Scope and Limitation of the Study..................................................................................................... 6 2. REVIEW OF LITERATURE ............................................................................. 8 2.1. The Concept of Adoption ................................................................................................................... 8 2.2. The Concepts of Food and Nutrition Security ................................................................................... 9 2.2.1 Food Security Situation in Ethiopia ................................................................................................. 9 2.2.2. Nutrition Security Situation in Ethiopia ....................................................................................... 10 2.3. Dietary Diversity ............................................................................................................................... 11 2.3.1. The Dietary Importance of Potatoes ............................................................................................ 11 2.4. Theoretical Models .......................................................................................................................... 12 2.4.1. Intensity of Improved Potato Varieties Adoption ....................................................................... 12 2.5.2. Impact Evaluation ......................................................................................................................... 13 2.6. Empirical Findings of Adoption ........................................................................................................ 15 2.7. Conceptual Framework ................................................................................................................... 19 3. METHODOLOGY OF THE STUDY .............................................................. 21 3.1. Description of the Study Area .......................................................................................................... 21 3.2. Sampling Techniques and Research Design .................................................................................... 23 3.2 Types and Sources of Data ................................................................................................................ 24 iii 3.3. Methods of Data Analysis ................................................................................................................ 24 3.3.1. Descriptive Statistics ..................................................................................................................... 25 3.3.1.1. Household Dietary Diversity Score (HDDS) ............................................................................... 25 3.3.1.2. Food Consumption Score (FCS) .................................................................................................. 26 3.3.1.3. Food Security Scale (FSS) ........................................................................................................... 26 3.3.2. Specification of Econometrics Model ........................................................................................... 27 3.3.2.1. Tobit Model ................................................................................................................................ 27 3.3.2.2. Propensity Score Matching Model (PSM) ................................................................................. 28 (i) Generating propensity scores p(x) ................................................................................................. 29 (ii) Choose a Matching Algorithm .................................................................................................... 30 (iii) Estimate the average treatment effect (ATT) ............................................................................ 32 (iv) Sensitivity Analysis ...................................................................................................................... 33 3.3.2.3. Endogenous Switching Regression Models (ESRM) .................................................................. 34 3.3.2.3.1. Switching Regression Model Specification ............................................................................ 35 3.4. Definition of Variables and Hypothesized Relationships ............................................................... 38 3.4.1. Dependent Variables .................................................................................................................... 38 3.4.2. Outcome Variables ........................................................................................................................ 39 3.4.3. Independent Variables .................................................................................................................. 39 4. RESULT AND DISCUSSION .......................................................................... 44 4.1. Descriptive Statistics of Sample Households’ Characteristics ........................................................ 44 4.2. Econometric Analysis ....................................................................................................................... 48 4.2.1. Factor Affecting Adoption of Improved Potato Varieties ............................................................ 48 4.2.2. Effects of Changes in Explanatory Variables ................................................................................ 51 4.2.3. Estimation of Propensity Scores ................................................................................................... 54 4.2.4 Matching Improve Potato Adopter with Non-Adopter Households ............................................ 56 4.2.5. Average Treatment Effect on the Treated (ATT) .......................................................................... 60 4.2.7. Sensitivity analysis of ATT estimation .......................................................................................... 61 4.2.7. Endogenous Switching Regression Model Result ........................................................................ 62 4.2.7.1. Nutritional Impacts of Improved Potato Varieties Adoption ................................................... 65 5. SUMMARY, CONCLUSION AND RECOMMENDATIONS ..................... 69 5.1 Summary and Conclusion .................................................................................................................. 69 iv 5.2. Recommendation ............................................................................................................................. 71 6. REFERENCE ..................................................................................................... 73 7. APPENDIX ................................................................................................... 85 BIOGRAPHY ....................................................................................................... 105 v LIST OF TABLES List of Tables………………………………………………………………………………page Table 1: Distribution of Sample households by Kebele ................................................................ 24 Table 2. Summary of independent variables and their expected signs .......................................... 43 Table 3: Descriptive statistics of sample households (continuous variable)................................. 45 Table 4: Descriptive statistics of sample households (dummy variables) ..................................... 47 Table 5: welfare outcome and categories of sample farmers ......................................................... 48 Table 6: Estimated result using Tobit model ................................................................................. 51 Table 7:change in probabilities of adoption and intensity of use due to change in explanatory variables ......................................................................................................................................... 53 Table 8:Variables used for PS generation and logit result of households IPV production participation ................................................................................................................................... 55 Table 9: Distribution of estimated propensity score ...................................................................... 56 Table 10: Performance of different matching estimator ................................................................ 58 Table 11: Propensity score and covariance balancing test ............................................................. 59 Table 12: Chi-square test for the joint significance of variables ................................................... 60 Table 13: ATT estimation result of households’ food security scale, dietary diversity score and food consumption score ................................................................................................................. 61 Table 14: Result of sensitivity analysis using Rosenbaum approach ............................................ 62 Table 15: Determinants of nutrition security status (second stage) ............................................... 65 Table 16: Expected conditional and average treatment effect of IPV on dietary diversity, food consumption, and food security of the household. ........................................................................ 68 vi LIST OF FIGURES List of Figures………………………………………………………………………. …...Page Figure 1: conceptual framework of the study ................................................................................ 20 Figure 2: Location of the study area. ............................................................................................. 22 Figure 3: Kernel density of propensity score distribution before matching .................................. 56 Figure 4: Kernel density of propensity scores of treated households ............................................ 57 Figure 5: Kernel density of propensity scores of control households. ........................................... 57 vii APPENDIXS Appendix 1: Scale Values and Food Status Categories for The Core Scale .................................. 85 Appendix 2: Correspondence Between Scale Values and Food Security Status ........................... 86 Appendix 3: Conversion Factors Used to Estimate Tropical Livestock Unit (TLU) .................... 86 viii LIST OF ABBREVIATION ADB African Development Bank ATE Average Treatment Effect ATT Average Treatment Effect of the Treated ATU Average Treatment Effect of Untreated CAADP Comprehensive Africa Agriculture Development Programme CFS Committee on World Food Security CIP International Potato Center CSA Central Statistical Agency EDHS Ethiopian Demographic and Health Survey ESR Endogenous Switch regression Model FAO Food and Agricultural Organization FCS Food Consumption Score FSS Food Security Scale FGD Focused Group Discussion GDP Gross Domestic Product GRAD Graduation with Resilience to Achieve Sustainable Development Ha Hectare HDDS Household Dietary Diversity Score IFPRI International Food Policy Research Institute ILRI International Livestock Research Institute ix LIST OF ABBREVIATION IPV Improved Potato Varieties KM Kilometer NGO Nongovernmental Organization PSM Propensity Score Matching Method PSNP Productive Safety Net Program RISING Research in Sustainable Intensification for the Next Generation SPSS Statistical Package for Social Science VAM Vulnerability Analysis and Mapping 1 1. INTRODUCTION 1.1. Background and Justification The cultivated potato (Solanum tuberosum L.) originated in South America where it has been used for food for over 10,000 years (CDC, 1999a) and globally, potato is a crop of world’s major economic importance and number one non-grain food commodity (Rykaczewska, 2013). It is the third most important food crop in terms of consumption after rice and wheat (Hielke et al., 2011; Birch et al., 2012; Hancock et al., 2014). Potato cultivation is exceeding 18.6 million hectares of land in more than 157 countries in the world with an estimated annual production of 330 million tons (Singh, 2008; FAO, 2009, 2010). According to FAO (2008) potato is a good source of income, and employment opportunity in developing countries and it is a good source of dietary energy and some micronutrients, and its protein content is very high in comparison with other roots and tubers. Due to its correct balance between protein and calories, it is considered as a good weaning food and these traits make it an efficient crop in combating world hunger and malnutrition (Berga et al., 1994). The commercial value of potatoes has increased considerably when it is processed into edible products (Kirkman, 2007). Potato consumption has increased in the developing world, and over the last decade world potato production has increased at an annual average rate of 4.5 percent Furthermore; Kirkman (2007) has estimated that global consumption in its processed form will increase from 13% of total food use in 2002 to nearly 18% by 2020. According to Mazengia (2016) potato has multiple benefits for low-income households where the land shortage is a constraint. It grows quickly, has a high yield, and contains more energy and protein per unit area when compared to cereal crops. Therefore, Potato can provide nutritious food for the poor and hungry in the developing countries and it is the most important crop to address food and nutrition security (Hussain, 2016), which is a major concern for countries like Ethiopia. High potato yield at critical food shortage periods could help households to get cash from the sale of potato and spend on a diversity of food types Mugisha et.al. (2017). Thus, potato production could increase the food availability and diversity. According to (FAO, 2008), potato has been declared as a Future Food crop and the 2 United Nation during the international potato year (2008), call the crop a “hidden treasure” (Hussain, 2016). In Ethiopia, potato has been cultivated for over 150 years (Kolech et al., 2015); currently it is grown in many parts of the country. The production area has reached 59,504 ha cultivated by over one million households in the main cropping season of 2011 (CSA, 2012). There is a high potential to expand the cultivation area of the potato crop, as most arable land is in principle suitable for cropping potato. Joshi, et al. (2009) indicated that the potential yield of potato in Ethiopia can reach up to 50 t/ha, but the average national potato production is 10.5 tones/ha, while progressive farmers who use improved agronomic practice attained yields of 25 tones/ha. According to CIP and ILRI (2016), improved potato varieties and with improved agronomic practice provided better tuber yields; Belete, Jalene, Gudene, and Gera varieties provide 46.93, 40.01, 38.93, and 32.98 tones of tuber yield per hectare respectively while the local variety provides only 14.4 t/ha. The results of the partial budget analysis also revealed that the use of improved potato varieties with its packages resulted in the net benefit of 122,535 birr/ha compared to the use of local variety and practice 43,920 birr/ha. The production problems that account for low yields and tuber quality are similar to the problems that were identified in many developing countries including Ethiopia. Limited supply of high-quality seed tubers (Gildemacher et al., 2009), inappropriate agronomic practices and inadequate storage (Tekalign, 2005), poor knowledge of seed tubers selection (Adane et al,.2010) are reported as much contributing factors to the low yields and poor- quality seed tubers production. According to Berga and Woldegiorgis (1994), one of the major factors attributed to the low productivity of potato is limited access to improved varieties. The main constraints to access improved varieties are shortage of improved and quality seed, damaged and spoiled seed due to poor transporting and handling (Emana & Ngussie, 2011). The available set of local varieties has been developed through a constant process of farmer experimentation, evaluation and selection of introduced varieties or clones from outside 3 sources. Varieties previously selected by farmers referred to as “local varieties”; while varieties developed by the research system over the past 28 years since their first release in 1987 referred to as “new varieties” (Kolech et al., 2015). Improved potato varieties are obtained through breeding process for disease resistance, drought tolerance, good yield, attractive color, size and shape, good cooking quality and other desirable characteristics (Asakaviciut et al., 2008). In Tigray, the production of potato as food security crop and source of income has long a history. Starting 2013 the potato tuber seed multiplication and demonstration of new varieties both at seed producer cooperatives and model farmers’ field was started in southern Tigray, particularly in Emba Alaje, Ofla and Endamehoni woreda supported by Government organization (Office of Agriculture) and different non-governmental organizations (NGOs) such as, Africa RISING, CIP (International Potato Center), and Graduation with Resilience to Achieve Sustainable Development (GRAD). However, the seed tubers and varieties promoted and delivered by these organizations were not sufficient for the areas due to the high increase of demand from farmers (Getachew, 2016). Farmers in Emba Alaje wereda produced Gudene, Jalene and Belete potato varieties. These varieties provide different yields, varying from farmers to farmers. This is mainly because of the differences in adoption of potato production technology packages; recommended seed rate and seed size, spacing between plant and row, fertilizer rate, chemical application to control fungal disease and storage facilities (Getachew, 2016). Therefore, raising the efficiency among the growers is essential element for getting the desired return from the potato cultivation. In Ethiopia few studies are conducted in SNNP, Oromia and Amhara region on adoption of potato varieties; ketema et al. (2016), Tesfay et al. (2006), and Endris (2003) studied the adoption rate and intensity of potato varieties, and other socio economics characteristics of farmers that affect the adoption of potato varieties and production packages. However, these studies did not address the impact of adoption of potato varieties on households’ food security and nutrition. In Tigray region there was no adoption study carried out on potato and related production technologies. The adoption and intensity of use of improved potato varieties and 4 its impact on household nutrition were not analyzed well. Hence, this study investigated the demographic and socio-economic characteristics of farmers that influence the probability and level of adoption of improved potato varieties, and assess the impact of adoption of potatoes varieties on farm household nutrition in through household dietary diversity score, food consumption score and food security scale. Using this multiple food and nutrition security indicators, this study measures the quality, diversity and quantity aspects of households’ food consumption. 1.2. Statement of the problem Increasing population pressure, soil nutrient loss, land degradation, and shrinking land holdings necessitated intensification of production practices to meet the increasing food demand of the population. The agricultural sector suffers from poor cultivation practices and frequent drought. CAADP (2016) indicated that the Joint efforts by the Government of Ethiopia and donors have strengthened Ethiopia's agricultural resilience, contributing to a reduction in the number of Ethiopians threatened with starvation. However, the number of food insecure people in the country increases from time to time; estimated to 2.9 million in 2014 and 4.5 million in August, 2015 and by the end of the same year this figure had more than doubled to 10.2 million. Consequently, 27 million Ethiopian became food insecure as a result of 2015 El Niño drought and 18.1 million dependents on relief food assistance in 2016 (Abdulselam, 2017). Under nutrition has long history and remains one of the major and most pressing health problem in Ethiopia. Nearly 40% of the rural farm family cultivate land less than half a hectare from where they produce only half of their annual food demand. Moreover, these farm family do not have enough purchasing power to buy from the market and children who have come from such a family member are almost malnourished (CAADP, 2016). High yield potential, early mature, and drought and disease resistance improved crop varieties play a vital role to increase the food crop production in the changing environment. One of the greatest advantages of potato production is high productivity per unit of area. Potato is one of the most productive food crops in terms of yields of edible energy and good quality protein 5 per unit area and per unit of time fitting into, intensive cropping systems (Rizov et al., 2018). According to Joshi et al. (2009) Potato can yield maximum tuber yield 50 t/ha and compared to cereals it is short duration crop which can be harvested from 3-4 months (Endale et al., 2008b). In Emba Alaje potato reach for harvest at critical food shortage period, when the other food crops are finished from the storage and not matured in the field, usually at the end of September and October. Emba Alaje is one the chronically food insecure woreda targeted by the Productive Safety Net Program (PSNP). The program started in 2005 with 23780 beneficiaries and currently number of PSNP beneficiaries raised to 30498. The figure shows an increase of 6718 beneficiaries. Thus, potato is a very important crop for the study areas where population pressure, food and nutrition insecurity are increasing. In addition, the study area has good climatic and edaphic conditions for higher potato production and productivity. According to MOA (2012), in Ethiopia 29 improved potato varieties have released to enhance its productivity. However, in the highlands of Southern Tigray Zone, particularly in Emba Alje woreda only four improved potato varieties were introduced and promoted by government and non-governmental institutions in limited number of kebeles kebeles. Even though potato is a good pathway for enhancing food and nutrition security Mugisha et.al. (2017), in the study areas the adoption of the improved varieties is very low. No in-depth studies have been carried out on the factors and level of adoption of potato varieties and its impact on households’ nutrition. Therefore, this study was to generate evidence on the major factor of the adoption and level of adoption of improved potato varieties, and its impact on households’ nutrition. 1.3. Objectives of the Study 1.3.1 General objective The overall objective of this study is to analyze the adoption of improved potato varieties and its impacts on smallholder farmers’ nutrition in the study area. 6 1.3.2 Specific objectives The specific objectives of the study are: i. To identify socio-economic determinant factors that influence the probability and level of adoption of improved potato varieties ii. To analyze the impact of improved potato varieties on households’ nutrition by using food security scale (FSS), food consumption score (FCS), household diet diversity score (HDDS). 1.4 Research Questions i. What are the socio-economic determinant factors that influence the probability and level of adoption of improved potato varieties by smallholder farmers? ii. What are the impacts of improved potato varieties adoption on households’ nutrition? 1.5. Significance of the Study The findings of this study are expected to provide a comprehensive understanding on the level of adoption of improved potato varieties and farmers characteristics which determine the adoption of potato varieties and its’ impacts on the farming household nutrition. The study findings could be used for policymaker, agricultural extension service providers, researchers, NGOs, farmers and potato seed producer cooperatives to design appropriate strategies and enhance the potential benefits from potato production and utilization. The results of this study would assist development activities underway and to be planned in the future in areas of potato. Extensions and other development practitioners would use the information to develop the appropriate extension agendas and to raise the awareness level of farmers on potato production and its role in food and nutrition security, employment and income generation. The results could be a basis for further investigation and setting of research agendas. 1.6. Scope and Limitation of the Study The study was carried out by surveying a sample of randomly selected farm households from two kebeles (the lowest administration unit) of Emba Alaje woreda where the improved 7 potato varieties have been widely produced. Since improved potato grower farmers were few in number as compared to the non-producer groups and to increase the share of producer farmers an equal number of samples were selected randomly from both groups. There are seasonal differences in dietary patterns and for a more complete assessment of usual diet, dietary diversity should be measured in different seasons. Due to time and budget constraints, the study was limited geographically to one woreda and collected cross-sectional data. However, the results of the study are applicable to other areas with similar physical and socioeconomic settings. 8 2. REVIEW OF LITERATURE In this chapter, the concept of adoption, food and nutrition security, dietary diversity and impact evaluation are explained. In addition, the past research works are critically reviewed in relevance to the present study objectives and the evidences from the reviews enables better understanding of the subject. 2.1. The Concept of Adoption Adoption is defined as a decision to apply an innovation and to continue to use it over a reasonably long period of time (Ban and Hawkins, 1996) and according to Feeder et al. (1985) defined adoption as the degree of use of a new technology in a long run equilibrium when a farmer has all of the information about the new technology and its potential. Ban and Hawkin, (1985) further noted that adoption is not a permanent behavior. An individual may decide to discontinue the use of innovation for a variety of personal, institutional, or social reasons, one of which might be the availability of an idea or practice that is better in satisfying his/her need. Therefore, adoption at the farm level describes the realization of a farmer’s decision to implement a new technology. If innovations are modified periodically, however, the equilibrium level of adoption will not be achieved. This situation requires the use of econometric procedures that can capture both the rate and the process of adoption. As the new technology is introduced, some farmers will experiment with it before adopting. The “rate of adoption” is defined as the proportion of farmers who have adopted a new technology at a specific point in time (e.g., the percentage of farmers using fertilizer). Nkony et al. (1997) also defined the rate of adoption as the level of adoption of a given technology. Furthermore, the “intensity of adoption” is defined as the level of adoption of a given technology, for example, by the number of hectares planted with improved seed or the amount of fertilizer applied per hectare (Degu, 2000). Markee (1992) defined adoption as the process of spreading of a new technology within a region, diffusion represents the cumulative process of adoption measured in successive times. Fliegel, (1984) noted that the rate of 9 diffusion depends, among other things, on extension communication and the extent to which farmers discuss agricultural issues among themselves routinely. According to Amit et al. (2017), the adoption process is the mental process through which an individual pass from first hearing of an innovation to its final adoption, while adoption is a decision to continue the full use of an innovation. Generally, the farmers do not adopt package of practices fully. There is only a partial adoption by them. As a result, the gap always appears between the recommended production technology and their use at the farmer’s field. 2.2. The Concepts of Food and Nutrition Security Food security is a concept that has evolved considerably over time and its definitions developed and diversified by different researchers, scholars and organizations (Abdulselam, 2017). Food security is a situation that achieved at the individual, household, national, regional and global levels when all people, at all times, have physical and economic access to sufficient, safe and nutritious food that meets their dietary needs and food preferences for an active and healthy life (FAO, 2008). Hussein (2013) defined food security as adequate availability of and access to food for households to meet the minimum energy requirements as recommended for an active and healthy life. Micronutrient deficiencies in diets are widespread and have long-term consequences, reflected in a wide range of health outcomes, including stunting, cognitive abilities and non- communicable diseases. Yet, unlike insufficient energy intakes, which translate quickly into sensations of weakness and hunger, these deficiencies are not immediately apparent and are therefore often referred to as ‘hidden’ hunger (Kennedy et al., 2003). For the most food- insecure households, some of the most widespread deficiencies involve inadequate levels of vitamin A, iron and zinc, but many important micro- and macro-nutrients may be insufficiently (Lele et al., 2016). 2.2.1 Food Security Situation in Ethiopia Ethiopia is facing a massive drought and food insecurity crisis over the years. According to ADB (2014), Ethiopia is one of the most food-insecure and famine affected countries. Drought, recurring food shortage and famine are great challenges faced by Ethiopian people. 10 A large portion of the country’s population has been affected by chronic and transitory food insecurity. According to Care Ethiopia (2014) findings chronic and acute food insecurity are prevalent, especially among rural populations and smallholder farmers. The findings indicated that about 10 percent of Ethiopia’s citizens are chronically food insecure, and this figure rises to more than 15 percent during frequent drought years. The El Nino -driven drought has greatly expanded food insecurity and malnutrition, and devastated livelihoods of the poorest and vulnerable people across the country (FAO, 2016). Food Security and Hunger/ Undernourishment Multiple Indicator Scorecard indicated that, Ethiopia ranked as first in having the highest number of people in state of undernourishment/ hunger which is 32.1 million people. This makes it, the fourth African country scoring (37.1%) of the population being undernourished/ in hunger. According to VAM (2008) there is no single way to measure food security, the concept itself being rather elusive. Food consumption measured in kilocalories is the gold standard for measuring consumption, and often considered to be one of the gold standards for food security. The food consumption score (FCS) as score which able to capture both Dietary Diversity and Food frequency. 2.2.2. Nutrition Security Situation in Ethiopia The term of nutrition security emerged with the recognition of the necessity to include nutritional aspects into food security. Nutrition security as a condition when all people at all times consume food of sufficient quantity and quality in terms of variety, diversity, nutrient content and safety to meet their dietary needs and food preferences for an active and healthy life, coupled with a sanitary environment, adequate health and care (Belton and Thilsted, 2014) Ethiopia has shown some progress in reducing under nutrition in recent years. The EDHS (2011) indicated that nationally, 44% of children under the age of five are found to be stunted, 33% are underweight, and 12% are wasted (measures the more immediate effect of malnutrition). However, it is still a major public health problem and remains a serious concern and a drawback to its rapid economic development. According CAADP (2013), to under nutrition has long history and remains one of the major and most pressing health 11 problems in Ethiopia. Chronic under nutrition as measured by stunting and underweight, anemia, iodine, zinc and vitamin A deficiency indicates major nutritional problems of Ethiopia. The childhood deaths associated with malnutrition reaches 57%. 2.3. Dietary Diversity According to IFPRI (2002), dietary diversity is the number of different foods or food groups consumed over a given reference period. Dietary diversity is a qualitative measure of food consumption that reflects household access to a variety of foods and is also a proxy for nutrient adequacy of the diet of individuals (FAO, 2010). The rationale for emphasizing dietary diversity in developing countries stems mainly from a concern related to nutrient deficiency and the recognition of the importance of increasing food and food group variety to ensure nutrient adequacy. Lack of dietary diversity is a particularly severe problem among poor populations in the developing world, because their diets are predominantly based on starchy staples and often include little or no animal products and few fresh fruits and vegetables (IFPRI, 2002). 2.3.1. The Dietary Importance of Potatoes Potatoes are prepared by consumers in variety of means. Potatoes are usually eaten cooked, and most often eaten boiled and unpeeled in many regions of the world. Baking, boiling, dehydrating, and frying are employed world-wide. According to Englyst et al. (1992) cooking or processing of potatoes greatly improves the digestibility of potato starch, which has very low digestibility in the raw state since potato starch granules have a β-crystalline structure that is resistant to amylase digestion. Unpeeled potatoes that undergo cooking have better nutrient retention than do peeled potatoes and size reduction brings about further losses. Mary et al. (2009) indicated that boiling cut or peeled potatoes leads to loss of water-soluble vitamins and minerals due to their leaching out into the cooking water but baking, roasting, and frying generally result in lower losses of vitamins than boiling. On the other side baking cause’s slightly higher losses of vitamin C than boiling, due to the higher oven temperatures, but losses of other vitamins and minerals during baking are lower (FAO, 2008). 12 According to FAO (2008) Potatoes, nutrition and diet report freshly harvested potato contains about 80 percent water and 20 percent dry matter. About 60 to 80 percent of the dry matter is starch. On a dry weight basis, the protein content of potato is similar to that of cereals and is very high in comparison with other roots and tubers. In addition, the potato is low in fat content. Potatoes are rich in several micronutrients, especially vitamin C – eaten with its skin, a single medium sized potato of 150 gram provides nearly half the daily adult requirement (100 mg). The potato is a moderate source of iron, and its high vitamin C content promotes iron absorption. It is a good source of vitamins B1, B3 and B6 and minerals such as potassium, phosphorus and magnesium, and contains folate, pantothenic acid and riboflavin. Potatoes also contain dietary antioxidants, which may play a part in preventing diseases related to ageing, and dietary fiber, which benefits health. In order to keep glycol-alkaloid content low and to insure healthy eating, potatoes should be stored in a dark and cool place. Under exposure to light potatoes turn green in color due to increased levels of chlorophyll. Since glycol-alkaloids are not destroyed by cooking, cutting away green areas and peeling potatoes before cooking ensures healthy eating (FAO, 2008). 2.4. Theoretical Models 2.4.1. Intensity of Improved Potato Varieties Adoption Limited dependent variable model provides a good framework to study adoption behavior of farmers. The most commonly used qualitative models to study the adoption behavior are the logit and the probit models (Feder et al., 1985). These models specify a functional relationship between the probability of adoption and various explanatory variables (Bekele et al., 2000). However, this approach does not capture intensity of adoption. The tobit model overcome this problem by measuring both adoption and intensity (Mazvimavi &Twomlow, 2009). Intensity of adoption of potato varieties is the average size of land occupied by improved varieties, whereas the adoption rate of improved potato varieties is the percentage of farmers growing potatoes varieties (Tesfay et al., 2006). Logit model: Logistic regression sometimes called the logistic model or logit model, analyzes the relationship between multiple independent variables and a categorical dependent 13 variable, and estimates the probability of occurrence of an event by fitting data to a logistic curve. There are two models of logistic regression, binary logistic regression and multinomial logistic regression. Binary logistic regression is typically used when the dependent variable is dichotomous and the independent variables are either continuous or categorical. Tobit model: Tobit is appropriate model to deal with such censored data and used to analyze the intensity of use of improved varieties in preference to multiple regression model, when significant number of observations on dependent variable having a value zero (Endris, 2003). Area planted with improved variety of potato represents a censored distribution since half of the sample farmers assume a value of zero for not adopting (non-users). Accordingly, there is a cluster of households with zero adoption of the improved technology at the limit. The application of Tobit analysis is preferred in such cases since it employs both data at the limit as well as those above the limit (Gairhe et al., 2017). 2.5.2. Impact Evaluation According to Amare et al. (2012) estimation of the impact of technology adoption on household welfare outcome variables based on non-experimental observations is not trivial because of the need of finding on counterfactual of intervention. The observed one is the outcome variable for adopters, in the case that they did not adopt. That is, we do not observe the outcome variables of households that adopt, had they not adopted (or the converse). Improved varieties are not randomly distributed to the two groups of households (as adopters and non-adopters), but rather to the household itself deciding to adopt given the information it has, therefore the two group may be systematically different. Estimation of impact of improved potato varieties adoption on farm household nutrition based on non-experimental observations is significant because of the need of finding counterfactual of intervention. To address this missing counterfactual Khonje et al. (2015) use propensity score matching (PSM), endogenous switching regression (ESR), and inverse probability weighting (IPW) models used for impact analysis. 14 Propensity score matching: is an alternative method to estimate the effect of receiving treatment when random assignment of treatments to subjects is not feasible (Rizov et al., 2008). Propensity score matching is the most widely used type of matching, in which the comparison group is matched to the treatment group on the basis of a set of observed characteristics or by using the “propensity score” (predicted probability of participation given observed characteristics); the closer the propensity score, the better the match. The method tries to pick an ideal comparison that matches the treatment group from a larger survey. A good comparison group comes from the same economic environment (Baker, 2000). PSM tries to create the observational analogue of an experiment in which everyone has the same probability of participation. The difference is that in PSM it is the conditional probability (P(Z)) that is intended to be uniform between participants and matched comparators, while randomization assures that the participant and comparison groups are identical in terms of the distribution of all characteristics whether observed or not. PSM does not require a parametric model linking outcomes to program participation. Thus, PSM allows estimation of mean impacts without arbitrary assumptions about functional forms and error distributions (Ravallion, 2005). Endogenous switching regression model: Endogenous switching regression model use when both observable and unobservable characteristics are accounted for, thus controlling for a 'hidden bias' which can arise when unobservable variables are not taken into account. Ignoring the endogeneity of adoption of improved potato varieties would result in biased estimated parameters (Wabwile et al., 2016). According to Khonje et al. (2015) the average treatment effect on the treated (ATT) computes the average difference in outcomes of adopters category with and without a technology. Most commonly used methods to calculate ATT such as PSM ignore unobservable factors that affect the adoption process, and also assumes the return (coefficient) to characteristics to be same for adopters and non-adopters. The differences in welfare outcome variables between those farm households that did and those that did not adopt improved technology could be due to unobserved heterogeneity. Not distinguishing between the casual effect of technology adoption and the effect of unobserved heterogeneity could. 15 Endogeneity of the adoption decision could account (that is, for the heterogeneity in the decision to adopt or not to adopt new technology and for unobservable characteristics of farmers and their farm) by estimating a simultaneous equations model with endogenous switching by full information maximum likelihood estimation (Solomon et al., 2010). The ESR framework proceeds in two stages: the first stage is the decision to adopt improved varieties, and this is estimated using a probit model; in the second stage an Ordinary Least Squares (OLS) regression with selectivity correction is used to examine the relationship between the outcome variable and a set of explanatory variables conditional on the adoption decision (Khonje et al., 2015). 2.6. Empirical Findings of Adoption The adoption process is conceptualized to include several mental stages through which an individual pass after first hearing about an innovation and finally deciding to accept or reject it. This process generally includes five stages: awareness, interest, evaluation, trial, and adoption (Endris, 2003). Feder et al. (1985) noted that farmers are classified according to their tendency to adopt an innovation as innovators, early adopters, followers, and laggards. As noted by Feder et al. (1985), a complete analytical framework for investigating adoption process at the farm level should include farmer's decision making model determining the extent and intensity of use of a new technology at each point throughout the adoption process and a set of equations of motion describing the time pattern of parameters which affect the decision made by the farmer. Ban and Hawkins (1996) indicated that people who are quick to adopt an innovation may be characterized by having many contacts with extension agents, active participation in many organizations, being well educated, and having a relatively high level of income and standard of living. The intensity of adoption of new technologies that are divisible (such as high yielding varieties or new variable inputs) can be measured at the individual farm level in a given time period by the amount or share of farm area utilizing the technology (Feder et al., 1985). According to Patel et al. (2012), farmers do not adopt a package of practices fully. There is only a partial adoption by them. As a result, the gap always appears between the 16 recommended production technology and their use at the farmer’s field. Endris (2003) on his Adoption of Improved Sweet Potato Varieties revealed that; experience, the value of livestock, farmer's perception of yield, earliness in maturity and establishment performance of the improved varieties and extension contact positively influenced the probability of adoption and intensity of use of improved sweet potato varieties. On the other hand, marginal changes in farm size and distance from the research center to the farm negatively influenced the adoption and use intensity of improved sweet potato varieties. For instance, if farming experience increased by one percent adoption and intensity of use of improved sweet potato varieties would increase by about 0.009% of which 0.005% is attributed to the increase in the intensity of use of improved varieties by those farmers already adopted new varieties. One percent increase in farm size decreases the probability of adoption and intensity of use of improved sweet potato varieties by 0.21% and 640.30%, respectively. The estimated increase in the probability of adoption and intensity of use of improved sweet potato varieties resulting from a one percent change in the value of livestock owned is 0.00004% and 0.00005%, respectively, which were very small as compared to the changes resulting from other significant variables. A change in the perception of the farmer on the yield of improved variety to be higher than that of local variety (i.e. a change from 0 to 1) brings about 0.20% increase in the probability of adoption and 0.29% increase in the intensity of use of improved sweet potato varieties. Farmer's perception of the earliness of the improved varieties increases the probability of adoption and intensity of use by about 0.095% and 0.14%, respectively. Establishment performance of improved variety being better than that of local variety brings about 0.07% increase in the probability of adoption and 0.10% increase in the intensity of use of improved sweet potato varieties. A marginal change in extension contact increases of the probability of adoption and intensity of use of improved sweet potato varieties by about 0.14% and 0.20%, respectively. 17 A one percent increase in the distance from research center to the farm reduces the probability of adoption and intensity of use of improved varieties by about 0.005% and 0.002%, respectively. Mwanga et al. (1998) in Tanzania has indicated that household size, farm size and education level significantly affected the adoption of improved wheat varieties. They indicated that the adoption of fertilizer was significantly affected by the number of livestock owned, farm size, extension contact, hired labor and credit availability. Other studies on the adoption of improved technologies at farm level have been conducted in Ethiopia by Legese (1992) indicated that profitability is a function of elements of agro-climatic and socioeconomic environments and these factors indirectly affect the adoption patterns. He has pointed out that the probability of adoption of improved varieties and intensity of adoption of fertilizer and herbicide was influenced by experience, credit, cash down payment, participation in farmer organization as a leader and close exposure to technology. Yohannes et al. (1990) reported that debt had a negative effect on the adoption of fertilizer and pesticides. Itana (1985) explained that distance to the extension center, education, farm size and adequacy of rainfall as major factors that affect the adoption of fertilizer and improved variety. Alene et al. (2000) confirmed the importance of farmers' access to resources, extension services, and the availability of improved seed. Creating more opportunities for off-farm employment and income will enhance the financial ability of smallholder farmers to acquire external inputs. The fact that extension services are making a difference, it follows that policy makers need to focus on targeting resource-poor farmers who represent the farming communities in many areas of the country. At the same time, the availability of improved seed proved to be a major constraint for adoption, a fact that calls for improvements in improved seed delivery to effectively cope with the demands of small farmers. The results of research by Legesse et al. (2001) showed that it is structural factors, in particular, oxen ownership and distance to market, that determining the adoption and intensity 18 of use of technologies compared to personal characteristics, extension activity, attitudes to prices or risk. Farm size and farmers’ perception of input prices were found to be significant with positive and negative effects, respectively, but these effects were not particularly robust across technology or crop mixes nor across model they specified. Farm income is another significant factor differentiating users from non-users and hence has implications for changing the existing input credit scheme (Alene, et al., 2000). These adoption studies undertaken in Ethiopia have extensively examined the factors that influenced adoption of improved technologies in few localities and most of them are centered on the adoption of new varieties of cereal crops, pesticides and fertilizer. No attempt has been made to study the adoption of root crops that feed many populations in Ethiopia (Endris, 2003). According to Ketema et al. (2016), access to irrigation, extension contact frequency, farm size, membership to cooperatives, and annual income were found significantly determining the adoption of the potato technology package. Farm size was hypothesized to positively influence the adoption of the potato technology package. However, the current result is against this expectation. The result shows that farm size was negatively affecting adoption of potato technology package. This could happen as the production of potato, unlike other crops, requires more intensive production managements that fit into smaller farms. This intensive management could in turn result in relatively higher productivity that further intensifies adoption of the package. Membership to cooperative institutions was found positively driving the adoption of the potato technology package. This could happen given the fact that cooperatives are among the strongest social institutions that play an important role in the adoption of technologies. Irrigation is an important factor that explains the production of potato. Farmers in the study area utilize irrigation for potato production and hence it enabled them to fetch a higher price 19 on the market. In line with this, farmers who used irrigation were found to be better adopters of potato technology package as compared to those who are not using irrigation. Alene et al. (2000) on his adoption of recommended potato production technology study indicated that education, experience in potato cultivation, Social participation, land holding, annual income, irrigation facilities, extension contact, mass media exposure, participation in extension activity, economic motivation, scientific orientation, risk orientation, and knowledge levels are important factors in determining the adoption level of potato production technologies. 2.7. Conceptual Framework This sub section clearly shows, the different factors that affect the adoption probabilities and use intensity of improved potato varieties, and the nutrition welfare outcomes. The adoption and use intensity of improved potato varieties determined by demographic characteristics (sex, age, family and size), socio-economic factors (education, cultivated land size, livestock holding, farmer experience, perception on IPV yield and maturity period) and institutional factors (access to extension service, plot distance, woreda market distance access to irrigation and availability of prior neighbor adopter). Similarly, the household’s nutrition security is the outcome of several interactions of farmer internal and external factors. These factors are, demographic characteristics (sex, age, family size and dependency ratio), socio-economic factors (education, cultivated land size, livestock holding, farmer experience, number of fertile plots, number of plots with recommended rate fertilizer and off-farm income ), institutional factors (woreda market distance) and adoption of improved potato varieties. Diagrammatic relationship of factor of IPV adoption, and nutrition outcomes are presented as follow. 20 Figure 1: conceptual framework of the study Source: Own design from literature reviewed, 2019. Nutrition security outcomes  Food security scale  Food consumption score Socio-economic factors  Education  Cultivated land  Livestock holding  Farmer experience  Perception on IPV maturity  Perception on IPV yield  Number of fertile plots  Number of plots with recommended rate fertilizer  Off-farm income Institutional factors  Access to extension advice  Plot distance  Woreda market distance  Availability of irrigation  Availability of neighbor adopter Demographic factors  Age of household heads  Sex of household heads  Family size  Dependency ratio Adoption & use intensity of IPVs 21 3. METHODOLOGY OF THE STUDY 3.1. Description of the Study Area This study was conducted at Emba Alaje woreda, one of the five districts of the southern zone of Tigray regional state. The woreda is selected relatively based on its widespread adoption of improved potato varieties. Emba Alaje is located about 90 km far from Mekelle, the capital city of Tigray Regional National State. Geographically, Emba Alaje Woreda is located 12o50’-13o 0’ N latitude and 39o15’-39o 40’30”E longitude (figure2). The woreda is bordered with Hintalo wajirat in the north, Endamehoni in the south, Raya Azebo in the south east, and Amhara region in the south-west. The Woreda covers a total land area of 1677 square kilometer (WARDO, 2009). Emba Alaje is among one of the highlands districts in Tigray having an average altitude of 2400 m.a.s.l. The Woreda is one of the densely populated areas and thus, small land-holding similar to most highlands of Ethiopia. According to the WARDO (2009) the area lies within three agroecological zones including highland (72%), mid-latitude (21%), and lowland (7%). The Woreda has bimodal rain fall pattern, summer is the main rainy season June to August (with its peak in July) and short rain season in from February to April. Moreover, rain falls has almost the same coverage in the districts sub districts with an average of 380 mm annually .The maximum temperature ranges from 24-degree cent grade to 36-degree cent grade while the minimum temperature ranges up to-6-degree cent grade on the peaks of EmbaAlage mountain (the second biggest mountain in Tigray with an altitude of 3956 m.a.sl. (Haylu, 2014). 22 Figure 2: Location of the study area. Source: (Authors compilation, 2019) From the CSA (2007) population data, the current populations number of Emba-Alage Woreda is projected around 145746, from these 71331 are males and 74414 of them are females. In addition to this, the report showed that, there are about 33427 households, with an average family size of 4.36. Furthermore, 98.18 % of the woreda population is Tigray ethnic group; 1.4 % constitutes by AgawKamyr ethnic group and other ethnic groups made up of 0.42 of the population. In addition to this, only 10.46 % of the populations are considered literate. Meanwhile, 99.68% of the population follows Ethiopian Orthodox Christianity as their religion. Agriculture is the most dominant means of livelihood of the population of the district. There are also a considerable number of people engaged in selling livestock, petty trading, livestock products and fuelwood selling. The main crops grown include are Wheat, Teff, faba bean, barley and potato where wheat is the dominant crop. The Meher cropping season begins late 23 June and continues up to end of December. Crops like wheat, Teff, Faba bean, Potato and arley planted in from June up to July and harvest up to end of December (Haylu, 2014). 3.2. Sampling Techniques and Research Design The survey was conducted on Emba Aleje Woreda, where potato varieties are relatively widely produced. Two Sample kebeles namely, Tekiha and Ayba were selected based on potato production potential and accessibilities of sub-kebeles. The number of improved potato varieties producer households are very few as compared to the non-producers (households who did not produce any types of potato varieties). The total number of households in Tekiha kebel are 1837, out of this only 182 households were only producing potato, whereas in Ayba Kebele the total number of households are 1835, but only 200 households were improved potato producer in 2018. Therefore, instead of following proportional sampling for each group, the researcher has found it more useful to take a sample size of 50% from Improve potato producer and 50% from non-producer, which was done to increase the share of improve potato producer in the sample for the analysis. Since the majority of potato producers use improved varieties, the control groups were selected from non-producer households. Farmers in each kebeles were further stratified in to two groups, potato producer and non-producer. For each stratum, the sampling frame was prepared, and sampling units selected randomly. During the analysis, the sampling proportion was corrected by applying corresponding sampling weights for all observations. The assigned weights are calculated for each observation based on the calculation of quotient of percentage in the population and percentage in the sample. Accordingly, the improved potato user received a weight of 0.198( భఴమ భఴయళ ௫ ଵ଴଴ ఱబ భబబ௫ଵ଴଴ ) and 0.218 ( మబబ భఴయఱ ௫ ଵ଴଴ ఱబ భబబ௫ଵ଴଴ ) in Tekiha and Ayba kebeles, respectively. Using similar procedure, the non-producers receive weights of 1.8 ( భలఱఱ భఴయళ ௫ ଵ଴଴ ఱబ భబబ௫ଵ଴଴ ) and 1.78 ( భలయఱ భఴయఱ ௫ ଵ଴଴ ఱబ భబబ௫ଵ଴ ) . Since STATA has an advantage of incorporating the weight option, the corresponding sampling weights have been included during the analysis of impact estimation. 24 Table 1: Distribution of Sample households by Kebele Sample kebele Participant households Non-Participants households Total Sample Total Sample Total Sample Tekiha 182 88 1655 94 182 Ayba 200 97 1635 91 188 Total 382 185 3290 185 370 3.2 Types and Sources of Data Primary data was the main source of data for this study. The required data was collected through farm household survey using structured questionnaire. The survey was conducted on May 2018 when most farmers had enough time for interview. Experienced enumerators were recruited and trained in the class on each parts and questions for common understanding of questions and ethical issues during before and interview. Structured questionnaire was prepared and pre-tested, and the necessary modifications were made before it was used for the actual survey. Trained enumerators under the supervision of the researcher interviewed the sample farmers. The supervisor was responsible for the spot data editing and crosschecking to control the data quality. The interview also was supplemented by key informant interview and focus group discussion to obtain in-depth information. In addition to the primary data, secondary data were collected from review of different document which include research works and reports from the woreda office of agriculture, GRAD and Africa RISING projects. 3.3. Methods of Data Analysis Descriptive and econometrics analysis were employed to analyze the collected data by using SPSS and STATA software. The most econometrics models commonly used in adoption and intensity of adoption are qualitative choice models including the linear probability function, logistic distribution function (logit), and normal distribution function (probit) and the Tobit model (Degu, 2000). In this study, Tobit model was applied to identify factors affecting the adoption and intensities of use of improved Potato varieties since advantage over other 25 adoption models in dealing with a dependent variable with censored distribution and generating information for both probabilities of adoption and intensity of use of the technology (Endris, 2003). Half of our sample households had zero value both in adoption probabilities and percentage of areas covered with improved potato varieties. To analyze the impact of improved potato varieties on households’ nutrition propensity score matching and switching regression model. Household dietary diversity score (HDDS) and food consumption score (FCS), food security scale (FSS) outcome variables were used to measure the impact on nutrition as nutrition cannot be measured directly. 3.3.1. Descriptive Statistics Descriptive statistics such as mean, percentage and standard deviation were used to characterize the farming system of the areas and analyze farmers' responses and their implications for adoption of improved potato varieties, proportion of households who consume a particular food group or nutrient-rich food, and the food consumption group (FCG). The frequency of DDS, FCS and FSS categories were also used to make comparisons between improved potato producer and no-producers. The t-test was employed for the comparison of different continuous variables or characteristics of farm households. Whereas, chi-square was used to compare categorical or dummy variables among IPV adopter and non- adopter households. 3.3.1.1. Household Dietary Diversity Score (HDDS) According to FAO (2010) the HDDS provide an indication of household economic access to food and it was calculated by summing the number of food groups consumed in the household respondent over the 24-hour recall period. Respondents were asked whether they consumed the 12 food groups and their “yes” responses were coded as 1 and the negative response “no” coded as 0. The next step is summing the dietary diversity variable values of all new food groups and, the potential score ranges from 0 to 12 for HDDS. The higher score indicated that households consumed more diversified food groups. The HDDS of ≤3, 4-5 and ≥ 6 implies low, medium and high dietary diversity respectively (FAO, 2010). 26 3.3.1.2. Food Consumption Score (FCS) According to VAM (2008) the FCS is a composite score based on dietary diversity, food frequency, and relative nutritional importance of different food groups and it can be calculated using the frequency of consumption of different food groups consumed by a household during the 7 days before the survey. The following four procedure are important to calculate the FCS, these are: (i) group all the food items (the 16 food items) into specific food groups (9 food groups), (ii) sum all the consumption frequencies of food items of the same group, and recode the value of each group above 7 as 7, (iii) multiply the value obtained for each food group by its weight (the standard weights for main staples 2, pulses 3, vegetables 1, fruit 1, meat and fish 4, milk 4, sugar 0.5, oil 0.5, condiments 0) and create new weighted food group scores and, (iv) sum the weighed food group scores, thus creating the food consumption score (FCS). FCS 0-21, 21.5-35, and >35 indicated poor, borderline, and acceptable household consumption respectively. 3.3.1.3. Food Security Scale (FSS) According to Bickel, et al. (2000) the full range of food insecurity and hunger cannot be captured by any single indicator. Instead, a household’s level of food insecurity or hunger must be determined by obtaining information on a variety of specific conditions, experiences, and behaviors that serve as indicators of the varying degrees of severity of the condition. Food insecurity cannot be measured directly. Therefore, to measure the food insecurity and hunger the 18 food security questions found to provide the statistically strongest set of indicator items for constructing a 12-month measurement scale. The sum of affirmative (“Almost every month”, “Often true”, “Sometimes true”, and “Yes” coded as 1) and negative responses (“Never true”, “only one or two months”, “No”, and questions that a household does not answer because it has been screened out, coded as 0) provide the FSS. According to Price et al. (1997) this measure expresses the household's level of food security or insecurity in terms of a numeric value that ranges between 0 and 10. The scale values of 0, indicating that household did not experience in the past year any of the conditions of food insecurity and the scale value 10 indicates the most severe level of food insecurity. Households with children having a scale value of 0-1.6 (0-2 affirmative response), 2.3-4.3 (3-7 affirmative 27 responses) and 4.7-6.4 (8-12 affirmative response) 6.8-10 (13-18 affirmative responses) out of the 18 food insecurity questions, classified into four food security status categories; these are, food secure, food insecure without hunger, food insecure with hunger, food insecure with severe hunger respectively. Household without children having a scale value of 0-2 (0-2 affirmative responses), 2.8 - 4.3 (3-5 affirmative responses), 5-6.5 (6-8 affirmative responses) and 7.5-8.2 (9-10 number of affirmative responses) also were classified in to food secure, food insecure without hunger, food insecure with hunger, food insecure with severe hunger respectively. 3.3.2. Specification of Econometrics Model 3.3.2.1. Tobit Model According to McDonald and Moffit (1980), to identify factors of the probability and level of adoption of improved potato varieties the Tobit model was used, and it can be expressed mathematically as, 𝑌∗ = 𝛽଴ + 𝛽ଵ𝑥ଵ + 𝛽ଶ𝑥ଶ + 𝛽௡𝑥௡ … … + 𝜇௜ = 𝑓(𝑥௜) Y = 𝑌∗, 𝑖𝑓 𝑌∗ > 0 and Y= 0 𝑖𝑓 𝑌∗ ≤ 0 … … … … ………………………………. … … . 𝑒𝑞 (1) Where Y = the observed dependent variable, in this case the proportion of area under improved potato varieties in the total cultivated areas since farmers in the study areas totally replaced the local varieties. Y* = is the latent dependent variable, which is not observable, X = explanatory variable, 𝛽= a vector of Tobit maximum likelihood estimates, μi = an independently and normally distributed error term with mean zero and constant variance. According to Tobin (1958), the expected value of use intensity of improved potato varieties across all observations was estimated by: 𝐸(𝑌ூ) = 𝑋𝛽𝐹 ൬ 𝑌௠ 𝜎 ൰ + σf ൬ 𝑌௠ 𝜎 ൰ … … … … … … … … … … … … … … … … … … … … … … … . 𝑒𝑞(2) 28 Where 𝐹 ቀ௒೘ ఙ ቁ and ௫ఉ ఙ , f(௒೘ ఙ ) are the cumulative normal distribution function and the value of the derivative of the normal curve respectively. ௫ఉ ఙ , (௒೘ ఙ ) represents the normalized index at the mean values of all explanatory variables and the Z-scores for the area under the normal curve. 𝛽 and σ represents are the Tobit maximum likelihood estimates and the standard error of the error term respectively. The marginal effect of an explanatory variable on the expected value of the dependent variable (proportion of area under improved potato varieties) is: డா(௒) డ௫೔ = f ቀ௒೘ ఙ ቁ 𝛽ଵ … … … … … … … … … … … … … … … … … … … … … … … … … … . . (3) The change in the probability of using a technology as independent variable Χi changes is: ப୊(ೊ೘഑ ) డ௫భ = f ቀ௒೘ ఙ ቁ ఉభ ఙ … … … … … … … … … … … … … … … … … … … … … … … … … . . 𝑒𝑞(4) 3.3.2.2. Propensity Score Matching Model (PSM) To address the second research objective, the Propensity score matching method was applied. In an experimental design, randomization ensures uniform/equal distribution of all relevant characteristics between treatment and control group and, because of this, the difference in mean outcomes correctly estimates the impact of the intervention. In the absence of randomization, however, the groups may differ not only in their treatment status, but also in their values of socioeconomic characteristics. In this case, it is necessary to account for these differences to avoid potential biases. Therefore, to avoid this potential bias PSM were applied and this method allows to create the comparable nonparticipants or counterfactuals to participants (Heinrich et al., 2010). The counterfactual was be identified by matching participant (improved potato grower) with nonparticipants (non-grower) which have similar pre-intervention characteristics, it is equally valid to match on the propensity score. The Method measures the impact of the intervention as the difference between the potential outcome in case of treatment and the potential outcome in the absence of treatment- 29 (Heinrich et al., 2010). For this study, the outcomes variables are HDDS, FCS and FSS. According to Heinrich et al. (2010) to apply the PSM for estimating the impacts of the intervention, in this case adoption of improved potato varieties, the following four procedures was used, these are: (i) Generating propensity scores p(x) The propensity score is estimated using various socio-demographic characteristics of farmers. These scores are probabilities that represent the households for adopting improved potato varieties given characteristics (X). The probability of participation summarizes all the relevant information contained in the X variables and as it allows for matching on a single variable (the propensity score) instead of on the entire set of covariates (Heinrich et al., 2010). According to Gujarati (2003), the propensity score matching was generated using logit model and the model mathematically specified as follows: 𝑝𝑖 = 𝐸(𝑦 = 1|𝑋௜ = 1 1 + 𝑒 − (𝛽଴ + 𝛽ଵ𝑥௜) … … … … … … … … … . . … … . (1 ) 𝑍௜ = 𝛽଴ + 𝛽ଵ𝑥௜ … … … … … … … … … … … … … … … … … … … . . … … … . (2) 𝑝௜ = 1 1 + 𝑒 − 𝑧௜ = 𝑝௜ = 𝑒௭೔ 1 + 𝑒௭೔ . … … … … … … … … … … … … . . … … … (3) 𝑍௜ = 𝛽଴ + ∑ 𝛽ଵ𝑥௜ + 𝜇௜௡௜ୀଵ … … … … … … … … … … … … … … … . . . … . (4) Where, p is the probability of participation in improved potato varieties production i =1, 2, 3 -- n (number of observations) β0 = the intercept βi = the slope of regression coefficients to be estimated xi = intervention characteristic of households µi = disturbance term or error term or stochastic variable The probability that a household belongs to non-improved potato producer is: 1-Pi= ଵ ଵା௘೥೔ --------------------------------------------------------------------------------------------(5) As indicated above, using the explanatory variable the logit model for this study was specified as follows. 30 Yi=α+β1Sex +β2Age +β3EduHH +β4HHsize +β5Sizeofland +β6TLU +β7Farmexpr +β8Extensiona +β9Closerpd+ β10Irracces + β11Neighbor + β12ipvmaturityP + β13ipvyield+ủ-------------------------------------------------------------------------------------------(6) (ii) Choose a Matching Algorithm The idea of matching is identifying control and treated individuals with the same or similar propensity score. Once an estimated propensity score is obtained, different matching algorism was used to match comparison units with treated units. The most commonly employed matching algorisms are the nearest neighbor, kernel matching, stratification matching, caliper matching and radius matching (Heinrich et al., 2010). For this study, the PS of treated households (user) was matched with counterfactual households (non-user) using the nearest neighbor, kernel, caliper and Radius matching estimator methods. To do the matching, three important tasks should be done first. The first task is, generating propensity score (probability of participation) based on the selected covariates. The second task is imposing the common support condition on the propensity score distribution of the sample households. The common support region is region between the higher value of the minimum and the lower value of maximum propensity score of the treated or control groups. The last task before matching is discarding observations whose propensity score is outside common support region. Nearest neighbor matching - one of the most straightforward matching procedures. An individual from the comparison group is chosen as a match for a treated individual in terms of the closest propensity score (or the case most similar in terms of observed characteristics). The nearest neighbor matching with replacement methods was used to match untreated individual more than once as a match. Using nearest neighbor ensure the use of the most similar observation to construct the counterfactual. Kernel matching - compare the outcome of each treated person to a weighted average of the outcomes of all the untreated persons, with the highest weight being placed on those with scores closest to the treated individual. One major advantage of this approach is the lower variance, which is achieved because more information is used. 31 Radius matching -specifies a “caliper” or maximum propensity score distance by which a match can be made. It uses all of the comparison group members within the caliper. According to Heinrich et al. (2010) to estimate the impact of a program correctly; PSM requires two main conditions, the conditional independence assumption and the common support condition. Conditional independence assumption: The assumption assesses the quality of matching to perform tests that check whether the propensity score adequately balances characteristics between the treatment and comparison group units. It verifies the treatment is independent of unit characteristics after conditioning on observed characteristics: D X | p(X). After the application of matching, there would not be statistically significant differences between covariate means of the treatment and comparison units. The inclusion of the variables is based on the conditional independence assumptions. Relevant variables related to the intervention and outcome were considered in the propensity score function (Heckman et al.,1997). According to Caliendo and Kopeinig (2005), the inclusion of non-significant variables would not bias the estimates or make them inconsistent. On the other hand, including the full set of covariates in small samples might cause problems in terms of higher variance, since either some treated have to be discarded from the analysis or control units have to be used more than once. The Common Support Condition Assumption: It helps to investigate the validity or performance of the propensity score matching estimation to verify the common support or overlap condition. The assumption is critical to estimation, as it ensures that units with the same X values have a positive probability of being both participants and nonparticipants: 0