i UNIVERSITY OF CAPE COAST TECHNICAL EFFICIENCY IN RICE PRODUCTION ON THE WETA IRRIGATION SCHEME IN THE KETU NORTH DISTRICT OF THE VOLTA REGION, GHANA BY FRANCIS KASTRO KAVI Thesis submitted to the Department of Agricultural Economics and Extension, School of Agriculture, College of Agriculture and Natural Sciences, University of Cape Coast, in partial fulfilment of the requirements for the award of Master of Philosophy degree in Agricultural Economics. JULY 2015 ii DECLARATION Candidate’s Declaration I hereby declare that this thesis is the result of my own original work and that no part of it has been presented for another degree in this university or elsewhere. Signature: ................................................ Date ............................... Candidate’s Name: Francis Kastro Kavi Supervisors’ Declaration We hereby declare that the preparation and presentation of the thesis were supervised in accordance with the guidelines on supervision of thesis laid down by the University of Cape Coast Signature ................................................... Date .............................. Principal Supervisor’s Name: Prof. Henry De-Graft Acquah Signature ............................................... Date ............................ Co-Supervisor’s Name: Dr. William Ghartey iii ABSTRACT This study investigated technical efficiency in rice production on the Weta Irrigation Scheme in the Ketu North District of the Volta Region of Ghana during the 2014/2015 cropping season. A two-stage sampling procedure was used to select a sample of 290 rice farmers from a population of 1,024. Primary data, collected from 285 respondents using structured interview schedule, were used for the study. A translog stochastic frontier production function which incorporates a model for inefficiency effects, using the Maximum Likelihood Method was employed in the analysis of the data. Results indicated that the major input factors that significantly influenced the output of rice were land area under cultivation, fertiliser input, irrigation cost and equipment. The socio-economic characteristics of rice farmers which were significant determinants of technical efficiency in the study area were age, sex, farming experience and membership of a farmer based organisation. Low purchasing price of rice, lack of government support, difficulty in accessing capital and erratic rainfall patterns were identified as the major constraints faced by rice farmers in the study area. Furthermore, the mean technical efficiency index was estimated at 70.7 per cent which implies that the rice farmers were not fully technically efficient. Thus there was the opportunity for them to increase their output by 29.3 per cent via efficient reallocation of available resources. Also the results indicated decreasing returns to scale technology among the rice farmers. Finally, the study recommended among others that the Ministry of Food and Agriculture should introduce fertiliser subsidies and establish guaranteed prices for rice to encourage rice farmers to produce. iv ACKNOWLEDGEMENTS This thesis would not have been conducted and completed without the inputs of several personalities to whom I am ever grateful. I would like to place on record my appreciation and sincere acknowledgement of the invaluable assistance of my supervisors, Dr. Henry De-Graft Acquah and Dr. William Ghartey whose guidance, suggestions and constructive criticisms made it possible for me to complete this thesis successfully. I would like to appreciate the services and support provided by the following personalities: Mr. Emmanuel Laryea Tetteh, the Ketu North District Director, Ministry of Food and Agriculture for readily providing me with information on the study area; Mr. Theophilus Kofi Hagan, the manager of the Weta Irrigation Scheme, for giving me the sampling frame; Mr. Holy Ahiabu and Mr. Philip Amable who devoted their time to help in the collection of data; the rice farmers on the Weta Irrigation Scheme for enduring long periods of interviews that led to the collection of primary data and Miss Beatrice Bempomaa for her immense contributions in the analysis of data. Finally, I would like to express my gratitude to my wife and children for enduring my frequent absence from home due to this thesis and to my good friend, Mr. Eugene Yaw Milledzi for his encouragement, advice and useful suggestions during the preparation of this work. v DEDICATION To my family vi TABLE OF CONTENTS Page DECLARATIONS ii ABSTRACT iii ACKNOWLEDGEMENTS iv DEDICATION v TABLE OF CONTENTS vi LIST OF TABLES x LIST OF FIGURES xi LIST OF ABBREVIATIONS xii CHAPTER ONE: INTRODUCTION 1 Background to the study 1 Statement of the problem 4 Objectives of the study 6 Research questions 6 Significance of the study 7 Delimitations of the study 7 Limitations of the study 8 Organisation of the study 8 CHAPTER TWO: REVIEW OF RELATED LITERATURE 9 Theoretical Framework 9 Definition of technical efficiency 10 Forms of efficiency 11 vii Measurement of technical efficiency 12 Requirements for functional form specification 16 Output elasticities 22 Determinants of technical efficiency 23 Age of farmer 23 Education 24 Farming experience 25 Farmers’ household size 25 Sex of farmer 26 Access to credit 26 Off-farm activities 27 Extension services 27 Membership of farmer based organisations (FBOs) 28 Empirical review 28 Overview of the rice sector in Ghana 39 Status of rice in national policies 41 Rice consumer preferences, per capita consumption and demand Projections 42 Typology and number of rice farmers 43 Potential for rice sector development 44 Rice research program 46 History of irrigation development in Ghana 47 Irrigation schemes in Ghana 47 viii Rice irrigation schemes in Ghana 48 Constraints to rice production 48 Conceptual framework 52 CHAPTER THREE: METHODOLOGY 54 Introduction 54 Study area 54 Research design 60 Population 61 Sampling and sampling technique 61 Instrumentation 62 Pre-testing of instruments 63 Data collection procedure 63 Description of output and input variables for the study 64 Description of farmer specific variables for the study 65 Data analysis 67 Specification of the models 68 CHAPTER FOUR: RESULTS AND DISCUSSION 75 Description of socioeconomic characteristics of rice farmers 75 Technical efficiency levels of rice farmers 84 Diagnostic statistics and gamma parameters 86 Frequency distribution of technical efficiency of rice farmers 89 Elasticity of output 90 Determinants of technical efficiency 92 ix Constraints to rice production 95 CHAPTER FIVE: SUMMARY, CONCLUSIONS AND RECOMMENDATIONS 99 Summary of Main Findings 100 Conclusions 102 Recommendations 103 REFERENCES 105 APPENDICES 121 A Interview Schedule for Rice Farmers in the Ketu North District 121 B Technical Efficiency Scores of Individual Rice Farmers on the Weta Irrigation Scheme 127 C Sample Size Determination Table 130 D Dam at Dekpor 131 E Weta Rice Farms 132 x LIST OF TABLES Table Page 1. Likelihood ratio test 70 2. Distribution of respondents by age 76 3. Distribution of respondents by educational background 78 4. Distribution of respondents by household Size 78 5. Frequency distribution of rice farmers by marital Status 79 6. Summary statistics of production parameters and farmer Characteristics 83 7. Maximum likelihood estimates of the mean scaled translog Stochastic frontier production function parameters 87 8. Frequency distribution of technical efficiency of rice farmers in the Ketu North District 90 9. Elasticity of output and return-to-scale 91 10. Maximum likelihood estimates of the inefficiency model 95 11. Kendall’s W test statistics 96 12. Ranking of constraints faced by rice farmers 97 xi LIST OF FIGURES Figure Page 1. Input-oriented measure of technical efficiency 10 2. Conceptual framework 53 3. Ketu North District in national context 55 4. Distribution of respondents by sex 77 5. Distribution of respondents on their access to extension Services 79 6. Distribution of respondents by off-farm work 80 7. Distribution of respondents by membership of a farmer based organization. 81 8. Distribution of respondents by the variety of rice cultivated 82 xii LIST OF ABBREVIATIONS AAGS Accelerated Agricultural Growth and Development AEA Agricultural Extension Agent CARD Coalition for Africa Rice and Development CRI Crop Research Institute CSRI Council for Scientific and Industrial Research DADU District Agriculture Development Unit DEA Data Envelopment Analysis DIP Dawhenya Irrigation Project FASDEP I&II Food and Agriculture Sector Development Policy FBO Farmer – Based Organisation GIDA Ghana Irrigation Development Authority GPRS I Ghana Poverty Reduction Strategy I GPRS II Growth and Poverty Reduction Strategy II GRIB Ghana Rice Interprofessional Body HHI Herfindahl Index IFPRI International Food Policy Research Institute J. H. S. Junior High School LRT Likelihood Ratio Test METASIP Medium Term Agriculture Sector Investment Plan MiDA Millennium Development Agenda MLE Maximum Likelihood Estimate MOFA Ministry of Food and Agriculture xiii MRTS Marginal Rate of Technical Substitution NERICA New Rice for Africa NRDS National Rice Development Strategy NRI Natural Resources Institutes OLS Ordinary Least Squares PRA Participatory Rural Appraisal PTIPs Pump-type Irrigation Projects RTS Returns – To – Scale S.H.S Senior High School S.M.C.D Supreme Military Council Decree SARI Savanna Agricultural Research Institute TE Technical Efficiency TICAD Tokyo International Conference on Africa Development US United States 1 CHAPTER ONE INTRODUCTION Background to the study Rice is considered as the second most important grain food staple in Ghana, next to maize, and its consumption keeps increasing as a result of population growth, urbanization and change in consumer habits [Ministry of Food and Agriculture (MoFA, 2009)].. The total rice consumption in Ghana in 2005 amounted to about 500,000 tonnes and this is equivalent to per capita consumption of 22 kilograms per annum. According to MoFA, per capita consumption of rice per annum is estimated to increase to 63 kilograms by 2018 as a result of rapid population growth and urbanization. Between 1996 and 2005, paddy rice production in Ghana was in the range of 200,000 and 280,000 tonnes (130,000 to 182,000 tonnes of milled rice) with large annual fluctuations. In 2010, rice was the 10 th agricultural commodity in Ghana by value of production while it ranked 8 th in terms of production quantity for the period 2005-2010 (MoFA, 2010). Average rice yield in Ghana is estimated to be 2.5 tonnes/hectare while the achievable yield is 6–8 tonnes/hectare. This significant yield potential can be tapped through improvements in agronomic practices and adoption of underutilized beneficial technologies. Also, rice occupies roughly 4 per cent of the total crop harvested area, although it accounts for about 45 per cent of the total area planted to cereals (MoFA, 2009). In addition to being a staple food mainly for high 2 income urban populations, rice is also an important cash crop in the communities in which it is produced. Rice is also the most imported cereal in the country accounting for 58 per cent of cereal imports (CARD, 2010) accounting for 5 per cent of total agricultural imports in Ghana over the period 2005-2009. Ghana largely depends on imported rice to make up for the deficit in rice supply. On the average, annual rice import in Ghana is about 400,000 tonnes (MoFA, 2009). It is therefore important for stakeholders in the food and agriculture sector to ensure increased and sustained domestic production of good quality rice for food security, import substitution and savings in foreign exchange. Domestic rice production satisfies around 30 to 40 per cent of demand with a corresponding average rice import bill of US $450 million annually (MoFA, 2010). The massive dependency on rice imports has always been a concern for policy makers, especially after food prices soared in 2008. However, import duties and other taxes as well as interventions to boost productivity and quality of local rice do not seem to produce any substantial impact on Ghana’s import bill. In May, 2008, Ghana was one of the first countries within the Coalition for Africa Rice and Development to launch its National Rice Development Strategy (NRDS) for the decade 2009 – 2018. The main objective of the NRDS is to double domestic production by 2018, implying a 10 per cent annual growth rate and enhance quality to stimulate demand for domestically produced rice. These increases will most likely come from utilizing potential irrigable lands and valley bottoms with water supply, promoting rice production, and increasing the productivity of existing growers. 3 The inability of local rice production to meet domestic demand can be attributed to the inability of the rice farmers to obtain maximum output from the resources committed to the enterprise (Kolawole, 2009). According to Rahman, Mia and Bhuiyan (2012), farm level performance can be attained in two alternate ways: either by maximizing output with the given set of inputs or by minimizing production cost to produce a prescribed level of output. The former concept is known as technical efficiency which is a measure of a farm firm’s ability to produce maximum output from a given set of inputs under certain production technology. It is a relative concept in so far as the performance of each production unit is usually compared to a standard. The standard may be used on farm-specific estimates of best practice techniques (Herdt & Mandac, 1981) but more usually by relating farm output to population parameters based on production function analysis (Timmer, 1971). A technically efficient farm operates on its frontier production function. Given the relationship of inputs in a particular production function, the farm is technically efficient if it produces on its production function to obtain the maximum possible output, which is feasible under the current technology. Put differently, a farm is considered to be technically efficient if it operates at a point on an isoquant rather than interior to the isoquant. Technical efficiency in agriculture production is an important element in the pursuit of output growth. A high level of technical efficiency implies that output is being maximized given the available technology. In this situation, output growth will be achieved through the introduction of new technology that will shift the production frontier outward. A low level of technical efficiency, on the other hand, indicates that output growth can be achieved given current inputs and available technology. Therefore, it is 4 important to determine the degree of technical efficiency among farmers, and if low technical efficiency is found, to investigate the factors that will increase efficiency. Measurement of productive efficiency of a farm relative to other farms or to the best practice in an industry has long been of interest to agricultural economists. From an applied perspective, measuring efficiency is important because it is the first step in a process that might lead to substantial resource savings. These resource savings have important implications for both policy formulation and farm management (Bravo-Ureta & Rieger, 1991). For individual farms, gains in efficiency are particularly important in periods of financial stress since the efficient farms are likely to generate higher incomes and thus stand a better chance of surviving and prospering. Statement of the problem In an economy where resources are scarce and opportunities for new technologies are lacking, further increase in output can best be brought about through improvement in the productivity of the crop. In this context, technical efficiency in the production of a crop is of paramount importance. Measurement of technical efficiency (TE) provides useful information on competitiveness of farms and potential to improve productivity, with the existing resources and level of technology (Abdulai & Tietje, 2007). Moreover, investigating factors that influence technical efficiency offers important insights on key variables that might be worthy of consideration in policy-making, in order to ensure optimal resource utilisation. 5 Several studies have been conducted on rice production in Ghana. However, most of these studies on rice focused on other areas rather than the technical efficiency of production. Examples are: “Impact of improved varieties on the yield of rice producing households in Ghana” (Wiredu, et al., 2010); “Cooking characteristics and variations in nutrient content of some new scented rice varieties in Ghana”(Diako et al., 2011), “Rice price trends in Ghana” (Amanor-Boadi, 2012) and “Patterns of adoption of improved rice technologies in Ghana’’(Ragasa et al., (2013). Even though some studies have been conducted on technical efficiency in rice production in Ghana, most of these studies are concentrated in Northern Ghana; especially on the Tono Irrigation Scheme, for example Technical efficiency in rice production at the Tono irrigation scheme in Northern Ghana (Donkoh, Ayambila & Abdulai,2012). Besides, in the Ketu North District rice is a major food crop and its production serves as a source of employment for many years. Yet, not much study has been conducted to determine the technical efficiency of rice farmers. The aforementioned reasons informed a study to be conducted to determine the technical efficiency in rice production on the Weta irrigation scheme in the Ketu North District of the Volta Region of Ghana. This would fill the knowledge gap and inform policy decisions. 6 Objectives of the study Generally, the study seeks to measure the technical efficiency in rice production in the Ketu North District of the Volta Region. The specific objectives include: 1. To estimate the level of technical efficiency of rice farmers in the district. 2. To analyse the determinants of technical efficiency. 3. To compute the output elasticities in rice production. 4. To identify and rank the constraints in rice production with respect to technical efficiency. Research questions The following research questions have been developed to guide the study. 1. What are the levels of technical efficiency in rice production in the district? 2. What are the output elasticities in rice production? 3. What are the determinants of technical efficiency? 4. What are the constraints to rice production in the district? Hypotheses The study seeks to test the following hypotheses: 1. H0: The translog functional form does not represent the data more adequately than the Cobb-Douglas H1: The translog functional form represents the data more adequately than the Cobb-Douglas 2. H0: Rice farmers are not fully technically efficient. H1: Rice farmers are fully technically efficient. 3. H0: The socio-economic characteristics of rice farmers have no significant influence on technical efficiency. 7 H1: The socio-economic characteristics of rice farmers significantly influence technical efficiency. 4. H0: There is no concordance among rice farmers regarding the constraints to rice production. H1: There is concordance among rice farmers regarding the constraints to rice production. Significance of the study Farmers are rational and tend to make production decisions in favour of crops that yield the most benefits to them. Therefore, information on technical efficiency in rice production is essential to rice farmers. The findings of this study would help to better understand where public investments can best be directed to effectively increase technical efficiency of rice farmers in Ghana. Increased rice output will address food security issues in the country as well as improve farmers’ livelihoods by increasing their incomes. Furthermore, determining the factors that influence technical efficiency will serve as a basis for policy and strategy development so as to enhance rice production in the country. Also, increased output of rice farmers will help reduce rice imports to save foreign exchange and strengthen the Ghana Cedi. Finally, the findings of this study would help fill the knowledge gap identified during literature review. Delimitations of the study The focus of this study was to estimate the levels of technical efficiency and its determinants among rice farmers on the Weta irrigation scheme in the Ketu North District of the Volta Region. It did not include other rice farmers in the district who were not on the irrigation scheme. Also, the 8 scope of the study is delimited to the estimation of technical efficiency and its determinants among the rice farmers but not the allocative efficiency which takes into consideration the respective prices of the inputs. Limitations of the study In this study, data were collected from rice farmers most of whom were illiterates or had only primary education and hence could not keep accurate records. Also, all the items on the instrument had to be translated from English to Ewe for the respondents. Furthermore, production technology was assumed to be constant for rice farmers on the Weta irrigation scheme. All the above could negatively affect the quality of data collected as well as the accuracy of results obtained. Organisation of the study The study is organised into five chapters. The first chapter is the introductory chapter and it covers the background of the study, statement of the problem, objectives of the study, research questions, significance of the study, delimitations of the study, limitations of the study and the organisation of the study. Chapter two provides a review of related literature and the relevant variables of the study. The review also covers empirical studies related to the study. Chapter three describes the research methodology. The discussion of results of data analyses are presented in chapter four while summary, conclusions and recommendations are presented in chapter five. 9 CHAPTER TWO REVIEW OF RELATED LITERATURE In this chapter, the relevant literature relating to the study has been reviewed. The review covers the following areas: theoretical framework, definition of technical efficiency, forms of efficiency, measurements of technical efficiency, requirement for a functional form, ou`tput elasticity, the determinants of technical efficiency, overview of the rice sector in Ghana, the constraints to rice production and conceptual framework. Theoretical Framework Kumbhakar and Lovell (2000) define technical efficiency as a measure of how well individuals convert inputs to output with a given level of technology and economic factors. Technical efficiency is given as the ratio of observed output to the corresponding frontier output ( with given levels of inputs and technology. Thus, ( ).Therefore, technical inefficiency exists if a farmer produces below the production frontier. This phenomenon is also influenced by inefficiency factors such as level of education, age of farmer, household size, and farming experience. In this study, technical efficiency is considered from the angle where a farmer uses a minimum combination of inputs to produce a given level of output (an input-oriented measure of technical efficiency). This is shown in figure 1: From figure 1, the firm uses the combination of inputs(x1, x2) defined by point A to produce a given level of output ( , ). The same level of 10 output could have been produced by combining smaller amounts of the inputs (Isoquant ( , )). This is defined by point B and it lies on the isoquant. Therefore, the input-oriented level of technical efficiency (TEI) is defined by OB/OA. The distance BA shows the technical inefficiency of the farm firm and it represents the amount by which all inputs could be proportionally reduced without a decrease in output. Farell (1957) defines allocative efficiency as the ability of a firm to choose the optimal combination of inputs given input prices. At point C, the marginal rate of technical substitution (MRTS) is equal to the ratio of the input prices ( ).This implies that the input allocative efficiency is given by the fraction OD/OB (Coelli, 1995). Figure 1: An input- oriented measure of technical efficiency Source: Kumbhakar and Lovell (2000). Definition of technical efficiency One of the basic thrusts of economics of agricultural production at the micro level is to assist individual farmers or a group of farmers to attain their objectives through efficient intra-farm allocation of resources at a particular time or over a period. Efficiency is achieved either by maximizing output from 0 A B C D 11 given resources or by minimizing the resources required for producing a given output (Varian, 1992).The first analyses of efficiency measures were initiated by Farrell (1957). Drawing from Debreu (1951) and Koopmans (1951), Farrell proposed a division of efficiency into two components: technical efficiency, which represents a firm's ability to produce a maximum level of output from a given level of inputs, and allocative efficiency, which is the ability of a firm to use inputs in optimal proportions, given their respective prices and available technology. The combination of these measures yields the level of economic efficiency. In microeconomics of production, technical efficiency is defined as “the maximum attainable level of output for a given level of inputs, given the current range of alternative technologies available to the farmer” (Ellis, 1993). Technical efficiency can be analysed using two approaches. These are the output-oriented and input-oriented approaches. The first one has output augmenting orientation, whereas the second one is targeted to conserve inputs (Koopmans, 1951). An output-oriented technical efficiency occurs when the maximum amount of an output is produced for a given set of inputs while an input-oriented technical efficiency occurs when the minimum amount of inputs are required to produce a given output level (Farrell, 1957). Therefore, technical efficiencies are derived from production functions or production possibility frontiers. Forms of efficiency Apart from the components of economic efficiency (technical and allocative efficiency), Leon (2001) came out with five additional categories of efficiency. These include: 12 1. Technological efficiency: the ability to produce an output by using the best technology available. 2. Pure technical efficiency: with the given amount of technology, output should not be produced in a manner such that inputs used would be more than necessary. The pure technical efficiency measure is derived by estimating the frontier output under the assumption of variable returns to scale. 3. Scale efficiency: the ability of the firm to produce at a suitable level by exploiting scale economies. Thus, management is able to choose the scale of production such that the optimum output is obtained. Inefficiency can arise when the size of a firm is too small or too large. 4. Dynamic efficiency: operating at the optimum level by incorporating innovations in products and processes. 5. Approach efficiency: it defines the ability of the firm to choose the appropriate technology with respect to the challenges that arises in the market. Measurement of technical efficiency Previous studies on efficiency of a farm can be classified broadly into three categories, namely; deterministic parametric estimation, non-parametric mathematical programming and the stochastic parametric estimation (Udo & Akintola, 2001). The use of non-parametric techniques are limited in efficiency measurement in agriculture despite the fact that non parametric methodologies can be used in situations where data are more limited and where production techniques are less well understood (Llewelyn & William, 1996). The non-parametric approach, which is the Data Envelopment Analysis (DEA) was initiated by Farrell (1957) and transformed into estimation techniques by Charnes, Cooper and Rhodes (1978). DEA is based on linear 13 programming and consists of estimating a production frontier through a convex envelope curve formed by line segments joining observed efficient production units. No functional form is imposed on the production frontier and no assumption is made on the error term. The DEA is also useful for multiple- input and multiple- output production technologies (Khai & Yabe, 2011). However, the non-parametric approach is limited because it cannot separate the effects of noise and inefficiency during the calculation of technical efficiency, and is less sensitive to the type of specification error (Kebede, 2001). Thus it assumes the absence of measurement or sampling errors and deviations from the production frontier are under the control of the production unit being considered. Also, it is very sensitive to extreme values and outliers and lacks the statistical procedure for hypothesis testing (Nchare, 2005). . The deterministic parametric estimation of technical efficiency for a cross-section of producers was first introduced by Farrell (1957).The deterministic frontier approach also assumes that any deviation from the frontier is due to inefficiency. The limitations of this approach arise due to the fact that it ignores any random factors that may influence the efficiency of a firm. Thus, the results of a deterministic frontier approach are highly sensitive to the selection of variables and data errors (Bravo-Ureta & Pinheiro, 1997). Farrel suggested a method of measuring technical efficiency of a farm by estimating the production function of a farm that are fully efficient, that is a frontier production function. The efficiency however, is bounded between zero and one, where scores of one indicate full efficiency (Tanko, 2009; Ezekiel, Adedapo & Olapade, 2009; Kolawole, 2009; Udoh & Akintola, 2001; Bagi, 1984; Battesse & Coelli, 1995). 14 Aigner and Chu (1968) developed stochastic frontier models, which acknowledge the influence of random errors and data noise on agricultural production. This approach assumes that deviations from the production frontier may not be entirely under the control of farmers. In so doing, it helps distinguish the effects of stochastic noise from the effects of other inefficiency factors. It also allows hypothesis testing on the structure of production and efficiency. Also,Aigner, Lovell and Schmidt (1977) and Meeusem and Van den Broeck (1977) independently developed a stochastic frontier approach to overcome some of the limitations of the deterministic frontier approach. The stochastic frontier approach, an econometric estimation, is made up of an error term which consists of two components, one being random noise, which is factors beyond the farmer’s control (e.g. weather conditions) and the other being a one-sided residual term representing inefficiency. The inefficiency component represents farmer-specific characteristics such as age of farmer and farming experience. A stochastic frontier production function representing a firm’s production is given by: =f( , β)+ (i) Where is the output of the farmer is a vector of the input quantities β is a vector of unknown parameters to be estimated = (v - u) is the composite error term, where v is a random error that captures the stochastic effects that are beyond the farmer’s control; such as weather conditions and diseases. The vs are assumed to be independent and identically distributed normal random errors having zero mean and unknown variance, 2v. The us are non-negative random 15 variables, called technical inefficiency effects, which are associated with technical inefficiency of production of the farmers which are assumed to be independent of the vs such that u is the non-negative truncation (at zero) of the normal distribution with mean ui, and variance,  2 u, where ui is defined by: ui=δ0+δ1Z1+δ2Z2+δ3Z3+…+δnZn (ii) Where, ui = Technical inefficiency of the ith farmer, δ1-δn = Unknown parameters to be estimated, Z1-Zn= Farmer-specific characteristics. The econometric approach, however, gives good results only for single output and multiple inputs technologies (Kebede, 2001).Criticisms of the stochastic approach resides in the need to specify beforehand, the functional form of the production function and a distributional form of the inefficiency term. Since rice production in the study area is an example of single output and multiple inputs production, this study focuses on the use of an econometric approach, which is the Stochastic Frontier Approach for measuring technical efficiency. This choice was made on the basis of the variability of agricultural production which is attributable to climatic hazards, plant pathology and insect pests on the one hand, and, on the other hand, because information gathered on production is usually inaccurate since small scale farmers do not have updated data on their farm operations. Furthermore, in analysing farm level data where measurement errors are substantial and weather is likely to have a significant effect, the stochastic frontier method is usually recommended (Coelli, 1995). 16 Requirements for functional form specification In order to analyse a production function, it is always necessary to specify a particular functional form that will depict the production technology. In most cases, the production function is known to be decomposed into a deterministic and a stochastic part. The deterministic part which is made up of observed inputs variables and unknown parameters become the algebraic function. Some criteria have been formulated by Lau (1978 & 1986) to help choose a particular functional form suitable for measuring certain economic relationships. These criteria have been categorized into: 1. Theoretical consistency: Estimating the parameters of a particular economic relationship involves the ability to choose an appropriate functional form that can represent observations of a production set. This implies that the functional relationship must be single valued, monotonic (additional units of an input will not decrease output) and concave (all marginal products are non-decreasing). 2. Domain of applicability: This relates to how the algebraic functional form fulfils all the theoretical conditions given the set of values of the independent variables. It further shows that a functional form must be well behaved in a range of observations, coherent with upheld hypotheses and acknowledge computational techniques to check those properties. Also, in order to predict relations, functional forms should be well matched with upheld hypotheses outside the range of observations. 3. Flexibility: A functional form is said to be flexible if the derived input demand functions and the derived elasticities of the chosen parameters are able to assume arbitrary values subject to only theoretical consistency with any given set of non-negative inputs. This principle permits available data the chance to give information about the critical parameters. 17 4. Computational facility: This principle implies that the functional form must be linear, uniform, parsimonious and representative. Thus, for estimation purposes functional forms should be linear-in-parameters and possible restrictions should also be linear. Again, any input demand function derived from the functional form should be represented in an explicit closed form so as to make calculation easy. To prevent methodological problems like multi- collinearity and loss of degree of freedoms, diverse functions should vary in parameters but should have the same ‘uniform’ algebraic form. 5. Factual consistency: This relates to how the modelled economic problem is consistent with other recognized empirical facts. Lau (1978) noted that one should not expect a particular algebraic functional form to meet all the requirements of the above criteria. Generally, a functional form can be considered as a suitable description of the production possibility set when it is theoretically consistent and flexible. In literature, four functional forms namely; the linear production function, the quadratic production function, the Cobb-Douglas production function, and the transcendental logarithmic (translog) production function are normally used to represent the technology of a data set (Henningsen, 2013). 1. The linear production function: It is known to be first order flexible as it has adequate parameters to provide a first-order differential approximation to an arbitrary function at a single point. 2. The quadratic production function: It has a second-order flexibility functional form as it has enough parameters to give a second-order approximation. 3. The Cobb-Douglas production function: the Cobb-Douglas production function is a first-order flexible extension of the quadratic production function. 18 This function is popular because its derivatives are simple. It can be linearized by transforming both sides of the function to natural logarithm. The corresponding coefficients in the function are equal to the output elasticities of the inputs and the sum of all the output elasticities gives the scale elasticity. The production technology portrayed by the Cobb-Douglas production function shows that output quantity becomes zero whenever a particular input quantity is zero. A restrictive property of this function is that input level variations do not cause output elasticities to change (Debertin, 2012). 4. The translog production function: It was introduced by Christensen, Jorgenson & Lau in 1971. It is a more flexible extension of the Cobb-Douglas production function as the Cobb-Douglas production function is “nested” into the translog production function. This function can accommodate any number of input sets and the elasticity of substitution of each pair of inputs may vary. It can be seen as a combination of the Cobb-Douglas function and the quadratic function. The translog production function would not become an appropriate functional form for analysing data sets when the output quantity of the data is positive whiles at least one of its input quantities is zero. This renders the weak and strict essentiality assumption incapable. Also, all translog production functions are continuous in nature and can be differentiated twice continuously. Generally, second order differentials are preferred to the first order differentials but this can result in muticolinearity problems as there would be more parameters to be estimated. However, Kopp and Smith (1980) asserted that the effects of functional forms on efficiency analysis are limited. Econometric modelling of stochastic frontier methodology of Aigner, Lovell and Schmidt (1977) associated with the estimation of efficiency has 19 been an important area of research in recent years. Basically, the studies are mostly based on Cobb-Douglas function and transcendental logarithmic (translog) functions that could be specified either as production function or cost function. The first application of stochastic frontier model to farm level agricultural data was by Battesse and Corra (1977). But technical efficiency of farms was not directly addressed in the work. Kalirajan (1981) estimated a stochastic frontier Cobb-Douglas production function using cross-sectional data and found the variance of farm effects to be a highly significant component in describing the variability of rice yield. Bagi (1984) used the stochastic frontier Cobb-Douglas production function model to investigate whether there were any significant differences in the mean technical efficiencies of part-time and fulltime farmers. Results showed no apparent significance, irrespective of whether the part-time and full-time farmers were engaged in mixed farming or crop cultivation only. The main advantage of the translog is that it is flexible, which implies that it does not impose assumptions about constant elasticity of production nor elasticities of substitutions between inputs. The translog permits the partial elasticities of substitution between inputs to vary, that is, the elasticity of scale can vary with output and factor proportions, permitting its long-run average cost curve to take the traditional U-shape. The explicit translog stochastic frontier production function used in this study is given in equation (iii):  0 1 1 1 1 ln ln ln ln 2 n n n i i i ij i j i i i i j Y x x x v u            (iii) 20 where in is the natural logarithm, iY is the output of rice (kilograms) produced during the 2014/2015 cropping season by the thi farmer and the jth observation; x is a set of ‘n’ inputs used to produce a given output where n is number of inputs; iv denotes random shocks; iu is the one-sided non-negative error representing inefficiency in production. After estimating technical efficiency, based on the stochastic production frontier, literature proposes two main approaches to analysing the factors influencing technical efficiency sources: the two-stage estimation procedure and the one-stage simultaneous estimation approach. The two-stage approach requires one to first estimate the stochastic production function to determine technical efficiency indicators and then, secondly, regress the derived efficiency scores on explanatory variables which usually represent the farm’s specific characteristics using ordinary least square (OLS) method or Tobit regression. This two-step approach has been used by authors such as Pitt and Lee (1981), Kalirajan (1981), Parikh, Ali and Shah (1995), and Ben- Belhassen (2000) in their respective studies. The major drawback with this approach is in the fact that, in the first step, the inefficiency effects, µs are assumed to be independent and identically distributed in order to use the approach of Jondrow, Lovell, Materov and Schmidt (1982) to predict the values of technical efficiency indicators. In the second step however, the technical efficiency indicators thus obtained are assumed to depend on a certain number of factors specific to the farm, which implies that the µs are not identically distributed, unless all the coefficients of the factors considered happen to be simultaneously null. 21 After becoming aware that the two-step approach displayed these inconsistencies, Kumbhakar, Gosh and McGuckin (1991) and Reifschneider and Stevenson (1991) developed a model in which inefficiency is defined as an explicit function of certain factors specific to the firm, and all the parameters are estimated in one-step using the maximum likelihood procedure. By following this second approach, Huang and Liu (1994) developed a non- neutral stochastic frontier production function, in which the technical inefficiency effects are a function of a number of factors specific to the firm and of interaction among these factors and input variables introduced in the frontier function. Battese and Coelli (1995) also proposed stochastic frontier production function for panel data in which technical inefficiency effects are specified in terms of explanatory variables, including a time trend to take into account changes in efficiency over time. The technical inefficiency effects are expressed as: µj = Zjδ (iv) Where; µj refers to the mean of the normal distribution that is truncated at zero to define the truncated normal distribution associated with the inefficiency effect for farm j, z is a vector of observed explanatory variables and δ is a vector of unknown parameters. Thus, the parameters of the frontier production function are simultaneously estimated with those of an inefficiency model, in which the technical inefficiency effects are specified as a function of other variables including socioeconomic and demographic factors, farm characteristics, environmental factors and non-physical factors. The one-step approach has since been used by such authors as Ajibefun et al. (1996), Coelli and Battese (1996), Audibert (1997), Battese and Sarfaz (1998), and Lyubov 22 and Jensen (1998) in their respective studies to analyse the factors affecting the technical efficiency (or inefficiency) of agricultural producers. The one-stage simultaneous estimation approach was adopted to analyse the factors that influence technical efficiency. The single stage approach was adopted because, relative to the two-step approach, the one-step approach presents the advantage of being less open to criticism at the statistical level, and helps in carrying out hypothesis testing on the structure of production and degree of efficiency (Nchare, 2005). Output elasticities The output elasticities calculated from the translog production function are given as: ln ln ln n i i ij j ji Y x x          (v) (Henningsen, 2013). The return to scale of the technology, RTS can be calculated by the sum of the elasticities of scale, given by i i   (vi) If the technology has increasing returns to scale (ε >1), total factor productivity increases when all input quantities are proportionally increased, because the relative increase of the output quantity y is larger than the relative increase of the aggregate input quantity x in equation (v). If the technology has decreasing returns to scale (ε <1), total factor productivity decreases when all input quantities are proportionally increased, because the relative increase of the output quantity y is less than the relative 23 increase of the aggregate input quantity x . If the technology has constant returns to scale (ε = 1), total factor productivity remains constant when all input quantities change proportionally, because the relative change of the output quantity y is equal to the relative change of the aggregate input quantity x. If the elasticity of scale (monotonically) decreases with firm size, the firm has the most productive scale size at the point, where the elasticity of scale is one (Henningsen, 2013). Determinants of technical efficiency The ability of farmers to combine inputs in an efficient manner to produce a maximum output is influenced by the decisions taken during the production process. These decisions which are influenced by farmer specific characteristics are examined to ascertain their impact on technical efficiency. The farmer specific characteristics include age, education, farming experience, sex, access to credit, household size, access to extension services, off-farm work and membership of a farmer based organization. Age of farmer In literature, the influence of age on technical efficiency tends to have conflicting views. It has been argued that a positive relationship exists between older farmers and technical inefficiency. Older farmers have a tendency to stick to their old methods of production and are usually unwilling to accept change. However, the younger generations of farmers prefer taking risks. On the other hand, older farmers are considered to be more technically efficient. Older farmers are known to be wealthier than the younger farmers as a result of the wealth they accumulate over the years in farming. So they are 24 able to buy the necessary inputs to undertake production. Adequate inputs coupled with long years of farming enable them to produce efficiently (Ali, Imad & Yousif (2012). Battese and Coelli (1995) asserts that expected sign for age with respect to technical efficiency is not clear. Therefore, it is anticipated that age could have either a positive or negative influence on technical inefficiency. Education Education is viewed as an important stock of human capital. It enhances the literacy and skills of farmers and this helps them to process agricultural information in their production activities. Evidence suggests that higher levels of schooling lead to higher levels of productivity as education has been found to have significant positive relationship with productivity (Hayami & Ruttan, 1985). Lin (1991) has pointed out that better educated farmers are more eager and faster when it comes to the adoption of new technologies and modern practices. They are not much afraid of the risk involved in such technologies. Similarly, Welch (1970) identified two different ways in which education can affect agricultural productivity. The first is the “worker effect” and it reflects how well educated farmers are able to use a given amount of resources more efficiently. The latter, “allocative effect”, also describes how an educated farmer obtains and deciphers information about costs and productive characteristics of other inputs. This improves farmers’ access to information enabling them to pay and receive better prices for inputs bought and outputs sold. Therefore, it is expected that education will have a positive relationship with technical efficiency. 25 Farming experience The age of a farmer is used as a proxy for measuring farming experience. Thus, experience gained by farmers increase as they advance in age. The numbers of years spent in farming can affect technical efficiency positively or negatively. Farmers with long years of farming are assumed to be more experienced in production activities. They are therefore able to make and take better decisions with regard to risk and inputs combination. Long years of farming experience can influence technical efficiency negatively in the sense that farmers may develop the habit of sticking themselves to the use of obsolete technology (Onyenweaku & Nwaru, 2005). It is anticipated that farming experience will either influence technical efficiency positively or negatively. Farmers’ household size Household size, be it large or small constitutes the family labour especially in most developing countries. The availability of labour especially during the peak periods in farming activities influences the technical efficiency of farmers. Families with large size normally depend on their members to carry out production activities and therefore may rely less on hired labour. It has been found that a positive relationship exists between household size and technical efficiency, indicating that families with large household size are technically efficient than those with small household size (Jema, 2007). It is expected that households with less family size would be technically inefficient in production. 26 Sex of farmer Regardless of women’s significant contribution to agricultural productivity, their output level is often constrained by lack of access to productive resources. Women’s ability to gain access to agricultural resources is influenced by socioeconomic factors such level of education, access to credit and extension service. In most developing countries, men hold much power and control when it comes to decision making in the household level. Therefore, decision making of most women are based on what their male counterparts think is best (Balk, 1997). The variable sex is known to be an important determinant of technical efficiency. It has been found that male farmers are technically efficient compared to the female household heads. This could be explained by the fact that male household heads are wealthier and therefore are able to acquire technologies that are costly (Onumah &Acquah, 2010). Therefore, it is expected that there would be a negative relationship between the variable sex and technical inefficiency. Access to credit Financial institutions, both formal and informal are means by which farmers gain access to credit. However, in most developing countries, farmers tend to depend more on the informal institutions for financial assistance. The collateral security expected by the formal institutions and the high interest rates are barriers to farmers’ access to credit. Furthermore, uncertainties in agricultural production can affect the productivity of farmers negatively. These factors limit farmers in their ability to pay back loans (Heidhues, 1995). The ability of a farmer to adopt improved technologies depends on credit accessibility. Adequate and timely access to credit is important in agricultural 27 productivity. These will determine the farmers’ access to inputs such as machinery, improved seeds, labour and fertilizer among others. Therefore, it is expected that access to credit will have a negative influence on technical inefficiency. Off-farm activities Off-farm activity is an additional work engaged in by farmers apart from farming to supplement household income. Studies have shown that farmers who engage in other non-agricultural activities are likely to be less efficient as they may fail to pay much attention to farm production (Ali & Flinn, 1989). Therefore, it is expected that off-farm work will positively influence technical inefficiency. Extension services The sources of information available to farmers have been divided into two. These are interpersonal, the face to face exchange of information and impersonal sources where one or more persons are able to get into contact with many at a time such as the use of the mass media (Okunade, 2007). In today’s agricultural development, the access to agricultural information is necessary for increasing production. It has been argued by Singh, Priya and Singh (2011) that farmers should have access to information regarding new methods of production, seeds, fertiliser use to enhance their output. Although, access to extension services enhances efficiency, it has been found that some extension systems perform poorly as a result of organisational inefficiencies, unsatisfactory program designs and ineffective system of information delivery (Binam, Toyne, Wandji, Nyambi & Akoa, 2004). Farmers who have extension contacts are expected to improve their efficiency levels. 28 Membership of farmer based organizations (FBOs) Membership to FBO’s serves as a platform by which information regarding the availability of inputs and market prices are disseminated. Also, members of FBO’s benefit from education on good agricultural practices. Membership of farmers’ association is known to reduce the inefficiency level of farmers (Idris, Siwar & Talib, 2013). The active participation of farmers in their various groups also gives them the chance to share modern agricultural practices with the other farmers. Therefore, a negative relationship between membership of FBO’s and technical inefficiency is expected. Empirical review Various researchers have undertaken studies to estimate the efficiency levels in the agricultural sector (Russell & Young, 1983; Battese & Coelli, 1988);Parikh et al., 1995; Ajibefun et al., 1996; Heshmati & Mulugeta, 1996;Wang , Cramer & Wailes ,1996; Seyoum , Battese & Fleming, 1998; Obwona, 2000; Chirwa, 2003; Ajibefun & Aderinola, 2004; Nchare, 2005). These studies found positive correlations between the degree of technical inefficiency and education levels of farmers, age of farmers, land size, proportion of hired labour used, per capita net income and negative correlations between the degree of technical inefficiency and farmining experience and off farm employment. A study conducted by Amaechi, Ewuziem and Agumanna (2014) estimated the technical efficiency of the small/semi-mechanized oil palm produce millers in Nigeria using the translog stochastic frontier production function model. A multi-stage sampling method was used to select 30 mills in the study area and cost route approach used in data collection. The estimates 29 for the mills showed firm level mean technical efficiency of 70.62 per cent with a range of 37.48 per cent to 93.46 per cent. This wide variation in oil palm produce output of millers from the frontier output was found to have arisen from differences in miller’s management practices rather than random variability. This also implies that even under the existing technology, potentials exist for improving productive efficiency with proper utilization of available resources. Also, the estimated gamma value was 0.96; indicating that 96 per cent of the variation in output was attributed to technical inefficiency. Education, processing experience, membership of cooperative society, credit, capital, fruits throughput, petroleum energy and water were significant and positive determinants of technical efficiency while age, household size and interest on loans were negatively related to technical efficiency. Policies geared towards the enhancement of productive efficiency of this category of producers should appropriately address such issues as education, cooperativeness, and access to credit/capital, oil palm plantation rehabilitation, sustainable petroleum energy and supply of other necessary facilities. In the Eastern Region of Ghana, Kuwornu, Amoah and Seini (2013) conducted a study to analyse technical efficiency of maize farmers. They estimated the coefficients of the inputs variables by employing the translog functional form. Some of the variables they hypothesised to influence technical efficiency included educational level, extension visits, credit in cash or kind, household size, farm experience, farmer based organisation (FBO) membership among others. Findings of the study revealed that FBO membership, frequency of meeting by members of FBOs, credit in cash or kind and formal training in maize farming were the key factors that influenced technical efficiency levels of maize farmers in that region. The coefficient of 30 farmers who belonged to one group or the other was positive implying that they were less technically efficient than their colleagues who did not belong to any farmer based organisation. This outcome was associated with the fact that groups that meet throw less light on factors such as agronomic practices but spend much time on institutional factors such as favourable market outlets. Again, it was shown that farmers who had access to credit in cash or kind perform better in terms of production than their counterparts who had none. The availability of credit enables farmers to acquire necessary inputs and take certain management decisions on time. Further, the study indicated that the mean technical efficiency level of maize farmers was 0.51 or 51 per cent. This implies that the difference between the actual and potential output is 0.49 or 49 per cent. Also, the output elasticity for maize production was estimated to be 0.47 indicating decreasing returns to scale. Etwire, Martey and Dogbe (2013) used soybean farms to examine technical efficiency and its determinants in the Saboba and Chereponi Districts of Northern Ghana. The estimation of the technical efficiency level was carried out using the stochastic frontier model. It was found out that the Cobb- Douglas production function did not adequately fit the data. Therefore, it was concluded that the functional form that adequately represented the data was the translog production function. Further, the study showed a positive relationship between output levels of soybean and family labour as well as land. Based on this finding, output levels of soybean will increase if land under cultivation and family labour employed are increased up to a certain point, all other things being equal. Also, the output elasticities with respect to hired labour, seed and ‘other’ inputs were found to be negative. Thus, any further increase in seed, hired labour and ‘other’ inputs would influence output 31 negatively. Farmer characteristics such as education and marital status had negative coefficients but were found to be statistically insignificant in determining technical efficiency. Age had a negative coefficient and was found to be statistically significant. This implies that younger farmers were less technically efficient as compared to older farmers. Njeru (2010) examined the factors influencing technical efficiency in wheat farming in Kenya using a stochastic frontier production function in which technical inefficiency effects were assumed to be functions of both socioeconomic characteristics of the farmer and farm-specific characteristics. The paper used random sampling to interview 160 farmers comprising 97 large-scale farmers and 63 small-scale farmers. The results revealed existence of significant levels of technical inefficiencies in wheat production, especially among the large-scale farmers. The study found that the magnitude of technical efficiency varied from one farmer to another and ranged from 48.9 per cent to 95.1 per cent with a mean of 87.2 per cent. This implied that farmers lost the average, 13 per cent of their potential output to technical inefficiencies. There was variation depending on the size of farm with small- scale farmers attaining higher technical efficiency than the large-scale farmers. The main factors that influenced the degree of inefficiency were education levels, access to credit, and ownership of the capital equipment. Higher levels of education (12 years and above or secondary and above) significantly reduced inefficiency as did access to credit facilities and owning the farm equipment. The study recommended that farmers be educated on the use of better techniques such as use of certified seeds and application of recommended levels of fertilizer. 32 Ogunniyi and Ajao (2011) presented the analyses of cross-sectional data on swine production in Oyo state of Nigeria, a Cobb Douglas production function, in which the technical inefficiency effects were specified in terms of age, level of education and farming experience of the farmers was estimated for swine farmers in Oyo State of Nigeria. The stochastic frontier production function involved feed, family and hired labour, drug and total machinery cost. The maximum likelihood estimate results revealed that feed, family labour, hired labour and drugs were the major factors associated with changes in output of swine. Given the specification of the Cobb Douglas stochastic frontier function, the results indicated that the technical inefficiencies of production of swine farmer were specifically related to age, level of education and farming experience. Furthermore, the coefficient for the age variable found to be positive implying that the older farmers are more technically inefficient than the younger farmers. Also, the coefficient for the level of education was positive implying that the higher the level of formal education of farmers, the more their technical inefficiency. However, the coefficient for farming experience was negative indicating that farmers with more experience tend to be less inefficient. It was evident from the Maximum likelihood Estimates of the parameters of the stochastic frontier production function that the sigma square (2) was statistically significant and more than zero at 5 per cent (0.05). This indicates a good fit and the correctness of the specified distributional assumption of the composite error term. More so, the variance ratio, gamma (ɣ) was estimated to be as high as 93 per cent, suggesting that 93 per cent of the variation in output level of the swine farmers was due to the technical inefficiency of the sample farmers. 33 Maganga (2012) adopted a stochastic production function approach to investigate technical efficiency and its determinants in Irish potato production in Malawi. The unknown frontier parameters were estimated using the method of maximum likelihood. It was reported that technical efficiencies of individual farmers ranged from about 0.45 to 0.98. This suggests that a wide difference exists among Irish potato farmers in their level of technical efficiency. The average level of technical efficiency was valued at 83 per cent, suggesting that farmers could improve their output level by 17 per cent with the existing technology. It was also concluded that non-farm employment, education, farm experience, household size, degree of specialization and frequency of weeding were significant in determining the variations in technical efficiency levels. A study conducted by Awunyo-Vitor, Bakang and Cofie (2013) investigated the determinants of small-scale cowpea production in Ejura/Sekyedumase Municipality in the Ashanti Region using a stochastic frontier production function that incorporates inefficiency factors. Data for the study were collected from 200 randomly selected cowpea farmers within the district. A maximum likelihood technique was used to analyse the data. The results indicated that small-scale cowpea farmers were not fully technically efficient as the mean efficiency was 66 per cent. Farm size, seed, pesticides and labour were the major input factors that influenced changes in cowpea output. The result also shows that a farmer’s educational level, membership of farmer based organization and access to extension services significantly influenced their efficiency positively. The implications are that policies that would encourage cowpea farmers to join farmer based organizations and provide them with easy access to extension services are options that would 34 improve the efficiency of the farmers. Also, the estimated gamma value was 0.71 or 71 per cent. This means that 71 per cent of the total variations in output were due to technical inefficiency in the study area. Khai and Yabe (2011) used an econometric approach to analyse technical efficiency of rice production in Vietnam. A total of 3,733 rice farmers were interviewed. The Tobit model was used to determine the factors that influence technical efficiency. The study revealed that one variable that had an incremental effect on technical efficiency was the intensity of labour use. Thus, farmers tend to be less technically inefficient when more labour are utilised in rice production. It was also found that farmers who had access to irrigational facilities were more technically efficient as compared to those without irrigation. In addition, farmers who had no education or primary education were more technically inefficient than those who had secondary or higher education in Southeastern Nigeria. They found the average age of farmers to be 43 years. The estimated coefficients of land, labour and material inputs were found to have a positive relationship with output. Also, results from the maximum likelihood estimates showed that younger farmers, farmers who had access to credit and farmers with more years of schooling were more technically efficient. The estimated gamma value and the mean efficiency were 0.99 and 0.41 respectively. Haider, Ahmed and Mallick (2011) in assessing technical efficiency of agricultural farms in Bangladesh used the Ordinary Least Square (OLS) and the Maximum Likelihood Estimation (MLE) methods to estimate the Cobb- Douglas production function parameters. Some of the major findings from this study were that the OLS as an estimation technique was inadequate in representing the data and that the intercept values of MLE were greater than 35 the OLS estimates. In addition, the mean technical efficiencies of crop cultivating farms, fish cultivating farms and livestock cultivating farms were 69 per cent, 29 per cent and 66 per cent respectively. These suggest that all three areas of agricultural farming can improve their output levels with the given resources and technology. Again, in fish farming, only farming experience and access to credit were the factors that influenced efficiency although the coefficients of age, schooling years and family size were positive. The Cobb-Douglas production function was employed by Adeyemo, Oke and Akintola (2010) to analyse the economic efficiency of small scale farmers in Ogun State, Nigeria. Two hundred cassava farmers were randomly selected for the study. It was revealed that about 90 per cent of the farmers were males and majority of the farmers had more than 10 years of farming experience. Farm size and quantity of planting stakes influenced output positively whilst quantity of fertiliser had a negative influence on output. The factors that influenced efficiency in cassava production were age, cost of fertiliser, cooperative, farming experience and educational level. The average technical efficiency was 89.04 with a minimum of 85.69 and a maximum of 100. Thus, cassava farmers were producing closely to the production frontier with small variations in output. The farmers were producing in the stage one of production as the return to scale was found to be 2.62. Rahman, Mia and Bhuiyan (2012) found fertiliser, manure, irrigation cost, insecticide cost, area under production and experience to be the important factors to increase rice production in Bangladesh. In the technical inefficiency effect, age, education and family size had positive impact on efficiency effect, whereas land under household had negative impact on efficiency effect. 36 A study conducted by Enwerem and Ohajiana (2009) analysed the technical efficiency and the sources of inefficiency in large scale and small scale rice production in Imo State, Nigeria during the 2009 cropping season, using a stochastic frontier production function which incorporates a model for inefficiency effects. A sample of 160 farmers selected using the multi-stage stratified random sampling techniques were used to generate primary data with structured and validated questionnaire through the cost-route approach. Results showed that factors that affected the output of rice farmers were labour, capital, land and planting materials. Low capital base for investment, poor extension contact and poor access to credit were the major factors that influenced farmers’ level of technical efficiency. The mean technical efficiency scores were 0.65 and 0.69 for large and small scale farmers respectively, which implies that the mean technical efficiency index could be increased by 35 per cent and 31 per cent for large scale and scale farmers respectively through efficient reallocation of the available resources. The values of returns to scale were 2.255 and 2.011 for the large scale and small scale rice famers respectively, which implies increasing returns to scale, indicating that an increase in the use of the selected variables would result in more than proportionate increase in the production of rice. The value of sigma squared were 0.651 and 0.593 for large scale and small scale farmers which were equally statistically significant at 5 per cent level of probability indicating a good fit and correctness of the distribution assumption specified. The variance ratio (Gamma) which measures the effects of technical efficiency in the variation of observed output had values of 0.324 and 0.353 for large scale and small scale farmers respectively, which means 37 that 32 per cent and 35 per cent of the total variations in the output of rice farmers were due to technical inefficiency. Oyewo (2011) used the stochastic frontier model to estimate technical efficiency of maize production in Oyo State, Nigeria. In this study, it was concluded that a positive relationship exists between farm size, seeds and the level of maize output. This presupposes that as the use of these two inputs increases, output would also increase. It was also observed that level of education and years of farming experience had an inverse relationship to maize output. The results from this study showed that farmers who had formal education were not many and so this could have accounted for the negative relationship with technical efficiency. Furthermore, the mean technical efficiency was estimated at 96 per cent and 12 per cent of the variations in maize output was due to inefficiencies. From Oyewo’s results, it can be concluded that about 82 per cent of the differences in maize output was due to stochastic noise which is beyond the farmer’s control. Baten, Kamil and Haque (2009) used panel data to model technical inefficiency effects in a stochastic frontier production function. They assumed that the random variables followed a truncated normal distribution with a mean of zero and the parameters of the stochastic frontier function and that of the technical inefficiency effects were simultaneously estimated by the maximum likelihood method. The coefficients of labour and land area were found to be statistically significant in production. The mean level of technical efficiency was 0.51 from the period 1990 to 2004 and the technical efficiencies of individual producers ranged between 0.29 and 0.92. It was also established that the variable time and Herfindahl index (HHI) had a negative 38 relationship with inefficiency. Smaller plantation sizes were found to be more efficient than the larger ones. Al-hassan (2008) employed the stochastic frontier function to examine the technical efficiency level of rice farmers in Northern Ghana. A sample of 732 irrigators and non-irrigators rice farmers was used. The maximum likelihood estimator was used to estimate the parameters of the model. The average years of schooling showed that the highest level of education attained by the farmers was the primary school. It was also revealed that the average yield obtained by non-irrigators was higher as compared to the irrigators. Eighty six per cent of the differences in output were attributed to inefficiencies among the farms and 14 per cent accounted for random shocks which were beyond the farmers’ control. Again, the mean technical efficiency was higher for non-irrigators. For both irrigators and non-irrigators, the main factors that influenced technical efficiency were education and extension contact. Findings of a research conducted in Cross River State of Nigeria indicates that rice farmers operated at an increasing returns to scale as the value of the estimated output elasticity was 1.57. Also, the estimated gamma value was 0.77 suggesting that 77 per cent of deviations from the production frontier were as a result of inefficiencies in production and the remaining 23 per cent was due to factors that were beyond the farmers’ control. It was also concluded that variables that had a positive and significant impact on technical efficiency were years of schooling, access to credit and membership of association. However, the coefficients of sex, age and household size had negative signs but were not significant. This investigation was undertaken by Idiong (2007) to estimate farm level technical efficiency in small-scale swamp 39 rice production. The multistage random sampling method was used to select 56 respondents for the study. Tijani (2006) estimated efficiencies on rice farms in Osun State, Nigeria and identified some socioeconomic factors which influenced productive efficiency. These technical efficiencies were estimated using the stochastic Frontier Production Function Approach applied to primary data. A tanslog production function was used to represent the production of the rice farms. The study showed that the levels of technical efficiency ranged from 29.4 per cent to 98.2 per cent with a mean of 86.6 per cent. This suggests that the average rice output fell 13.4 per cent short of the maximum possible level. Therefore, output could be increased by 13.4 per cent in the short run under the existing technology. The study also showed that these efficiencies are positively and significantly correlated with the application of traditional preparation methods and with off-farm income. The variance ratio (gamma) was estimated to be 51.3 per cent, implying that about 51 per cent of the discrepancies in between observed output and the frontier output were due to technical inefficiency. This study adds to the existing body of literature by using the stochastic frontier approach to analyse the factors influencing technical efficiency among rice farmers on the Weta irrigation scheme in the Ketu North District in the Volta region of Ghana. Overview of the rice sector in Ghana Rice is fast becoming a cash crop for many farmers in Ghana (MiDA 2010; Osei-Asare, 2010). National and agricultural development plans and strategies, such as the Ghana Poverty Reduction Strategy (GPRS I), Growth 40 and Poverty Reduction Strategy (GPRS II), Food and Agricultural Sector Development Policy (FASDEP) I and II, Medium Term Agriculture Sector Investment Plan (METASIP), and Accelerated Agricultural Growth and Development Strategy (AAGDS), have featured rice as one of the targeted food security crops. Annual per capita consumption of rice is growing rapidly, from 17.5 kilograms in 1999–2001 to 22.4 kilograms in 2002–2004 and 24 kilograms in 2010–2011 (MoFA 2011a), and rice demand is projected to grow at a compound annual growth rate of 11.8 per cent and maize at 2.6 per cent in the medium term (MiDA, 2010). Several estimates show very high levels of imports valued at US$500 million annually (Osei-Asare, 2010), putting much pressure on foreign currency reserves and food security in Ghana. Estimates show that imported rice comprises about 70 per cent of the quantity consumed in Ghana, or a 174 per cent import penetration ratio (Amanor‐Boadu, 2012). Following the Tokyo International Conference IV on Africa Development (TICAD IV) in May, 2008 an initiative, Coalition for Africa Rice Development (CARD) was launched. This strategy is an outcome of Ghana Government’s subscription to the vision of the initiative to double rice production in Africa. The Ghana National Rice Development Strategy which covers the period 2008 to 2018 is a response to forestall the effects of the global food crisis. The strategy proposes to double rice production taking into consideration the comparative production capacities of the three major ecologies (rain fed upland, rain fed lowland and irrigated). Over the last 10 years (1999-2008), per capita rice consumption increased from 17.5 kilograms to 38.0 kilograms. By 2018, it is estimated that it will grow to 63 kilograms as a result of rapid population growth and urbanization. 41 In developing this strategy, National experts from multi-sectorial backgrounds with inputs from donor groups operating in Ghana have variously impacted on the development of production and manpower requirements and set objectives. The major constraints especially land development and land tenure arrangements, seed, fertilizer, inadequate human resource capacity, inadequate harvesting and post-harvest management technology, weak local rice marketing system and the role of Government and related agencies have been considered. A governance structure with many of the key actors in the rice sector has been considered and may be modified as the project unfolds. The role of government, public sector, private sector, NGOs have been considered crucial for the attainment of the goals of the strategy. The implementation plan which will be developed subsequent to the adoption of the NRDS will indicate the details of the action plans and funding. Global rice imports have increased by 80 per cent from 2.5 billion tonnes (grain) in the early 1990s to 4.5 billion tonnes in 2004. During the same period, African countries increased rice imports by 140 per cent from 5 million tonnes in the early 1990s to 12 million tonnes in 2004. This is equivalent to about a quarter of the world import, with an import value estimated at US$2.5 billion. West African countries show the same increasing trend of rice import, increasing from 4 million tonnes (US$ 0.8 billion) in early 1990s to 8 million tonnes (US$1.6 billion) in 2004-2005, accounting for two-thirds of Africa’s rice import (Amanor-Boadu, 2012). Status of rice in national policies Policy strategies over the years as captured in FASDEP I, GPRS I & II, METASIP, AAGDS and Ministry of Food and Agriculture policy documents have 42 sought to promote rice production to address food security and poverty reduction. FASDEP II, which is an agriculture sector development policy guideline (2008 – 2010), targets reducing rice imports by 30 per cent through increasing production levels to 370,000 tonnes per annum to ensure food security and import substitution. Specific measures, among others, to reach this level of production are increased mechanization, increased cultivation of inland valleys and efficient utilization of existing irrigation systems. In addition, varietal improvement and increased seed production and utilization are to be pursued vigorously (MoFA, 2011). Rice consumer preferences, per capita consumption and demand projections There is a wide variation in rice consumer preference in Ghana on the basis of grain characteristics. However, most consumers prefer long grain perfumed rice of good taste, good appearance, and with whole grains, although broken grains have their place in specific local dishes. Health-conscious consumers patronize local brown rice while parboiled rice is preferred in the Northern regions of Ghana. Annual per capita rice consumption during 1999-2001 was 17.5 kilograms on average. This increased to 22.6 kilograms during 2002-2004. In the same period, per capita rice consumption increased to around 8.9 per cent per annum, higher than the population growth of 2.5 per cent per annum. Assuming the same trend continues, per capita rice consumption will increase to 41.1 kilograms in 2010 and 63.0 kilograms in 2015. Based on population growth rate alone the current demand of about 500,000 tonnes per year is projected to increase to about 600,000 tonnes per year in 2015. However, taking both population growth and increase in per capita consumption together, rice demand will increase to 1,680,000 tonnes per year (Ragasa, et al., 2012). 43 Typology and number of rice farmers Rice producers in Ghana are categorized by agro-ecologies namely: irrigated, rain fed lowland and rain fed upland. In general, the lowland rain fed system covers 78 per cent of the arable area; the irrigated system covers 16 per cent while the upland system covers 6 per cent. The majority of local rice production comes from the Northern (37 per cent), Upper East (27 per cent), and Volta regions (15 per cent). Production in the Northern and Upper East regions decreased in 2011 due to poor weather condition, but production in Volta continued to increase and did not seem to be affected by less rain in 2011. In general, rice production and the area cropped with rice are increasing. Since 2007, production has been increasing at a faster rate than areas of cultivation; a proof that yields during this period have been trending upward. This growth is encouraging and may have been the result of the various initiatives to develop the rice sector in Ghana, including passage of the National Rice Development Strategy in 2009; various donor-funded projects, the majority of which were implemented in the period 2004–2009; and the national fertilizer subsidy program introduced in 2008, to which rice farmers had responded. There was a jump in production and acreage starting in 2008, which could be a compounded result of these various initiatives. However, the national average yield has remained low, at 2.5 tonnes/hectare/year or 2.2 tonnes/hectare/season according to a survey by the Crops Research Institute (CRI), Savannah Agricultural Research Institute (SARI), and International Food Policy Research Institute (IFPRI), indicating significant opportunity to reach potential achievable yields of 6–8 tonnes/hectare (MoFA, 2011). 44 Potential for rice sector development There is potential to develop the rice sector in Ghana. Rice demand is projected to grow (MiDA, 2010), and prices have been trending upward over time. The average wholesale price of local milled rice (100 kilogram bag) more than doubled, from GH₵55.00 in 2006 to about GH₵120.00 by 2011, while that of imported milled rice nearly tripled, from about GH₵60.00 to nearly GH₵169.00 (Amanor‐Boadu, 2012). Considering a monthly salary for a middle-income earner in Accra of about GH₵400.00 per month, rice purchases account for a substantial portion of household income. These statistics indicate the economic viability and attractiveness of rice production, as confirmed by a policy analysis matrix calculated by Winter-Nelson and Aggrey-Fynn (2008) and Akramov and Malek (2011), although profitability becomes negative when subsidies and trade protection are removed and when family labour is included in the calculation (Amanor-Boadu, 2012). Imported rice is priced higher than local rice, by about 15–40 per cent on average, and is mainly associated with better-quality long-grain perfumed rice of good taste and good appearance (translucent and with whole grains, although broken grains have their place in specific local dishes). Interviews among farmers in Ashanti region suggest that farmers sell a 50-kilogram bag of Jasmine 85 (perfumed and long grain) at GH₵90.00, while Sikamo (unperfumed local rice) sells at GH₵60.00 to GH₵70.00 per 50 kilograms bag (Amanor-Boadu, 2012). The rice sector in Ghana is segmented into two distinct target markets of local and imported rice. Imported rice is more popular in urban centres in general. In Accra, there is a heavy preference for imported rice; 95 per cent of sample consumers were more familiar with imported varieties, and 71 per cent 45 consumed only imported rice and never tried local rice (Diako et al., 2010). However, the adoption of fragrant local varieties by growers (e.g., Jasmine 85, Togo Marshall, and Aromatic short) nearer to Accra is giving access to consumers in the capital (Osei-Asare, 2010). In Accra, Kumasi, and Tamale, 86 per cent of sample consumers prefer imported rice, while a niche segment (14 per cent) prefers local rice (Tomlins et al., 2005). While the appearance of raw rice is critical to consumers’ choice, taste and aroma determine consumer preference for cooked rice (Diako et al., 2010). The reasons given for not purchasing locally cultivated rice were poor postharvest handling, unavailability, and the generally perceived poor quality. The 29 per cent who tried local rice did so because it is relatively cheaper than imported rice and is perceived it to be more nutritious than imported rice (Diako et al., 2010). A study by Diako et al. (2011) confirms that local varieties have a higher mineral content than imported varieties, although imported varieties have the advantage of being easier to cook and the expansion ratio is greater, which is another feature preferred by many consumers. However, local rice is preferred in many rural areas where there is local production, especially in northern Ghana. In certain niche segments, health-conscious consumers purchase local brown rice while parboiled local rice is preferred in the Northern Region of Ghana. The study by Acheampong, Marfo and Haleegoah (2005) in Hohoe and Bibiani Districts in the Volta Region and Western Region respectively suggests a strong preference for local varieties, as consumers perceive that local rice contains more nutrients than imported rice. Moreover, sample consumers interviewed reported that local rice was consumed because it was more readily available than imported 46 varieties. These findings suggest that there is existing demand for local varieties, that greater promotion of the nutritional advantages of local varieties could further boost purchases of local rice, and that improved post-harvest handling and quality standards could enable several local perfumed rice varieties to directly compete with imported rice. On the supply side, vast potential irrigable lands, valley bottoms with water supply, and water bodies throughout the regions are available (Osei- Asare, 2010). It is also said that because rice has been grown in Ghana for centuries, there is indigenous knowledge of rice that can be tapped in developing suitable agronomic practices (Osei-Asare, 2010). In addition, the policy environment is also advantageous for rice production. The development of the rice sector seems to have received plenty of attention in Ghana over the years, as evidenced by numerous projects and programmes supporting the sector. Rice research program Rice research in Ghana is performed primarily by the Crops Research Institute (CRI) and the Savannah Agricultural Research Institute (SARI) of the Council for Scientific and Industrial Research (CSIR). The University of Ghana and other universities also conduct both varietal research and testing and socioeconomic research on rice. Scientists at CRI and SARI reported that about 80–90 per cent of research work is on varietal improvement and testing (MoFA, 2011). 47 History of irrigation development in Ghana Small scale irrigated agriculture practice across the country dates back as early as 1880 in Keta areas on land above flood level between the lagoon and the sandbar separating it from the sea. The method was so adopted in the area, in the sense that the natural condition was not in support of the principle of shifting cultivation in agriculture as practised elsewhere. The first irrigation scheme in the country was set up by the government in 1920. This was an integral part of the then Winneba Water Supply Project (Smith, 1969). Soon after independence in 1959, the first national irrigation project in Dawhenya was established. However, available records reveal that Asutuare Irrigation Project was the pioneer scheme in 1967. The 1950s and 1960s witnessed some water development schemes accounting for about 240 earth dams and dug-outs in the North and about 66 in the Ho-Keta plains of the South purposely to provide water for dry season irrigated agriculture. Currently, irrigation schemes are managed by the Ghana Irrigation Development Authority (GIDA) set up by Supreme Military Council Decree (S.M.C.D) 85, 1977 and their function is to provide for the development of irrigation and other related matters (Owusu, Nyantakyi & Borkloe, 2013). Irrigation schemes in Ghana There are about forty official irrigation schemes across the length and breadth of the country. The Ghana Irrigation Development Authority (GIDA) has 22 irrigation schemes cultivating rice under its jurisdiction covering about 14,100 hectares out of which 53 per cent are developed and about 5478 hectares is actually put under irrigation (Owusu, Nyantakyi & Borkloe, 2013). 48 Rice irrigation schemes in Ghana The self-sufficiency policy, adopted in Ghana during the 1970s, resulted in the creation of the Ghana Irrigation Development Authority (GIDA) of the Ministry of Agriculture as a semi-autonomous organization in 1977. In GIDA, a lot of emphasis was placed on the development of large- scale irrigation projects for the production of rice. The ultimate aim was to reduce the country’s dependency on imported rice. Rice yields on the irrigation projects vary between 4.0–6.0 tonnes/hectare, with an average yield of about 4.6 tonnes/hectare. Yields of rice on irrigation schemes, and cropping intensity, are directly related to the amount of water available from season to season. It is estimated that the contribution of irrigated agriculture to the total national rice production needs to be about 24 per cent in order to satisfy national demand. Only seven of the schemes are seriously into rice irrigation. The others are shared schemes which incorporate mix cropping; involving rice and other crops especially vegetables (Owusu, Nyantakyi & Borkloe, 2013). Constraints to rice production The inability of farmers to produce maximum output from their resources can partly be attributed to production constraints. This section provides an empirical review of the constraints faced by rice farmers so as to propose government policies that seek to overcome the constraints to domestic production of rice. A study conducted by Thahn and Singh (2006) surveyed 100 farmers in Punjab and West Bengal states of India and Giang and Vinh Long provinces of Vietnam. It found that the agro-ecological constraints faced by farmers, 49 ranked from more to less serious were related to dependence on monsoon; land/soil problems; environmental pollution; lack of water and small land holdings. Under technical constraints, it was found that diseases (sheath blight, blast, and stem rot); pests; lack of proper varieties; post-harvest technology constraint; storage problems were the most serious constraints perceived by large percentage of respondents. Fertilizer problems; plant protection constraints; weed problems; lack of labour and poor processing were found to be other constraints as perceived by farmers. In case of socio-economic constraints, the study found that poor infrastructures; high cost of inputs; credit problems; low rice price; inadequate inputs and lack of trainings were the most important constraints as perceived by large percentage of farmers. Other constraints as perceived by lower percentages of farmers were poor extension services; lack of information and lack of helpfulness from local authorities/governments. Obiri-Opareh (2008) analysed some factors that adversely affect rice production in Ghana as a whole and at some irrigation projects in particular, and measures put in place to address them. The study was based mainly on a field survey conducted at the Dawhenya Irrigation Project (DIP) between 2004 and 2007 as part of a policy study on irrigation agriculture in Ghana. The findings showed that the local rice industry had suffered from, among others, high cost of inputs and production constraints at the pump-type irrigation projects (PTIPs); particularly the high electricity tariffs that had resulted in closure of some irrigation schemes, including those at Dawhenya and Weija, difficulties in accessing credit, use of poor-yielding seed varieties, inappropriate agronomic practices, limited mechanisation, poor processing methods, and poor marketing strategies. 50 Omofonmwan and Kadiri (2007) examined the problems and prospects of rice production in the Central District of Edo State, Nigeria. It was found that the aged and illiterates dominate rice farming. The farming system was characterized by the use of crude implements, small farm holdings and subsistence. Other findings of the study were no extension workers, farmers’ unawareness of new varieties of rice seedlings, high cost of fertilizers, diseases and attack from pests like birds, rodents. Yield per farm was low and financial return was correspondingly low. Matanmi, Adesiji, Owawusi and Oladipo (2011) investigated the perceived factors limiting rice production in five selected villages in Patigi Local Government Area of Kwara State, Nigeria. The study area was purposively selected based on their known potentials for rice production. One hundred and ten rice farmers were selected for the study. An interview schedule was used to obtain information from the farmers. Frequency counts, percentage and means were used to analyze the data. The study revealed that the perceived limiting factors in rice production include lack of rice processing machines, financial constraints, illiteracy, poor access to inputs, pests and diseases, poor transportation, fluctuations in climate, lack of extension services and lack of storage facilities. It is recommended that government should assist the rice farmers with the provision of rice processing machines such as threshers and destoners and credit facilities so as to improve the quality and quantity of rice produced in the study area. Alarima Adamu Masunaga and Wakatsuki (2011) conducted a study to examine the constraints to the adoption of sawah system of rice production in Nigeria. Data were collected from 124 randomly selected sawah-rice farmers. 51 Data were analysed using correlation and regression to determine the relationships between the study variables. The results showed that respondents were predominantly male (98.80 per cent), married (98.80 per cent) and had Quranic education (62.70 per cent). Farm size ranged from 0.03 to 10 hectares, mean yield was 4.65 tonnes/hectare, and mean income was $1,041.38. Production and on-farm constraints affecting sawah development were water management and flood. Major economic constraints faced by sawah farmers were lack of viable financial agencies to support production, poor capital base and non-availability of loan. Regression analysis showed that the yield of sawah was negatively related to land acquisition constraints (b=-0.34, p<0.05) and technological constraints (b = -0.43, p < 0.01). This study concluded that problems faced by farmers were interwoven in which existence of one relates to the other. Addressing these problems will lead to increase in the rate of adoption of sawah rice production technology and ultimately rice productivity in Nigeria. Musiime, et al. (2005) carried out a Participatory Rural Appraisal (PRA) to identify the major constraints to rice production which led to the increased rice yield gap in Bugiri district, Kampala, Uganda. Two rice growing villages; Nkaiza and Bupala for upland and paddy rice, respectively were used for the PRA based on two hypotheses: major constraints contributing to rice yield gap are both socio-economic and biophysical; the traditional banana-coffee crops still dominate the cropping systems. Informal discussions were held with farmers’ groups in each of the villages on issues regarding major rice constraints, existing coping mechanisms, rice production, and the existing cropping systems. The major constraints were identified and ranked in their order of importance using a pair-wise ranking tool as water 52 management (28.6 per cent), soil nutrient depletion (23.8 per cent), weeds (19.0 per cent), labor (11.9 per cent), pests and diseases (11.9 per cent), poverty (4.8 per cent) for wetland rice while weeds (33.3 per cent), change in rainfall patterns (23.3 per cent), soil nutrient depletion (20 per cent), labor (16.7 per cent), pests and diseases (6.7 per cent) was the order of importance of constraints for upland rice production. Generated coping mechanisms to the identified constraints were; draught animal power (for labor constraint), resistant varieties (insect pests & diseases), pests (bird chasing, rat traps), water management (planting at different periods), soil nutrient depletion (Fertilizer application and proper straw management), weeds (maintenance of recommended water levels, herbicide control) and changing rainfall patterns (early planting). Conceptual framework The conceptual framework represented by Figure 2 presents the relationships between the main variables of the study. The input factors such as land area under cultivation, labour, fertilizer, irrigation facilities and equipment directly affect the state of production of rice which includes farm size, farming methods used, the technology employed, among others. The socio-economic characteristics of the farmer such as age of farmer, farmer’s educational background and household size also determine the state of production. For instance, the age of the farmer and household size influence the size of the farm. Constraining factors such as erratic rainfall patterns, high cost of labour, inadequate capital and low output price of rice directly affect the state of production. The state of production directly influences output of rice which, in turn, determines the level of technical efficiency of the farmer. 53 Figure 2: Conceptual framework Source: Author’s construct, 2015 Input Factors: Land, fertilizer, labour, irrigation, equipment (farm tools), etc. Socioeconomic Characteristics: Age, education, household size, farming experience, etc. Production Constraints: erratic rainfall, high cost of labour, inadequate capital, low output prices, etc. State of Production: farm size, technology employed, farming methods Output Technical efficiency 54 CHAPTER THREE METHODOLOGY Introduction This chapter describes how the study was conducted. It talks about the study area, research design, population and sampling, research instrument, data collection procedure and the analysis of data. Study area The study was conducted in the Ketu North District of the Volta Region. The Ketu North District was one of the thirty two newly created districts in the Volta Region in 2008. It was carved out of the original Ketu District by a legislative instrument, L118 41 in 2008 in the district. The Ketu North District is located between latitudes 6 o 03”N and 6o 20”N and longitudes 0o 49’E and 1o 05’E. It shares boundaries with the Akatsi North District to the North and the Republic of Togo to the East. To the South, it is bounded by the Ketu South District and Keta Municipality and to the west it is bounded by the Akatsi South District. The district capital, Dzodze is about 80kilometres from Ho, the capital of Volta Region. The district has a surface area of about 754 square kilometres. Figure 3 shows the map of the district in national context. 55 Figure 3: Ketu North District in national context. Source: MoFA, Ketu North District, 2014 The area of the district is underlain by three main geological formations; namely, the Dahomenyan formation to the North made up of soil such as Tropical Grey and Black Earths, the Regosolic Groundwater Laterites, the Recent Deposits of the littoral consisting of the Tertiary formation comprising Savannah Ochrosols for its soil type. These soil types are suitable for the cultivation of different types of crop. (Ministry of Food and Agriculture, Ketu North District, 2014) In terms of relief, Ketu North District is relatively low lying with altitudes around 66 metres. The plain nature of the terrain makes movement within the district easy. The Drainage of the district is towards the South and is dominated by several seasonal streams that flow in wide valleys between 56 Ohawu and Ehie to end in the swamplands of Afife. The major rivers include Kplikpa and Tsiyi. There are about six large fresh water reservoirs (dams) - Ohawu, Kporkuve, Dzodze, Tadzewu, Dekpor-Adzotsi and Lave as well as a few small community dugouts in the district. The district experiences the dry Equatorial type of climate. The average monthly temperatures vary between 24℃ and 30℃, which are generally high for plant growth throughout the year. The mean annual rainfall for the district is around 800 millimetres. The rainfall is of double maxima type occurring from April to July and September to October. The dry season, which is mainly dominated by the dry harmattan winds, extends from December to February in the district. Generally, rainfall in the district is considered low and erratic particularly during the minor season (MoFA, Ketu North District, 2014). The original vegetation of the district is Savannah woodland made up of short grassland with small clumps of bush and trees as well as mangrove forests in the marshlands are found in the district. However, the extensive farming activities in the district have, over the years, reduced the natural vegetation. Amid these are cultivated holdings of cassava, maize, coconut, oil palm, and velvet tamarind, the occasional baobab and fan palm. The decimation of the vegetation by population pressure may have adversely affected rainfall in the district. The physical characteristics of the Ketu North District contain a basket of potentials that can be tapped for the socio- economic development of the area. In terms of relief and drainage, the vast expanse of flat land is a potential for large scale mechanized farming. Road construction and other activities are also relatively less costly. 57 The water resources in the district could also be harnessed for irrigation purposes, especially for rice cultivation and dry season gardening aside its current use for the supply of potable water for some communities in the district. The high intensity of the sun in the area provides abundant solar energy, which is already being used by farmers for preservation and storage purposes (MoFA, Ketu North District Assembly, 2014). The economic activities in the area include Ewe-Kente (the traditional cloth) produced by numerous weavers in the district. Handicrafts like works of metal; clay, fiber and wood dominate the relics and jewellery of the shrines in Ketu North. There are small-scale workshops with basic equipment located in the urban centres to produce metal products such as hand tools for farming and cassava graters. The linkage between processing and manufacturing on one hand, and the agriculture sector on the other is however weak (MoFA, Ketu North District, 2014). In the area of financial services, there is one Commercial Bank in the district namely, the Ghana Commercial Bank located at Dzodze. There is also a branch of the Avenor Rural Bank located at Dzodze. These banks are expected to provide credit to promote agricultural production as well as commercial activities in the district. A large section of the population of the district, however, lacks access to banking services (MoFA, Ketu North District, 2014) Based on estimates from the 2010 population census, the Ketu North District has a population of 99,913. The population of the district has always been experiencing growth over the years. From 1970 to 1984, the district experienced a population growth rate of 1.9 per cent. This growth rate remained unchanged for the1984 – 2000 censual year. 58 Another significant feature of the district population is its large labour force. The cohort that falls within the active labour force constitutes 52.5 per cent of the district population. This is a bit lower than the national active labour force of 55.2 per cent and higher than that of the regional figure of 52.2 per cent. This large active labour force could be positioned to harness and maximize the vast agricultural potentials of the district (MoFA, Ketu North District, 2014). Agriculture is the mainstay of the Ketu North District economy. It employs about 70 per cent of the economically active labour force. Nearly every household in the District is engaged in farming or agricultural related activity. There are over 27,781 (13,752 males 14,029 females) members of household who are into agriculture. Farming in the district is largely carried out on small-scale basis. The average acreage cultivated ranges between 0.4- 0.8 hectares for the major staples like maize, cassava, rice, cowpea and sweet potato, whilst the area under vegetable production is considerably smaller. Most farming households keep small ruminants like goats and sheep (MoFA, Ketu North District, 2014). The crop sub-sector accounts for about 60 per cent of agricultural activities in the District. The crops in the sub-sector can be categorized as arable crops, plantation crops and vegetables. The soils in the area favour the production of a variety of crops. Currently, crops grown in commercial quantities in the district include maize, cassava, sweet potato, cowpea and rice (MoFA, Ketu North District, 2014). Irrigated Rice production in the District, which is solely on the Weta Irrigation project, is under the management of Ghana Irrigation Development Authority (GIDA) and the developed area under production is 880 hectares out 59 of the total land size of 950 hectares. There are a total of 1,024 farmers on the project with an estimated yield of 4.5-5.0 tonnes/hectare. From 2010, production on the Weta Irrigation Project has increased dramatically due to the double season cropping pattern adopted by the scheme (MoFA, Ketu North District, 2014). The Weta Irrigation Project is located approximately between latitudes 6 o 04’ and 6° 08’ and longitudes 0o 45’ and 0o 55’ East at a distance 162 kilometres east of Accra. The project consists essentially of two (2) earthen dams situated on the lower reaches of the Agali and Kplikpa rivers that deliver water to the field by gravity. The Agali reservouir was constructed by the Soviets in 1962 and later rehabilitated in 1982 by the Chinese during the construction of the Kplikpa reservouir. The project headwork’s is an earth fill dam of height 11.5 metres and a crest level of 18.80 metres. The spillway level is at an elevation of 15.49 metres and the surface area at the spillway elevation is 544 ha. The reservouir capacity is 29.45 cubic millimetres. The dead storage is about 12.21 metres and the surface area at this elevation is 2.73 square kilometres. The dead storage capacity is 3.32 cubic metres. The spillway type is open with length of 50 metres and a design discharge of 320 cubic metres/second (Ghana Irrigation Development Authority, Ketu North District, 2014). The essence of extension services is to upgrade the knowledge of farmers in improved techniques and modern methods of agriculture with a view to improving upon incomes and output. Extension services have not been within easy reach of the prospective farmer. According to the District Directorate of Agriculture, the district is zoned into eighteen agricultural operational areas. For an efficient extension service delivery, one operational 60 area needs to be manned by One Agricultural Extension Agent (AEA).There are currently eight Agricultural Extension Agents (AEA) in the district expected to man the eighteen operational areas. If the entire district should be covered, then one operational area would be just too big for one AEA to cover. It is no wonder therefore that majority of the farmers do not have access to their services. Farmer- Extension ratio is too high (3,500:1) in those areas that have access to their services. These are in areas where the AEAs are resident. In these cases, they attend to other farmers outside their operational areas based on demand through phone calls as well as home visits. (MoFA, Ketu North District, 2014). Research design The study employed cross-sectional survey research design to measure the technical efficiency of rice farmers. In this design, a sample of the population was selected from which data were collected to answer research questions of interest. It is called cross-sectional because the information that were gathered about the phenomenon represent what existed at only one point in time. A cross-sectional study is one that produces a ‘snapshot’ of a population at a particular point in time. The single ‘snapshot’ of the cross- sectional study provides researchers with data for either a retrospective or a prospective enquiry (Louis, Lawrence & Keith, 2007). This research design is appropriate as it makes inference about the effect of one or more explanatory variables on the dependent variable by recording observations and measurements on a number of variables at the same point in time (Gay, 1992). 61 Population Amedahe (2002) defines population as the target group about which the researcher is interested in gaining information and drawing conclusions. The target population was all rice farmers in the Ketu North District of the Volta Region. The accessible population for the study included all the 1,024 rice farmers on the Weta irrigation scheme in the Ketu North District. Sample size and sampling technique Hummelbrunner, Rak and Gray (1996) explain sampling as selecting a portion of the population that is most representative of the population. The study employed a two-stage sampling technique to select the participants for the study. The rice farmers on the Weta Irrigation Scheme were grouped into 11 sections by the Irrigation Development Authority. Considering each section as a cluster, six sections were selected at random at the first stage. At the second stage, a total of 290 rice farmers were chosen from the six sections using proportionate random sampling technique to form the sample for the study. A sampling frame was obtained from the Irrigation Development Authority. The computer software programme known as excel was then used to generate a list of randomly selected numbers within a specified range. Rice farmers with those randomly selected numbers were then identified and interviewed. This sample size was determined using the sample size determination table produced by Krejcie and Morgan (1970). However, out of the 290, only 285 rice farmers were reached, giving a response rate of 98.3 per cent. 62 Instrumentation The structured interview schedule was developed by the researcher and used to collect data relating to technical efficiency in rice production from the respondents (farmers). It contained both open-ended and close ended questions. A structured interview is an interview in which the specific questions to be asked and the order of the questions are predetermined and set by the researcher. It is based on a strict procedure and a highly structured interview guide which is no different from a questionnaire. The structured interview is, in reality, a questionnaire read by the interviewer as prescribed by the researcher. The rigid structure determines the operations of this research instrument and allows no freedom to make adjustments to any of its elements such as content, wording or order of questions (Amedahe, 2002). Data were collected on socio-economic characteristic of farmers, input and output quantities and the constraints faced by rice farmers. The structured interview schedule comprised four sections; namely, A, B,C and D. Section A covered the farm and farmer- specific characteristics such as the age of the farmer, sex of farmer, household size, educational level, marital status, off- farm work, farming experience and years of formal education. Section B of the interview schedule dealt with the production activities of the farmer such as methods of weed control, access to technical training, access to credit, access to agricultural extension services and number of times of producing rice in a cropping year. Section C of the interview schedule provided information on the inputs used and the output obtained by the farmer. These included information on land, labour, materials used for planting, fertilizer, equipment, chemical use and output obtained. The last section, D covered the constraints 63 that farmers face in the production of rice in the district. This included input, production, and marketing constraints. Pre-testing of instrument Sarantakos (1997) defines a pre-test as small tests of single elements of the research instrument which are predominantly used to check eventual ‘mechanical’ problems of the instrument. The instrument was pre-tested before it was used for data collection. The pre-test was undertaken in November, 2014 using 30 respondents who cultivated rice in the South Tongu District. This helped to check the adequacy of response categories, ambiguity and respondents’ interpretation of certain questions, thereby making it possible for adjustments to be made where necessary. Inaccuracies identified during the pre-testing were corrected before the actual data collection took place. The reliability of the instrument was established using the Cronbach’s alpha reliability coefficient. The reliability coefficient was estimated at 0.75. According to Cohen, Manion and Morrison (2007), the widely acceptable minimum standard of internal consistency is 0.70. Therefore, the reliability coefficient of 0.75 is interpreted as high; implying that the individual items or sets of items on the instrument would produce results consistent with the overall instrument. Data collection procedure Data were collected by the researcher and two field assistants during the 2014/2015 cropping season. The selection of the field assistants took into consideration their level of education and their ability to speak the local language of the farmers. A visit was paid to the study area by the researcher with an introductory letter from the Department of Agricultural Economics 64 and Extension, University of Cape Coast to inform the District Director of Agriculture, the Irrigation Scheme Manager, the Sectional Heads and the rice farmers about the study a month ahead of the data collection date. A two-day training programme was organised to equip the field assistants with interviewing skills and to explain to them the various items on the instrument. A second visit was paid to the study area to agree on the date and duration for data collection with the rice farmers and their Sectional Heads a week before data collection began. Data collection was done for a period of two months. Description of output and input variables for the study Output: This refers to the total quantity of rice harvested during the 2014/2015 cropping season and it is measured in kilogram per hectare. Land: It is the area of the farm allocated to the production of rice and this variable was measured in hectares. The amount of land used was expected to have a positive influence on output. Labour: This includes both family and hired labour and it was measured as person-days per hectare of farm from land preparation to harvesting. It was expected that labour would have a positive influence on output. Equipment: The cost of farm tools and machinery involved in the production process. It is measured in Ghana cedi (GH₵) per hectare. The use of equipment was anticipated to increase output. Seed: The quantity of rice seeds planted and it was measured in kilograms (kg) per hectare. The plant population of rice is influenced by the quantity of seeds planted per hectare of land which will, in turn, influence output. 65 Pesticide: The quantity of agrochemicals (fungicides and insecticides) used per hectare of land and it was measured in litres per hectare. Its influence on output could be positive or negative. Weedicide: This is the quantity of chemicals applied to control weeds before and after planting. It was measured in litres per hectare of farmland. Like pesticides, the use of weedicides can influence output positively or negatively. Fertilizer: The amount of fertilizer applied on rice plots in kilograms per hectare during the 2014/2015 cropping season. It was expected that fertiliser will have a positive effect on yield. Irrigation cost: This was measured as the amount (in Ghana cedi) spent on irrigation per hectare per cropping season. This was expected to increase output. Description of farmer- specific variables for the study Educational level: It is measured by the number of years of schooling by the farmer. Education promotes the adoption of better management practices and resource use which contributes to the efficiency levels of farmers. Findings from a study by Ahzar (1991) show that education enables one to make better choices regarding input combination and use of existing resources. Hence, it is anticipated that education would influence technical efficiency positively. Age: It was measured in years and it is used as a proxy for farming experience. Chukwuji, Inoni and Ike (2007) have indicated that older farmers are less efficient than the younger ones. This has been attributed to the fact that older farmers are less willing to adopt new ideas in their production activities. 66 Sex: It was measured as a dummy variable; one, if the farmer was a male and zero, if the farmer was a female. Male farmers were expected to be more technically efficient than female farmers. Extension services: It shows whether the farmer had access to extension services during the cropping season. It was measured as a dummy variable; one, if farmer had access to extension service and zero, if otherwise. Extension services provided to farmers enable them to learn better farm management practices and efficient use of resources. Off-farm activities: It indicates whether the farmer engaged in other economic activities aside from rice farming during the 2014/2015 cropping season. Those who engage in different economic activities at the same time are not fully committed to any of the activities thereby leading to technical inefficiency. However, the additional income earned from off-farm activities can be used to purchase farm inputs. It was measured as a dummy variable and had a value of one, if the farmer engaged in other off farm work whilst a value of zero indicates that rice production was a full time occupation. Access to credit: It was measured as a dummy variable; one if the farmer had access to credit, and zero if otherwise. Access to credit, defined as the availability of loans and other financial aids to the farmer helps to ease the financial constraints faced by farmers. Farmers who have access to credit tend to have higher technical efficiency than those who do not have access to credit (Binam et al., 2004). Household size: It includes the number of people who were living with the farmer during the 2014/2015 cropping season. It was expected that large family size would have a positive relationship with technical efficiency as they provide labour for farming activities. 67 Experience: The number of years engaged in rice farming. Bozoglu and Ceyhan (2007) have concluded that farmers with more years of farming experience reduce their technical inefficiency level by ensuring the optimal usage of time and inputs. Therefore, it was expected that farming experience would have a significant relationship with rice output. Membership of farmer based organization. This indicates whether the rice farmer belongs to a farmer based organization or not. It was measured as a dummy variable and had a value of one if the farmer belonged to an organization and zero if otherwise. This variable was expected to increase the yield of rice. Data analysis Descriptive statistics, including the mean, frequencies, charts and standard deviation were used to describe the socio-economic characteristics of farmers. The stochastic production frontier analysis, a parametric approach in measuring technical efficiency was employed in this study. The transcendental logarithmic (translog) form of production function was then fitted to the production function to estimate technical efficiency level of rice farmers and the determinants of technical efficiency simultaneously (Research questions one and two). Research question three was also analysed using the translog stochastic frontier production function. The Kendall’s coefficient of concordance was employed to analyse the constraints to rice production (Research question four). Data were analysed using the SPSS version 21 and the R programming software. 68 Specification of the models The explicit translog stochastic frontier production function used in this study is given in equation (vii):   8 8 8 0 1 1 1 1 ln ln ln ln 2 i i i ij i j i i i i j Y x x x v u            (vii) Where; iY is the output of rice (kilograms) produced in 2014/2015 cropping season by the thi farmer; x is a set of eight input categories namely: land area (hectares), labour (person-days), seed (kilograms), weedicides(litres), pesticides(litres), equipment(GH₵), fertiliser(kilograms) and irrigation cost(GH₵);  denotes the unknown parameters to be estimated; iv denotes a random error that captures the stochastic effects that are beyond the farmer’s control. ; iu is the one-sided non-negative error representing inefficiency in production. The inefficiency model of the stochastic frontier function is given by: 9 0 1 i i i i u Z     (viii) 0 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 iu Z Z Z Z Z Z Z Z Z                     (ix) Where, iu denotes farm specific inefficiency  denotes a set of parameters to be estimated 1Z denotes farmers’ educational level (years of schooling) 2Z denotes age of the farmer (years) 69 3Z denotes sex of the farmer (dummy variable, 1 = male, 0 = female) 4Z denotes access to extension services (dummy variable; if yes = 1, no = 0) 5Z denotes off-farm work (dummy variable; if yes =1, no = 0) 6Z denotes access to credit (dummy variable; if yes =1, no =0) 7Z denotes household size (number) 8Z denotes farming experience (number of years of rice production) 9Z denotes membership of farmer based organisation (dummy variable, if yes =1, no =0) The translog function was adopted in order to estimate the level of technical efficiency in a way consistent with the theory of production function after preliminary testing for the most suitable functional forms of the model under the data set available using the generalised likelihood ratio test(Griffiths, Hill & Judge, 1993). The generalised likelihood- ratio test statistic is of the form:      0 12LR ln L H ln L H     (x) where,  0L H and  1L H are values of the likelihood function under the null and alternative hypotheses respectively. Asymptotically, the test statistic has a Chi-square distribution with the degree of freedom equal to the difference in the number of parameters between the models. Here, the null tested was that the Translog functional form does not represent the data more adequately than the Cobb-Douglas. The results of the likelihood ratio test presented in Table 1 show a p-value of 0.05676 which is statistically significant at the 10 per cent significance level indicating the rejection of the Cobb-Douglas functional 70 form in favour of the more flexible translog. Thus the null hypothesis was rejected. Table 1: Likelihood ratio test Model Log Likelihood value Degree of freedom Chi-square P- value Cob-Douglas -18.9452 - - - Translog 4.4345 33 46.76 0.05676 ** **-------Significant at 10% Source: Field survey data, 2015 The Cobb-Douglas production function is not used in this study because although it is simpler and easier to estimate, it makes several restrictive assumptions. It is assumed that the elasticity coefficients are constant, implying constant shares for the inputs. The elasticity of substitution among factors is unity in the Cobb-Douglas form. Moreover, this being linear in logarithm, the output is zero if any of the inputs is zero (Henderson& Quandt, 1971). Also, in order to reduce the chance of using a too restrictive functional form, the translog functional form was used in this study. In this study, the Maximum Likelihood Estimation (MLE) approach was used to obtain the estimates for equations (vi) and (ix), using the computer programme,’R’ by the simultaneous estimation procedure proposed by Reifschneider and Stevenson (1991) and subsequently by Battesse and Coelli (1995). The maximum likelihood (ML) method is defined as the value of the parameter that maximises the probability (or likelihood) of randomly drawing a particular sample of observations (Coelli, Rao, O'Donnell & Battese, 2005). It makes some distributional assumptions about the two error terms. Thus, it 71 helps to model the impact that external factors may have on the distribution of the inefficiencies. The Maximum Likelihood estimator is preferred to other estimators such as the ordinary least squares and the corrected ordinary least squares because it is asymptotic. Thus, it has many desirable large sample properties. With the maximum likelihood estimation, a value is chosen for  such that the value makes the observations the most likely observations and that there is a high concord between the model and the observations. This makes the method more unique, nearly unbiased with large sample, and consistent as it brings the estimated parameter very close to the true value of the parameter. Aside from the estimate of the  value, the ML estimation also generates the gamma ( ) value. The gamma computes the total variation of observed output from the frontier output. It is expressed as the ratio of the variance of the error associated with inefficiency ( 2 u ) to the total variation in the model ( 2 ). The total variation of the model is defined as the addition of the variance of the error associated with inefficiency ( 2 u ) and the errors associated with the stochastic noise, 2 v , that is, 2 = 2 u + 2 v (xi) The gamma estimate is specified as: 2 2     u (xii) Gamma ( ) takes a value between zero and one, that is, 0<= <=1. Variations in the observed output are attributed to inefficiency factors if the gamma value is equal to one. On the other hand, deviation from the frontier 72 output is entirely attributed to statistical noise (random factors) if the gamma value is equal to zero (Battese &Corra, 1977; Coelli, 1995). Therefore, results would be equal to that of the ordinary least square results if the parameter gamma becomes zero whereas the noise term becomes irrelevant if the value of the gamma becomes one. The Kendall’s coefficient of concordance was used to rank the constraints faced by farmers in rice production. Kendall and Smith (1939) provide a descriptive measure for which the concordance between rank orders within an individual rank structure can be assessed. This measure which is known as the Kendall’s coefficient of concordance is a non-parametric statistic. It is a measure of agreement among several “judges” who assess a given set of objects. These “judges” could be variables or characters. It is used to identify a given set of constraints from the most critical to the least so as to measure the degree of agreement among respondents. In this study, the most important constraints to rice production were identified and assessed for severity on a scale of one to five (1 = very high constraint; 2 = high constraint; 3 = low constraint, 4 = very low constraint, 5= no constraint). The rankings were then subjected to the Kendall’s coefficient of concordance measure so as to know the degree of agreement of rankings by different rice farmers. After computing for the total rank score for each constraint, the constraint with the least score was interpreted as the most pressing constraint whereas the constraint with the highest score was ranked as the least constraint. Mathematically, it is expressed as: 73   2 2 2 2 12 ( 1) T T n W nm n            (xii) Where (W ) represents the coefficient of concordance which is defined as the ratio of the sum of squared deviations of rank totals from the average rank total to the maximum possible value of the sum of squared deviations of rank totals from the average rank total; T represents the sum of ranks for constraints being ranked; m represents the number of rice farmers; n represents the number of constraints being ranked. The F distribution was used to test for the significance of the Kendall’s coefficient of concordance (Tetteh, Adjetey & Abiriwie, 2011). Mathematically, the F-ratio is given as:    1 / 1F m W W     (xiv) From the above equation, the degree of freedom for the numerator is given as:    21n m  (xv) Likewise, the degree of freedom for the denominator is given as:   21 1m n m       (xvi) On the other hand, one can compute for the Kendall’s coefficient of concordance by using the sum of squares of rank totals instead of the sum of squared deviations of rank totals from the average rank total. It can be expressed in the form given by (Legendre, 2005) as: 74   2 1 2 2 12 1 3 11 m j j R n S nm n n       (xvii) When perfect agreement exists between the values of the ranking variable, W = 1. When W = 0, then it means that there is maximum disagreement between the values of the ranking variable. Kendall’s coefficient of concordance does not take negative values. It takes a value between zero and one ( 0 1W  ). Here, the null hypothesis tested is that there is no agreement among farmers in the ranking of the constraints. The null hypothesis is rejected if the computed F-value exceeds the tabulated, showing that the respondents are in agreement with each other on the ranking of the constraints. 75 CHAPTER FOUR RESULTS AND DISCUSSION This chapter presents the results of the analysis of data collected from the respondents. The chapter consists of four sections; namely description of the socioeconomic characteristics of rice farmers, the levels of technical efficiency of rice farmers, the determinants of technical efficiency, the elasticities of inputs and constraints faced by farmers in rice production in the study area. Finally, the chapter presents the results of tests of hypotheses. Description of socioeconomic characteristics of rice farmers The socio-economic characteristics presented in this section include age, sex, educational background, marital status, household size, number of years engaged in rice farming, access to extension, off-farm work and membership of a farmer-based organization. Table 2 shows the distribution of respondents by age. As indicated in Table 2, 28.8 per cent of respondents were less than 40 years. Also, it can be observed that majority of the rice farmers representing 63.5 per cent fell within the age group of less than 50 years. The mean age of 46.8 years could be classified as youthful, active and productive age that can contribute immensely to rice production in the study area and in the nation as a whole. This would generate income for the youth because rice is a cash crop. This finding lends support to the finding of Enwerem and Ohajianya (2013) who found that 77.5 per cent of small scale rice farmers in the Imo State of Nigeria fell within the age bracket of 41-50 years. The finding is also consistent with the finding of Donkoh, Ayambila and Abdulai (2012) 76 that the mean age of rice producers on the Tono irrigation scheme was 38.2 years; which was described as a youthful adult population. Table 2: Distribution of respondents by age Age of Farmers (Mean=46.8) Frequency Percent (%) 20 – 29 11 5.6 30 – 39 52 23.2 40 – 49 107 34.7 50 – 59 72 25.6 60 – 69 43 10.9 Total 285 100 Source: Field survey data, 2015. The sex distribution of the respondents was also analysed. The results in figure 4 show that majority of the respondents; representing 59.6 per cent were males while females were in the minority, representing 40.4 per cent. From the results, rice production was dominated by males in the study area. The finding is also consistent with that of Donkoh, Ayambila and Abdulai (2012) that males outnumbered females in the production of rice on the Tono irrigation scheme in the northern region of Ghana. In addition, the finding is consistent with that of Enwerem and Ohajianya (2013) who found that rice production among small scale farmers in the Imo state of Nigeria was male dominated. 77 Figure 4: Distribution of respondents by sex Source: Field Survey Data, 2015 The educational background of farmers which plays an important role in the adoption of technologies by farmers was also analysed. The results are presented in Table 3. As shown in Table 3, 37.5 per cent of the respondents had primary education while 33.0 per cent had middle school or junior high school education. Formal education is seen as a foundation for farmers to become better managers of agricultural resources. It is known to influence the kind of information accessed and the planning that takes place at the household level (Moyo & Chambati, 2009). The results further indicate that more than half of the respondents, representing 56.4 per cent had no formal education or attained only primary school education. This could negatively affect record keeping on farm operations. 78 Table 3: Distribution of respondents by educational background Educational background of farmer Frequency percent (%) No formal education 54 18.9 Primary education 107 37.5 Middle school / JHS 94 33.0 O’ level / SHS 27 9.5 Tertiary 3 1.1 Total 285 100.0 Source: Field survey data, 2015 The household sizes of the respondents were also analysed. The results are presented in Table 4. As shown in Table 4, majority of the rice farmers representing 56.5 per cent had house hold sizes of more than five persons. The mean household size of six persons implies that most of the labour could be supplied by the household members. Table 4: Distribution of respondents by household Size Household size (mean=6) Frequency Percent (%) 1-5 124 43.5 6-10 130 45.6 11-15 25 8.8 16-20 5 1.8 21-25 1 0.3 Total 285 100.0 Source: Field survey data, 2015 Another socioeconomic characteristics analysed was marital status. Results in Table 5 show that most of the rice farmers, representing 90.2 per cent were married. 79 Table 5: Frequency distribution of rice farmers by marital Status Marital Status Frequency Percent (%) Single 9 3.2 Married 257 90.2 Divorced 1 0.4 Widowed 18 6.2 Total 285 100.0 Source: Field survey data, 2015. Respondents were asked whether they had access to extension or not. The distribution of respondents by their access to extension services is illustrated in Figure 5. Figure 5 shows that 69.8 per cent of the respondents had access to extension services in the study area. This implies that majority of the farmers had access to information on rice production which could help enhance their output. Figure 5: Distribution of respondents on their access to extension services Source: Field survey data, 2015 80 Figure 6 reveals that 158 of the respondents, representing 55.4 per cent did not engage in off-farm work which could earn them extra income. This could affect farm work positively since much time would be spent on the farm instead of on non-farm activities (Ali & Flinn, 1989). Figure 6: Distribution of respondents by off-farm work Source: Field survey data, 2015 It can be observed from Figure 7 that majority of the respondents, representing 88.1 per cent did not belong to any farmer based organization. This might affect their efficiency negatively since membership of a farmer based organization is known to reduce the inefficiency of farmers since information on appropriate farming practices would be shared among the farmers (Idris, Sirwar & Talib, 2013). 81 Figure 7: Distribution of respondents by membership of a farmer based organization. Source: Field survey data, 2015 Respondents were also asked to state the variety of rice cultivated. The results presented in Figure 8 show that only two varieties of rice, namely; Jasmine 85 and Togo marshal were cultivated by farmers in the study area. As indicated in Figure 8, majority of the rice farmers, representing 63.2 per cent cultivated Jasmine 85 while only 36.8 per cent cultivated the Togo marshal. Jasmine 85 was more popular among the rice farmers because it was more high -yielding and had a better taste and aroma than the Togo marshal (MoFA, Ketu North District, 2014). 82 Figure 8: Distribution of respondents by the variety of rice cultivated Source: Field survey data, 2015 Table 6 presents the summary statistics of farmer-specific characteristics as well as production parameters. It can be observed from Table 6 that on average, rice farmers in the study area had 19 years of farming experience, with a minimum of 2 years and a maximum of 36 years. Table 6 also shows that the mean number of years of formal education was 5 years with a minimum of zero and a maximum of 13 years. Also, the mean extension contacts was twice a year. This is relatively low considering the importance of extension in agriculture. The low extension contacts imply that not much information got to the farmers as far as innovations and technologies are concerned. Table 6 also indicates that on average, rice farmers in the study area produced an output of 6059.9 kilograms of rice per hectare using an average of 1.66 hectares of land, 21.15 litres of weedicide per hectare, 492.33 kilograms of fertilizer per hectare, 16.98 litres of pesticide per hectare, 625 person days of labour per hectare, 275 kilograms of seeds per hectare, GH¢608.50 worth of irrigation facilities per hectare and GH¢40.75 worth of 83 equipment per hectare. The minimum output of rice was 3250 kilograms/hectare and the maximum was 22000 kilograms/hectare. The large variation in output of rice in the study area can be attributed to variations in their levels of technical efficiency. Table 6: Summary statistics of production parameters and farmer characteristics Variable Minimum Maximum Mean Standard deviation Output(kg)/ha 3250.00 22000.00 6059.85 4082.75 Land area(ha) 0.80 4.00 1.66 1.77 Fertilizer(kg)/ha 187.50 1000.00 492.33 163.55 Seed(kg)/ha 75.00 600.00 275.00 101.88 Pesticide(litres)/ha 2.50 40.00 16.98 7.78 Weedicide(litres)/ha 10.00 35.00 21.15 6.38 Labour(person days)/ha 195.00 1350.00 625.01 282.95 Irrigation cost(Gh¢)/ha 150 1240 608.50 236.65 Equipment(Gh¢)/ha 17.50 70.00 40.75 12.05 Farming experience (years) 2.00 36 18.58 1.77 Years of formal education(years) 0.00 13.00 5.58 3.58 Extension contacts(number) 0.00 6.00 2.34 1.77 Source: Field survey data, 2015. 84 Technical efficiency levels of rice farmers The maximum likelihood estimates of the stochastic frontier translog production parameters are presented in Table 7. It can be observed from Table 7 that only inputs of land, equipment, fertilizer and irrigation costs were statistically significant at 5 per cent. This implies that, among the eight inputs, only land, equipment, fertilizer and irrigation cost were important factors that influenced the output of rice in the study area. The other inputs, namely; labour, seed, weedicide and pesticide were not important factors that affect the yield of rice in the study since they were statistically insignificant. Among the significant inputs, however, only equipment cost has a negative sign. The negative sign on equipment cost implies that an increase in equipment cost would result in a decrease in rice output. In other words, rice output in the district would increase when equipment cost decreases. This finding is contrary to the finding of Bempomaa (2014) that equipment cost contributed positively to maize output in the Ejura-Sekyedumase District of the Ashanti Region. This could be due to the use of heavy equipment such as tractors and power tillers by the rice farmers on small land holdings. This is because heavy farm equipment such as tractors and tillers could not be utilised to their full capacity on small landholdings such as 0.8 hectares. Moreover, it was found that most of the farmers had too many equipment relative to the size of their farms. Therefore, these equipment could not be put to optimum use. Fertilizer input has a positive sign and this implies that an increase in the quantity of fertilizer would result in an increase in output of rice in the study area. This finding confirms that of Rahman, et al. (2012) who found fertilizer to be significant with a positive coefficient among marginal, small 85 and medium scale rice farmers in Bangladesh. Also, the variable, land is significant with a positive sign. This implies that an increase in the area of land under cultivation would lead to an increase in the output of rice. Irrigation cost was also significant with a positive sign. This means an increase in irrigation cost would result in an increase in output. This is because the amount paid for irrigation was proportional to the land area under cultivation which contributed positively to output. These findings are consistent with those of Rahman, et al. (2012) who found that land area under cultivation and irrigation cost were significant and positively contributed to output of rice among marginal, small, medium and large scale rice farmers in Bangladesh. In the case of the squared values of the input variables, none of them was significant. However, three of them, namely land, fertilizer and irrigation cost had positive signs while the remaining five inputs had negative signs. The squared values in a translog model shows the long term effects of the input variables on output. For instance, the fact that land and land squared were both positive imply that an increase in the area of land under cultivation would lead to an increase in output both in the short and long term. This finding is contrary to that of Donkoh, Ayambila and Abdulai (2012) who found that continuous increase in land area under cultivation would lead to a decrease in output of rice on the Tono irrigation scheme both in the short and long term. Also, an increase in fertilizer and irrigation cost would result in increase in output both in the short and long term since the square of these variables were positive. On the other hand, since labour and labour squared were negative it can be said that increasing the quantity of labour would lead to a decrease in output both in the short and long runs. Similarly, an increase 86 in seed rate, weedicide, pesticide and equipment cost would lead to a decrease in output. The interaction terms explain the substitutability or complementarity of the variables. A parameter with a positive sign implies that the two variables are complementary, while a parameter with a negative sign means that the two variables are substitutes. From Table 6, the statistically significant parameters with a positive sign are the interactions between seed and irrigation cost at 5 per cent; and weedicide and equipment at 10 per cent. Those with negative signs are land and seed at 5 per cent; seed and weedicide at 10 per cent and equipment and irrigation at 10 per cent. The positive sign on the interactions between seed and irrigation cost implies that seed was complementary to irrigation therefore the two inputs would be more productive when used together. Weedicide use was also complementary to equipment in the study area. The implication is that these pairs of inputs jointly contribute positively to output in the study area. On the other hand, seed was a substitute to land and weedicide as equipment and irrigation costs were substitutes. The interactions between these pairs of inputs yielded less productive results when used together in the study area. Diagnostic statistics and gamma parameters Table 7 also shows that the estimate of sigma-squared ( 2 ) value of 0.097 is statistically significant and different from zero at 0.1 per cent. This indicates a good fit and the correctness of the specified distributional assumption of the composite error term. The gamma value (γ) is a measure of inefficiency in the variance parameter and ranges from zero to one. From Table 7, the translog model used in the study area estimated gamma at 87 approximately 0.92 or 92 per cent. This implies that 92 per cent of the variations in output of rice were due to technical inefficiency in the study area. In other words, the presence of technical inefficiency among the sample farmers explains about 92 per cent of the variation in the output level of rice in the study area. The results of the diagnostic statistics therefore confirm the relevance of the stochastic parametric production function and the maximum likelihood estimate employed in the study. Table 7: Maximum likelihood estimates of the mean scaled translog stochastic frontier production function parameters Variable Parameter Coefficient standard error z – value Intercept β0 0.3745*** 0.01147 3.2651 ln(mland) β1 0.4421 ** 0.2267 1.9501 ln(mlabour) β2 - 0.0509 0.0735 - 0.6928 ln(mseed) β3 - 0.0320 0.0824 - 0.0185 ln(mweedicide) β4 - 0.0014 0.0771 -0.0185 ln(mpesticide) β5 -0.0915 0.0704 - 1.2986 ln(mequipment) β6 - 0.1415 ** 0.0804 -1.7592 In(mfertilizer) β7 0.1460 ** 0.0796 1.8343 ln(mirrigation) β8 0.3722 ** 0.2158 1.7248 ½ [In(mland)] 2 β9 4.0325 2.5886 1.5578 ½ [ln(mlabour)] 2 β10 -0.4927 0.3277 -1.3774 ½ [ln(mseed)] 2 β11 -0.4819 0.4463 -1.0797 ½[ln(mweedicide)] 2 β12 -0.5294 0.4882 -1.0846 ½ [lnmpesticide)] 2 β13 -0.1739 0.1390 -1.2424 ½[ln(mequipment)] 2 β14 -0.8909 0.5639 -1.2424 ½ [ln(mfertilizer)] 2 β15 0.2037 0.4236 0.4810 ½ [ln(mirrigation)] 2 β16 -2.2893 1.7236 -1.3282 ln(mland)*ln(mlabour) β17 -0.6767 0.7013 -0.9649 88 Variable Parameter Coefficient standard error z – value ln(mland)*ln(mseed) β18 -1.4844* 0.6942 -2.1383 ln(mland)*ln(mweedicide) β19 0.0178 0.8931 0.0199 ln(mland)*ln(mpesticide) β20 0.2296 0.3181 0.7218 ln(mland)*ln(mequipment) β21 0.9574 0.9948 0.9624 ln(mland)*ln(mfertilizer) β22 -0.4546 0.7729 -0.5881 ln(mland)*ln(mirrigation) β23 -2.8567 1.7729 -1.6098 ln(mlabour)*ln(mseed) β24 0.4329 0.2725 1.5885 ln(mlabour)*ln(mweedicide) β25 0.1598 0.3252 0.4915 ln(mlabour)*ln(mpesticide) β26 -0.0202 0.2059 -0.0983 ln(mlabour)*ln(mequipment) β27 -0.2544 0.3709 -0.6858 ln(mlabour)*ln(mfertilizer) β28 -0.0283 0.3723 -0.0761 ln(mlabour)*ln(mirrigation) β29 0.4979 0.6004 0.8293 ln(mseed) *ln(mweedicide) β30 -0.5679** 0.3155 -1.7998 ln(mseed)*ln(mpesticide) β31 0.1011 0.1643 0.6155 ln(mseed)* ln(mirrigation) β34 1.5351* 0.6115 2.5104 ln(mweedicide)*ln(mpesticide) β35 -0.0239 0.2084 0.1148 ln(mweedicide)*ln(mequipment) β36 0.6058** 0.3686 1.6473 ln(mweedicide)*ln(mfertilizer) β37 0.2742 0.3086 0.8889 ln(mweedicide)*ln(mirrigation) β38 0.0224 0.7602 0.0295 ln(mequipment)*ln(mfertilizer) β39 0.1992 0.3035 0.6564 ln(mequipment)*ln(mirrigation) β40 -1.3375 ** 0.7098 -1.8844 ln(mfertilizer)*ln(mirrigation) β41 0.1897 0.6354 0.2986 Variance parameters Sigma squared Gamma Log likelihood value 2 γ 0.0973*** 0.9191*** 4.4345 Significant codes: ***--------Significant at 0.1% **-- ------ Significant at 10 % *---------- Significant at 5% Source: Field survey data, 2015. Table 7 continued 89 Frequency distribution of technical efficiency of rice farmers Table 8 shows the frequency distribution of technical efficiency of rice farmers. The mean level of technical efficiency of rice farmers was 70.7 per cent with a minimum of 29.6 per cent and a maximum of 96.3 per cent. This shows that there was a wide disparity among rice farmers in their level of technical efficiency. This, in turn, indicates that there was an opportunity to improve the existing level of production of rice in the study area through enhancing the level of technical efficiency of rice farmers. The mean level of technical efficiency further implies that the level of output of rice in the study area could be increased on an average by about 29.3 per cent if appropriate measures are taken to improve the level of efficiency of rice farmers. In other words, there was a possibility of increasing the yield of rice by about 29.3 per cent using the available resources in an efficient manner without introducing a new technology. The results in the table 8 also show that 45.26 per cent of the respondents operated below the mean level of technical efficiency. Thus the null hypothesis that rice farmers in the Ketu North District are not fully technically efficient is not rejected. 90 Table 8: Frequency distribution of technical efficiency of rice farmers in the Ketu North District Technical Efficiency (%) Frequency Percentage (%) <50 26 9.12 50 – 59 44 15.44 60 – 69 59 20.70 70 – 79 61 21.40 80 – 89 63 22.11 >89 32 11.23 Total 285 100 Mean technical 70.7 Efficiency Minimum 29.6 Maximum 96.3 Source: Field survey data, 2015 Elasticity of output Determination of elasticity is necessary for the estimation of the responsiveness of output to input. Table 9 shows results of the translog stochastic frontier production function. It can be observed from Table 8 that input elasticity of land area under cultivation was 0.44. This means that a 1 per cent increase in land area under cultivation would increase yield of rice by 0.44 per cent. Also, a 1 per cent increase in the quantity of fertilizer would increase output by 0.15 per cent and a 1 per cent increase in irrigation cost would increase output by 0.37 per cent. However, coefficients of elasticity of labour, seed, weedicide, pesticide and equipment were negative. The implications are that a 1 per cent increase in labour would decrease output by 0.05 per cent; a 1 per cent increase in the quantity of seed planted would 91 decrease output by 0.03 per cent; a 1 per cent increase in the quantity of weedicide would decrease output by 0.001 per cent; a 1 per cent increase in the quantity of pesticide would decrease output by 0.09 per cent and finally, a 1 per cent increase in the cost of equipment would decrease output by 0.14 per cent. Moreover, all the inputs used in the production of rice in the study area were found to be inelastic; a 1 per cent increase in each input resulted in less than proportionate increase in output. Furthermore, the return- to-scale indicated in Table 9 is 0.642. The return-to-scale of the technology is given by the sum of the elasticities of all the inputs. If all inputs are varied by the same proportion, the return-to -scale indicates the percentage by which output will increase. The return- to -scale of 0.642 is less than one and indicates a decreasing return to scale. This implies that if all inputs are proportionally increased by 1 per cent, output of rice would increase by only o.642 per cent. It is called a decreasing return to scale because the relative increase in output is less than the relative increase in the aggregate input quantity. The result further indicates that rice farmers were operating at the irrational stage of production (stage III) in the study area. Table 9: Elasticity of output and return-to-scale Input variable. Elasticity Return -To- Scale (RTS) ln(mland) 0.442 0.642 ln(mlabour) -0.051 ln(mseed) -0.032 ln(mweedicide) -0.001 ln(mpesticide) -0.092 ln(mequipment) -0.142 In(mfertilizer) 0.146 ln(mirrigation) 0.372 Source: Field survey data, 2015. 92 Determinants of technical efficiency The determinants of the technical efficiency are discussed using the estimated (δ) coefficients associated with the inefficiency effects in Table 10. Variables with negative coefficients have negative relationships with inefficiency while those with positive coefficients have positive relationships with inefficiency. The results show that age, sex, farming experience and membership of farmer based organization were the only significant determinants of the level of technical efficiency in the study area. Education, extension contact, off farm occupation, access to credit and household size were not statistically significant. This means that these factors were not important determinants of technical efficiency in the study area. Furthermore, the insignificance of extension visits has not come as a surprise. This is because even though most of the rice farmers had access to extension services in the study area, the mean extension contacts was found to be only twice in a year. This might be due to the low extension agent-farmer ratio in the study area (1:3500) which could make extension services ineffective (MoFA, Ketu North District, 2014). However, the positive sign on extension means that rice farmers who had access to extension services were less technically efficient than those who had no access to extension. This is, however, contrary to the apriori expectation. Among the significant inefficiency sources, only age has a positive sign. This implies that older farmers were less technically efficient than younger farmers. This finding lends support to the finding of Maganga (2012) who found that older farmers were less technically efficient than younger farmers in the production of Irish potato at the Dedza District of Malawi. The finding is also consistent with that of Njeru (2010) who found that older wheat 93 farmers were less technically efficient than younger ones in the Uasin Gishu District of Kenya. This could be explained by the fact that older farmers have the tendency to stick to their old methods of production and are usually unwilling to accept change. This finding is however contrary to the position of Ali, Imad and Yousif (2012) that adequate inputs coupled with long years of farming enable older farmers to produce more efficiently. Furthermore, the finding contradicts that of Etwire, Martey and Dogbe (2013) who found younger soybean farmers in the Saboba and Chereponi Districts of Northern Ghana to be less technically efficient than older ones. The negative sign on the sex variable means that male farmers were more technically efficient than female farmers in the study area. This finding is consistent with that of Donkoh, Ayambilah and Abdulai (2012) who found male farmers to be more technically efficient than female farmers in the production of rice on the Tono irrigation scheme in the Northern Region of Ghana. This could be explained by the fact that male farmers are wealthier and therefore are able to acquire technologies that are costly (Onunmah & Acquah, 2010). Female farmers are also known to have more responsibilities at home than male farmers. These, coupled with their biological child bearing role could reduce efficiency in production because less time would be devoted to management of the farm. The negative sign on experience implies that farmers with more years of farming experience are more technically efficient than those with less farming experience. Again, this finding confirms that of Maganga (2012). Farmers with longer years of farming may combine inputs more optimally leading to high technical efficiency than the less experienced farmers. 94 Finally, the coefficient of the dummy variable for membership of a Farmer-Based Organisation (FBO) is negative and statistically significant at 10 per cent. This implies that rice farmers who belonged to a farmer based organisation were more technically efficient or less technically inefficient than those who did not belong to any farmer based organisation. This finding lends support to that of Awunyo-Vitor, Bakang and Cofie (2012) who found that cowpea farmers who belonged to a farmer based organisation were more technically efficient than those who did not. Membership of a farmer based organization is part of social capital. It also affords farmers the opportunity to share information on modern rice production practices through interactions with other farmers (Awunyo-Vitor, Bakang & Cofie, 2012). The finding is however contrary to that of Kuwornu, Amoah and Seini (2013) who found membership of farmer based organization to positively influence technical inefficiency in maize production in the Eastern Region of Ghana. The maximum likelihood estimates of the inefficiency model indicate that the socio-economic characteristics of the rice farmers significantly influence technical efficiency in the study area. Thus the null hypothesis that the socioeconomic characteristics of rice farmers has no significant effect on technical efficiency is rejected. 95 Table 10: Maximum likelihood estimates of the inefficiency model. Variable Parameter Coefficient Standard Error Z-Value Intercept 0 0.0346 0.2723 0.1269 z- Education 1 0.0034 0.0063 0.3623 z- Age 2 0.0115* 0.0047 2.4381 z- Sex 3 -0.1807* 0.0715 -2.5262 z- Extension 4 0.0262 0.0701 0.3733 z- Off farm 5 0.1005 0.0686 1.4662 z – Credit 6 -0.0589 0.1135 -0.5195 z-Household size 7 0.0051 0.0090 0.5637 z- Experience 8 -0.0125* 0.0056 -2.2385 z-FBO membership 9 -0.2119 ** 0.1224 -1.7326 Significant codes: * ------- Significant at 5% **------- Significant at 10% Source: Field survey data, 2015 Constraints to rice production The views of respondents were sought on the seriousness of the constraints faced in rice production in the study area. The constraints were tabulated and ranked on a 5-point Likert scale as follows: 1-Very high constraint; 2-High constraint; 3-Low constraint; 4-Very low constraint; 5- No constraint. The result is presented in Table 11. The Kendall’s coefficient of concordance (Kendall’s W) of 0.598 shows that an agreement exists among the rice farmers on the ranking of constraints to rice production at 1 per cent significant level. This implies that rice farmers in the study area agreed on the rankings of the constraints that limit them in the production of rice. Therefore the null hypothesis that there is no concordance among the rice farmers on the ranking of the constraints they faced in rice production is rejected in favour of the alternative. 96 Table11: Kendall’s W test statistics W 285 Kendall’s Wa 0.598 Chi-square 2896.49 Df 17 Asymp.sig. 0.000*** Significant codes: ***------- Statistically significant at 1% a-----------Kendall’s coefficient of concordance. Source: Field survey data, 2015 It can be observed from Table 12 that the first three constraints faced by rice farmers were low purchasing price of rice, lack of government support and difficulty in accessing capital. These constraints were interrelated. For instance, lack of government support in the form of provision of capital for production compels farmers to access loans from unscrupulous moneylenders at exorbitant interest rates. Most of these money lenders also buy and sell rice for profit therefore they dictate the price. Moreover, the dire need for money to settle debts owed by the farmer to friends and relatives compels them to sell their produce at the prevailing market price which is usually low. The farmers had, therefore, become price takers. Further, the low price offered could not cover production cost, making it difficult for farmers to purchase the required inputs for the next season. This finding confirms the finding of Matanmi et al. (2011) who cited financial constraints and access to inputs as serious challenges facing irrigated rice productions at the Kwara State, Nigeria. The finding is also consistent with those of Alarima, et al. (2011) and Bempomaa (2014) who found that lack of financial agencies to support production; poor 97 capital base and low purchasing price of rice were the major constraints facing rice and maize production in Nigeria and Ghana respectively. The next constraint faced by rice farmers was the erratic nature of rainfall in the study area. Although the rice farmers were on an irrigation scheme, inadequate rainfall makes irrigation water unavailable since the water level in the dams becomes very low during the minor raining season. On the other hand, excessive rainfall leads to flooding. This finding lends support to that of Musime et al. (2005) that change in rainfall pattern was a major constraint limiting rice production in the Bugiri District, Kampala, Uganda. At the bottom of the table is the constraint of lack of storage facilities. In fact, almost all the respondents did not see this as a problem because all the sections on the irrigation scheme had a place for storage. Also, there was ready market for rice and the financial pressure on farmers to sell their produce to enable them pay debts could not permit them to store their produce for some time to attract higher prices. Table 12: Ranking of constraints faced by rice farmers Constraints Mean Rank Low purchasing price of rice 4.07 Lack of government support 4.41 Difficulty in accessing to capital 4.70 Erratic rainfall 5.50 Price fluctuations 6.43 High cost of labour 6.00 Difficulty in land preparation 8.45 Pests and diseases 9.32 Lack of marketing associations 9.27 Difficulty in weed control 9.45 98 Constraints Mean Rank Difficulty in acquiring fertilizers 9.47 Difficulty in acquiring pesticides 9.66 Difficulty in applying chemicals 12.20 Difficulty in land acquisition 12.37 Difficulty in Planting of seeds 13.99 Difficulty in acquiring planting materials 14.39 Lack of market for rice 15.32 Lack of storage facilities 16.01 Source: Field survey data, 2015 Table 12 continued 99 CHAPTER FIVE SUMMARY, CONCLUSIONS AND RECOMMENDATIONS This chapter presents a summary of the study; that is, the research problem and the main findings of the study. The chapter also presents the conclusions drawn based on the findings and recommendations based on the conclusions. Overview of the research problem The research problem investigated was the technical efficiency of rice production on the Weta Irrigation scheme in the Ketu North District of the Volta Region of Ghana. Generally, the study sought to estimate the level of technical efficiency and the determinants of technical efficiency in rice production. Data were collected from 285 rice farmers through interviews and this constituted the unit of analysis. The cross-sectional survey research design was used for the study. Primary data were collected from the respondents using the structured interview schedule. A pre-test was done in the South Tongu District of the Volta Region to establish the reliability of the instrument. Descriptive statistics, namely; percentages, frequency, mean and standard deviation as well as the translog stochastic frontier analysis, an econometric model were employed in the analyses of the research questions and tests of hypotheses. The analyses were done using the SPSS version 21 and the ‘R’ programming software. 100 Summary of main findings It was found from the results of the analyses that majority of the rice farmers, representing 63.5 per cent were within the age group of less than 50 years and the mean age was 46.8 years. Rice production in the district was dominated by males, representing 59.6 percent. Most of the respondents representing 90.2 per cent were married and the mean household size was six persons. Also, majority of the rice farmers, representing 56.4 per cent could not attain the middle school or junior high school level of education. This figure comprised 18.9 per cent of the respondents who had no formal education and 37.5 per cent who had only primary education. Furthermore, the mean extension contacts was twice a year even though majority of the farmers, representing 69.8 per cent had access to extension services. Most of the respondents, representing 88.1 per cent did not belong to any farmer based organization while 55.4 per cent did not engage in any off-farm economic activity. Two major rice varieties were cultivated in the study area namely; Jasmine 85 and Togo marshal, with the majority of farmers, representing 63.2 per cent cultivating the Jasmine 85. On average, 1.66 hectares of land, 625 person days of labour per hectare, 492kilograms of fertilizer per hectare, 21 litres of weedicide per hectare, 17 litres of pesticide per hectare, 275 kilograms of seeds per hectare, Gh¢41 worth of irrigation and GH¢16 worth of equipment were combined to obtain a yield of 6060 kilogram of rice per hectare. Also, results indicated that rice farmers in the study area were not fully technically efficient. Variations in output were attributed to inefficiency in production as well as random variables outside the farmer’s control; with the 101 former contributing a higher percentage (92 per cent). There was a wide variation in technical efficiency with a minimum of 29.6 per cent and a maximum of 96.3 per cent. The mean technical efficiency score was 70.7 per cent. Therefore, 29.3 per cent of output was lost to inefficiency and other factors. Furthermore, the results indicated that the significant input factors that affected the output of rice were land, fertilizer, irrigation and equipment. Of these, only equipment affected rice output negatively while land area under cultivation, fertilizer and irrigation cost positively influenced output. Also, all the inputs used in the production of rice were found to be inelastic and the farmers were found to be producing at decreasing returns to scale. Results of the technical inefficiency model indicated that the socio economic variables that significantly influenced technical efficiency were age, sex, farming experience and membership of a farmer based organization. Of these, only age had a negative relationship with technical efficiency while sex, farming experience and membership of a farmer based organisation influenced technical efficiency positively. Finally, empirical results showed that low purchasing price of rice, lack of government support, difficulty in accessing capital and erratic rainfall patterns were the major constraints faced by rice farmers in the study area. 102 Conclusions Based on the findings, the following conclusions are drawn: Firstly, rice farming on the Weta irrigation scheme in the Ketu North District was male dominated and the male farmers were more efficient in rice production than female farmers. 56.5 per cent of the rice farmers had household sizes of more than five persons. This implies that most of the labour could be supplied by the household members. Majority of the rice farmers were in the age group of less than 50 years, implying a youthful adult population engaged in rice production in the study area. Most of the respondents, representing 56.4 per cent constituted those who had no formal education and those who attained only primary or elementary school education. This could negatively affect record keeping on farm operations as well as the farmers’ ability to understand and adopt innovations in farming. This could, in turn, affect farmers’ efficiency negatively. Majority of the farmers belonged to a farmer based organizations which would enable them to share information on modern practices in rice production. This could explain why farmers in the study area who belonged to a farmer based organization were more technically efficient in rice production than those who did not belong to any farmer based organization. Most of the rice farmers had access to extension services which was rather inadequate and the main variety of rice which was cultivated by the farmers was Jasmine 85. Secondly, rice farmers in the study area were not fully technically efficient. With the mean technical efficiency estimated at 70.7 per cent, there was an opportunity for the rice farmers to increase their output by 29.3 per cent through efficient reallocation of the available resources. 103 Thirdly, in the study area, land area under cultivation, fertilizer input and irrigation cost were significant and had positive effects on output while equipment cost was also significant but affected output of rice negatively. Furthermore, all the input variables were inelastic and this implies that 1percent increase in each input would result in less than proportionate increase in output in the study area. Also, rice farmers were producing at decreasing returns- to- scale. Fourthly, the socio-economic characteristics of rice farmers which were significant determinants of technical efficiency in the study area were age, sex, farming experience and membership of a farmer based organization. Finally, the major constraints faced by rice farmers in the study area were low purchasing price of rice, lack of government support, difficulty in accessing capital and erratic rainfall patterns. All other things being equal, if these constraints are removed or minimised rice farmers on the Weta irrigation scheme would be more efficient in production. Recommendations From the conclusions drawn, the study recommends that: 1. Since rice production in the Ketu North District was male dominated, the Ministry of Food and Agriculture should formulate policies targeted at empowering male farmers by improving their access to agricultural inputs, especially, land, fertilizer and irrigation facilities to increase their efficiency in rice production. Also, training of farmers on how to improve upon their production activities through the efficient combination of inputs should be intensified through demonstration farms within the vicinity of farmers since the farmers were not fully efficient in production. More Agricultural 104 Extension Agents should be employed by the Ministry of Food and Agriculture to facilitate the training of farmers. 2. Furthermore, the Ministry of Education should bring formal education to the doorstep of farmers by establishing basic and second cycle schools in the farming communities to equip the youth with knowledge and skills to get the maximum benefits from innovations in farming. 3. The Irrigation Development Authority should expand the land area under irrigation and provide adequate irrigation facilities to increase farmers’ land holdings on the irrigation scheme. This is because the study established that increased land area under cultivation and irrigation cost positively influenced output of rice. 4. Furthermore, the Ministry of Food and Agriculture should adopt appropriate measures such as introducing a fertilizer subsidy that will ensure the availability of fertilizers at affordable rates to farmers. This would increase fertilizer use resulting in increased yield. Also, farmers should be educated on optimum use of equipment so as to increase output. 5. In order to improve efficiency in rice production on the Weta Irrigation Scheme, farmers need to organise themselves into groups since group membership positively influenced their efficiency. 6. The government, through the Ministry of Food and Agriculture, should make rice production attractive to the youth by providing them with incentives such as soft loans that would enable them take farming as a business. This is because it was established from the study that younger rice farmers were more efficient than older ones. 7. The government should establish guaranteed prices for rice and provide credit facilities to encourage the rice farmers to produce. 105 REFERENCES Abdulai, A., & Tietje, H. (2007). Estimating technical efficiency under unobserved heterogeneity with stochastic frontier models: Application to northern German dairy farms. European Review of Agricultural Economics, 34(3), 393–416. Adeyemo, R., Oke, J. T. O., & Akinola, A. A. (2010). Economic efficiency of small scale farmers in Ogun State, Nigeria. Tropicultura, 28(2), 84-88. Ahzar, R. A. (1991). Education and technical efficiency during the Green Revolution in Pakistan. Economic and Cultural Change, 651-665. Aigner, D. J., Lovell, C. A. K., & Schmidt, P. (1977). Formulation and estimation of stochastic frontier production function models. Journal of Econometrics, 6, 21-37 Aigner, D.J., & Chu,S.F. (1968). On estimating the industry production function. American Economic Review, 58, 826-39. Ajibefun, A.I., & Aderinola, A. E. (2004). Determinants of technical efficiency and policy implications in traditional agricultural production: Empirical study of Nigerian food crops farmers. AERC final report. Ajibefun, I. A., Battese, G. E.,& Kadar, A. (1996). Efficiency and productivity analysis. A working paper. Department of Econometrics, University of New England: Armidale, Australia. Akramov, K., & Malek, M. (2011). Analyzing profitability of maize, rice, and soybean production in Ghana: Results of PAM and DEA analysis. GSSP working paper 28.International Food Policy Research Institute: Accra 106 Alarima, C.I., Adamu, C.O. Masunaga, T., & Wakatsuki, T. (2011). Constraints to sawah rice production system in Nigeria. Journal of Human Ecology, 36(2), 121-130 Al-hassan, S. (2008). Technical efficiency of rice farmers in Northern Ghana. African Economic Research Consortium: Nairobi. Ali, A. A., Imad, E. E., & Yousif, A. K. (2012). Economic efficiency of wheat and faba bean production for small scale farmers in the Northern State of Sudan. The journal of Animal and Plant sciences, 22(1) Ali, M., & Flinn, J. C. (1989). Profit efficiency among Basmati rice producers in Pakistan Punjab. American Journal of Agricultural Economics 71, 303-310. Amaechi, E.C.C, Ewuziem, J.E. & Agumanna, M.U. (2014). Estimation of technical efficiency in the translog stochastic frontier production function model. An application to the oil palm produce mills industry in Nigeria. Advances in Applied Science Research, 5(3), 230-236 Amanor‐Boadu, V. (2012). Rice price trends in Ghana (2006–2011).METSS‐ Ghana Research and Issue Paper Series 02‐2012.Accra. Amedahe, F. K. (2002). Notes on educational research. Unpublished lecture notes. University of Cape Coast. Cape Coast. Audibert, M. (1997). Technical efficiency effects among paddy farmers in the villages of the Niger, Mali, West Africa. Journal of Productivity Analysis, 8, 379-94 Awunyo-Vitor, D.,Bakang, J., & Cofie, S. (2013). Estimation of farm level technical efficiency of small scale cowpea production in Ghana. American-Eurasian Journal of Agriculture & Environmental Sciences, 13 (8), 1080-1087 107 Bagi, F. S. (1984). Stochastic frontier production function and farm level technical efficiency of fulltime and part-time farms in West Tennessee, N. Cent. Journal of Agricultural Economics, 6, 48-55 Balk, D. (1997). Change comes slowly for rural women in Bangladesh. Asia- Pacific population and policy. Baten, M. A., Kamil, A. A., & Haque, M. A. (2009). Modeling technical inefficiencies effects in a stochastic frontier production function for panel data. African Journal of Agricultural Research, 13, 74-1382. 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, 169-179. Battese, G. E., & Coelli, T. J. (1988). Prediction of firm-level technical efficiencies with a generalized frontier production and panel data. Journal of Econometrics, 38(3), 387-399. Battese, G. E., & Sarfraz, H. (1998). Technical efficiency of cotton fanners in Vehari District of Punjab, Pakistan. Center for Economic Policy Analysis Working Paper No. 8/98. Department of Econometrics, University of New England: Armidale, Australia. Battese, G. E., & Coelli, T. J. (1995). Model for technical inefficiency effects in a stochastic frontier production function for panel data. Empirical Economics, 20, 325-32. Bempomaa, B. (2014). Technical efficiency in maize production in Ejura- Sekyedumase District of the Ashanti Region. Unpublished master’s thesis. University of Cape Coast. Cape Coast. Ben-Belhassen, B. (2000). Measurement and explanation of technical efficiency in Missouri hog production. Selected paper. American 108 Agricultural Economics Association (A AH A). Annual meeting. Tampa, Florida, 30 July - 2 August. Binam, J. N., Toyne, J., Wandji, N., Nyambi, G., & Akoa, M. (2004).Factors affecting the technical efficiency among smallholder farmers in the slash and burn agricultural zone of Cameroon.Food Policy, 29, 531- 545. Bozoglu, M., & Ceyhan, V. (2007). Measuring the technical efficiency and exploring the inefficiency determinants of vegetable farms in Samsun Province, Turkey. Agricultural systems,94(3), 649-656. Bravo-Ureta, B. E.,& Rieger, L. (1991). Dairy farm efficiency management using stochastic frontiers and neoclassical duality. American Journal of Agricultural economics, 73(2), 421-428. Bravo-Ureta, B.E., & Pinheiro, A. E. (1997). Technical, economic and allocative efficiency in peasant farming: Evidence from the Dominican Republic. The Developing Economies, 35(1), 48-67. Charnes, A., Cooper, W.A.V & Rhodes, E. (1978). Measuring the inefficiency of decision making unit. European Journal of operational Research, 2, 429-44. Chirwa, E.W. (2003). Sources of technical efficiency among smallholder maize farmers in southern Malawi. Draft final report presented at the biannual research workshop of AERC, Nairobi. Christensen, L. R., Jorgenson, D. W., & Lau, L. J. (1971). Conjugate duality and the transcendental logarithmic production function. Econometrica 39 (4), 255-256. 109 Chukwuji, C. O., Inoni, O. E., & Ike, P. C. (2007). Determinants of technical efficiency in gari processing in Delta State Nigeria. Journal of Central European Agriculture,8(3), 327-336 Coalition for African Rice Development (CARD) (2010). Mapping of poverty reduction strategy papers (PRSPs). Sector strategies and policies related to rice development in Ghana. Coelli, T. & Battese, G.E. (1996). Identification of factors which influence the technical inefficiency of Indian farmers. Australian Journal of Agricultural Economics, 40(2), 103-28. Coelli, T. J., Rao, D. S., O'Donnell, C. J., & Battese, G. E. (2005). An introduction to efficiency and productivity analysis.Springer Inc. :New York, USA. Coelli, T.J. (1995).Recent developments in frontier modeling and efficiency Measurement. Australian Journal of Agricultural Economics, 39(3), 219-245. Cohen, L., Manion, L., & Morrison, K. (2007). Research methods in education. (6 th Ed.). Routledge: London Dalton, T., & Guei, R. (2003). Productivity gains from rice genetic enhancements in West African countries and ecologies. World Development, 31 (2), 359–374 Diako, C. E., Sakyi-Dawson, B., Bediako-Amoa, F. K. S., & Manful J. T. (2010). Consumer perceptions, knowledge and preferences for aromatic rice types in Ghana. Nature and Science, 8 (12), 12–19 Diako, C. E., Sakyi-Dawson, B., Bediako-Amoa, F. K. S., &Manful, J.T. (2011). Cooking characteristics and variations in nutrient content of 110 some new scented rice varieties in Ghana. Annals. Food science and Technology, 12 (1), 39–44. Debertin, D. L. (2012). Agricultural production economics.University of Kentucky Debreu, G. (1951). The coefficient of resource utilization.Econometrica, 19 (3), 273-292. Donkoh, S.A., Ayambila, S., & Abdulai, S. (2012). Technical efficiency of rice production at the Tono irrigation scheme in Northern Ghana. Unpublished master’s thesis. University of Development Studies. Tamale. Ellis, F. (1993). Peasant Economics: Farm households and agrarian development. (2 nd Ed.). Wye studies in Agricultural and Rural Development. Cambridge University Press: London. 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. Etwire, P. M., Martey, E., & Dogbe, W. (2013). Technical efficiency of soybean farms and its determinants in Saboba and Chereponi Districts of Northern Ghana: A stochastic frontier approach. Sustainable Agriculture Research, 4, 106-116. Ezekiel, A. A., Adedapo, K. D., &Olapade, F. (2009). Technical efficiency of rice farmers in Ilesa agricultural zone in Osun State, Nigeria. Food production in a developing economy. 1, 220-204 Farrell, M. J. (1957). The measurement of productive efficiency: Journal of Royal Statistical Society Series A., 120(3), 253-290. 111 Gay, L. R. (1992). Research methods for business management.Macmillan and Co. : New York. Ghana Irrigation Development Authority, Ketu North District (2014). District profile on irrigation. Ghana Irrigation Development Authority &Japan International Development Agency (2004). Strategies for effective utilization of existing small scale irrigation projects. Irrigated Agriculture Promotion Project-Follow UP (SSIAPP-FU). GIDA/JICA: Accra Ghana Irrigation Development Authority, Ketu North District (2014). District profile on irrigation. Ghana Statistical Service, [GSS] (2012). Revised Gross Domestic Product 2011. Retrieved July, 2013 from http: //pdf.usaid.gov/pdf_ docs/PNAB K614.pdf Griffiths, W. E., Hill, R. C., & Judge, G.G. (1993). Learning and practising econometrics. Wiley: New York Haider, M. Z., Ahmed, M. S., & Mallick, A. (2011). Technical efficiency of agricultural farms in Khulna, Bangladesh: A stochastic frontier approach. International Journal of Economics and Finance, 3(3), 248- 256. Hayami, Y., & Ruttan, V. W. (1985). Agricultural development: An international perspective. Baltimore: Johns Hopkins University Press. Heidhues, F. (1995). Rural financial markets- an important tool to fight poverty. Quarterly Journal of Agricultural Economics, 34, 105-108. Henderson, J. M. & Quandt, R. (1971). Microeconomic theory. McGraw-Hill: New York.106-113 112 Henningsen, A. (2013). Introduction to econometric production analysis with R (Draft version). Heshmati, A. &Mulugeta, Y. (1996). Technical efficiency of the Ugandan matoke farms. Applied Economic Letters, 3, 49-194. Herdt, R.W. & Mandac, A. M. (1981). Modern technology and economic efficiency of Philippines rice farmers. Economic Development and Cultural Change, 29,375-399. Huang, C. J. & Liu, J. T. (1994). Estimation of a non-neutral stochastic frontier production function. Journal of Productivity Analysis, 4,171 - 180. Hummelbrunner, S. A., Rak, L.J., & Gray, J. (1996). Contemporary business statistics with Canadian applications. (2 nd Ed.).Prentice Hall, Canada Inc.: Ontario Idiong, I. C. (2007). Estimation of farm level technical efficiency in small scale swamp rice production in Cross River State of Nigeria: A stochastic frontier approach. World Journal of Agricultural Sciences, 3(5), 653-658. Idris, M. N., Chamhuri, S., & Talib, B. (2013). Determinants of technical efficiency on pineapple farming. American Journall of Applied Sciences, 10(4), 4-6. Jema, H. (2007). Production efficiency of smallholders' vegetable-dominated mixed farming system in Eastern Ethiopia: A non-parametric approach. Journal of African Economies, 16(1), 1-27. Jondrow, J., Lovell, C. A. K., Materov, I. S., & Schmidt, P. (1982). On the estimation of technical inefficiency in stochastic frontier production function model. Journal of Econometrics, 19, 23-338 113 Kalirajan, K.P. (1981). An econometric analysis of yield variability in paddy production. Canadian Journal of Agricultural Economics, 29, 283-294 Kebede, T.A. (2001). Farm household technical efficiency: A stochastic frontier analysis. A study of rice producers in Mardi Watershed in the Western Development Region of Nepal. Unpublished master’s thesis. University of Bergen. Kendall, M. G., & Smith, B. B. (1939). The problem of rankings. Annals of Mathematical Statistics, 10(3), 275-287 Khai, H. V., & Yabe, M. (2011). Technical efficiency analysis of rice production in Vietnam.Journal of International society for Southeast Asian Agricultural Sciences, 17(1), 135-146. Kolawole, O. (2009). Technical, allocative and economic efficiency in upland rice production system in Nigeria: A frontier function approach. Food production in a developing economy, 1, 273-279. Koopmans, T.C. (Ed.). (1951). Analysis of production as an efficient combination of activities. Cowles Commission for Research in Economics: New York, Wiley. Kopp, R. J., & Smith, V. K. (1980). Frontier production function estimates for steam electric generation:A comparative analysis. Southern Economic Journal, 1049-1059. Krejcie, R.V.,& Morgan, D.W. (1970). Determining sample size for research activities. Educational & psychological measurement, 30, 607-610. Kumbhakar, S. C., Ghosh, S., & McGuckin, J. T. (1991). A generalized production frontier approach for estimating determinants of 114 inefficiency in U.S. dairy farms. Journal of Business and Economic Statistics, 9, 279-86. Kumbhaker, S. C. & Lovell, C. A. K. (2000). Stochastic frontier analysis. Cambridge University Press: Cambridge, UK Kuwornu, J. K., Amoah, E., & Seini, W. (2013). Technical efficiency analysis of maize farmers in the Eastern Region of Ghana. Journal of Social and Development Sciences, 4(2),84-99. Lau, L. J. (1986). Functional forms in econometric model building. Elsevier Science Publishers : New York. Legendre, P. (2005). Species associations: The Kendall coefficient of concordance revisited. Journal of Agricultural, Biological, and Environmental Statistics, 10(2), 226-245. Leon, J. V. (2001). Cost frontier analysis of efficiency: An application to the Peruvian municipal banks. Unpublished Doctoral thesis, Ohio State University: USA Lin, J. Y. (1991). Education and innovation adoption in agriculture: Evidence from hybrid rice in China. American Journal of Agricultural Economics 73(3), 713-23. Llwelyn, R. V., & Williams, J. C. (1996). Non-parametric analysis of technical and scale efficiencies for food production in East Java. Indonesian Journal of Agricultural Economics, 15, 113-126 Louis, C., Lawrence, M., & Keith, M. (2007).Research methods in education. Routledge Publishers: New York. Lyubov, A. K., & Jensen, H. H. (1998). Technical efficiency of grain production in Ukraine. Paper presented at the 1998 American 115 Agricultural Economics Association Annual Meeting, Salt Lake City, Utah, 2-5 August. Maganga, A. M. (2012). Technical efficiency and its determinants in Irish potato production: Evidence from Dedza District, Central Malawi. African Journal of Agricultural Research, 7(12), 1794-1799. Matanmi, B. M., Adesiji, G. B., Owawusi, W.O.,& Oladipo, F. O.(2011). Perceived factors limiting rice production in Patigi Local Government Area of Kwara State, Nigeria. Journal of Agriculture and Social Research 11(2), 99-102. Meeusen, W. &van den Broeck, J. (1977). Efficiency estimation from Cobb- Douglas production function with composed error. International Economic Review, 18, 435-44. Millennium Development Authority. (2010). Investment opportunity in Ghana: Maize, rice, and soybean. MiDA: Accra, Ghana Ministry of Food and Agriculture (2009). Evaluation of the Ghana rice campaign: A marketing campaign implemented by Engineers without Borders and the Ghana Ministry of Food and Agriculture to stimulate the rice value chain, April, 2009. Ministry of Food and Agriculture (2010). Agriculture in Ghana: Facts and figures. SRID: Accra. Ministry of Food and Agriculture, Ketu North District (2014). District Agriculture Profile. District Agriculture Development Unit: Dzodze. Ministry of Food and Agriculture, MOFA (2011). Agriculture in Ghana: Facts and figures. SRID: Accra. 116 Moyo, S., & Chambati, W. (2009). Fast track land reform baseline survey in Zimbabwe: Trends and tendencies, 2005/2006.African Institute for Agrarian Studies (AIAS): Harare. Musiime, O. Tenywa, M. M., Majaliwa, M.J. G., Lufafa, A. Nanfumba, D., Wasige, J., Woomer, P. L. & Kyondha, M.(2005). Constraints to rice production in Bugiri district. African Crop Science Conference Proceedings, 7, 1495-1499 Nchare, A. (2005). Analysis of factors affecting technical efficiency of Arabica coffee producers in Cameroon. AERC final report 2005. Njeru, J. (2010).Factors influencing technical efficiencies among selected wheat farmers in Uasin Gishu District, Kenya. African Economic Consortium Research Paper, 206 Natural Resources Institute, NRI. (2013). Contrasting rice value chains: A benchmarking study of rice in Ghana, Senegal and Thailand, draft report. Natural Resources Institute, University of Greenwich: Greenwich, UK. Obiri-Opareh, N. (2008). Socio-economic analysis of rice production in Ghana: Agenda for policy study. Ghana Journal of Agricultural Science, 41(2), 97-107 Obwona, M. (2000). Determinants of technical efficiency differentials amongst small and medium scale farmers in Uganda: A case of tobacco growers. Final report presented at the AERC biannual workshop, Nairobi, Kenya. Obwona, M. (2000). Determinants of technical efficiency differentials amongst small and medium scale farmers in Uganda: A case of 117 tobacco growers. Final report presented at the AERC biannual workshop, Nairobi, Kenya Ogunniyi, L.T., & Ajao, A. O. (2011). Investigation of the factors influencing the technical efficiencies of swine farmers in Oyo State of Nigeria. Journal of Human Ecology, 35(3),203-208 Okunade, E. O. (2007). Effectiveness of extension teaching methods in acquiring knowledge, skill and attitude by women farmers in Osun State. Journal of Applied Sciences, 3(4), 282-286. Omofonmwan, S.I., & Kadiri, M. A. (2007). Problems and Prospects of Rice Production in Central District of Edo State, Nigeria. Journal of Human Ecology, 22(2), 123-128 Onumah, E. E., & Acquah, H. D. (2010). Frontier analysis of aquaculture farms in the southern sector of Ghana. World Applied Sciences Journal, 9(7), 826-835. Onyenweaku, C. E., & Nwaru, J. C. (2005). Application of a stochastic frontier production function to the measurement of technical efficiency in food crop in Imo State, Nigeria. Nigerian Agricultural Journal, 36, 1-12. Osei-Asare, Y. (2010). Mapping of poverty reduction strategies and policies related to rice development in Ghana. Coalition for AfricanRice Development (CARD): Nairobi, Kenya. Owusu, P.A., Nyantakyi, E.K.,& Borkloe, J.K. (2013). Assessing Performance of Irrigation of Rice in Ghana. ARPN Journal of Science and Technology,3(7), 305-308 Oyewo, I. O. (2011). Technicalefficiency of maize production in Oyo State. Journal of Economics and International Finance, 3(4), 211-216 118 Parikh, Ali, A. F., & Shah, M. K., (1995). Measurement of economic efficiency in Pakistani agriculture. American Journal of Agricultural Economics, 77, 675-85. Pitt. M. M.& Lee, L. F. (1981). Measurement and sources of technical inefficiency in the Indonesian weaving industry. Journal of Development Economics, 9, 43-64.l 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. GSSP Working Paper, IFPRI, Accra, Ghana. 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. A scientific Journal of Krishi Foundation, 10(2), 9-19 Reifschneider. D., & R. Stevenson (1991). Systematic departures from the frontier: A framework for the analysis of firm inefficiency. International Economic Review, 32, 715-23. Russell, N. P., & T. Young. (1983). Frontier production functions and the measurement of technical efficiency. Journal of Agricultural Economics, 34, 13950. Seyoum, E.T., E.G. Battese, & E.M. Fleming. (1998). Technical efficiency and productivity of maize producers in eastern Ethiopia: A study of farmers within and outside the Sasakawa- Global 2000 project. Agricultural Economics, 19, 341-48. Singh R., Priya, A., & Singh, P. (2011). Role of ICT in rural empowerment. Journal of Social Science and Management, 1(5), 52-65. 119 Singh, P.K. & Varshney, J.G. (2010). Adoption level and constraints in rice production technology. Indian Research Journal of Extension Education, 10 (1) Smith, M. (1996). A historical sketch of water resources development in Ghana. WRI/CSIR: Accra Son, T.V.H., Coelli, T., & Fleming, E. (1993). Analysis of technical efficiency of state rubber farm in Vietnam. Journal of Agricultural Economics, 9,183-201 Tanko, L. (2009). Technical efficiency in arable crop production in kebbi State, Nigeria. Food production in a developing economy, 1,194-200 Thahn, N.C., & Singh, B. (2006). Constraints faced by farmers in rice production and export. Omonrice, 1497-110. Tetteh, A. B., Adjetey, N. A., & Abiriwie, S. A. (2011). Consumer preferences for rice quality characteristics and the effects on price in the Tamale Metropolis, Northern Ghana. International Journal of Agricultural Sciences 1(2), 67-74. Tijani, A.A. (2006). Analysis of the technical efficiency of rice farms in Ijesha land of Osun State, Nigeria. Agrekon, 45(2) Timmer, C. P. (1971). Using a probabilistic frontier production function to measure technical efficiency. Journal of Political Economy, 79, 776- 794. Udoh, E.J., & J.O. Akintola (2001). Land management and resource use efficiency among farmers in South Eastern Nigeria. Award winning paper presented to the African Real Estate Society and the Rice foundation. 120 Varian, R.H. (1992). Microeconomic Analysis. (3rd Ed.). Norton: New York. Wang, J., Cramer, J. G., &Wailes, E. J. (1996). Productive efficiency of Chinese agriculture: Evidence from rural household survey data. Agricultural Economics, 15, 1728 Welch, F. (1970). Education in production. Journal of Political Economy, 78, 35-59. Winter-Nelson, A., & Aggrey-Fynn, E. (2008). Identifying opportunities in Ghana’s agriculture: Results from a policy analysis matrix. GSSP background paper No. 12. International Food Policy Research Institute: Accra, Ghana. Wiredu, A.N., Gyasi, K.O., Marfo, K.A., Asuming-Brempong, S., Haleegoah, J. Asuming-Boakye, A., &Nsiah, B.F. (2010). Impact of improved varieties on the yield of rice producing households in Ghana. Council for Scientific and Industrial Research – Savanna Agricultural Research Institute, Tamale. 121 APPENDIX A UNIVERSITY OF CAPE COAST SCHOOL OF AGRICULTURE DEPARTMENT OF AGRICULTURAL ECONOMICS AND EXTENSION INTERVIEW SCHEDULE FOR RICE FARMERS IN THE KETU NORTH DISTRICT TOPIC: TECHNICAL EFFICIENCY IN RICE PRODUCTION IN THE KETU NORTH DISTRICT OF THE VOLTA REGION, GHANA. INTRODUCTION: This interview schedule is designed with the sole aim of gathering information relating to technical efficiency in rice production and its determinants. All information will be used for academic purposes only. Information obtained from this interview shall be treated as strictly confidential. Thank you in advance for your cooperation. Name of community: ……………………………………………… Date of interview: …………………………………… SECTION A: FARMER- SPECIFIC CHARACTERISTICS 1. Age of farmer at last birthday: ………… years 2. Sex of farmer: (i) Male (ii) Female 3. Marital status of farmer: (i) Single (ii) Married (iii) Divorced (iv) Widowed 4. Educational background of farmer: (a) No formal education (b) Primary (c) Middle/JHS [ ] (d) O’level/SHS (e) Tertiary 5. Years of formal education …………………….. 122 6. Household size: ……………. (number) 7. Number of years engaged in rice farming: …………… years 8. Do you engage in other economic activity apart from rice farming? (i) Yes (ii) No 9. If yes, what other economic activity are you engaged in?............................................ SECTION B: PRODUCTION ACTIVITIES 10. Do you have access to extension services? (i) Yes (ii) No 11. If yes, how often? …………………………… 12. Do you have access to credit? (i) Yes (ii) No If yes from where? (I) bank (ii) credit union (iii) money lenders (iv) family(v) friends Others (specify)……………………………. 13. Do you belong to any farmer organisation? (i) Yes (ii) No 14. If yes which farmer organization do you belong to?....................................... 15. Methods of land preparation: (i) Slash and burn (ii) Use of weedicides followed by farm implements (iii) Use of weedicides only (iv) Use of farm implements only 16. Methods of weed control: (i) Use of weedicides (ii) Use of cutlass (iii) Use of hoe (iv) Hand picking of weeds (iv) Other (specify)…………… 17. Have you received any technical training in rice production over the past three years? (i)Yes (ii) No 123 18. If yes, who provided the training? (i) Fellow farmers (ii) AEAs (iii) Media (iv)others(specify) …………...................... 19. Do you cultivate rice in both major and minor seasons? (i) Yes (ii) No, If no, why? …………………………………………………………… SECTION C: INPUT AND OUTPUT DATA FIXED INPUTS 26. LAND Kind of land ownership: (i) Farmer’s own land (purchased or gift) (ii) Family land (iii) Rent (iv) Leased (v) Sharecropping (vi) Others (specify) ……………………. Item Size (acres) Cost per acre (GH₵) Total cost (GH₵) Land 20. EQUIPMENT Tools Number No. of months/years used Unit cost (GH₵) Total Cost(GH₵) a. Cutlass b. Hoe c. Sprayer d. Tractor/Tractor service 21. IRRIGATION COST Item Cost per month(GH₵) Total(GH₵) 124 VARIABLE INPUTS 22. PLANTING MATERIALS Item Quantity (kg)/acre Unit cost (GH₵) Total cost (GH₵) Seeds 23. Source of rice seeds for production: (i) Farmers’ own seeds (ii) Certified seed growers (iii) Friends/Family (iv) Government (v) Others (specify) …………. 24. LABOUR INPUTS IN RICE PRODUCTION Please indicate the number of persons, hours and days used in production activities. Type of labour Number of persons Hours used per person per day Days worked per crop season Wage per person per day (GH₵) a.Family Male Female child (under18) b.Hired 25. FERTILIZER INPUT Type of fertilizer Quantity (kg)/acre Unit cost (GH₵) Total Cost (GH₵) a.NPK b.Ammonia c.Urea d.Others (specify) 125 26. PESTICIDE USE Type of pesticide Quantity used (litres) per acre Unit cost (GH₵) Total cost (GH₵) a. Fungicide b.Insecticide c.Rodenticide 34. WEEDICIDE USE Weedicide Quantity Litres/acre Unit Cost GH₵ Total Cost GH₵ a)Pre-planting weedicide b)Post-planting weedicide OUTPUT OF RICE Season Harvested area(acres) Quantity harvested (kg) Quantity sold (kg) Price/kg (GH₵) Major Minor 126 SECTION D: CONSTRAINTS TO RICE PRODUCTION 27. INPUT, PRODUCTION AND MARKETING CONSTRAINTS Input constraint Response Production constraints Response Marketing constraints Response Cost of labour Land preparation Purchasing price of rice Land acquisition Pest and disease control Price fluctuations Acquisition of fertilizer Weed control Availability of storage facilities Pesticide acquisition Agro- Chemical application Government support Access to capital Rainfall Formation of marketing association Acquisition of seeds Planting of seeds Market for Rice Others (specify) Others (specify) Others (specify) Scale: 1 = very high constraint; 2 = high constraint; 3 = low constraint; 4 = very low constraint, 5=no constraint 127 APPENDIX B TECHNICAL EFFICIENCY SCORES OF INDIVIDUAL RICE FARMERS ON THE WETA IRRIGATION SCHEME FARMER NO. TE (%) FARMER No. TE (%) FARMER NO. TE (%) 1 89.08 46 54.95 91 63.23 2 73.59 47 71.81 92 38.57 3 44.87 48 70.31 93 57.97 4 90.27 49 49.62 94 83.15 5 91.80 50 78.91 95 95.11 6 89.35 51 74.53 96 67.23 7 91.71 52 78.93 97 77.54 8 71.75 53 88.02 98 66.94 9 90.91 54 81.23 99 61.82 10 83.16 55 32.54 100 52.85 11 66.83 56 75.55 101 80.99 12 90.22 57 51.41 102 88.28 13 86.31 58 69.11 103 85.90 14 58.23 59 72.64 104 86.79 15 73.84 60 49.07 105 74.34 16 86.42 61 87.28 106 94.90 17 92.88 62 69.81 107 29.59 18 47.11 63 54.58 108 67.30 19 68.10 64 46.65 109 93.77 20 56.30 65 56.69 110 61.56 21 86.96 66 55.64 111 49.93 22 58.75 67 69.62 112 93.50 23 91.62 68 75.88 113 87.11 24 58.24 69 55.24 114 51.13 25 64.57 70 60.82 115 84.66 26 61.68 71 69.93 116 72.14 27 89.29 72 64.49 117 64.19 28 69.75 73 64.03 118 60.08 29 80.69 74 78.24 119 76.98 30 75.83 75 81.98 120 91.95 31 53.35 76 61.40 121 92.56 32 61.90 77 50.35 122 56.98 33 50.21 78 80.66 123 86.74 34 73.17 79 89.69 124 49.64 35 89.00 80 71.39 125 89.99 36 51.23 81 72.02 126 90.45 37 74.53 82 79.24 127 82.91 38 37.88 83 72.81 128 51.76 39 44.73 84 60.59 129 88.85 40 91.79 85 92.15 130 68.06 41 79.53 86 69.51 131 82.09 42 64.64 87 76.35 132 62.57 43 74.18 88 83.75 133 87.86 44 63.65 89 60.20 134 34.67 45 54.10 90 67.23 135 58.27 128 FARMER NO. TE (%) FARMER NO. TE (%) FARMER NO. TE (%) 136 62.96 181 59.76 226 75.00 137 60.04 182 86.25 227 74.62 138 93.51 183 52.06 228 72.76 139 43.15 184 68.85 229 85.03 140 64.71 185 89.59 230 61.87 141 47.69 186 56.93 231 66.91 142 59.01 187 87.63 232 52.37 143 70.87 188 68.71 233 60.19 144 95.12 189 84.67 234 74.23 145 87.63 190 86.55 235 69.55 146 67.65 191 64.65 236 86.12 147 43.82 192 72.45 237 80.71 148 44.27 193 61.76 238 69.00 149 66.80 194 76.03 239 58.10 150 57.02 195 82.91 240 61.34 151 41.96 196 91.90 241 89.41 152 54.98 197 94.09 242 81.59 153 81.26 198 71.18 243 83.03 154 48.12 199 87.40 244 55.82 155 63.65 200 61.37 245 77.43 156 85.61 201 96.27 246 49.03 157 38.18 202 69.76 247 83.25 158 76.25 203 50.78 248 88.69 159 42.78 204 74.10 249 81.03 160 71.12 205 63.09 250 83.90 161 68.33 206 85.59 251 60.46 162 89.09 207 52.84 252 62.43 163 64.03 208 77.09 253 43.10 164 90.41 209 66.55 254 60.51 165 52.34 210 85.47 255 47.55 166 87.38 211 71.75 256 87.53 167 92.01 212 90.58 257 69.37 168 62.87 213 81.80 258 96.33 169 59.72 214 71.34 259 95.32 170 71.17 215 78.75 260 73.00 171 85.84 216 63.16 261 87.39 172 72.04 217 68.30 262 74.07 173 70.61 218 60.60 263 88.08 174 49.27 219 53.12 264 75.48 175 64.11 220 74.47 265 86.68 176 60.32 221 77.61 266 78.68 177 71.62 222 84.14 267 53.40 178 83.05 223 93.77 268 79.47 179 93.67 224 85.74 269 78.52 180 86.35 225 88.52 270 40.09 129 FARMER NO. TE (%) 271 49.11 272 54.52 273 55.85 274 73.82 275 54.54 276 38.06 277 63.47 278 87.68 279 85.50 280 55.32 281 52.83 282 57.95 283 69.24 284 92.06 285 51.31 130 APPENDIX C SAMPLE SIZE DETERMINATION TABLE POPULATION SAMPLE POPULATION SAMPLE POPULATION SAMPLE 10 10 220 140 1200 291 15 14 230 144 1300 297 20 19 240 148 1400 302 25 24 250 152 1500 306 30 28 260 155 1600 310 35 32 270 159 1700 313 40 36 280 162 1800 317 45 40 290 165 1900 320 50 44 300 169 2000 322 55 48 320 175 2200 327 60 52 340 181 2400 331 65 56 360 186 2600 335 70 59 380 191 2800 338 75 63 400 196 3000 341 80 66 420 201 3500 346 85 70 440 205 4000 351 90 73 460 210 4500 354 95 76 480 214 5000 357 100 80 500 217 6000 361 110 86 550 226 7000 364 120 92 600 234 8000 367 130 97 650 242 9000 368 140 103 700 248 10000 370 150 108 750 254 15000 375 160 113 800 260 20000 377 170 118 850 265 30000 379 180 123 900 269 40000 380 190 127 950 274 50000 381 200 132 1000 278 75000 282 210 136 1100 285 1000000 384 131 APPENDIX D DAM AT DEKPOR 132 APPENDIX E WETA RIC EFARMS 133