PRODUCTION RISK AND TECHNICAL EFFICIENCY OF LOWLAND RICE FARMS IN ASHANTI REGION, GHANA BY BAAWUAH GEORGE THIS THESIS IS SUBMITTED TO THE UNIVERSITY OF GHANA, LEGON IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE AWARD OF MASTER OF PHILOSOPHY DEGREE IN AGRICULTURAL ECONOMICS DEPARTMENT OF AGRICULTURAL ECONOMICS AND AGRIBUSINESS SCHOOL OF AGRRICULTURE COLLEGE OF BASIC AND APPLIED SCIENCES UNIVERSITY OF GHANA, LEGON JUNE, 2015 University of Ghana http://ugspace.ug.edu.gh i DECLARATION I, BAAWUAH GEORGE, do hereby declare that except for the references cited, which are duly acknowledged, this thesis titled “PRODUCTION RISK AND TECHNICAL EFFICIENCY OF LOWLAND RICE FARMS IN ASHANTI REGION, GHANA” is the product of my own research work undertaken in the Department of Agricultural Economics and Agribusiness, University of Ghana, Legon. This research work has never been presented in a whole or in part for any other degree of this university or elsewhere. Signature:………………………….…. Date……………………………… Baawuah George (Student) Signature:……………………………… Signature:……………………… Dr. Akwasi Mensah-Bonsu Dr. Henry Anim-Somuah (Major Supervisor) (Co-Supervisor) Date…………………………………… Date…………………………….. University of Ghana http://ugspace.ug.edu.gh ii DEDICATION This work is dedicated to my brothers and sisters particularly my twin sister Georgina Baawuah for their help and support in different ways to make my studies successful. University of Ghana http://ugspace.ug.edu.gh iii ACKNOWLEDGMENT My utmost gratitude goes to the Lord Almighty God for gift of life to complete the study successfully. I wish to express my sincere gratitude and deep appreciation to my supervisors; Dr. Akwasi Mensah-Bonsu and Dr. Henry Anim-Somuah for their continuous and structured guidance, corrections, suggestions and constructive criticism in shaping the work. I acknowledge all the lecturers in the Department of Agricultural Economics and Agribusiness, University of Ghana for their generous support in diverse ways to successfully complete the study. To my colleagues and Ph.D. students in the Department of Agricultural Economics and Agribusiness also contributed in many ways to complete this research work, I say thank you. However, the following individuals deserve commendation for their unstinting support and encouragement; Boahen Atta Oppong, Jaures Cocou Amegnaglo, Samuel Ofosu Appiah, Francisca Ayerh, Eric Brako Dompreh, Daniel Ofori-Sasu, Michael Owusu Ansah and Ulysse Azannai. My appreciation also goes to extension workers at Ministry of Food and Agriculture in Ashanti region who assisted in data collection and the rice farmers for their patience and cooperation to make the data collection a reality. Finally, I extend my gratitude to Ministry of Food and Agriculture through West Africa Agricultural Productivity programme (WAAPP) for their financial support to further my studies. University of Ghana http://ugspace.ug.edu.gh iv ABSTRACT The study was conducted to analyse production risk and technical efficiency of lowland rice farms in Ashanti region using adjusted stochastic production frontier. Cross sectional data for 2014 cropping season was collected from 200 sampled rice farms using multistage sampling technique. The results of the study revealed very low utilization of improved seed (42.8 kg/ha) by famers as against the standard quantity of 100kg/ha. Among the farm inputs used, fertilizer and improved seed were significant at 1% and found to be risk reducing inputs. The partial elasticity estimates indicated that fertilizer (0.57) and improved seed (0.44) used by farmers contributed largely to lowland rice productivity in the study area. The computed return to scale for the lowland rice farms demonstrated decreasing return to scale. The average productivity of lowland rice in the study area was 3.7Mt/ha as compared to the national average of 2.5 Mt/ha. Approximately 33.6% of the sampled rice farms produced below the national average whereas 66.7% produced above it. About 85% of the sampled rice farms produce below the national potential yield of 6.5Mt/ha. The mean technical efficiency estimates (77%) of the study suggest that extra 23% of the rice output can be produced by the rice farmers without any additional farm inputs. The results of the inefficiency estimates indicates that improved seed-usage, shorter farm distance, farmer access to extension services, basic education, mechanization services and canal maintenance reduces technical inefficiency. The study therefore recommends that stakeholders in agricultural sector should increase input outlets to enable rice farmers have access and use more of fertilizer and improved seed to increase production. Ministry of Food and Agriculture should continue with the extension service provision to educate farmers on how to follow good agronomic management practices for rice production such canal maintenance to reduce production inefficiency. Factors that make credit to contribute to lowland rice production inefficiency should be investigated and addressed by stakeholders. University of Ghana http://ugspace.ug.edu.gh v TABLE OF CONTENTS CONTENT PAGE DECLARATION ...................................................................................................................... i DEDICATION ......................................................................................................................... ii ACKNOWLEDGMENT ........................................................................................................ iii ABSTRACT ............................................................................................................................. iv LIST OF ABBREVIATIONS ................................................................................................. x CHAPTER ONE ...................................................................................................................... 1 INTRODUCTION ................................................................................................................... 1 1.1 Background ............................................................................................................................. 1 1.2 Problem Statement .................................................................................................................. 4 1.3 Objective of the Study............................................................................................................. 7 1.4 Justification of the Study......................................................................................................... 7 1.5 Organization of the Study ....................................................................................................... 8 CHAPTER TWO ..................................................................................................................... 9 LITERATURE REVIEW ....................................................................................................... 9 2.1 Introduction ............................................................................................................................. 9 2.2 Lowland Rice Farming............................................................................................................ 9 2.3 Rice Production and Imports in Ghana (2002 – 2012) ......................................................... 10 2.4 Rice Consumption Preferences ............................................................................................. 11 2.5 National Policies to Promote Local Rice Production ............................................................ 12 2.5.1 Trade Policy ...................................................................................................................... 12 2.5.2 Agricultural Policies Relevant for Rice Development ...................................................... 13 2.5.3 National Rice Development Strategy (NRDS) ................................................................. 14 2.6 Performance Measure of Rice ............................................................................................... 15 2.7 Productivity Measurement .................................................................................................... 16 2.8 Efficiency Measurement Mechanisms .................................................................................. 18 2.8.1 Non-Parametric Approach of Efficiency Measurement .................................................... 19 University of Ghana http://ugspace.ug.edu.gh vi 2.8.2 Parametric Approach of Efficiency Measurement ............................................................ 20 2.9 Production Risk Measurement .............................................................................................. 23 2.9.1 Incorporation of Production Risk in the Stochastic Frontier Model ................................. 25 2.9.2 Review of Factors Influencing Production Risk and Technical Efficiency ..................... 26 2.9.3 Description of Technical Inefficiency Variables used in the Study .................................. 29 2.10 Conclusion ............................................................................................................................ 32 CHAPTER THREE ............................................................................................................... 34 METHODOLOGY ................................................................................................................ 34 3.1 Introduction ........................................................................................................................... 34 3.2 Conceptual Framework ......................................................................................................... 34 3.3 Theoretical Framework ......................................................................................................... 36 3.4 Method of Analysis ............................................................................................................... 39 3.6 Hypothesis of the Study ........................................................................................................ 43 3.7 Data Source and Collection................................................................................................... 44 3.8 Sample Size and Sampling Technique .................................................................................. 45 3.9 Geographical Setting of the Study Area ............................................................................... 45 CHAPTER FOUR .................................................................................................................. 47 RESULTS AND DISCUSSIONS .......................................................................................... 47 4.1 Introductions ......................................................................................................................... 47 4.2 Descriptive Statistics of Socio-economic Characteristics of the Respondents .................... 47 4.3 The Productivity Estimates of Lowland Rice Farms ............................................................ 50 4.4 Descriptive Statistics of Economic Variables in the Stochastic Frontier ............................. 51 4.5 Validation of Hypothesis of the Study .................................................................................. 52 4.6 Translog Stochastic Frontier ................................................................................................. 53 4.7 Estimated Stochastic Frontier Model with Risk .................................................................... 56 4.8 Technical Efficiency Indices of Lowland Rice Farms .......................................................... 57 4.9 Technical Inefficiency Estimates of Lowland Rice Farms .................................................. 58 University of Ghana http://ugspace.ug.edu.gh vii CHAPTER FIVE ................................................................................................................... 62 SUMMARY, CONCLUSIONS AND POLICY RECCOMMENDATIONS .................... 62 5.1 Introduction ........................................................................................................................... 62 5.2 Major Findings From the Research ....................................................................................... 62 5.3 Conclusion of the Study ........................................................................................................ 63 5.4 Policy Recommendation ....................................................................................................... 64 5.5 Limitation of the Study ......................................................................................................... 65 5.6 Suggestion for Future Study ................................................................................................. 66 REFERENCES ....................................................................................................................... 67 Appendix I: Questionnaires for the Study .............................................................................. 75 Appendix II: Tanslog and Cobb-Douglass Regression Results ............................................. 82 Appendix III: Technical Efficiency Estimation (No production risk considered) ................. 84 University of Ghana http://ugspace.ug.edu.gh viii LIST OF TABLES CONTENT PAGE Table 2. 1: Overview of Rice Production and Importation in Ghana (2002 – 2012) .............. 11 Table 3. 1 Technical Inefficiency Variables Influencing Lowland Rice Production ............ 43 Table 3. 2: Test for Statistical Assumptions of the Stochastic Frontier (hypothesis) ............. 44 Table 3. 3: Selected Lowland Rice Districts and Communities in Ashanti Region ................ 45 Table 4. 1: Summary of Socio-Economic Characteristic of the Respondents ...................... 49 Table 4. 2: Productivity Distribution of Lowland Rice Farms in Ashanti Region ............. 50 Table 4. 3: Summary of Economic Variables in The Stochastic Production Frontier .......... 52 Table 4. 4: Results of Tested Hypothesis for the Study ......................................................... 53 Table 4. 5: Maximum likelihood Estimates of Translog Mean Output Function .................. 54 Table 4. 6: Elasticity of Mean Output and Returns to Scale (RTS) ........................................ 55 Table 4. 7: Maximum Likelihood Estimates of Production Risk Function ............................ 57 Table 4. 8: Technical Efficiencies Scores of Lowland Rice Farms ........................................ 58 Table 4. 9: Determinants of Technical Inefficiency in Rice Production ................................ 61 University of Ghana http://ugspace.ug.edu.gh ix LIST OF FIGURES CONTENT PAGE Figure 3. 1: Conceptual Framework of the Study .................................................................. 36 Figure 3. 2: District Map of Ashanti Region .......................................................................... 46 University of Ghana http://ugspace.ug.edu.gh x LIST OF ABBREVIATIONS AFD Agence Francaise de Development CARD Coalition for African Rice Development CAADP Comprehensive African Agricultural Development Programme DEA Data Envelopment Analysis DPs Donor Partners ECOWAS Economic Community of West Africa States EDIF Export Development and Investment Fund FBOs Farmer Based Organizations FAO Food and Agricultural Organization FASDEP Food and Agricultural Sector Development Program GDP Gross Domestic Product GPRS I Ghana Poverty Reduction Strategy IFAD International Food and Agricultural Development IFPRI International Food Policy Research Institute JICA Japanese International corporation Agency MAFAP Monitoring African Food and Agricultural policies METASIP Medium Term Agricultural Service Investment Program MoFA Ministry of Food and Agriculture NHIS National Health Insurance Scheme NERICA New Rice for Africa NRDS National Rice Development Strategy NGO Non-Governmental Organization SRID Statistical Research and Information Directorate WARDA West African Rice Development Association University of Ghana http://ugspace.ug.edu.gh 1 CHAPTER ONE INTRODUCTION 1.1 Background Agricultural production in Ghana is basically a rural activity and the sector is characterised by smallholder farmers who engage in rudimentary farming practices (Ganidekam, 2013). Smallholder farmers account for 80% of total production and cultivate on less than 4 hectares of land (Al-Hassan, 2012). Most larger holder farmers cultivate cash crops such as oil palm, rubber and cocoa. Commodities produced by the farmers can be classified into three areas relevant for the national economy. These are: foodstuffs for local consumption; raw material for local industry and exportable commodities for foreign markets. However, the desire to survive makes food crop production the primary occupation of most farmers (Wayo & Nyanteng, 2003). Ghana is endowed with vast land with different agro ecological zones which makes it suitable to cultivate diversity of crops (Throup et al., 2011). Key among the diversified commodities produce by the farmers are maize, rice, cassava, yam, and cowpea. These commodities are described as food security crops but production is mainly rain-fed (MoFA, 2010). Their dependency on rainfall for production makes the sector prone to output variation and rice production is no exception (Opoku-Duah et al., 2000; Al-Hassan, 2012 & Yiadom-Boakye et al., 2013). Rice is considered as the second most important commodity after maize and Ghana Poverty Reduction Strategy (GPRS I), Food and Agricultural Sector Development Policy (FASDEP II) and Medium Term Agricultural Investment Plan (METASIP) have highlighted the commodity as one of the major food security crop (Ragasa et al., 2013). The recognition given to rice by these policy documents implies that rice has assumed a strategic position in the food basket of both the rural and urban households and is cultivated in University of Ghana http://ugspace.ug.edu.gh 2 virtually all the ten regions in Ghana (Angelucci et al., 2013). Rice production in Ghana has been categorized into three ecosystems based on adaptive mechanism which is characterized by water supply namely; irrigated, upland and lowland rice farming. Among these three ecosystems, lowland rice farming constitutes 78 percent of the arable land area, followed by the irrigated system contributing 16 percent while the upland rice farming system covers 6 percent (MoFA, 2009). Irrigated rice farming facilitates continuous farming all year round through intensification to boost food production and availability. However, high cost element in the construction of irrigation system and the necessary technical expertise required to manage the system does not allow majority of farmers to depend on irrigated rice production (Apori-Buabeng, 2009). In Ghana, domestic demand for rice has not been adequately met by growth in local production. The low productivity of rice has created substantial rice yield gap (MoFA, 2012). This phenomenon has created an opportunity for international food traders to supply rice to Ghana due to its high demand (Guisse, 2010). Current rice production in Ghana satisfied around 30 to 40 percent of the national demand with a corresponding average rice import bill of 450 million United State dollars annually (Angelucci, 2013; Kula & Dormon, 2009). Foreign exchange rate is an important factor affecting prices in global rice trade because international rice prices are quoted in US dollars. This implies that if the Ghanaian currency (cedi) is depreciating relative to the US dollar, rice import bill (in US dollars) increases and more cedis will be required to import. This makes rice imports an issue of national concern for policy makers because of its impact on domestic currency and local industries (Awudu & Huffman, 2000). Past intervention measures in the form of projects by government and Donor Partners (DPs) to reduce rice importation includes; inland valleys rice development project which was implemented in five regions, namely Ashanti, Brong Ahafo, Central, Eastern and Western University of Ghana http://ugspace.ug.edu.gh 3 regions of Ghana in 2001 at a total cost of $22.2 million dollars (JICA, 2008). The lists of lowland rice projects implemented within the period of 2001 to 2014 are as follows: 1. Inland Valleys Rice Development Project (2001-2009) 2. Improvement of Drought Tolerance of Rice through Gene Transfer (2007-2009) 3. NERICA Rice Dissemination Project (2005-2010) 4. Rice Seed Production (2008-2010) 5. Rice Sector Support Project (2008-2014) 6. Project for Sustainable Development of Rain-fed Lowland Rice Production (2009-2014) 7. Improving Yield, Quality and Adaptability of Upland and Rain-fed Lowland Rice Varieties in Ghana to Reduce Dependency on Imported Rice (2010-2012) 8. An Emergency Initiative to Boost Rice Production (USAID – SARI) (2008-2010) 9. Deficit and Improve Farmers Income in Ghana (2011-2014) The potential for lowland rice farming is much higher compared to upland rice farming because lowland presents the possibility of cultivating rice two or more times annually in the absence of irrigation which makes it economically more valuable compared to upland rice farming. According to MoFA (2009), National Rice Development Strategy (NRDS) was initiated in 2009 with the road-map to double domestic rice production from 14 to 28 million tons by 2018. To achieve this level of expected national rice production target will require improved access to farm inputs to facilitate their usage which is one of the important intervention strategies being pursued by government to increased rice productivity. According to Moro et al. (2008) and Issaka et al. (2014), low soil fertility has been identified as a major cause for low rice productivity in Ghana and the situation is compounded by high cost of fertilizer. To increase accessibility and agricultural inputs usage, MoFA in 2010, raised agricultural input outlet centres by 11%. However, unwarranted application of University of Ghana http://ugspace.ug.edu.gh 4 inorganic fertilizers and agro-chemicals on rice production has both environmental and cost implications. Therefore, efficient input usage by lowland rice farmers must always be considered in an effort to increase rice productivity. This makes technical efficiency estimation of lowland rice farms important for assessing farmers’ production risk with regard to input usage. According to Chang et al. (2008), Bokusheva & Hockmann (2006), production environment of lowland rice farms poses some magnitude of risk which has the potential to affect the output variability as well as the technical efficiency levels of the farmers. If the production environment is uncertain and affects producers input use decisions, then analyzing technical efficiency to account for production risk is needed to produce unbiased estimates of the technical efficiency. This makes technical efficiency estimation of lowland rice farms important for assessing farmers production risk with regard to input usage. Therefore, assessing production risk and technical efficiency in lowland rice growing ecology to formulate appropriate adaptive policies will to increase domestic rice output in the study area. 1.2 Problem Statement Rice production in Ghana is dominated by smallholder farmers and these farmers cultivate both on upland and lowland ecologies (MoFA, 2009). Low productivity coupled with stiff competition posed by importers over the years has restrained the farmers from earning significant returns from their investment and this has created rice production deficit. To minimize the effect of the rice production shortfalls on national demand, several efforts have been made by successive governments together with donor partners in a form of implemented projects to increase rice production. In spite of a number of lowland rice projects implemented within the period of 2001 to 2014 by government and donor partners, national average of rice production generates 2.5 Mt/Ha which is low compared to Cote D’ivoire (3.0 University of Ghana http://ugspace.ug.edu.gh 5 Mt/Ha), another developing country within West Africa using the similar production technology (WARDA, 2001). The gap between achievable yields under best farming practices and actual yields of rice ranges from 2.5 to 6.5 Mt/Ha. This therefore calls for the need to increase production using productivity enhancing approach such as the use of improved rice seed, fertilizer, reduce technical inefficiency and risk on the part of the farmers. Effort to increase rice productivity over the years has proved ineffective due to limiting factors such as: inadequate institutional support (access to credit, research and extension), inappropriate production system, inadequate basic infrastructures, post-harvest management technology, inappropriate marketing strategy, production risk and inefficiency on the part of the farmers (Apori-Buabeng, 2009; MoFA, 2009; Al-Hassan, 2012; & Yiadom- Boakye et al., 2013). In the mist of rice production challenges, national rice development strategy has outlined projections to double rice cultivation to reduce importation. The central question is that can lowland rice farmers improved on their technical efficiency to reduce production risk and help to increase rice output. Over the years, many researchers and policy makers in Ghana have focused their attention on the impact that technologies adoption have on increasing farm productivity and income (Ragasa et al., 2013). However, gains resulting from technology adoption by lowland rice farmers are likely to be minimized by losses arising from production risk relating to input adjustment and technical inefficiencies on the part of smallholder famers. In Ghana, several empirical efficiency studies have been carried out on rice. For instance, Al-Hassan (2012) in the Upper East region studied technical efficiency of smallholder rice farmers and found that they are technically inefficient because they produce with mean efficiency of 34%. Dessalegn (2005) modelled farm irrigation decisions under rainfall risk in the white-Volta basin found that, rainfall risk in particular is important in limiting the use of improved technologies such University of Ghana http://ugspace.ug.edu.gh 6 as fertilizer. However, expansion of irrigation infrastructure promotes the use of such input (fertilizer). Asante et al. (2013) examined the impact of NERICA rice variety adoption on technical efficiency of rice farmers in Ashanti region and concluded that adoption of improved rice variety improve efficiency by 69.1%. Yiadom-Boakye et al. (2013) examined Gender, resource-use and technical efficiency among rice farmers and found that there is inefficiency in resource-use but on the average, female rice farmers are relatively inefficient compared to their male counterparts in Ashanti region. Although considerable efforts have been made in examining technical efficiency of rice farmers, none of such studies has attempted to exclusively look at production risk and technical efficiency of these farmers with regard to input-use. The subject of production risk is important in reducing the level of biasness in efficiency estimation. As a step forward, present study focuses on production risk and technical efficiency since they form an integral component in agriculture production and influence farmers decision making (Guan & Wu, 2009). The lowland rice farmers’ ability to increase productivity depends on their ability to reduce production risk and inefficiencies to increase their efficiency levels. Since lowland rice farmers dominate the rice sector and the technology used by these farmers are not absolutely changing (Ragasa et al, 2013) but decision making associated with the use of input resources, socioeconomic and institutional factors keep on changing, the relevant questions relating to the study to provide answers to production risk and technical efficiency of lowland rice farms in Ashanti region are: What has been the productivity of rice with respect to key input factors in lowland rice growing areas in Ashanti region? What are the input factors influencing risk in rice production? What are the technical efficiency levels of lowland rice farmers? What are the determinants of technical inefficiency in the region? University of Ghana http://ugspace.ug.edu.gh 7 1.3 Objective of the Study The main objective of the study is to assess the production risk and technical efficiency of small holder lowland rice farmers in Ashanti region of Ghana. The specific objectives are: 1) To estimate the productivity of rice with respect to key input factors. 2) To estimate production risk with respect to factor inputs. 3) To estimate the technical efficiency levels of lowland rice farmers. 4) To investigate the determinants of technical inefficiency in lowland rice production. 1.4 Justification of the Study Investments in rice sector have been rising over the years as government and donor partners commitment to finance rice projects to increase productivity surges on. The ability of lowland rice farmers to increase productivity and reduce national imports depends on their level of technical efficiency as well as the knowledge and ability to reduce risk associated with rice production. Efficiency is a very important factor for productivity growth. In an economy where opportunities to use new technologies are limited, efficiency study will indicate the possibility to increase productivity by reducing inefficiency without necessarily developing new technologies or increasing the resource base. Based on that, the study will help to identify the relevant variables within the socioeconomic, institutional and managerial factors that can improve on the efficiency of rice farmers in the study area. Production risk affects farmers’ decision to optimized inputs usage or adopts technology to enhance production. Since policy makers are particularly interested in the impact of their policies on output and given the limited resource allocation for production and the importance of risk effect on input usage, the estimate of productivity of the factor inputs and production risk will help understand the relationship of the inputs to output. This will help to University of Ghana http://ugspace.ug.edu.gh 8 inform policy directives to increase productivity and mitigate risk in input allocation. Hence, the outcome of this study will contribute to improve on production efficiency of rice farmers and reduce risk associated with input usage to increase productivity in rice growing ecologies to reduce rice imports. 1.5 Organization of the Study The study consists of five chapters. Chapter 1 examines the overview of rice production in Ghana, problem, specific objectives and justification for the study. Chapter 2 reviews literature related to rice production, importation and policies in Ghana. Chapter 3 presents the model specification and detailed discussion of the variables and data set utilized in the study. Chapter 4 presents the results and discussion from the study, while conclusions of the major findings, recommendations and suggestions for further research are discussed in chapter five. University of Ghana http://ugspace.ug.edu.gh 9 CHAPTER TWO LITERATURE REVIEW 2.1 Introduction This chapter reviews literature relevant to the present study. Section 2.2 defines lowland rice farming, section 2.3 describes and anylise rice production and imports in Ghana from 2002 to 2012. This is followed by section 2.4 which reviews literature on rice varietal research to improve yield and adaptation. Trade policy on rice, agricultural policy and strategy to reduce imports are presented in section 2.5. Finally, section 2.6 reviews and synthesized existing literature on productivity, production risk and technical efficiency measurements. 2.2 Lowland Rice Farming According to Hara et al. (2001), lowland rice farming system is operated in low laying areas of land with hydromorphic soil properties (swamps flood plains). The rice farms depend mainly on rainfall and climate change manifest itself through reduction or increment in rainfall, temperature, pests and diseases incidence which affect agricultural production and food security particularly in sub-Saharan Africa (McCarthy, 2011). This system of rice farming receives runoff from hills and mountains. The water capture systems are structures such as bunds for water retention and canals for gravity delivery. Bunds and spillways on the fields improve infiltration and allow drainage when required. Much rainfall causes disease infestation in crops while reduction in rainfall reduces the availability of water for lowland rice cultivation which can be detrimental to crop yield at crucial growing stage (Molua et al., 2007). According to Dessalegn (2005) water control is the most important management practice that determines the efficacy of all other production inputs (fertilizer, herbicide, pesticide, etc ) as a mitigation factor for the risk associated with lowland rice farming. This University of Ghana http://ugspace.ug.edu.gh 10 makes it important to consider managerial variables as a contributory factor to reduce inefficiency in the rice production. 2.3 Rice Production and Imports in Ghana (2002 – 2012) Rice production has been the focus of attention in Ghana over the years to increase its output. The commodity accounts for 58 percent of cereal imports and 5 percent of total agricultural imports over the period 2005 to 2009 (CARD, 2010). The production has bimodal system of cultivation (major and minor season) and is consumed mainly by humans rather than fed to livestock because it is an efficient food for supplying carbohydrates, vitamins, and nutrients in diets (Angelucci et al., 2013). The development of rice industry in Ghana over the years has been inconsistent in character but in general, production has increased considerable whiles at the same time, serious output variations were recorded (MoFA, 2012). Paddy rice production estimates rose from 280.0 Mt in 2002 to 481.1 Mt in 2012 with equivalent area expansion of 54.3 percent associated with the change in rice output during the same period as seen in Table 2.1. Like many other West African countries that use similar production technology, rice production in Ghana has been driven mostly by expansion in harvested area and producer price (MoFA, 2009; & Boansi, 2013). The tremendous growth in rice yield (9%) during the period is attributed to farmers’ response to number of programmes and projects that were implemented by the government and donor partners within 2004 to 2009. This includes fertilizer subsidy programme introduced in 2008 and National Rice Development Strategy in 2009. The bulk of both rice production area and output came from the Upper East, Northern and Volta Regions of Ghana (Ragasa et al., 2013). However, the volume of rice imports per hectare basis within the same period also grew by 11% at a corresponding average rice import bill of USD $ 218.7 Million. This has financial implication on the prosperity of the smallholder farmers, local rice industry as well as the University of Ghana http://ugspace.ug.edu.gh 11 food security as it makes the overall production process uncompetitive for domestic rice farmers to thrive. Table 2. 1: Overview of Rice Production and Importation in Ghana (2002 – 2012) Year Area (000Ha) Production (MT) Yield* Quantity (Mt) Imported Value of Imported rice $ million 2002 122.8 280.0 2.3 297.0 68.85 2003 117.7 239.0 2.0 797.7 124.66 2004 119.4 241.8 2.0 253.9 119.15 2005 120.0 236.5 2.0 484.5 138.94 2006 125.0 250.0 2.0 389.6 159.47 2007 108.9 185.3 1.7 442.1 157.86 2008 132.8 301.9 2.3 395.4 187.28 2009 162.4 391.4 2.4 384.0 218.5 2010 162.4 491.6 3.0 320.2 200.88 2011 197.5 464.0 2.3 543.5 391.17 2012 189.5 481.1 2.5 509.0 639.4 Total 1558.4 3562.6 24.5 4816.9 2406.16 Source: (MoFA, 2012). Yield figures were computed by author* 2.4 Rice Consumption Preferences The physical dimensions of rice are of vital interest to consumers and rice industry in Ghana. These dimensions are important in influencing production, consumer choices, marketing, grading and processing. Rice of different sizes adversely affects the value chain process and yield (Richman et al., 2006). To minimize the problem of low rice yield and climatic adaptation to enhance rice productivity whiles maintaining both consumer and industrial qualities, African rice center has played an important role in germplasm enhancement for rice adaptation into different ecologies such as upland, lowland, irrigated, mangrove and deep water (Sié et al., 2011). Based on morphological and agronomic characteristics (high yielding and resistance to drought, pest and diseases), new rice variety has been developed by African rice by crossing O. glaberrima and O. sativa ssp. using conventional biotechnology University of Ghana http://ugspace.ug.edu.gh 12 such as anther culture and embryo rescue techniques to overcome sterility barriers between the two species. The hybrid variety is called New Rice for Africa (NERICA) and it is well noted for it good adaptation in sub-Saharan African environment and soil conditions (Diagne et al., 2011). According to MoFA (2009) & Ragasa et al. (2013) considerable number of varietal research has also been conducted by council for scientific and industrial Research (CSIR) and currently the following rice varieties (Sikamo, Jasmine 85, Digang, NERICA1, NERICA 2, Bouake 189, Marshall, KRC-Baika and ITA 304) are been promoted for wide adoption under suitable ecologies in the short to medium term to address the varietal needs. The knowledge in breeding programs is important to people who are engaged in agricultural planning and development because it forms the basis to improve rice yield together with good management practices. Therefore, this makes it important to capture information on improved rice usage in the study to determine whether it is significant in contributing to efficiency. 2.5 National Policies to Promote Local Rice Production 2.5.1 Trade Policy Trade policies to develop the rice industry in Ghana date back to the 1970’s. The removal of input subsidies as part of the Structural Adjustments Programs in the 1980’s brought about a decline in local rice production because it rendered the producers uncompetitive as the result of high cost associated with production as compared to the rice imported (Kula & Dormon, 2009). Since then, several programs have been developed to boost the industry. In 1982, the government intervened heavily in the rice sector by imposing stiff restrictions on imports to encourage domestic production, tax on imported rice amounted to 36%. Government again made concerted effort to reduce rice imports by 30% in 2001 using Tariffs and quotas as the University of Ghana http://ugspace.ug.edu.gh 13 main trade policy instruments to control rice imports (Hara et al., 2001; & Angelucci et al., 2013). The use of such trade policy measures as a means to reduce rice imports are meant to stabilize domestic rice industries as well as protect and motivate local producers on the assumption that such interventions will make the required impact on domestic rice production to increase output. Tariff on imported rice could also aid in increasing the competitiveness of locally produced rice, however a high tariff without a corresponding increase in the quality of locally produced rice would only increase the price burden on consumers (Boansi, 2013). Food and Agricultural Sector Development Policy (FASDEP II) on a different front, seeks to regulate agricultural imports using a different technique by applying non-tariff barriers through the use of subsidies on agricultural inputs (fertilizer, certified seeds, mechanization, block farm program and national buffer stock company program) for production, storage and exports instead of quotas and import tariffs. This will help to reduce cost of production for farmers to increase their competitiveness at the domestic market. 2.5.2 Agricultural Policies Relevant for Rice Development Policy strategies over the years, as captured in Ministry of Food and Agricultural sector implementing documents (METASIP) have sought to promote rice production to address food security and poverty reduction (MoFA, 2010). The policy objectives of food security and emergency preparedness as well as improved growth in income of MoFA places emphasis on rice as one of the key commodities to be promoted to increased food security and import substitution. Priority intervention measures to achieve this include (i) provision of irrigation infrastructure (ii) research (iii) enhancing access to credit and inputs (iv) improving access to mechanized agriculture and (v) increasing access to extension services. According to Ganidekam (2013) developing countries in general and Ghana in particular has neglected agricultural research for over decades and this has however led to poor yield and University of Ghana http://ugspace.ug.edu.gh 14 reduction in food crop production. This is because inadequate productive research has denied farmers some knowledge that could have been useful in production. Serious attention is been given to rice research and projects because of the impact of global food insecurity and rice has been identified as food security crop (MoFA, 2009; & Ragasa et al., 2013). Rice production according Boansi & Ganidekan (2013) is influenced by price which is considered as one of the agricultural policy instruments and non-price factors. The non price factors include irrigation and rural infrastructure such as road, storage facility and market. The under development of such infrastructure has adverse effect on food price and food security in general. Governments over the years had faced considerable challenges in bridging the infrastructural gap to facilitate agricultural production. Key among them is the irrigation sector development which is recognized as a key in modernizing agriculture (MoFA, 2011). As a solution, National Irrigation Policy was prepared and adopted in 2010. Key among the recommendation is the preparation of a National Irrigation Master Plan as a core strategy to enhance rice production (Dessalegn, 2005). 2.5.3 National Rice Development Strategy (NRDS) The rice development strategy is an outcome of Ghana Government contribution to the vision to double rice production within the period 2008 to 2018 to reduce rice imports. The strategy was developed by taking into consideration the comparative production capacities of the three major rice ecologies (upland, lowland and irrigated) in Ghana as well as growth of rice consumption (MoFA, 2009). The ultimate goal of the strategy is to contribute to national food security, increased income and reduced poverty through rice self-sufficiency which is expected to be achieved by: University of Ghana http://ugspace.ug.edu.gh 15 (i) Increasing domestic rice production by 10% annually using gender sensitive and productivity enhancing approaches for small holders, commercial producers and entrepreneurs along the value chain. (ii) Promoting consumption of domestic rice through quality improvement by targeting domestic markets. (iii) Enhancing capacity of stakeholders to utilize rice by-products, contributing to sound environmental management practices. (iv) Promoting dialogue among rice stakeholders within the value chain towards building efficient information sharing and linkages. According to MoFA (2009) the strategy seeks to increase national rice output by increasing area expansion for three rice growing ecologies through the use of: improved seed, fertilizer, Post-harvest management technology as well as water control and management. Access to mechanization, extension service; research and technology development; farmer-based organizations and credit management for smallholder farmers are other vital service areas to complement existing efforts to achieve higher yield in lowland rice production. Most of the variables found in the strategy to increase production are consistent with efficiency variables found in other empirical studies conducted by Ogundari, (2008); Enwerem & Ohajianya, (2013). 2.6 Performance Measure of Rice Performance measure is guided by development strategies in agriculture (Rahman, 2012). At farm level, rice performance measures can be obtained in two alternate ways by either profitability or efficiency analysis (Loukoianova, 2008). These two performance measures have different approach to measurement. Profitability analysis indicates the ability of a firm to earn substantial profit and return on investment and usually measured as ratios. These ratios provides sound basis to assess the financial status of a firm and how it is managing its assets (Lesáková, 2007). Technical efficiency on the other hand measures the ability to University of Ghana http://ugspace.ug.edu.gh 16 produce the maximum output from a given set of inputs under certain production technology which is usually compared to a standard which may be used on farm-specific estimates of best practice techniques (Herdt & Mandac, 1981). The current study focuses on technical efficiency because one needs to be technically efficient (Fernandez & Nuthall, 2009) before one can be profit efficient. 2.7 Productivity Measurement According to Coelli et al. (1998) and Watkins et al. (2014) productivity is the ratio of output over input. This measure is easy to calculate if a producer uses a single input to produce a single output. But when multiple inputs are used to produce several outputs, the outputs and inputs will have to be combined in a way consistent with economic theory so that productivity remains the ratio of two scalars. Rice production involves the use of several inputs and variability in productivity is not only limited to inputs but other exogenous factors as well. According to Fried et al. (1993), the causes of disparities in productivity has been categorized into production technology, scale of operation, operating efficiency and the environment in which production occurs among different producers. Productivity is useful as a measure of evaluating the performance at producer level using productive efficiency through production frontier, a concept which compares the transformation process of converting input into output. Aigner, Lovell & Schmidt (1977) used stochastic frontier model to measure productivity of a firm. The production function according to their approach is represented as:  ; ( )........................................................................................(2.1)i ii iY f x u v   According to Aigner, Lovell & Schnidt (1977), = productivity (kg/ha) of the firm, = vector of firm inputs, β = parameters estimate. The and represent the technical inefficiency and random error term of the firms respectively. Coelli et al (1998) indicated that University of Ghana http://ugspace.ug.edu.gh 17 the random error term ( ) captures the factors outside the control of the production unit whereas inefficiency term ( ) takes care of factors within the control of the farmer. The two error terms account for the difference between actual output and the potential output known as stochastic element in production or disturbance term. Ajetomobi (2011), measured productivity improvement in ECOWAS rice farming using parametric (stochastic) approach. The productivity measure was based on the estimated parameters of the conventional inputs {land, fertilizer, and seed capital, labour and technology (irrigation)}. The findings of his study revealed that improvement in productivity of ECOWAS countries (Cote d’Ivoire, Ghana, Guinea, Mali, Nigeria and Senegal) was due to technological change which had the greatest impact on productivity. Kuwornu et al. (2012) examined productivity of tomato and watermelon farms in Dangme East District of Ghana using total factor productivity method. Their study measured the total value of tomato and watermelon as well as the value of inputs used for producing each specific crop per hectare and expressed as a ratio to measure productivity. The study found concluded that productivity of watermelon is higher compared to that of tomato and attribute it to difference in output prices, labour and material input cost of the respective commodities. Kang (1997) analysed the productivity of trucking industry using stochastic frontier function in Korea. According to the Kang, the basis for using stochastic frontier was to test for the efficiency of each individual firm against the average firm to establish the technical efficiency changes among then. The findings of the study indicated that technical efficiency increased by 5% for the tested period. University of Ghana http://ugspace.ug.edu.gh 18 The present study adopt stochastic production frontier using Translog to measure the productivity of farm inputs based on its ability to account for stochastic element in production as compared to total factor productivity (Aigner, Lovell & Schmidt (1977). 2.8 Efficiency Measurement Mechanisms Efficiency is the relationship between what a farmer produces and what it could feasibly produce which is based on the assumption of optimizing behavior of the producer subject to constraint (Kokkinou, 2010). According to Fare et al. (1985) & Farrell (1957) efficiency can be measured either from the output side or input side. On the output side, observed output is compared to potential output obtainable from the inputs whereas from the input point of view, observed input levels are compared to minimum potential input required to produce the output. Farrell (1957) categorized efficiency into two component; technical and allocative efficiency. The two components combine to form economic efficiency which is the ability of a firm to obtain maximum output from optimal set of input, given their respective prices. According to (Aigner et al., 1977) technical efficiency refers to the ability to avoid waste by producing as much output as technology will allow or by using as little input as required by technology to produce maximum output. Thus the analysis of technical efficiency can have an output augmenting orientation or an input conserving orientation. The measurement of technical efficiency is important because technical efficiency reduces production costs and makes a farmer more competitive (Alvarez & Arias, 2004). The allocative component refers to the ability to combine inputs in an optimal proportion at prevailing prices (Farrell, 1957). According to Chiona (2011), allocative efficiency measures firms’ ability to choose input combination that reduces cost given the best available technology. This implies that intensifying the use of certain inputs based on their prices will University of Ghana http://ugspace.ug.edu.gh 19 require accurate and timely information about the price movement or trend of that input to reduce or minimizes inefficiencies that may arise from unobserved prices. Efficiency measurements involve the use of frontier to compare actual performance with optimal performance but since the true frontier is unknown, an empirical approximation is used to represent the true frontier called best practice. Approximation of the best practice frontier can be done using parametric or non-parametric techniques. Both techniques put emphasis on optimizing behavior subject to constraints but each of the two alternative approaches have different strengths and weaknesses. The distinguishable elements in these methods lie in the differences on the assumptions made on the following during estimation: a) The functional form of the frontier (parametric or a nonparametric ) b) Whether a random error is included and c) The distributional assumption underlying the error terms 2.8.1 Non-Parametric Approach of Efficiency Measurement Non-parametric frontier approach of measuring efficiency was first introduced by Farrell (1957), which has been extended by Charnes et al. (1978) as well as Fare, Schmidt & Lovell (1986) to include multiple output- input technologies. The approach has traditionally been assimilated into Data Envelopment Analysis (DEA) which is mathematically not demanding as compared to the parametric approach (Kovács, 2010). It builds a linear function from empirical observations of inputs and outputs, without assuming any a priori functional relationship between the inputs and outputs for efficiency measurement. It comes up with a single scalar value as a measure of efficiency. Schmidt & Lovell (1986) argued that technical efficiency results obtained by non-parametric approach might be less precise because non-parametric approach makes less use of information (observations) than the parametric approach. This makes Farrell’s model University of Ghana http://ugspace.ug.edu.gh 20 sensitive to extreme observations and measurement error (Forsund et al., 1980). Another limitation of this approach is that, it is conceptually difficult to separate the effects of uncontrollable environmental variables and measurement error from the effect of differences in farm management (Jaforullah & Whiteman, 1999). In addition to that, Data Envelopment Analysis (DEA) do not permit tests of hypothesis in relation to differences in technical efficiency to be performed statistically as required in scientific study (Schmidt, 1986). Chiona (2011) examined technical and allocative efficiency of smallholder maize farmers in Zambia using non-parametric approach. The study revealed that smallholder maize farmers operate with low levels of technical (15%) and allocative efficiency (12%). This implies that on average, the level of maize inputs used by farmers and cost of operation can be reduced by 85 and 88 percent respectively without reducing output. The significant determinants of economic efficiency of maize farmers were: hybrid seed, farm size and household size, access to extension services and education attainment of household head. Technical inefficiency among farmers was also found to be influenced by: involvement in community agricultural activities, use of organic or chemical fertilizers and livestock ownership. Yusuf & Malomo (2007) also studied technical efficiency of poultry production in Nigeria using non-parametric approach and OLS regression. They identified that most farmers are technically efficient (87.3%) in resources use and farmers with large farm size were the most efficient compared to medium and small farm sizes, years of experience and education according to their findings reduce technical inefficiency. 2.8.2 Parametric Approach of Efficiency Measurement The parametric frontier approach defines the functional form of the efficient frontier and uses either cross sectional data or panel data for technical efficiency analysis (Aigner et al., 1977). The approach imposes restriction on production function but allows for statistical inferences University of Ghana http://ugspace.ug.edu.gh 21 that can be tested as well as different hypotheses on the estimated parameters of the production frontier. Parametric frontier is divided into deterministic and stochastic (Kibirige, 2008). The deterministic production frontier assumes that any deviations (due to unfavorable weather conditions, socio-economic and demographic factors and uncertainties) from the efficient frontier are under the control of the farmer and are considered as inefficiency (Constantin et al., 2009). The principal disadvantages with the deterministic frontiers are that, any measurement errors, approximation errors, specification problems and other sources of variation in output are attributed to inefficiency whereas the stochastic frontier model captures the effect of random shocks to the production frontier (Mastromarco, 2008 and Al- Hassan, 2012). This makes the estimations of technical efficiency sensitive to extreme values (Greene, 2008). Therefore, measures of technical efficiency using deterministic frontiers can produce misleading information. The deterministic frontier model can be represented as;    ; exp , ........... ..............................................................................(2.2)i ii iY f X u  Where Yi (kg/ha) represents the productivity for the i-th farm, with the deterministic part  ;i if X  common to all producers, Xi denote vector of inputs for the i-th farm, βi is unknown parameters to be estimated. The ui is a non-negative random variable representing inefficiency with the following distributional assumption for different specifications such as Half-normal, Truncated, Exponential and Gamma distribution (Songsrirote & Singhapreecha, 2007). STATA 12.0 has an integral statistical package that can be used to estimate half- normal, truncated-normal and exponential models. The choice of distributional assumption with respect to the inefficiency term (ui) is based on computational and theoretical expediency. The truncated normal and gamma models allow for a wider range of observations which is associated with cost and computational complexity (Boahen, 2012). University of Ghana http://ugspace.ug.edu.gh 22 For deterministic frontier, the technical efficiency (TEi) of individual farmers is defined as the ratio of observed output    ; expi ii iY f X u to the corresponding potential output  ;i iiY f X   where there is no inefficiency. Therefore, technical efficiency of the deterministic frontier is represented as: ( ; ) exp( ) exp( )...........................................................................(2.3)( ; ) i i i i i i i Y f X uTE uY f X        These parametric and non-parametric approaches of measuring efficiency are two competing methodologies. No formulation has yet been made to unify the two in a single analytical framework. However, stochastic frontier analysis of the parametric approach is widely used in literature based on its ability to account for factors that are beyond the control of the farmers (Al-Hassan, 2012; Jeffrey & Xu, 1998). Stochastic frontier analysis has been applied in many studies relating to production data in many industries including agricultural sector especially the rice sub-sector. The first application of the stochastic frontier production model to farm level agricultural data was presented by Battese & Corra (1977). It was applied to panel data to model rain-fed rice production in Philippines. Kalirajan (1981) also applied a stochastic frontier using Cobb-Douglas production function for 70 rice farmers in a district in India. Although stochastic Cobb-Douglas production function provides the opportunity for returns to scale to be measured and interpret elasticity coefficients with relative ease, however it is restrictive in terms of return to scale. The defects with deterministic and Cobb- Douglas models provide the basis for stochastic production frontier to be adopted for the present study. The stochastic frontier model is symbolized as:  ; exp( )............................................................................................(2.4)i i i iY f X v u  University of Ghana http://ugspace.ug.edu.gh 23 The definition of the terms in the deterministic model as described earlier is also applicable to the stochastic frontier model except the symmetric random error (vi) term which account for statistical noise in the stochastic frontier. The symmetric random error term is assumed to be identically and independently distributed with zero mean and constant variance i.e. 2~ (0, )i vv N  . Technical efficiency estimation base on this model can be represented as:           ; exp exp ....................................................................(2.5); exp i i ii i i i i i f x v uYTE uY f x v       The difference between the two technical efficiency estimations is embodied in the deviation associated with the inefficiency term (ui). If the inefficiency term is equal to zero (ui=0), producers become technically efficient and production lies on the frontier. On the other hand if the inefficiency term is greater than zero, then the producers become technically inefficient because production lies below the frontier. The inefficiency term (ui) associated with stochastic frontier is always smaller as compared to the deterministic frontier model because the stochastic frontier accounts for statistical noise (Battese & Corra, 1977). 2.9 Production Risk Measurement Production risk in agriculture occurs as the results of uncertainties about the performance of crop due to factors such as weather, pest and diseases (Akinbogun, 2010). The choice of inputs by farmers according to Bokusheva & Hockmann (2006) and Villano & Fleming (2004) also affect output variance. Their study showed that the effect of inputs on production could be risk increasing, risk decreasing or risk neutral depending on their marginal effect which is consistent with Just & Pope (1978) framework. According to Kumbhakar (2002) if an input influences output positively, that input is expected to influence output variance positively and vice versa. Just & Pope risk analysis in 1978 indicated that production function has separate effect on inputs, mean output and the variance of output (output risk) therefore, University of Ghana http://ugspace.ug.edu.gh 24 there should be no a priori restrictions on the risk effects of inputs. This imply that marginal production risk of inputs could take on positive, zero or negative values. In other words, the production function should be general enough to accommodate both increasing and decreasing output risk from inputs (Bokusheva & Hockmann, 2006). Just & Pope (1978) analysed production risk by separating production process into mean output and output risk components. The mean output specifies the effects of inputs on output whereas output risk embodied the effect of input on the variance of the output. This is represented as:    ; ; .......................................................................................................(2.6)i i i iY f x g x v   Where  ;if x  is the mean output function,  ;ig x  is the output risk function, ix are the input variables, and   are the parameters estimates in the mean output and output risk function respectively and iv is the random noise effect in the model. Given the mean and variance of output for the i-th farmer, and the values of the inputs and the technical inefficiency effect ( the mean output function is represented as    , ;i i iE Y x u f x  (Bokusheva & Hockmann, 2006). Given the variance of the output from productivity estimate (Kumbhakar, 2002) risk function is defined as:    2, ;i i iVar Y x u g x  The marginal production risk with respect to the i-th input is defined to be the partial derivative of the variance of production with respect to .      , 2 ; ; 0...........................................................................(2.7)i i j j Var Y x u g x g x x       University of Ghana http://ugspace.ug.edu.gh 25 The marginal risk can be positive as well as negative or zero, depending on the signs of  ; ,jg x  and  ,jg x  , where the latter is the partial derivative of the  ;jg x  with respect to inputs. A positive marginal risk means that the input has an increasing effect on output risk and a negative value means that the input has a decreasing effect on the output risk. A zero marginal risk means that the input is a risk neutral (Just & Pope, 1978). Akinbogun (2010) modeled technical efficiency and production risk in fish farms in Nigeria using stochastic production frontier with flexible risk properties. The measure of risk in his study was output variance and inputs such as fertilizer and fish feed were estimated to be risk increasing whereas labour was risk reducing input. Villano & Fleming (2004) investigated production risk and technical efficiency of rain-fed lowland rice farms in Philippines. Their study regressed the variance of rice output on input variables to obtain their marginal risk effects; fertilizer and labour were identified to be risk increasing inputs whereas herbicide was reducing. The present study adopted their approach to measure risk in lowland rice farms in Ashanti region using rice output variance as a measure of production risk. 2.9.1 Incorporation of Production Risk in the Stochastic Frontier Model Incorporation of production risk in a stochastic frontier model is done to account for technical efficiency estimation biasness and this can be done in two ways. According to Battese et al. (1997) and Just & Pope (1978), the disturbance term or stochastic element ( in production process is multiplied by production risk function i.e. to generate equation (2.8). ) The second approach of incorporating risk is the flexible functional form suggested by Kumbhakar (2002), where an additional function is introduced into equation (9) to account for technical inefficiency. University of Ghana http://ugspace.ug.edu.gh 26 The parameters of mean production, inefficiency and output risk are represented by  ,  and  respectively. Following equation (10) in line with equation (11) proposed by Battesse & Coelli (1998) and given the values of the inputs ( ) and its technical inefficiency effect ( ) to the corresponding mean production i.e. E ( ), if there was no technical inefficiency ( ) of production (Battese & Coelli, 1988), technical efficiency can be estimated by: Where technical inefficiency (TI) is defined as potential output loss and represented as: Therefore technical efficiency of the i-th farmer is predicted by: 1 .......................................................................................(2.12)iiTE TI   In the absence of production risk { } in the technical inefficiency model (equation 2.11), the technical inefficiency estimate ( ) will be small which will lead to bias estimates of the individual technical efficiencies ( ) of the firm (Songsrirote & Singhapreecha, 2007). 2.9.2 Review of Factors Influencing Production Risk and Technical Efficiency The presence of risk does not only influence production output but also producers’ behaviour, primarily with regard to input use. Risk mitigation measures by farmers play a critical role in decision-making which in turn contribute to technical inefficiency. Therefore, technical efficiency assessment by considering producer's response to uncertainty (risk) is important in an area of production where risk on input-use decisions is concerned. University of Ghana http://ugspace.ug.edu.gh 27 Few empirical studies have attempted to analyse production risk and technical efficiency in a single framework. Kumbhakar (1993) used a method to estimate production risk and technical efficiency using a flexible production function to represent the production technology. The model was adopted by Villano & Fleming (2004), to analyzed technical efficiency in a rain-fed lowland rice environment in Central Philippines using stochastic production frontier. The risk increasing variables in the study were identified to be area of land, fertilizer and labour whereas herbicide was risk decreasing. This implies that farm size, fertilizer and labour have the potential to increase the variance of rice output. The mean technical efficiency was found to be 79%. Bokusheva et al. (2006) conducted empirical study to identify the possible causes of production volatility that characterized Russian agriculture from 1995 to 2001 using stochastic frontier approach to analyse production risk and technical inefficiency. A total of 447 agricultural enterprises from three regions in Central, Southern and Volga Russia were selected for the study using panel data. The following key input variables (labour, seed, fertilizer and depreciation of capital assets) were used as a function of production, production risk and technical inefficiency in the production process. However, all the variables were found to be risk increasing. This implies that most input factors enhance farms production volatility. Rola & Quintana-Alejandrino (1993) used a Cobb-Douglas stochastic production frontier to estimate the technical efficiency of rice farmers in selected regions of the Philippines. The estimated mean technical efficiencies of 0.72, 0.65 and 0.57 for irrigated, rain-fed and upland environments were respectively found for the regions. Education and access to capital were the common factors that affected the levels of technical efficiency of farmers in the three different environments. University of Ghana http://ugspace.ug.edu.gh 28 Balde et al. (2014) used stochastic frontier approach to study the technical efficiency of mangrove rice production in Guinea using 140 rice farmers belonging to three different groups (traditional mangrove, improved mangrove and salt marsh). The analysis indicated that farm size and depreciation cost of farm tools contributed significantly to enhance the mangrove rice production. The farm-specific variables used to explain inefficiencies indicate that elderly mangrove rice farmers with farming experience, significant household size and access to off-farm income and remittance tend to be more efficient. However, education level, improved seed, extension services provided by the government and access to credit had a negative influence on technical inefficiency in the mangrove rice farming system. In general, the study revealed a wide variation in the level of technical efficiency of mangrove rice farmers with an average of 23%. This implies that the level of inefficiency associated with mangrove rice production was 77%. Tuffour (2012) examined profit efficiency and its determinants in broiler production in greater Accra using Cobb Douglas profit function by taken into consideration production risk. The results indicated that price of labour reduced profit but the price of day old chick increased profit. The study also revealed that broiler producers were able to achieve on the average 54% of their frontier profit and experience in broiler production was found to reduce inefficiency. Al-Hassan (2012) conducted a study on technical efficiency analysis of smallholder paddy rice farms in Ghana based on different farming systems (irrigators and non-irrigators) and gender using stochastic frontier analysis. The results indicated that irrigator farmers were more technically efficient compared to the non-irrigators but in general, the smallholder farms were found to be technically inefficient because they obtain 34% of their rice output. Among the variables used for the study, credit, education, family size and non-farm activity significantly influenced technical efficiency of smallholder farmers. University of Ghana http://ugspace.ug.edu.gh 29 Yiadom-Boakye et al. (2013) examined the differences in resource use and technical efficiency between male and female rice farmers in the Ashanti region using stochastic production frontier. The study identified that female rice farmers were inefficient compared to their male counterpart. The mean technical efficiency of the rice farmers in general was found to be 24%. Farm inputs such as farm size, labour, fertilizer and seed were identified to influence rice production in the study area. On the other hand, technical inefficiency was found to be reduced by credit, family size and off-farm. Although Yiadom-Boakye et al. (2013) and Al-hassan (2012) estimated technical efficiency and examined the factors contributing to rice production inefficiency, theirs studies failed to incorporate production risk which according to Songsrirote & Singhapreecha (2007) lead to upward efficiency biasness. 2.9.3 Description of Technical Inefficiency Variables used in the Study The explanations for the farm specific variables within the inefficiency model are outlined as follows: Gender ( ): This variables was dummied as one if the farmer is a male and zero otherwise. In Ghana, production activities that are labour intensive are associated with men whereas agro-processing activities are noted to be women domineering. Technical efficiency study by Al-Hassan (2012) in the north revealed that male rice farmers are technically efficient compared to their female counterparts. Gender in the current study was included to determine who is efficient in rice production in terms of gender in the study area under the assumption that both male and female have equal access to farm inputs. Farm distance ( ): This was measured as the number of kilometres the farmer travels from house to the farm. Distance from a farmer’s residence to his farm is included in order to determine whether the frequency of farmers visit to the farm has a significant impact on the University of Ghana http://ugspace.ug.edu.gh 30 value of output produced by the farmer. A study by Kuwornu et al. (2012) on productivity and resource-use efficiency in tomato and watermelon farms in Ghana indicated that the shorter the farm distance, the more frequent and efficient it is for the farmer to visit the farm. This is because it reduces the farmer’s time and energy spent in travelling to conduct farm operations. Experience ( ): This variable has been included to determine whether the number of years of rice farming has the potential to reduce inefficiency. According to Yiadom-Boakye et al. (2013), the frequency of farming allows the famer to specialize and avoids repeated mistakes in the area of soil and crop management that are likely to contribute to inefficiency. Therefore, this variable is expected to relate negatively to inefficiency to increase lowland rice productivity. Credit ( ): To capture the effect of credit on the efficiency of farmers, it was measured as a dummy variable with the value one if the farmer had access to credit for the 2014 major season for rice production period and zero otherwise. Agricultural production in general is capital intensive and the availability of credit as an institutional variable will assist the rice farmers to get inputs on a timely basis for production. According to Dessalegn (2005), a significant increase in agricultural credit to farmers increases their demand for labour to increase productivity. Therefore, credit is important and is expected to increase the efficiency of the farmers by relating negatively to technical inefficiency. Agricultural Mechanization ( ): It refers to the replacement of manual labour and simple hand tools with machinery for production. According to Abdulquadri & Mohammed (2012), traditional method of farming holds farmers investment in land, labour and capital below a profitable margin whereas mechanization increases productivity and farmers profit. The lowland rice farmer access to agricultural mechanization services is expected to reduce University of Ghana http://ugspace.ug.edu.gh 31 inefficiency in performing farm activities such as rice harvesting and milling to increase output. Agricultural mechanization was dummied, one if the farmer had access to mechanization services and zero if otherwise. Farmer access to extension services ( ): Extension officers play significant role in disseminating new technologies and technical advice to farmers that enhance technology adoption to improve productivity (Chiona, 2011). This is because farmers who have access to extension services are expected to respond fast to new technologies and appreciate best farm management practices such as timely planting and weeding, correct quantity of fertilizer and seed for application. This minimizes their inefficiency levels and impact positively on productivity. Education level of the farmer ( ): Education is an important tool to sharpen the managerial competences of rice farmers (Abedullah et al., 2007). This is because educated farmers are expected to access information easily to make an informed decision about the cost and the necessary combination of farm inputs for production, thus improving the efficient use of inputs resource. It was measured as dummy, one if the farmer highest level of schooling is Basic education and zero otherwise. This was repeated for secondary and no education levels. Canal maintenance ( ): In lowland rice farming, canal maintenance can have important consequences on water availability and flooding condition. Low or high rainfall often results in field conditions that are too dry or too wet respectively. Rice is extremely sensitive to prolong water shortage or flood. When soil water content drops below saturation, growth and yield are severely affected. Lowland rice is adapted to water logging. However, complete submergence can also be harmful. Most lowland rice varieties can survive complete submergence for only few days. The unpredictable nature of the rain makes it important to University of Ghana http://ugspace.ug.edu.gh 32 take precautionary measures to improve technical efficiency by maintaining canals to regulate water inflow and outflow to increase paddy rice yield (Sakurai, 2006). Improved seed-use ( ): The type of rice seed planted by the lowland farmers determines their level of productivity holding other factors constant, improved seeds have high yield potential and are able to resist harsh environmental conditions such as pest and diseases as compared to local variety (Kibirige, 2008). The improved seed variety used by the farmers was dummied as one if the farmer use improved seeds and zero if otherwise. Planting method ( ): Row planting enhances efficiency by facilitating farm operations such weeding, spraying and harvesting. It provides quick access to perform task as compared to the broadcasting method where no space is created to be used as a walk way to carry other activities out thereby creating a room for inefficiency. The variable was measured as dummy one if the farmer uses row planting method and zero if otherwise. Farmer based organizations (FBOs) ( ): According to Addai et al. (2014), farmers membership to FBO helps to improve on rural service delivery particularly in the areas where there is institutional failure in the public or private sectors. For many donor and NGO projects, a famer membership to FBO guarantees support from a project, with no consideration given to farmers who do not belong to such groups (Tinsley, 2004). The rice farmers membership of FBO related to rice was measured as dummy, one if the farmer belong to FBO and zero otherwise. 2.10 Conclusion The existing literatures on rice production and related efficiency studies were reviewed to identify the production risk and inefficiency variables for the current study. These were classified into conventional and non-conventional factors. Non-conventional factors capture University of Ghana http://ugspace.ug.edu.gh 33 the impacts of macroeconomic variables such as investment in research and irrigation infrastructure for rice production. Conventional factors are traditional input variables in the farmers’ production decision process such as: seed, fertilizer, herbicide, labour and capital assets. However, variables found to reduce technical inefficiency for the literature reviewed are; improved seed, credit, education, family size, non-farm activity, extension service and experience. For each of these studies reviewed, Cobb-Douglas and Translog models were the predominant methods employed in estimating productivity using parametric approach (stochastic production frontier) compared to non-parametric (Data Envelopment Analysis). Therefore the present study adopt Translog model to estimate for productivity based on it wide application in empirical research. The lapses identified in the domestic efficiency studies related to rice were bridged in the current study by incorporating production risk to account for efficiency estimation biasness in analyzing the productivity of lowland rice farms in the study area. University of Ghana http://ugspace.ug.edu.gh 34 CHAPTER THREE METHODOLOGY 3.1 Introduction Section 3.2 of this chapter describes the conceptual framework of the study. Section 3.3 presents the theoretical framework. Finally, the method of analysis for address the specific objectives as well as the overview of the study area are presented by section 3.4 and 3.5 respectively. 3.2 Conceptual Framework The conceptual framework provides a summary structure of the study based on literature and personal experience drawing linkages on how production risk, mean output and technical inefficiency affect lowland rice output in Ashanti region. The concept of the study is consistent with the production function stated by Kumbhakar (2002) who used a single model to represent production process that comprised of; mean output, production risk and technical efficiency. Production risk as indicated in Figure 3.1 is influenced by both external and internal factors (farm inputs) associated with the production environment (Bokusheva et al., 2006; Villano & Fleming, 2004 & Hasanthika et al., 2013). The external factors contributing to production risk takes into account the lowland rice farmers exposure to erratic rainfall (climatic condition), agricultural policies, market environment, pest and diseases. Erratic rainfall according to Samal and Pandey (2005) brings about flood and drought that affects production process by inhibiting the potency of farm input usage and chemical application for rice cultivation. The intensity of flood and submergence of rice affect harvest and contribute to production losses which increase output variation among farmers. University of Ghana http://ugspace.ug.edu.gh 35 According to Ganidekam (2013), Agricultural policies in Ghana over the past years has place more emphasis on export crops (cocoa, oil palm, cotton and timber) since they generate more revenue for the state, given little attention to other policies to improve on the production of food security crops such as rice and maize and it market environment. Such policies defects have affected rice production and have contributed to its output variation. The internal factors contribute to production risk when lowland rice farmers decision to used farm inputs in the study area is affected by technical inefficiency as depicted in the Figure 3.1. When the lowland rice farmer decision to use farm inputs is not influence by production inefficiency, a farmer can subdue the effects of production risk with some inputs to reduce output variation. A typical example is that agrochemicals (weedicide and pesticide) can be used to control the risk posed by some external elements (weeds, pest and diseases) on output. Different production environment creates different atmosphere for pest and weed population. The farmer managerial ability determines the amounts and types of agro- chemicals that can be used to combat weeds, pest and disease infestation. In modern rice production systems relatively few weedicides are used because improved rice variety used by farmers and ponding water are effective weed control measures (Nwilene et al., 2011). The technical inefficiency of lowland rice farmers are influenced by socio-economic, institutional and managerial factors according to studies (Korwunu, et al., 2012; Dessalegn, 2005; Chiona, 2011 and Abdulquadri & Mohammed, 2012) which interact to affect lowland rice production as indicated in Figure 3.1. When lowland rice farmers are able to combine the right proportion of farm inputs given a production technology and overcome their technical inefficiencies and production risk, rice output in the study area will increase and output variation will reduce among farmers. University of Ghana http://ugspace.ug.edu.gh 36 Figure 3. 1: Conceptual Framework of the Study Source: Author’s construct 3.3 Theoretical Framework Production frontier can be estimated using either parametric or non-parametric frontier approach. The choice between these techniques depends on the underlying reasons for estimating productive efficiency (Padilla-Fernandez & Nuthall, 2009). According to Onumah et al. (2010), the parametric approach using stochastic production frontier incorporates random error (factors beyond farmers’ control) as compared to non-parametric method that assumes a deterministic frontier which does not account for deviation outside the control of the farmers. The stochastic frontier has been used in a study conducted by Jeffrey & Xu Rice Output Production risk Technical Inefficiency Farm inputs  Seed  Fertilizer  Agro-chemical  Family Labour  Hired Labour Socioeconomic variables  Gender  Farm distance  Experience Institutional factors  Access to credit  Access to Extension  Access to mechanization Managerial variables  Improve seed-use  Education  Planting method  Water canal management  FBO External factors  Climatic condition  Pest and disease  Policies  Market environment Production function University of Ghana http://ugspace.ug.edu.gh 37 (1998); Constantin et al. (2009); Al-Hassan (2012) & Ligeon et al. (2013) based on it comparative advantage to the deterministic frontier. The Stochastic production frontier is selected for the study due to it corrective ability to account for random error. An estimated production function indicates the average levels of outputs that can be produced from a given level of inputs. One of the basic assumptions in the estimation of a production function of lowland rice farms in the study area is that all farmers have homogeneous production function and are technically efficient which allow them to be on the production frontier. Another assumption underlying the specification of a production frontier is that farmers engaged in production using best management practices which allow them to obtain the maximum potential output for a given set of inputs. Based on that, any level of output below the maximum output is attributed to production risk and technical inefficiency (Bokushev et al., 2006). The mean output function of lowland rice production is represented as: The error term ( ) in the above equation can be decomposed into two; ui and vi; and these two terms are assumed to be identical and independently distributed from each other (Songsrirote & Singhapreecha, 2007). The is a standard random variable which captures the random variation in output due to statistical noise that arises from (a) unintended omission of relevant variables (b) measurement errors and approximation errors associated with the choice of the functional form; (c) unexpected stochastic changes in production and (d) other factors that are not under the control of the farmer (Al-Hassan, 2012). The on the other hand account for variation in output due inefficiency on the part of rice farmers.  ; ...........................................................................................................(3.1)i iiY f x    University of Ghana http://ugspace.ug.edu.gh 38 Risk in production process is influence by factor inputs and exogenous factors. Unlike the exogenous factors, the factor inputs can be controlled to increase farm efficiency. However, there are risks associated with resource adjustments (Ligeon et al., 2013). The study by Just & Pope (1978) provides an important outline for production risk analysis. The Just & Pope stochastic production function consist of two components and it is represented as: Where  ;if x  is the deterministic component,  ;i ig x v is the risk component. According to Just & Pope in estimating for risk, the marginal production risk function should be liberal enough to accommodate both increasing and decreasing output risk from inputs. Given that the values of inputs and inefficiency effects ( ) the mean output function of the i-th farmer is given by E , and the variance of output  20( ) ;u iVar Y g x   , the marginal production risk which measures the effect of inputs on output is given by:      2 ; ; 0...................................................................................(3.3)i i i j u oVar Y g x g x x       The defect of Just and Pope approach of measuring risk is that it failed to capture the effect of inefficiency term on mean output. To address this defect, Kumbhakar (2002) came up with flexible functional form that has the three components {mean output  ; iif x  , production risk ( ; )i i ig x v and inefficiency ( ; )i i iq x u } to represent production process.  ; ( ; ) ( ; ) ..................................................................................(3.4)i ii i i i i iiY f x g x v q x u     Where i is the estimated inefficiency parameter. Maximum likelihood estimate of the total output variation due to technical inefficiency (λ) is parameterized as 2 2u v   0. University of Ghana http://ugspace.ug.edu.gh 39 The estimation of lambda (  ) is necessary because it indicates whether variability associated with rice output in production process is coming from the variance of technical inefficiency ( ) or random error ( ). If the value of lambda is equal to zero ( ) it means that technical inefficiency is absent and deviations of the observed output from the frontier output is entirely due to pure noise effect. On the other hand if then technical inefficiency is present in the data and deviations from the frontier output is as a result of technical inefficiency and pure noise. Based on the marginal production risk function, the expected outcomes of the study are that: if the estimated coefficient of the input variable for rice production is negative then the input is risk decreasing whereas positive coefficient implies risk increasing input. On the other hand, if the coefficient is zero than the input is risk neutral. 3.4 Method of Analysis 3.6.1 Estimation of Rice Productivity Function As a first in rice productivity estimation, rice output and farm inputs were standardized using the area of the rice farm which was then normalized by dividing both the inputs and output by their respective sample means before it was log transformed. Hypothesis was formulated and tested using the generalized likelihood ratio statistic to identify the model specification for 2014 cross sectional data. Based on the maximum likelihood ratio test, Translog model was selected for discussion as against it corresponding Cobb-Douglas production functions. The Translog model has been used in similar studies by Sauer et al. (2006); Balde et al. (2014) and Renato & Fleming, (2004). The estimated Translog model is represented as:  5 5 5*0 0 1 1 1 1 ......................................................... 3.52i i j i ji i ji i jInY D InX InX InX          University of Ghana http://ugspace.ug.edu.gh 40 Where = productivity (kg/ha), ln = natural logarithm, = parameter estimates, = seed (ka/ha), = fertilizer (kg/ha), = Agro-chemical (Litre/ha), =Family labour (Man- days/ha), = Hired labour (Man-days/ha) = random errors from the stochastic frontier production and = vector of non-negative technical inefficiency component of the error term. To deal with zero cases in to avoid the biasness of the estimates, dummy variable was introduced for fertilizer which has value of one if fertiliser was applied and zero if otherwise to account for change in intercept. This data collection defect with regard to fertilizer and corresponding corrective procedure has been used by Battese (1996) and Onumah et al. (2009) in their respective research. According to Sauer et al. (2006), when the output and input variables have been normalized by their respective means, the first-order coefficient for tranlog can be interpreted as elasticity of output with respect to the different inputs i.e. ( ). The sum total of the output elasticities is the estimated scale elasticity (ε) which is defined as the percentage change in output from 1% change of all input factors and it measures return to scale for the lowland rice farm. The decision rule for return to scale when (ε) >1 implies increasing return to scale, (ε) <1 implies decreasing return to scale and (ε) = 0, indicates constant return to scale. 3.6.2 Estimation of production risk To estimate the production risk of lowland rice farms, the deviation from the input variables was obtained from productivity estimate and squared to get the variance of rice output. The variance was then regressed on the inputs variables. The risk function is presented as follows:    50 1 ; ............................................................................................... 3.6i ji i i V g X InX       University of Ghana http://ugspace.ug.edu.gh 41 Partial derivative of the variance of risk function  ;ig X  with respect to the input variables ( ) was taken to determine the risk nature of the input variables. The marginal production risk function is represented as: The decision rule is that, if the estimated risk parameter for the input variable is negative then it is risk increasing, if positive then is risk increasing and if zero then is risk neutral. The risk reducing farm inputs reduces output variation whereas risk increasing inputs increase rice output variation of the farmers. The variables within the risk function are defined as: =Output, j = estimated risk parameters, = seed (kg/ha), = fertilizer (kg/ha), = Agro-chemicals (litre/ha) and = Family Labour (Man-days/ha) =Hired Labour (Man-days/ha). 3.6.3 Estimation of technical efficiency levels of lowland rice farmers The technical efficiency was estimated using the output orientation of the production frontier associated with the sampled lowland rice farms in the study area. This is represented as: where technical efficiency ( iTE ) relative to stochastic production frontier is given by * exp( )ii i i YTE uY   . The difference between the actual output ( ) and the potential output ( ) among the rice farmers is captured by the one-sided error component e x p ( )iu whereas other factors beyond the rice farmers control are embodied in . The definition of the terms as well as the statistical assumptions underlying the Ui and Vi are applicable to earlier definitions within the study. According to Jondrow et al. (1982); Battese & Coelli (1988), when , then the expected value of conditioned on the error term becomes . When the conditional estimate of is obtained from the deviation of the rice output, then the technical efficiency level of the individual rice farmers University of Ghana http://ugspace.ug.edu.gh 42 for the current study was predicted using STATA based on the conditional expectation of which is represented as The estimated mean technical efficiency for the rice farmers lied between zero and one because the farm had inefficiency index greater than zero ( . 3.6.4 Determinants of technical inefficiency in lowland rice production The inefficiency effect ( ) obtained from the productivity estimate was regressed on the farm specific variables hypothesized by literature to influence technical inefficiency. These farm specific variables are represented in the model as:    120 1 ; ....................................................................................................... 3.7ii i q z z       Where is the intercept, and z’s are the farm specific variables that accounts for technical inefficiency in the study and have been classified into socio-economic (gender, farm distance and experience), institutional (credit, mechanization and extension) and managerial variables (Education level, canal maintenance, improved seed used and planting method). The δ’s are the estimated technical inefficiency parameters. University of Ghana http://ugspace.ug.edu.gh 43 Table 3. 1 Technical Inefficiency Variables Influencing Lowland Rice Production Variables Description Measurement Apriori Exp. Socio-economic Profile Gender Dummy (1=male, 0=Female) +/- Farm distance Kilometres - Experience Years - Institutional factors Credit Dummy (1=yes, 0= No) - Mechanization Dummy (1=yes, 0= No) - Extension access Dummy (1=yes, 0= No) - Managerial factors Education Dummy 1=Basic Education., 0=otherwise 1=Secondary Educ., 0=otherwise - Canal maintenance Dummy (1=yes, 0= No) - Improved seed used Dummy (1=yes, 0= No) - Row Planting Dummy (1=row planting, 0= broadcasting) - FBO Dummy (1=yes, 0= No) -/+ 3.6 Hypothesis of the Study The following hypotheses were formulated to determine the potency of the translog model adopted for the study and assumptions made under the parameter estimates to determine whether production risk in inputs and technical inefficiency significantly explain output variability. The null hypotheses was tested using the generalized likelihood ratio statistic i.e. evaluated at the restricted ( RL ) and the unrestricted ( UL ). The likelihood ratio test statistic has chi-square distribution ( 2 ) with degree of freedom ( )J equal to the number of parameters specified under the null hypothesis. University of Ghana http://ugspace.ug.edu.gh 44 Table 3. 2: Test for Statistical Assumptions of the Stochastic Frontier (hypothesis) Null hypothesis Description Cobb Douglas model is an adequate representation of the data. Cobb Douglas model is not and adequate representation of the data. Output variability is explained by production risk in inputs. Output variability is not explained by production risk in inputs. Coefficients of the explanatory variables do not explain technical inefficiency (truncated normal distribution). Coefficients of the explanatory variables do explain technical inefficiency (truncated normal distribution). 4 The inefficiency effects are absence. The inefficiency effects are not absence. 3.7 Data Source and Collection The study was conducted using cross-sectional data which was collected at the farm level from the farmers in Ashanti region by selected enumerators who were extension officers. The questionnaires were pre-tested to correct mistakes, evaluate the importance of a given data, add relevant information and exclude irrelevant information to improve the standard of the data in line with the objectives of the study. These enumerators had both good experience in terms of relaying the relevant information and ability to extract the required information form the respondents. Both open and close ended questionnaires designed to address the specific objectives were used. The data described rice producer and production characteristics under socio-economic, institutional and managerial variables. Lowland rice is grown following the two rain patterns received in a year for the region. The first season covers the period of March- June and the second from August – December (main season). Because this crop requires a considerable amount of rainfall, most farmers in the study area produced rice in the main season of August – December since it has a longer rainy period. Both quantitative and qualitative data were collected for the last farming year (2014). University of Ghana http://ugspace.ug.edu.gh 45 3.8 Sample Size and Sampling Technique The cross sectional data used in the study was obtained through a field survey across 5 districts in the Ashanti region. A multi-stage sampling technique was used for the study. The selection of the districts as well as the communities was carried out using systematic random sampling procedure. In the first stage, 5 districts as seen from Table 3.2 were selected from a list of 10 rice producing districts in the region. This was followed by random selection of two communities under each of the districts. A total of 200 rice producing households were sampled and approximately 20 respondents from each community were randomly selected. However, one questionnaire was completely discarded during the data cleaning process due to data defect beyond control whereas eight others were treated as outliers for the purpose of data quality and so they were not included in the regression analysis. Table 3. 3: Selected Lowland Rice Districts and Communities in Ashanti Region No Name of District Name of Communities Number of respondents 1 Ahafo Ano North Tepa and Kyekyewere 41 2 Ejura Sekyeredumasi Aframso and Kobriti 40 3 Adansi south Asarekrom and Akutreso 40 4 Asante Akim North Akutuoase and Wraponso 48 5 Asante Akim central Ohene Nkwanta and Atonsu 41 Total 200 3.9 Geographical Setting of the Study Area Ashanti region occupies a total land area of 24,389 km² and forms 10.2 % of the total land area in Ghana (MoFA, 2012). The region has a total rural farm population of 2,201,405 people and lies between longitudes 0.15W and 2.25W, and latitudes 5.50N and 7.46N. The region is surrounded by Brong-Ahafo in the North, Eastern region in the East, Central region in the South and Western region in the South. About 55.5 percent of the economically active University of Ghana http://ugspace.ug.edu.gh 46 population engaged in agriculture except fishing. Using Figure 3.2, the focus of the study was centred on the following districts in Ashanti region: Ahafo Ano North, Ejura Sekyeredumasi, Adansi south, Asante Akim North and Asante Akim central which were selected randomly. The study was carried out in Ashanti region because the region is considered as one of the emerging lowland rice growing areas in Ghana which has vast area of untapped lowlands that needs to be exploited for rice cultivation (Hara et al., 2001). In addition, the location of the region makes it a strategic hub for business activities compared to other lowland rice producing areas which also creates market potential for rice output. Figure 3. 2: District Map of Ashanti Region Source: Department of Geography, University of Ghana University of Ghana http://ugspace.ug.edu.gh 47 CHAPTER FOUR RESULTS AND DISCUSSIONS 4.1 Introductions This chapter focuses on the presentation and discussion of field results to address the specific objectives of the study. Section 4.2 and 4.3 provides description of the socio-economic characteristics of the respondents and productivity estimates of lowland rice farms respectively. Section 4.4 describes the economic variables in the stochastic production frontier whereas 4.5 outline the results of tested hypotheses of the study. The stochastic frontier model with Translog and risk estimates are presented by section 4.6 and 4.7 respectively. Finally, section 4.8 and 4.9 describes the estimates of technical efficiency and inefficiency of lowland rice farms. 4.2 Descriptive Statistics of Socio-economic Characteristics of the Respondents The average age and experience of the respondents are about 40 years and 12 years respectively as presented in Table 4.1. This indicates that farmers were moderately matured and experienced to take informed decisions to improve their performance. About 66% of the rice farmers interviewed were males which indicate that lowland rice production is a male dominant activity in the study area. Considerable number of lowland rice farmers also had access to extension officers (67%) for technical advice in broad areas of farming activities but the most prominent services received from the extension officers were planting (56), chemical application process (6%), land preparation (5.5%), refilling of seedlings (5%), weeding(2.5%) and credit application process (1.5%) etc. These farmers on average travel 3km to carry out their farm activities. According to Kuwornu et al. (2012) and Masaku & Xaba (2013), long distance from a farmer’s residence to the farm has negative impact on University of Ghana http://ugspace.ug.edu.gh 48 productivity as it creates room for inefficiency. Short distance reduces transportation costs, tracking time and transportation damages to agricultural produce. Education in this study represents the maximum level of formal schooling attained by each rice farmer. About 48% of the rice farmers had no education background, 45% had basic education whereas 7% had secondary education. To ascertain the importance of education level required to improve rice productivity, both basic and secondary education levels were captured and included in the inefficiency model. According to Schutz (1976) cited in Antle et al. (1990), for a modernized agricultural system, the ability to read and use modern inputs makes education important whereas in static traditional agriculture, experience is relatively more valuable asset for farm management than education. Another point of interest is the effect of institutional factors on agricultural development. Agricultural mechanization and credit for farming plays complementary role to increase productivity by improving the timeliness and efficiency of farm operations (Abdulquadri & Mohammed, 2012). Unfortunately, the availability of credit to the Ghanaian agricultural sector has over the years remained inadequate and where they are available, their releases are untimely due to high default rate (MoFA, 2011). For the 2014 major rice farming season, farmers who had access to credit and agricultural mechanization for the purpose of rice farming were 16% and 25 % respectively. About 41% of the sampled rice farmers engage in non-farm activities out of which salary workers and traders constitutes 4% and 27% respectively whereas about 8% participated in other unclassified non-farm activities. According to the farmers, the risky nature of agricultural activities due to erratic rainfall, price fluctuation, pest and diseases allows them to diversify their resources by engaging in non-farm activities as a means of survival in an event of unforeseen crop failure. Non-farm activities also provide farmers extra capital to pre-finance their operations (Kuwornu et al., 2012). University of Ghana http://ugspace.ug.edu.gh 49 Lowland rice is inundated in water for most of it growing periods except when it is matured and ready to be harvested. However, due to the excessive depth of rain water in the major season, regulating the water by canal is essential for the healthy growth. Given the importance of canal management, the study found out that 50% and 61% of the respondents did canal maintenance and row planting respectively whereas 30% did both row planting and canal maintenance. According to the farmers, row planting creates walk-way on the paddy field that facilitates farm operations and also ensures the optimum plant population and yield. Table 4. 1: Summary of Socio-Economic Characteristic of the Respondents Variable Unit Observation Minimum Mean Maximum Std. Dev. Age Years 199 21 40.06 76 11.54 Experience Years 199 1 11.86 50 9.37 Farm distance Km 199 0 2.51 10 2.18 Gender Dummy 199 0 0.66 1 0.48 Non-farm activity Dummy 199 0 0.41 1 0.49 Credit Dummy 199 0 0.16 1 0.37 Mechanization Dummy 199 0 0.25 1 0.43 Extension agent Dummy 199 0 0.67 1 0.47 No Education Dummy 199 0 0.48 1 0.50 Basic Education Dummy 199 0 0.45 1 0.50 Sec. Education Dummy 199 0 0.07 1 0.26 Canal maintenance Dummy 199 0 0.50 1 0.50 Improved seed Dummy 199 0 0.55 1 0.50 Row Planting Dummy 199 0 0.61 1 0.49 Source: Authors Computation (2015) University of Ghana http://ugspace.ug.edu.gh 50 4.3 The Productivity Estimates of Lowland Rice Farms The productivity of lowland rice farms based on the results of the study are presented in Table 4.2. The average lowland rice productivity for farms sampled was 3.7 Mt/ha as against the national average of 2.5 Mt/ha (MoFA, 2012). Approximately 33.7% of the rice farms sampled produced below the national average whereas 66.3% of the lowland rice farms were above. Farms that produced below the average yield (3.7Mt/ha) in the study area constituted 56% whereas those above were 29%. The highest percentage of sampled farms (22%) produced within the range of 2.6 - 3.5Mt/ha as seen in Table 4.2. Productivity varied greatly among the rice farmers below and above the range of 2.6 - 3.5Mt/ha. Approximately 85% of the rice farms produced below the national potential yield of 6.5Mt/ha whereas 15% produced above it. Farms producing lower than 0.6Mt/ha and within the range of 8.6- 9.5Mt/ha were treated as outliers and therefore were not included in the regression analysis. Table 4. 2: Productivity Distribution of Lowland Rice Farms in Ashanti Region Productivity (Mt/Ha) Number of farms Percentage (%) <0.6 5 2.5 0.6-1.5 23 11.6 1.6-2.5 39 19.6 2.6-3.5 44 22.1 3.6-4.5 30 15.1 4.6-5.5 18 9.0 5.6-6.5 11 5.5 6.6-7.5 18 9.0 7.6-8.5 8 4.0 8.6-9.5 3 1.5 Observation (N) 199 100 Source: Author’s computation (2015) University of Ghana http://ugspace.ug.edu.gh 51 4.4 Descriptive Statistics of Economic Variables in the Stochastic Frontier The mean output obtained for 2014 major rice production season in the study area was 3705.4Kg/Ha. For every hectare of rice cultivated, an average of 42.8 Kg of improved seed, 138.7Kg of fertilizer, 5L of agro-chemicals, 372.9 Man-days of family labour and 52.1 man- days of hired labour were used as indicated in Table 4.3. The recommended amount of seeds per hectare for rain-fed rice production has been estimated to be at 100 kg/ha (IRRI, 1995 cited in Ogundari, 2008 & Nwilene et al., 2011).This study revealed that on average 42.8kg/ha of seed was used by the famers which are less than the recommended rate. The average man-days for the combined labour (424.9 man-days) used by the farmers was comparatively closer to a similar study conducted by Ogundari (2008) in Nigeria where average labour used for rice farming was 624 man-days. Source of labour for rice farming in Ghana is mainly from family and hired labour, with family labour contributing 87.7% of the total labour used in production. The hired labours used to compliment family labour were identified to perform difficult task such as land preparation, harvesting and praying. The major means of pests and diseases control common to most farmers in the study area was through the use of agro-chemicals. Sizeable proportion of the farmers (59%) applied pesticide whereas about 97% of the farmers used weedicide to control weeds in their farms. The high percentage of farmers using weedicide in the study area has bearing on both family and hired labour usage for the purpose of weeding. University of Ghana http://ugspace.ug.edu.gh 52 Table 4. 3: Summary of Economic Variables in The Stochastic Production Frontier Variable Unit Observation (N) Min Mean Max Std. Error Output Kg/Ha 191 687.50 3705.40 8050.00 1961.64 Improved Seed Kg/Ha 191 10.00 42.82 210.00 29.96 Fertilizer Quantity Kg/Ha 191 0 138.69 750.00 156.64 Fertilizer-use Dummy 191 0 0.55 1.00 0.50 Agro-chemical L/Ha 191 0.60 5.04 17.00 2.96 Family Labour Man-days/Ha 191 20.60 372.79 1343.80 276.17 Hired Labour Man-days/Ha 191 5.20 52.08 165.80 36.24 Source: Author’s computation (2015) 4.5 Validation of Hypothesis of the Study The likelihood ratio (LR) test results for the hypothesis of the study are presented in Table 4.4. The Translog production function was found to be an adequate representation of the 2014 lowland rice major season data over the Cobb-Douglas, hence Translog model was used in the analysis. The null hypothesis that lowland rice output variability in Ashanti region is not explained by risk and technical inefficiency were rejected based on the tested hypothesis 2 and 4 as seen in Table 4.4. These imply that rice output variation is influenced by production risk and technical inefficiency. The estimated value for lambda (0.604) is greater than zero, implies that the variations in the observed output from the frontier output are due to technical inefficiency (u) and random noise (v). University of Ghana http://ugspace.ug.edu.gh 53 Table 4. 4: Results of Tested Hypothesis for the Study Null Hypothesis Log-likelihood Value LR Test Value (α=0.001) Decision -100.77 51.36 37.7*** Reject H0 2 -86.37 22.56 20.52*** Reject H0 -104.09 58.00 32.9.*** Reject H0 0.604*** Reject H0 *** Statistically significance at 1% significance level. Finally, the null hypothesis that exogenous variables of the model do not explain technical inefficiency i.e. truncated normal distribution ( ) was rejected at 1% significant level. Therefore the exogenous variables jointly explain the technical inefficiency in rice production. This makes it possible to identify the relevant policy variables to explain technical inefficiency in lowland rice production. The understanding of technical inefficiency is important to improve upon technical efficiency in the lowland rice production process since efficiency estimates are obtained base on technical inefficiency indices. According to Coelli (1995), all critical values for testing the hypothesis can be obtained from appropriate Chi-square distribution. However, if the test include the hypothesis that inefficiency effects are non-stochastic ( ), then the asymptotic distribution necessitates mixed Chi-square distribution. 4.6 Translog Stochastic Frontier The Translog frontier estimate is represented in Table 4.5 and was jointly estimated together with production risk and technical inefficiency by a single step maximization of the loglikelihood function using Stata 12.0. However, discussion is done using the output elasticity, this is because all the input variables including the output were normalized by their respective means and log transformed so that the first order coefficient of the parameter estimates can be interpreted as partial elasticity of output at the sample mean. University of Ghana http://ugspace.ug.edu.gh 54 Table 4. 5: Maximum likelihood Estimates of Translog Mean Output Function Stoc. Frontier normal/half-normal model Number of obs = 191, Wald chi2 (21) =188.38 Log likelihood = -75.09 Prob > chi2 = 0.00 Variable Parameter Estimates P-value Constant 0.235757 0.033** Improved seed 0.437416 0.000*** Fertilizer quantity 0.571648 0.000*** Fertilizer dummy 0.000346 0.997 Agro-chemicals 0.028747 0.754 Family Labour -0.18213 0.015** Hired Labour 0.064965 0.390 0.5*(Lnseed)sq -0.29077 0.085* 0.5*(LnFert)sq -0.63739 0.003*** 0.5*(LnAgro)sq -0.08545 0.514 0.5*(LnFlab)sq -0.25721 0.000*** 0.5*(LnHlab)sq -0.1357 0.145 Lnseed*lnFert -0.25377 0.064* Lnseed*lnAgro 0.234797 0.080* Lnseed*lnFlab -0.07113 0.278 Lnseed*LnHlab 0.205008 0.013** LnFert*lnAgro -0.18222 0.095* LnFert*LnFlab 0.008178 0.892 LnFert*LnHlab 0.085298 0.298 Lnagro*lnFlab -0.0122 0.804 Lnagro*lnHlab 0.128758 0.022** LnFlab*lnHlab 0.064258 0.107 sigma_v 0.400 sigma_u 0.242 sigma squared 0.219 lambda ( ) 0.604 Source: Author’s computation (2015) University of Ghana http://ugspace.ug.edu.gh 55 The Maximum-likelihood estimates of the stochastic frontier translog model are presented in Table 4.5. However, discussion of the parameters is based on output elasticities evaluated at means with respect to the various inputs as indicated in table 4.6. Table 4. 6: Elasticity of Mean Output and Returns to Scale (RTS) Variable Parameter Estimates P-value Improved seed 0.437416 0.000*** Fertilizer quantity 0.571648 0.000*** Fertilizer dummy 0.000346 0.997 Agro-chemicals 0.028747 0.754 Family Labour -0.18213 0.015** Hired Labour 0.064965 0.39 RTS 0.92 The output elasticities with respect to factor inputs as indicated in table 4.6 shows the degree of responsiveness of rice output to changes in the various input variables. The partial elasticity for improved seed, fertilizer, family and hired labour were 0.437, 0.572, 0.028, - 0.182 and 0.065 respectively. The use of improved seed and fertilizer by rice farmers’ were significant at 1% and had positive impact on productivity. These imply that 1% increase in the use of improved seed and fertilizer by rice farmers will increase output by 0.437% and 0.572% respectively. This is consistent with a study conducted by Yiadom-Boakye (2013) on gender and rice production efficiency in Ashanti region where fertilizer and seed were found to have positive effect on output. According to Balde et al. (2014), Japan and China farming system has shown that smallholder famers can obtain the levels of output per unit area of land which is equal or greater than those achieved by large-scale farmers anywhere in the world due to their access to farm inputs such fertilizer and improved seed. Family labour was significant at 5% and had negative effect on productivity as indicated in Table 4.5. This is because family labour was freely available to the rice farmers in the study area and therefore they over-utilized their services. This is consistent with a study conducted University of Ghana http://ugspace.ug.edu.gh 56 by Msuya et al. (2008) who indicated that the presence of many family labours in Tanzania small holder farms had negative impact on farm output. The unproductive nature of family labour can be experienced in mostly rural areas where the avenue for income generating activities are only limited to farming. Hence, the teaming family members are often channelled to assist in production whether willing or unwilling. The unwilling family labours exhibit unproductive behaviours to work such as laziness, lateness and over-eating leading to production inefficiencies and consequently affecting output. The findings also conforms to a similar study conducted by Kloss & Petrick (2014) which revealed that farms with higher share of hired labour are more productive than areas traditionally characterised by family labour. Agro-chemical and hired labour were not significant to influence rice productivity in the study area. The estimated return to scale (0.92) for lowland rice production is the sum of all the output elasticities which indicate that on the average, lowland rice farms in the study area have decreasing return to scale. This imply that in the long run as all factor inputs are increased by 1%, output will also increase but less than the proportionate increase in inputs. These suggest that the farmers are on the second stage of production frontier. The estimated return to scale is consistent with the estimate of 0.99 by Villano & Fleming (2004). 4.7 Estimated Stochastic Frontier Model with Risk The parameter estimates of the risk function are presented in Table 4.7. Output variability in lowland rice production is explained by the input factors which reveal information for production risk management. Some of the inputs were risk reducing while others were risk increasing and this provides vital information to stabilize lowland rice output. The estimates for the marginal input risk of seed, fertilizer, agro-chemical and hired labour on rice output were negative and imply that they are risk reducing (reduces output variability) inputs although agro-chemical and hired labour were not significant. These imply that risk averse University of Ghana http://ugspace.ug.edu.gh 57 farmer can use more of improved seed and fertilizer to stabilize rice output. Family labour was noted to be risk increasing but not significant to influence output variation. Improved seed used by the lowland rice farmers reduce output variance because they have high germination rate and are able to withstand harsh environmental conditions such as erratic rainfall, pest and diseases. The finding of the study is consistent with Guan & Wu (2009) who examined off-farm work, technical efficiency and production risk in Taiwan and found that fertilizer was risk reducing. Similarly, Guttormsen & Roll (2013) found seed to be risk reducing input in accordance with existing empirical findings that improved seed reduces output variation (production risk). Table 4. 7: Maximum Likelihood Estimates of Production Risk Function Input Variable Parameter Estimate P-Value Constant ω0 -2.166 0.000 Seed ω1 -0.837 0.056* Fertilizer quantity ω2 -2.131 0.000*** Agro-chemical ω3 -0.106 0.720 Family Labour ω4 0.163 0.421 Hired Labour ω5 -0.132 0.581 Source: Author’s Computation (2015) *** Significant 1% and * Significant 10% 4.8 Technical Efficiency Indices of Lowland Rice Farms The distribution of the predicted technical efficiencies of the lowland rice farms in Ashanti region is depicted in Table 4.8 in the form of frequency. The mean technical efficiency was estimated to be 77% with the maximum and minimum been 99% and 26% respectively. Similarly, the mean technical efficiency estimate without taking into consideration production risk factors (production inputs) was 81% with 37% and 100% as the minimum University of Ghana http://ugspace.ug.edu.gh 58 and maximum (see Appendix III). The economic interpretation of the efficiency estimates are that on average rice farmers in the study area can improve on their efficiency of production by 19% (for the conventional estimates) and 23% (for the flexible risk estimates) without requiring any additional resources to produce the same output as the efficient rice farms on the frontier. The average efficiency for the flexible risk specification and conventional stochastic frontier model estimates imply that technical efficiency estimation without considering the effect of risk on input-use decisions by the rice farmers in the study area will produce misleading results (bias estimate) for policy decision. The mean technical efficiency (77%) of rice farms for the current study compared to other similar rice efficiency study by Asante et al. (2014) in the study area with empirical efficiency estimates of 69.1% indicates significant improvement in the mean efficiency levels of the rice farms in terms of converting inputs to output. Table 4. 8: Technical Efficiencies Scores of Lowland Rice Farms Interval Number of farms Percentage (%) ≤0.50 16 8 0.51-0.60 21 11 0.61-0.70 26 14 0.71-0.80 29 15 0.81-0.90 47 25 0.91-1.00 52 27 Observation (N) 191 100 Std. Dev. .17 Maximum .99 Mean .77 Minimum .26 4.9 Technical Inefficiency Estimates of Lowland Rice Farms The factors that were considered to influence technical inefficiency of lowland rice farms in the study area are presented in Table 4.9. The coefficients of basic education and farm distance were significant at 1% and had the expected a priori expectations. Similarly, gender, University of Ghana http://ugspace.ug.edu.gh 59 extension, canal maintenance and improved seed also met their expected a priori expectation and were significant at 5%. However, credit and agricultural mechanization were significant at 10% although the expected a priori expectation of credit was not met. The negative signs of the estimated variables in the inefficiency model indicate positive effects on technical efficiency which imply that such variables reduce rice production inefficiency. The parameter estimate for gender which has a value of one if the lowland rice farmer is a male and zero otherwise, indicates that lowland rice farmers who are males operate efficiently than their female counterpart. Basic education is one of the important variables in the inefficiency model that serve to improve on the managerial ability of the lowland rice farms. The variable basic education positively and strongly relate to technical efficiency, implying that the level of education that is required to at least reduce inefficiency of rice production in the study area is basic education as compared to secondary level which was not significant. According to Hyuha (2006), primary level of education in Uganda was necessary to increase profit efficiency of rice farmers. This makes primary education an important factor to augments the capacity of farmers to obtain and optimize the use of information in relation to production inputs to carry out production and managerial tasks on rice farms. The finding of the study with regard to the effect of education on rice farming is consistent with a study by Al-Hassan (2012) and Villano & Fleming (2004). Farmers’ access to credit is significant at 10% and negatively related to technical efficiency. This means that farmers who use credit are less efficient than farmers who do not use credit, probably because the farmers who use credit either do not get the credit at the time required to buy inputs for production or mismanage the credit by diverting it from it intended focus. The results of the study agrees with the findings of a similar study conducted by Balde et al. (2014) and concluded that farmers who have access to credit for mangrove rice production in coastal guinea exhibit higher level of technical inefficiency. University of Ghana http://ugspace.ug.edu.gh 60 The farm distance from the farmers’ residence to the rice field was significant at 1% and related negatively to inefficiency. It is rational to assume that the shorter the farm distance, the more frequent it is to visit the farm and consequently the farm receives the needed attention in terms of management from the farmer. According to Xaba & Masuku (2013), shorter farm distance reduces transportation cost. Adhikari (2009) also revealed that other cost of production which transportation cost is a component have effect on output when the farm distance is short. Therefore, farms located closer to the residence of the farmers reduces production inefficiency as compared to one located further away from home. Canal maintenance and improved rice seed variety used by the farmers related negatively to technical inefficiency of rice production. Maintenance of canal helps to regulate excess water movement during peak rainy season to prevent flooding whilst improved riced seed helps to resist pest and disease infestation that may occur aftermath of the receding water. According to Dessalegn (2005) and Nwilene1 et al.(2011), the success and failure of rice production have always been attributed to environmental factors such as pest, diseases and rainfall hence the mechanisms for managing the risk imposed by these factors have remained the subject of extensive studies in various disciplines such as inefficiency and risk to improve on rice yield. Farmers’ access to extension service was significant and had the expected a priori expectation to reduce inefficiency. Extension service helps to improve productivity in diverse ways through training and technical advice to farmers on various farm practices to address the problem of information asymmetry on input usage. These allow farmers to adopt modern agricultural technology for farming and also control pest and diseases using the required quantity of chemicals. According to MoFA (2011), extension - farmer ratio in Ghana is 1: 762. Despite the gap that exists between farmers and extension officers, their presence had positive impact on production efficiency. This phenomenon (low extension to farmers’ ratio) University of Ghana http://ugspace.ug.edu.gh 61 is very synonymous with developing countries and has also been reported in similar studies in analysis of technical efficiency of mangrove rice production in Guinea (Balde et al., 2014). Table 4. 9: Determinants of Technical Inefficiency in Rice Production Inefficiency Variable Parameter Estimate P-value Constant 1.553 0.056* Socio-economic Profile Gender -1.034 0.025** Farm distance -0.469 0.003*** Experience 0.021 0.390 Institutional factors Credit 0.893 0.053* Mechanization -0.981 0.086* Extension access -0.994 0.048** Managerial factors Basic Education -1.535 0.003*** Secondary Education -3.564 0.174 Canal maintenance -0.947 0.035** Improved seed -0.852 0.045** Planting method -0.286 0.476 FBO 0.535 0.217 Source: Author’s computation (2015) *** Significant 1%, ** Significant 5% and * Significant 10% University of Ghana http://ugspace.ug.edu.gh 62 CHAPTER FIVE SUMMARY, CONCLUSIONS AND POLICY RECCOMMENDATIONS 5.1 Introduction This chapter presents the summary of major findings, conclusions and policy recommendations of the study. 5.2 Major Findings From the Research The study considered the stochastic frontier model with flexible risk property to analysed the production risk and technical efficiency of lowland rice farms in Ashanti region, Ghana using 200 farms. The study considered a number of hypotheses to investigate adequacy of the models, the choice of the functional forms adopted and explanation of variation in inefficiency. The STATA 12 software was used for the analysis. The major findings of the study include: i. The Translog functional form was accepted as the best fit for the model, whilst the frontier model with the risk component was preferred against the conventional technical efficiency model and the average response function. ii. Inclusion of the intercept dummy for fertilizer in the model was important to prevent any bias estimation. iii. The input variables considered in the frontier model were all positive except family labor. iv. The average productivity of lowland rice farms in the study area was 3.7mt/ha as compared to the national average of 2.5 mt/ha. v. Partial elasticity indicated that fertilizer and improved seed used by the farmers were the major contributors to rice production in the study area. University of Ghana http://ugspace.ug.edu.gh 63 vi. Fertilizer and improved seed were identified to be risk reducing inputs. vii. The mean technical efficiency was estimated to be 77%. viii. This study found that on average 42.8kg/ha of rice seed was applied by lowland rice famers compared to the recommended rate of 100 kg/ha. ix. The estimated return to scale for the lowland rice farms demonstrated a decreasing return to scale. x. Results of the inefficiency estimates indicates that improved seed-usage, shorter farm distance, farmer access to extension services, basic education, mechanization services and canal maintenance reduces technical efficiency. xi. It was established that conventional efficiency studies that do not capture the issue of risk have higher technical efficiency than improved conventional efficiency study that captures risk. 5.3 Conclusion of the Study Based on the findings of the study, the following conclusions are made:  The joint contribution of conventional farm inputs (seed, fertilizer, agro-chemicals, family labour and hired labour) played an important role in achieving the average rice productivity of 3.7Mt/ha for the sampled lowland rice farms.  Rice farmers on average under-utilised seed as an input resource for production (42.8 Kg/ha) as against the recommended quantity of 100kg/ha. This implies that farmers are probably not taking advantage of the seed subsidy programme by the government to increase seed utilization. The impact of the seed subsidy programme can only be manifested in higher yield if farmers understood the need to optimized input usage. University of Ghana http://ugspace.ug.edu.gh 64  Farm inputs such as fertilizer and improved seed were identified to reduce rice output variation among farmers (risk reducing) therefore, lowland rice farmers can use more of such inputs to increase yield.  The estimated technical efficiency value of 77% suggests that current rice output in the study area can be increased by 23% without any additional expansion in farm inputs if the prevailing access to credit inefficiency in rice production is addressed.  Lowland rice productive efficiency can be increased if farmers at least attain basic level of education to adopt improved seed usage and good farm management practices (canal maintenance & shorter farm distances) coupled with improvement in extension and mechanization service delivery by MoFA to the farmers. 5.4 Policy Recommendation For agriculture in general and rice farms in particular to increase productivity to ensure sustained domestic food security, reduce importation and foreign exchange outflows, the results of the study are important to inform policy decision on how to reduce the existing inefficiencies and mitigate the effects of production risk in rice production. Based on the findings of the study, it is therefore recommended that:  Stakeholders (MoFA, DPs, FBOs, etc.) supporting and promoting agricultural programmes should educate farmers through extension officers to use the recommended quantity of improved rice seed per hectare of land for sowing. To further enhance input-use by the farmers, the individual organizations mentioned can also develop food policy measure that can be monitored to ensure that farmers in the country follows best farming practices to allow for efficient use of farm inputs particularly improved seed as a vital step to increase output. University of Ghana http://ugspace.ug.edu.gh 65  Stakeholders in Agricultural sector should strengthen extension and mechanization service delivery to farmers to reduce lowland rice production inefficiency since they reduce technical inefficiency.  Farmers should be educated by extension officers to develop and maintain canals on paddy fields to regulate excess water to enhance input use efficiency and yield to reduce rice production inefficiency.  The government should strengthen the improved seed and fertilizer subsidy programme to ensure sustainable rice production since this inputs are risk decreasing. The improved rice variety has intrinsic features that enable rice to resist harsh environmental conditions (drought, pest and diseases). When rice farmers cultivate the same variety over a period of time, it potency to endure such conditions reduces and consequently affect output. This highlights the importance of breeding institutions to be resource adequately to develop new improved varieties to reduce produce risk.  Factors that make credit to contribute to technical inefficiency in lowland rice production should be investigate in the study area and address them accordingly. 5.5 Limitation of the Study The study focused on Ashanti region due to its central location which is bordered by four other regions namely Brong Ahafo, Western, Central and Eastern regions and has linkages to the rest of the country. The production risk and technical inefficiency associated with rice production in the study area may be different from the other regions due to certain peculiar practices or character of farmers that pertains in those areas. Therefore the recommendations of the study to reduce production risk and technical inefficiencies can only help to improve rice production in Ashanti region. University of Ghana http://ugspace.ug.edu.gh 66 5.6 Suggestion for Future Study Future research into the topic should extend the scope of the study area to establish the effect of production risk and technical inefficiencies in other regions since they jointly influence rice production and it corresponding level of importation in Ghana. University of Ghana http://ugspace.ug.edu.gh 67 REFERENCES Aigner, D., Lovell, C. K., & Schmidt, P. (1977). Formulation and estimation of stochastic frontier production function models. Journal of Econometrics, 6(1), 21-37. Al-Hassan, S. (2012). Technical efficiency in smallholder paddy farms in Ghana: An analysis based on different farming systems and gender. Journal of Economics and Sustainable Development, 3(5), 91-105. Apori-Buabeng F. (2009). Market Organisation and Consumers’ Perception of Locally Produced Rice in the Ashanti Region of Ghana (Master’s dissertation). Abdulquadri, A. F., & Mohammed, B. T. (2012). The Role of Agricultural Cooperatives in Agricultural Mechanization in Nigeria. World Journal of Agricultural Sciences, 8(5), 537-539. Akinbogun, K. O. (2010). Modeling Technical Efficiency with Production Risk: A Study of Fish Farms in Nigeria. Marine Resource Economics, 25(3), 295-308 Ajetomobi, J. O. (2011). Productivity improvement in Ecowas millet farming. Agricultura Tropica Et Subtropica, 44(1), 1-7. Angelucci, F., Asante-Poku, A., & Anaadumba, P. (2013). Analysis of incentives and disincentives for rice in Ghana. Technical notes series, MAFAP, FAO, Rome. Adhikari, C. B., & Bjørndal, T. (2009). Measuring the extent of technical inefficiency in Nepalese agriculture using SDF and DEA models. Working Paper No. 28/09. Addai, K. N., Owusu, V., & Danso-Abbeam, G. (2014). Effects of farmer–Based- organization on the technical efficiency of maize farmers across Various Agro- Ecological Zones of Ghana. Journal of Economics and Development Studies, 2(1), 141-161. Abedullah, Kouser, S., & Mushtaq, K. (2007). Analysis of technical efficiency of rice production in punjab (Pakistan): Implications for future investment strategies. Pakistan Economic and Social Review, 45(2), 231-244. Asante, B. O., Wiredu, A. N., Martey, E., Sarpong, D. B., & Mensah-Bonsu, A. (2014). NERICA Adoption and Impacts on Technical Efficiency of Rice Producing Households in Ghana: Implications for Research and Development. American Journal of Experimental Agriculture, 4(3), 244-262. Awudu , A., & Huffman, W. (2000). Structural adjustment and economic efficiency of rice farmers in northern Ghana. Economic Development and Cultural Change, 48(3), 503-520. University of Ghana http://ugspace.ug.edu.gh 68 Alvarez, A., & Arias, C. (2004). Technical efficiency and farm size: a conditional analysis. Agricultural Economics, 30(3), 241-250. Antle, J. M., & Crissman, C. C. (1990). Risk, efficiency, and the adoption of modern crop varieties: Evidence from the Philippines. Economic Development and Cultural Change, 517-537. Bokusheva, R., & Hockmann, H. (2006). Production risk and technical inefficiency in Russian agriculture. European Review of Agricultural Economics, 33(1), 93-118. Balde, B. S., Kobayashi, H., Nohmi, M., Ishida, A., Esham, M., & Tolno, E. (2014). An Analysis of Technical Efficiency of Mangrove Rice Production in the Guinean Coastal Area. Journal of Agricultural Science, 6(8), 179. Battese, G. E., & Corra, G. S. (1977). Estimation of a production frontier model: with application to the pastoral zone of Eastern Australia. Australian journal of agricultural economics, 21(3), 169-179. Battese, G. E., & Coelli, T. J. (1988). Prediction of firm-level technical efficiencies with a generalized frontier production function and panel data. Journal of econometrics, 38(3), 387-399. Battese, G. E., & Coelli, T. J. (1992). Frontier production functions, technical efficiency and panel data: with application to paddy farmers in India. Springer Netherlands, 149- 165. Battese, G.E., Malik, S.J. and Gill, M.A. (1996). An investigation of technical inefficiencies of production of wheat farmers in four districts of Pakistan. Journal of Agricultural Economics 47(1), 37-49. Boansi D. (2013). Rice yields in Ghana: Macro-level response and some prescriptions. International Journal Agriculture. Policy Res.1 (9): 270-276. Boahen A. O. (2012). Production Risk and Technical Efficiency of Maize Farms in the Brong Ahafo Region of Ghana. Unpublished master’s dissertation. Department of Agricultural Economics, Agribusiness, University of Ghana. Constantin, P. D., & Martin, D. L. (2009). Cobb-Douglas, translog stochastic production function and data envelopment analysis in total factor productivity in Brazilian agribusiness. Journal of operations and supply chain management, 2(2), 20-33. CARD, (2010). Mapping of Poverty Reduction Strategy Papers (PRSPs, Sector Strategies and Policies related to rice development, Ghana. Charnes, A., Cooper, W. W., & Rhodes, E. (1978). Measuring the efficiency of decision making units. European journal of operational research, 2(6), 429-444. University of Ghana http://ugspace.ug.edu.gh 69 Chang, H. H., & Wen, F. I. (2008). Off-farm Work, Technical Efficiency, and Production Risk: Empirical Evidence from a National Farmer Survey in Taiwan. In 2008 Annual Meeting, July 27-29, 2008, Orlando, Florida (No. 6164). American Agricultural Economics Association (Agricultural and Applied Economics Association). Coelli, T. J., & Battese, G. E. (1988). Prediction of firm-level technical efficiencies with a generalized frontier production function and panel data. Journal of econometrics, 38(3), 387-399. Coelli, T. (1995). Estimators and hypothesis tests for a stochastic frontier function: a Monte Carlo analysis. Journal of productivity analysis, 6(3), 247-268. Coelli, T. J. (1995). Recent developments in frontier modelling and efficiency measurement. Australian Journal of Agricultural Economics, 39(3), 219-245. Coelli, T., & Rao, D. G. Battese (1998). Introduction to Efficiency and Productivity Analysis. Chiona, S. (2011). Technical and allocative efficiency of smallholder maize farmers in Zambia (Master’s dissertation, University of Zambia). Ragasa, C., Dankyi, A., Acheampong, P., Wiredu, A. N., Chapoto, A., Asamoah, M., & Tripp, R. (2013). Patterns of adoption of improved rice technologies in Ghana. International Food Policy Research Institute Working Paper, 35. Dessalegn, T. Y. (2006). Modeling Farm Irrigation Decisions Under Rainfall Risk in the White-Volta Basin of Ghana: A Tool for Policy Analysis at the Farm-household Level. Cuvillier-Verlag. Enwerem, V. A., & Ohajianya, D. O. (2013). Farm size and technical efficiency of rice farmers in Imo state, Nigeria. Greener Journal of Agricultural Sciences, 3(2), 128- 136. Farrell, M. J. (1957). The measurement of productive efficiency. Journal of the Royal Statistical Society. Series A (General), 253-290. Forsund, F. R., Lovell, C. K., & Schmidt, P. (1980). A survey of frontier production functions and of their relationship to efficiency measurement. Journal of econometrics, 13(1), 5-25. Fried, Harold O. Lovell, C. K., & Shelton S.. Schmidt. (1993). The measurement of productive efficiency: techniques and applications. Oxford University Press. Fare, R., Grosskopf, S. & Lovell, C. A. K. (1985). The measurement of efficiency of production. Kluwer-Nijhoff, Boston. University of Ghana http://ugspace.ug.edu.gh 70 Farrell, M. J. (1957). The measurement of productive efficiency. Journal of the Royal Statistical Society. Series A (General), 253-290. Guisse, R. (2010). Post Harvest Losses of Rice (Oriza Spp) From Harvesting to Milling: A Case Study in Besease and Nobewam in the Ejisu Juabeng District in the Ashanti Region of Ghana (Master’s dissertation, Kwame Nkrumah University of Science and Technology, Kumasi). Guttormsen, A. G., & Roll, K. H. (2014). Production Risk in a Subsistence Agriculture. The Journal of Agricultural Education and Extension, 20(1), 133-145. Ganidekam, E. (2013). Political economy of food prices in Ghana. Ghanaian Journal of Economics, (1), 67-86. Guan, Z., & Wu, F. (2009). Specification and estimation of heterogeneous risk preference. In Contributed paper prepared for presentation at the 27th International Conference of Agricultural Economists (IAAE 2009), Beijing, China, August 16Á22. Guang H. Wan, & Battese, G. E. (1992). A stochastic frontier production function incorporating flexible risk properties. University of New England, Department of Econometrics. Greene, W. H. (2008). The econometric approach to efficiency analysis. The measurement of productive efficiency and productivity growth, 92-250. Herdt, R. W., & Mandac, A. M. (1981). Modern technology and economic efficiency of Philippine rice farmers. Economic Development and Cultural Change, 375-399. Hara M. W., Dosso H., & Lekorchi D. (2001). Inland Valleys Rice Development Project (Appraisal report No. GHA/PAAC/2001/01), Ghana. Hasanthika, W. K., Edirisinghe, J. C., & Rajapakshe, R. D. (2013). Climate Variability, Risk and Paddy Production. Journal of Environmental Professionals Sri Lanka, 2(2): 57-65. Hyuha, T., (2006). Profit efficiency among rice farmers in Uganda. Unpublished Ph.D thesis, Makerere University, Uganda. Issaka R. N., Buri M. M., Nakamura S., & Tobita S. (2014). Comparison of different fertilizer management practices on rice growth and yield in the Ashanti region of Ghana. Agriculture, Forestry and Fisheries; 3(5): 374-379 Jaforullah, M., & Whiteman, J. (1999). Scale efficiency in the New Zealand dairy industry: a non- parametric approach. The Australian Journal of Agricultural and Resource Economics, 43(4), 523-541. Jondrow, J., Lovell, C. A., Materov, V. & Smith, P. (1982). On the Estimation of Technical Efficiency in the Stochastic Frontier Production Function Model. Journal of Econometrics, 19(3), 233-238. University of Ghana http://ugspace.ug.edu.gh 71 Just, R. E., & Pope, R. D. (1978). Stochastic specification of production functions and economic implications. Journal of Econometrics, 7(1), 67-86. Jeffrey, S. R. & Xu X. (1998). Efficiency and Technical Progress in Traditional and Modern Agriculture: Evidence from Rice Production in China. American Journal of Agricultural Economics, 18(2), 157-165. JICA (2008). The study on the promotion of domestic rice in the republic of Ghana, 2- 9. Kang K. (1997). An Analysis of Trucking Industry Productivity: Using Stochastic Frontier Function. Joumal of the Eastem Asia Society for Transportation Studies, 2 (6). Kula O. & Dormon E. (2009). Global food security response Ghana rice study, United States agency for international development, USAID- Ghana. Kokkinou A. (2010). Theory of Productive Efficiency and Stochastic Frontier Models European Research Studies, Volume XIII, Issue (4). Kloss, M., & Petrick, M. (2014, August). The productivity of family and hired labour in EU arable farming. In 2014 International Congress, August 26-29, 2014, Ljubljana, Slovenia (No. 183041). European Association of Agricultural Economists. Kalirajan, K. (1981). An econometric analysis of yield variability in paddy production. Canadian Journal of Agricultural Economics/Revue canadienne d'agroeconomie, 29(3), 283-294. Kuwornu, J. K., Amegashie, D. P., & Wussah, C. E. (2012). Productivity and Resource Use Efficiency in Tomato and Watermelon Farms: Evidence from Ghana. Developing Country Studies, 2(2), 23-37. Kumbhakar, S. C. (1993). Production risk, technical efficiency, and panel data. Economics Letters, 41(1), 11-16. Kumbhakar, S. C. (2002). Specification and estimation of production risk, risk preferences and technical efficiency. American Journal of Agricultural Economics, 84(1), 8-22. Kibirige, D. (2008). Analysis of the Impact of the Agricultural Productivity Enhancement Program on the Technical and Allocative Efficiency of Maize Farmers in Masindi District (Master’s dissertation, Makerere University Kampala). Ligeon, C., Jolly, C., Bencheva, N., Delikostadinov, S., & Puppala, N. (2013). Production efficiency and risks in limited resource farming: The case of Bulgarian peanut industry. Journal of Development and Agricultural Economics, 5(4), 150-160. Lesáková, Ľ. (2007). Uses and limitations of profitability ratio analysis in managerial practice. In International Conference on Management, Enterprise and Benchmarking. University of Ghana http://ugspace.ug.edu.gh 72 Molua, E. L. (2007). The economic impact of climate change on agriculture in Cameroon. World Bank Policy Research Working Paper Series. McCarthy, N., Lipper, L., & Branca, G. (2011). Climate-smart agriculture: smallholder adoption and implications for climate change adaptation and mitigation. Mitigation of Climate Change in Agriculture Working Paper. Mastromarco, C. (2008). Stochastic Frontier Models. Department of Economics and Mathematics- Statistics, University of Salento. Masuku, M. B., & Xaba, B. (2013). Factors Affecting the Productivity and Profitability of Vegetables Production in Swaziland. Journal of Agricultural Studies, 1(2), 37-52. Msuya, E. E., Hisano, S., & Nariu, T. (2008). Explaining productivity variation among smallholder maize farmers in Tanzania. MoFA, (2009). National Rice Development Strategy (NRDS). Coalition for African Rice Development, Accra. MoFA (2010). Medium Term Agriculture Sector Investment Plan (METASIP) 2011 – 2015, Accra : 19-20. MoFA (2010). Annual Progress Report. Policy Planning Monitoring and Evaluation Directorate, Accra. MoFA (2011). Annual Report Part II Accra, 40-50. MoFA (2012). Statistics, Research and Information Directorate (SRID). Agriculture in Ghana. Facts and Figures: 6-52. Moro, B. M., Nuhu, I. R., & Toshiyuki, W. (2008). Determining optimum rates of mineral fertilizers for economic rice grain yields under the “Sawah” system in Ghana. West African Journal of Applied Ecology, 12(1). Nwilene, F. E., Onasanya, A., Togola, A., Youdeowei, A., Abo, E., Ogah, E., ... & Oyetunji, O. E. (2011). Is Pesticide Use Sustainable in Lowland Rice Intensification in West Africa?. INTECH Open Access Publisher. Ogundari, K. (2008). Resource-productivity, allocative efficiency and determinants of technical efficiency of rainfed rice farmers: A guide for food security policy in Nigeria. Zemedelska Ekonomika-Praha-, 54(5), 224. Onumah, E. E., Brümmer, B., & Hörstgen‐Schwark, G. (2010). Elements which delimitate technical efficiency of fish farms in Ghana. Journal of the World Aquaculture Society, 41(4), 506-518. University of Ghana http://ugspace.ug.edu.gh 73 Onumah, E. E., Hoerstgen-Schwark, G., & Brümmer, B. (2009). Productivity of hired and family labour and determinants of technical inefficiency in Ghana's fish farms (No. 0907). Department of Agricultural Economics, Czech. Opoku-Duah, S., Kankam-Yeboah, K., & Mensah, E. K. (2000). Determination of Water Requirements for Producing Irrigated Rice and other crops in the Afram River valley bottom at Aframso, Ghana. Ghana Journal of Science, 40(1), 15-24. Padilla-Fernandez M. D. & Nuthall P. L. (2009). Technical Efficiency in the Production of Sugar Cane in Central Negros Area, Philippines: An Application of Data Envelopment Analysis. Journal of ISSAAS, 15 (1): 77-90 Rola, A. C., & Quintana-Alejandrino, J. T. (1993). Technical efficiency of Philippine rice farmers in irrigated, rainfed lowland and upland environments: a frontier production function analysis. Philippine Journal of Crop Science, 18(2), 59-69. Rahman, K. M. M., Mia, M. I. A., & Bhuiyan, M. K. J. (2012). A Stochastic Frontier Approach to Model Technical Efficiency of Rice Farmers in Bangladesh: An Empirical Analysis. The Agriculturists, 10(2), 9-19 Villano, R., & Fleming, E. (2004). Analysis of technical efficiency in a rainfed lowland rice environment in Central Luzon Philippines using a stochastic frontier production function with a heteroskedastic error structure. Watkins, K. B., Hristovska, T., Mazzanti, R., Wilson, C. E., & Schmidt, L. (2014). Measurement of Technical, Allocative, Economic, and Scale Efficiency of Rice Production in Arkansas Using Data Envelopment Analysis. Journal of Agricultural and Applied Economics, 46(01), 89-106. WARDA (2007). Africa Rice Trends: Overview of recent developments in the sub-Saharan Africa rice sector. Africa Rice Center Brief, Cotonou, Benin. Wayo S., A., & Nyanteng, V. K. (2003). Afrint macro study: Ghana report (revised). Legon- Accra: Institute of Statistical, Social and Economic Research (ISSER). University of Ghana. Samal, P., & Pandey, S. (2005). Climatic Risks, Rice Production Losses and Risk Coping Strategies: A Case Study of a Rainfed Village in Coastal Orissa. Agricultural Economics Research Review, 18(1), 61-72. Schmidt, P. (1986). Frontier Production Functions, Econometric Review, 4(1), 289-328. Sié, M., Sanni, K., Futakuchi, K., Manneh, B., Mandé, S., Vodouhé, R., ... & Traoré, K. (2012). Towards a rational use of African rice (Oryza glaberrima Steud.) for breeding in Sub-Saharan Africa. Genes, genomes and genomics, 6(1), 1-7. University of Ghana http://ugspace.ug.edu.gh 74 Songsrirote, N., & Singhapreecha, C. (2007). Technical efficiency and its determinants on conventional and certified organic jasmine rice farms in Yasothon province. Thammasat Economic Journal, 25(2), 96-133. Schmidt, P. and Sickles, R. C. (1984). Production frontiers and panel data. Journal of Business and Economic Statistics, 2, 367-374. Sauer, J., Frohberg, K., & Hockmann, H. (2006). Stochastic efficiency measurement: the curse of theoretical consistency. Journal of Applied Economics, 9(1), 139-165. Sakurai, T. (2006). Intensification of rainfed lowland rice production in West Africa: present status and potential green revolution. The developing economies, 44(2), 232-251. Tinsley, R. L. (2004). Developing Smallholder Agriculture: A Global Prospective. Singapore: AGBE Publishing. Tuffour, M., & Oppong, B. A. (2012). Profit Efficiency In Broiler Production: Evidence From Greater Accra Region Of Ghana. International Journal of food and Agricultural Economics, 2(1), 23-32. Throup, D., Cooke, J. G., & Downie, R. (2011). Ghana: Assessing risks to stability. Center for Strategic and International Studies. Yiadom-Boakye, E., Owusu-Sekyere, E., Nkegbe, P. K., & Ohene-Yankyera, K. (2013). Gender, resource use and technical efficiency among rice farmers in the Ashanti Region, Ghana. Yusuf S.A. & Malomo O. (2007). Technical Efficiency of Poultry Egg Production in Ogun State, Nigeria: Data Envelopment Analysis Approach. Journal of Poultry Science, 6 (9): 622-629. University of Ghana http://ugspace.ug.edu.gh 75 Appendix I: Questionnaires for the Study STUDY ON LOWLAND RICE PRODUCTION IN ASHANTI REGION DEPARTMENT OF AGRICULTURAL ECONOMICS AND AGRIBUSINESS UNIVERSITY OF GHANA This questionnaire has been purposely designed to address the above topic in partial fulfilment for the award of Master of philosophy in Agricultural Economics at the University of Ghana, Legon. Any information provided by respondents will be purposely used for data analysis in respect of the above mentioned topic only and will be treated very confidential District………………………………………..Name of respondent………………………..………..Date…………/…………/2014 Town…………………………………………………..Telephone…………………………………………………………………. Identification code University of Ghana http://ugspace.ug.edu.gh 76 Please circle the appropriate option of the answers provided for closed ended questionnaires below A. Socioeconomic characteristics of the farmer What is your major occupation? 1=Farming 2=Trading 3=Salary worker 4=Voluntary worker 5=Others Marital status? 0=Single 1=Married 2=Separated 3=Widowed 4=Widower Gender of the farmer? 1=Male 0=Female Educational status 0=No Schooling 1=Basic Education 2=Secondary Education 3=Tertiary Education Age of farmer (years) Farming experience (years) Why are you interested in rice cultivation 1=For Consumption 2=For Income 3=Both Income and Consumption Do you engage in non- farm activity 1=Yes 0 = No If yes, what non-farm activities 1=Trade 2=Salary Worker 3 =Others Household size (head count) B. Institutional variables Do you belong to any farmer group (FBO)? 1=Yes 0= No Please indicate the importance of this group? 1=land preparation 2=chemical application 3=Credit application process 4=harvesting 5=weeding 6=others Did you use credit in 2014 rice major season? 1=Yes 0 = No If yes, where is the source of the credit? 1=Family 2=Friends 3=Financial institution 4= Others What was the amount received? How did you service the credit? 1= In-cash 2=In-Kind If in-cash, how much did you pay on monthly basis, for how long? (interest rate) Did you have contact with extension agents for the purposes of rice farming in the 2014 major season? 1 =Yes 0=No If yes, how many times did it happen What were the services received from extension officers? 1=planting 2=chemical application 3=Credit application process 4=refilling of seedlings 5=weeding 6=others University of Ghana http://ugspace.ug.edu.gh 77 C. Capital asset used for rice farming in 2014 major season Item Quantity Year of purchase Unit price purchased (Gh¢) Lifespan Hoe Cutlass Sprayer Sickle Willington Boot 1. 2. D. Managerial Variables (Plot level data) for 2014 rice major cropping season Plot No. Plot size (Acres) How did you acquire the farm Land? 1= Outright Purchase, 2=Leased/Rented, 3 Share Cropping, 4=Family Land, 5=Borrowed. If leased/rent, how much did you pay for 2014? (GH¢ per acre) Farm distance from home? (Kilometers) Planting method used? 0= Broadcasting 1=Row planting 2=Other planting methods. What was the cropping system used? 1= Mono Cropping 2= Intercropping 3= mixed cropping Did you maintain the canal and drainage systems for 2014 1= Yes 0 = No Did you use the services of mechanization centres 1=Yes 0=No Plot 1 Plot 2 Plot 3 University of Ghana http://ugspace.ug.edu.gh 78 Plot If yes, what is the distance to the mechanization centre (kilometres) What type of rice seed was used? 1= Improved seed 2= Local seed Quantity of improved seed used per (kg) or sachet? Unit cost of seed (GH¢) Source of rice seed used? 1= Local market seller 2= Certified seed dealer 3= Owned stored seed 4 = Others (projects, programs etc) What type of fertilizer did you use for 2014 major cropping season? 1. inorganic fertilizer 2. Liquefied fertilizer 3. Organic manure Plot 1 Plot 2 Plot 3 E. Please indicate the quantity and cost of fertilizer used for rice farming in 2014 major season. Plot Number Quantity of fertilizer used Unit cost per 50kg bag (Gh¢) (1000milegram) NPK (50kg bag) SOA (50kg bag) Urea (50kg bag) Liquid fertilizer (1000milegram) NPK SOA Urea Liquid fertilizer Plot 1 Plot 2 Plot 3 University of Ghana http://ugspace.ug.edu.gh 79 F. Please indicate the type of agro-chemical used for rice farming Plot Number Quantity and type of agro-chemical Unit cost per litre (Gh¢) Name of weedicide Quantity (Litres) Name of pesticide (Litres) Quantity (Litres) Weedicide Pesticide Plot 1 1. 2. Plot 2 1. 2. Plot 3 1. 2. G. Farm labor activity and requirements for 2014 major crop season (Plot 1) Number of times of performing the activity(if applicable) Family Labor Hired Labor Farm Activity Male Female Male Female Major season Quantity Hours worked Quantity Hours worked per day Quantity Hours worked per day Unit cost per day (Gh¢) Quantity Hours worked per day Unit cost per day (Gh¢) per day Nursery (if applicable) Land Preparation Sowing/planting Fertilizer application Spraying Weeding Bird scaring Harvest Threshing/winnowing Bagging Drying Transport University of Ghana http://ugspace.ug.edu.gh 80 H. Farm labor activity and requirements for 2014 major crop season (Plot 2) Number of times of performing the activity(if applicable) Family Labor Hired Labor Farm Activity Male Female Male Female Major season Quantity Hours worked Quantity Hours worked per day Quantity Hours worked per day Unit cost per day (Gh¢) Quantity Hours worked per day Unit cost per day (Gh¢) per day Nursery (if applicable) Land Preparation Sowing/planting Fertilizer application Spraying Weeding Bird scaring Harvest Threshing/winnowing Bagging Drying Transport University of Ghana http://ugspace.ug.edu.gh 81 I. Farm labor activity and requirements for 2014 major crop season (Plot 3) Number of times of performing the activity(if applicable) Family Labor Hired Labor Farm Activity Male Female Male Female Major season Quantity Hours worked Quantity Hours worked per day Quantity Hours worked per day Unit cost per day (Gh¢) Quantity Hours worked per day Unit cost per day (Gh¢) per day Nursery (if applicable) Land Preparation Sowing/planting Fertilizer application Spraying Weeding Bird scaring Harvest Threshing/winnowing Bagging Drying Transport J. Please indicate the rice output for only the 2014 Major Season. Plot Quantity of paddy rice output (kg) Quantity consumed Selling price 84kg bag/sack ( N4 bag) 110kg bag (N5 bag) 84kg bag/sack ( N4 bag) 110kg bag (N5 bag) 84kg bag/sack 110kg bag Plot 1 Plot 2 Plot 3 University of Ghana http://ugspace.ug.edu.gh 82 Appendix II: Tanslog and Cobb-Douglass Regression Results Translog Model _cons 1.553255 .8131358 1.91 0.056 -.0404622 3.146972 fbo .5352585 .4339447 1.23 0.217 -.3152576 1.385775 plantingmethod -.2860982 .4016668 -0.71 0.476 -1.073351 .5011543 improveseed -.8523659 .4256974 -2.00 0.045 -1.686717 -.0180144 canalmaintainance -.9472254 .4497949 -2.11 0.035 -1.828807 -.0656436 seceducation -3.5635 2.62039 -1.36 0.174 -8.699369 1.572369 basiceduc -1.534644 .5190395 -2.96 0.003 -2.551943 -.5173457 extensionagent -.9942703 .503615 -1.97 0.048 -1.981338 -.007203 mechanization -.9810873 .5711268 -1.72 0.086 -2.100475 .1383006 credit .8931211 .4621022 1.93 0.053 -.0125825 1.798825 experience .0212515 .0247359 0.86 0.390 -.0272299 .0697329 farmdistance -.4687008 .1578945 -2.97 0.003 -.7781682 -.1592333 gender -1.033626 .4606152 -2.24 0.025 -1.936415 -.1308372 lnsig2u _cons -2.165571 .2774177 -7.81 0.000 -2.709299 -1.621842 lhlab -.1321193 .239146 -0.55 0.581 -.6008369 .3365983 lnflab .1630795 .2024987 0.81 0.421 -.2338107 .5599697 lnagro -.1064397 .2971863 -0.36 0.720 -.6889141 .4760347 lnfert -2.131265 .4660568 -4.57 0.000 -3.044719 -1.21781 lnseed -.8366116 .4386068 -1.91 0.056 -1.696265 .0230421 lnsig2v _cons .2357568 .1104573 2.13 0.033 .0192644 .4522491 lnflablnhlab .0642582 .0398567 1.61 0.107 -.0138595 .142376 lnagrolnhlab .1287582 .0561484 2.29 0.022 .0187094 .2388069 lnagrolnflab -.0122006 .0491365 -0.25 0.804 -.1085063 .0841051 lnfertlnhlab .0852978 .0819438 1.04 0.298 -.0753092 .2459047 lnfertlnflab .0081776 .0604087 0.14 0.892 -.1102213 .1265766 lnfertlnagro -.1822209 .1092251 -1.67 0.095 -.3962982 .0318564 lnseedlnhlab .2050081 .0826272 2.48 0.013 .0430618 .3669545 lnseedlnflan -.0711323 .0655152 -1.09 0.278 -.1995397 .0572752 lnseedlnagro .2347974 .134082 1.75 0.080 -.0279986 .4975934 lnseedlnfert -.253768 .1369952 -1.85 0.064 -.5222738 .0147378 lnhlabsq -.1357043 .0930025 -1.46 0.145 -.3179859 .0465773 lnflasq -.2572092 .0553611 -4.65 0.000 -.3657149 -.1487035 lnagrosq -.0854503 .1308622 -0.65 0.514 -.3419355 .1710349 lnfertsq -.6373863 .2133194 -2.99 0.003 -1.055485 -.2192879 lnseedsq -.2907655 .1689822 -1.72 0.085 -.6219645 .0404336 lhlab .064965 .0756027 0.86 0.390 -.0832136 .2131436 lnflab -.1821257 .0745922 -2.44 0.015 -.3283237 -.0359277 lnagro .0287473 .0917371 0.31 0.754 -.1510541 .2085488 fertd .0003464 .0850284 0.00 0.997 -.1663062 .1669989 lnfert .5716482 .1614506 3.54 0.000 .2552107 .8880856 lnseed .4374157 .1082554 4.04 0.000 .225239 .6495923 lny lny Coef. Std. Err. z P>|z| [95% Conf. Interval] Log likelihood = -75.089694 Prob > chi2 = 0.0000 Wald chi2(21) = 188.38 Stoc. frontier normal/half-normal model Number of obs = 191 University of Ghana http://ugspace.ug.edu.gh 83 Cobb-Douglas Model _cons 2.798282 1.224692 2.28 0.022 .3979297 5.198634 fbo 1.153174 .664363 1.74 0.083 -.1489536 2.455301 plantingmethod -.1939813 .5818369 -0.33 0.739 -1.334361 .9463981 improveseed -1.184717 .6950845 -1.70 0.088 -2.547058 .1776235 canalmaintainance -1.989945 1.01855 -1.95 0.051 -3.986266 .0063762 seceducation -2.410227 2.645236 -0.91 0.362 -7.594795 2.774341 basiceduc -2.294488 .7439413 -3.08 0.002 -3.752586 -.8363902 extensionagent -1.676765 .6744615 -2.49 0.013 -2.998686 -.3548452 mechanization -.9412408 .9398979 -1.00 0.317 -2.783407 .9009253 credit .719857 .6975764 1.03 0.302 -.6473676 2.087081 experience -.0343439 .0494067 -0.70 0.487 -.1311794 .0624915 farmdistance -.2869228 .1936912 -1.48 0.139 -.6665506 .092705 gender -2.377168 .9497523 -2.50 0.012 -4.238648 -.5156879 lnsig2u _cons -1.666931 .1753222 -9.51 0.000 -2.010556 -1.323306 lhlab .3060465 .205789 1.49 0.137 -.0972926 .7093856 lnflab -.0113607 .1734896 -0.07 0.948 -.3513941 .3286726 lnagro .000317 .2168716 0.00 0.999 -.4247435 .4253775 lnfert -.9624181 .3162355 -3.04 0.002 -1.582228 -.3426079 lnseed -.2300414 .3214366 -0.72 0.474 -.8600456 .3999628 lnsig2v _cons .0691374 .0941177 0.73 0.463 -.11533 .2536048 lhlab .1312517 .0427334 3.07 0.002 .0474958 .2150075 lnflab .0061636 .0446808 0.14 0.890 -.0814092 .0937365 lnagro -.1453162 .0607677 -2.39 0.017 -.2644188 -.0262137 fertd -.0039856 .0882552 -0.05 0.964 -.1769625 .1689914 lnfert .2288425 .0802328 2.85 0.004 .0715892 .3860958 lnseed .199294 .1050636 1.90 0.058 -.006627 .4052149 lny lny Coef. Std. Err. z P>|z| [95% Conf. Interval] Log likelihood = -100.77058 Prob > chi2 = 0.0000 Wald chi2(6) = 45.86 Stoc. frontier normal/half-normal model Number of obs = 191 University of Ghana http://ugspace.ug.edu.gh 84 Appendix III: Technical Efficiency Estimation (No production risk considered) Interval Number of farms Percentage (%) ≤0.50 13 7 0.51-0.60 10 5 0.61-0.70 21 11 0.71-0.80 24 13 0.81-0.90 59 31 0.91-1.00 64 34 Observation (N) 191 100 Std. Dev. 0.15 Maximum 1 Mean 0.81 Minimum 0.37 University of Ghana http://ugspace.ug.edu.gh