THE EFFECT OF FARMERS’ EDUCATION ON FARM PRODUCTIVITY AND INCOME IN GHANA: IMPLICATION FOR FOOD SECURITY BY EDWIN AFARI A THESIS SUBMITTED TO THE DEPARTMENT OF AGRICULTURAL ECONOMICS AND AGRIBUSINESS, UNIVERSITY OF GHANA, LEGON, IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE AWARD OF THE MASTER OF PHILOSOPHY DEGREE IN AGRICULTURAL ECONOMICS. JULY, 2001 University of Ghana http://ugspace.ug.edu.gh ( p _ 3 6 8 2 1 3 University of Ghana http://ugspace.ug.edu.gh DECLARATION I, EDWIN AFARI, author of this thesis, do hereby declare that the work presented in this thesis: “THE EFFECT OF FARM ER’S EDUCATION ON FARM PRODUCTIVITY AND INCOME IN GHANA: IMPLICATION FOR FOOD SECURITY ” , was done entirely by me in the Department o f Agricultural Economy and Farm Management, University o f Ghana, Legon, from September, 2000 to September, 2001. This work has never been presented either in whole or in part for any other degree o f this University or elsewhere. Edwin Afari This work has been submitted for examination with our approval as supervisors. Dr. D. BTSarpeng (Major supervisor) Mr. A. Mensah- Bonsu (Co-Supervisor) University of Ghana http://ugspace.ug.edu.gh To God Almighty, Nana Akua Afriyie , Erne and all Ghanaian farmers. DEDICATION University of Ghana http://ugspace.ug.edu.gh ACKNOWLEDGEMENT I would like to first and foremost THANK THE ALMIGHTY GOD for the Grace, Mercy and strength that He gave me during the course of study and in carrying out this work in particular. Secondly, I wish to express my deep, humble appreciation to my supervisors, Dr. D. B. Sarpong and Mr. A. Mensah-Bonsu for their patience, criticisms and corrections made to this study. God richly bless you in your teaching carrier. This study would not have been successfully carried out without assistance from various individuals and institutions. In this connection, I wish to express my gratitude to all the Lecturers of the Faculty of Agriculture and Senior Research Fellows at the Institute o f Statistical, Social and Economic Research (ISSER). Many thanks go to them for their suggestions, criticisms and corrections. I wish to also express my appreciation to the entire staff of the Reseau Ghaneen o f SADAOC FOUNDATION for giving me the opportunity to use their computers to carry out this study. To the entire Ghanaian cassava and maize farmers interviewed, I say thank you and “Ayeekoo”. With reference to individuals, special thanks go to all my colleagues in the Department of Agricultural Economics and Farm Management especially Victor, Mina, Joe, Frank and Paul for being there always. I wish to express my sincere thanks to Mr. Kumah, Aunts and Uncles for all the moral and financial support they gave me. Finally a debt o f gratitude goes to Jennifer Dela Kumordjie, and Celestina Opoku-Ampomah for painstakingly typing this study for me. Your work is gratefully acknowledged. University of Ghana http://ugspace.ug.edu.gh ABSTRACT Low agricultural productivity has been identified as the major cause of food insecurity and food self-insufficiency in Ghana. Human capital improvement through farmer education (both formal and informal) is essential for increasing agricultural productivity and farm income o f rural farmers. The study estimated and quantified the contribution of education and exposure to extension service contact to farm productivity and farm income. The findings show that the average schooling years of farmers sampled from the GLSS 4 survey data is just below the primary level. Weighted Least Squares estimates for cassava farm production function indicate that household heads’ education, as measured by years of formal schooling completed, had a positive but insignificant effect on farm productivity. We find similar results for maize farmers. Farm productivity increases ceteris paribus by 0.59 percent and 1.43 percent for cassava farmers with primary and secondary education respectively. Percentage increase in cassava output is as much as 3.7 for one (each) additional year of schooling above the mean educational level o f sampled farmers. For maize farmers, at least middle school education was found to be necessary for significant benefits of schooling and returns to schooling are highest for farmers with middle and post secondary education in the farm income estimation. Maize output is likely to increase by 3.1 percent for one extra year of schooling at the mean educational level o f the sampled farmers. Extension service contact was found not to significantly enhance productivity in any way for both cassava and maize farmers. The study recommends that higher levels o f investment in basic and secondary education should be a priority for the government, with special attention going to smallholder staple crop farmers in the area of informal education to enable them improve on their efficiency on the farm. Increase in farm productivity through improved education would ultimately contribute to the attainment of food security in the country. University of Ghana http://ugspace.ug.edu.gh TABLE OF CONTENTS Page Declaration................................................ 1 Dedication................................................ 11 Acknowledgement................................... 111 Abstract................................................... >v Table of Contents...................................... vi List of Tab les............................................. ix Abbreviations.......................................... xi CHAPTER ONE: INTRODUCTION 1.1 Background Statement.......................... 1 1.1.1 The Agricultural Sector and the Ghanaian Economy 1 1.1.2 Constraints and Problems Facing the Agricultural Sector 2 1.1.3 Farmer Education and Agricultural Productivity 4 1.2 Problem Statement................................. 5 1.3 Objectives.............................................. 7 1.4 Justification of the study ................... 8 1.5 Organisation of the study .................. 9 1.6 Scope and Limitations....................... 9 v University of Ghana http://ugspace.ug.edu.gh CHAPTER TWO: LITERATURE REVIEW 2.1 Introduction........................................... II 2.2 Education and Development 11 2.3 Determinants of Farm Productivity and Income 18 2.4 Effects o f Education on Farm Productivity: Empirical Evidence.... 19 2.4.1 Evidence from Developing Countries......................... 20 2.4.2 Evidence from G hana................................................... 25 2.5 Educational Effects: A Review of Competing Methodologies 27 2.5.1 Technical and Allocative Efficiency............................... 27 2.5.2 Estimation of the effect of Education on Agricultural Production: Production Function Approach................... 28 2.5.2.1 Non-Frontier (Direct) Production Function Approach... 29 2.5.2.2 Frontier Production Function Approach 31 2.5.3 Summary 33 CHAPTER THREE: METHODOLOGY 3.1 Introduction.............................................. 35 3.2 Theoretical and Analytical Framework 36 3.2.1 Production Function............................. 36 3.2.2 The Cobb Douglas Production Function.... 38 3.2.3 Profit Maximisation.................... 39 3.2.3.1 Short-Run Profit Maximisation....................... 40 vi University of Ghana http://ugspace.ug.edu.gh 3.3 Empirical Models.......... 3.3.1 Introduction 42 3.3.2 Farm Technology............... 42 3.3.3 Education as a factor of Production............... 43 3.3.3.1 Introduction......................................... 43 3.3.3.2 Non-Frontier Empirical Model: Worker Effect o f Schooling ................ 43 3.3.3.3 Hypothesis to be te s ted .......................... 44 3.3.4 Percentage gain per year of education..................... 45 3.3.5 Farm Profit (Income) Function.................................. 46 3.3.5.1 Hypothesis to be tested 47 3.4 The D a ta .................................................................................... 47 CHAPTER FOUR: RESULTS AND DISCUSSION 4.1 Introduction.............................................................. 49 4.2 Sample Selection and Variable Definition..................... 50 4.3 Descriptive Statistics and Demographic Profile of Farmers 53 4.3.1 Farmers Locality and Ecological Zone............................... 54 4.3.2 Age-Sex Structure.................................................................. 55 4.3.3 Formal School Attendance................................................... 56 4.3.4 Educational Attainment....................................................... 57 4.3.5 Extension Service Contact................................................... 59 vii University of Ghana http://ugspace.ug.edu.gh 4.4 Farm Productivity Regression Functions....................................... 59 4.4.1 Some Econometric Issues ................................................... 59 4.4.2 Basic Production Function Results: Cassava.................. 61 4.4.3 Basic Production Function Results: Maize...................... 63 4.4.4 Effect of Formal Education Contact on Farm Productivity: Cassava............................. 66 4.4.5 Effect of Formal Education Contact on Farm Productivity: M aize ................................ 70 4.5 Full Estimation Results for Farm Net Income.................. 76 4.6 Education, Productivity and Food Security Issues 78 CHAPTER FIVE: SUMMARY, CONCLUSION AND RECOMMENDATION 5.1 Summary and Conclusion 82 5.2 Policy Recommendation 84 REFERENCES 86 APPENDICES 98 viii University of Ghana http://ugspace.ug.edu.gh LIST OF TABLES Page 1.1 Average Yield per Hectare of Major Crops, 1987-1996 (Mt).... 1.2 Educational Levels o f the Active Population (both sexes) by Main Industry (Percent)......................................................................... 4.1 Variable Definitions and Mean Values....................................... 4.2 Classification of agricultural household heads into selected Cassava and Maize farmers by ecological zone and locality..................... 4.3 Age-Sex Distribution of Farmers by Farm Activity.................. 4.4 Frequency Distribution of Farmers attending Formal Schooling by Farm Activity..................................................... 4.5 Level o f Education completed by Sex and Farm Enterprise... 4.6 Extension Service Contact Response..................................... 4.7 OLS and WLS estimate of the C-D Production Function: Cassava (without education)............................................................. 4.8 OLS and WLS estimate of the C-D Production Function: Maize (without education)............................................................ 4.9 WLS estimates of the C-D Production Function: Cassava (with Education) 4.10 WLS estimates of the C-D Production Function: Maize (with Education)... 4.11 Results o f Farm Income Function Estimation: Maize................. A2a Correlation Coefficients between Variables: Cassava................. A2b Correlation Coefficients between Variables: Maize................. A3a OLS regression results: Cassava................................................. A3b OLS regression results: Maize.................................................... A4a WLS estimation results showing interaction effects for Extension Service And Years of Schooling completed: Cassava...................................... A4b WLS estimation results showing interaction effects for Extension Service ix J> 6 51 54 55 56 58 59 61 64 67 71 76 99 100 102 103 104 University of Ghana http://ugspace.ug.edu.gh And Years of Schooling completed: Maize....................................... Percentage Increase in Output associated with an extra year o f Schooling O f head fa rm er........................................................................................... University of Ghana http://ugspace.ug.edu.gh ABBREVIATIONS COCOBOD Cocoa Board of Ghana EAP Economically Active Population FAO Food and Agriculture Organisation GLSS Ghana Living Standards Survey ISSER Institute o f Statistical, Social and Economic Research MoFA Ministry of Food and Agriculture PPMED Policy Planning, Monitoring and Evaluation Department NGO Non-Governmental Organisation SADAOC Securite Alimentaire Durable Africain Occidental Centrale (Sustainable Food Security in Central West Africa) WLS Weighted Least Squares C-D Cobb - Douglas SOW-VU Stichting Onderzoek Wereldvoedselvoorziening van de Vrije Universiteit (Centre for World Food Studies at the Free University) MSLC Middle School Leaving Certificate BECE Basic Education Certificate Examination CES Constant Elasticity of Substitution xi University of Ghana http://ugspace.ug.edu.gh CHAPTER ONE INTRODUCTION 1.1 Background Statement 1.1.1 The Agricultural Sector and the Ghanaian Economy The Ghanaian economy is broadly divided into three main sectors, agriculture, services and industry. The agricultural sector is the dominant sector in the Ghanaian economy in terms o f its share o f Gross Domestic Product (GDP), employment and foreign exchange earning. In 1997, the sector employed about 70% o f the labour force, contributed about 47% to GDP and accounted for over 57% o f foreign exchange earnings (PPMED of MoFA , 1998). The broad agricultural sector, which includes fishing and forestry, contributes on the average over 40% o f the total GDP (Ghana-Vision 2020, 1995). In the 1980s, its contribution to GDP averaged about 52 percent (Nyanteng and Dapaah, 1997). A high proportion o f the economically active population (EAP) depends on farming as their main source of income. It offers job avenues to many people as farmers, farm labourers and workers, among others. The proportion o f EAP in agriculture was about 52% in 1987, 50% in 1990 and declined further to 48% in 1994 (Nyanteng and Dapaah, 1997), though the absolute number has been increasing over the years. According to FAO (1998), the EAP in agriculture in absolute numbers increased from 4.2 million in 1990 to 4.8 million in 1995 and to 5.2 million in 1998. At a population growth rate of about 3% and an expected increase in population from 18.5 million to about 36 million by the year 2020, the agricultural labour force is expected to increase substantially. The percentage o f the EAP in agriculture in Ghana is far higher than that recorded for Canada, USA, Brazil, Nigeria and some other African countries 1 University of Ghana http://ugspace.ug.edu.gh (FAO, 1998) but agricultural growth in Ghana has been low and disappointing. During the period 1991-1994, agricultural growth averaged 1.9% only, recording a negative value (-0.6%) in 1992 (ISSER, 1998). Between 1995 and 1998 however, agriculture GDP increased at an average o f 4.2% per year (MoFA, 1998). The performance o f the agricultural sector has serious implications not only for most o f the rural inhabitants, since fanning in Ghana is largely a rural phenomenon, but also for the agro-industrial sector growth. The role o f the agricultural sector is crucial for the attainment o f the goals set in Ghana’s Vision 2020. The sector is expected to ensure food security and adequate nutrition for all Ghanaians, to supply raw materials and other inputs to other sectors o f the economy as well as to provide producers with farm income comparable to earnings outside agriculture. It has been argued that if the agricultural sector does not grow at a rate higher than 4%, lower income in agriculture would be translated into lower income for unskilled labour throughout the economy (PPMED, 1998; Ghana-Vision 2020, 1995). 1.1.2 Constraints and problems facing the agricultural sector Though the agricultural sector has been the backbone of the Ghanaian economy, its contribution to GDP growth has fallen behind the contribution made by the services sector. According to Ghana Statistical Service national accounts, the agricultural sector recorded a real growth rate o f 2.3% in 1993 as against 5.4% by the industrial sector and 7.2% by the services sector. In 1995, the growth rate was 4.2% as against the 3.3% by the industrial sector and 4.9% by the services sector. However in 1998, the food and agriculture sector grew at a rate of 5.3% (ISSER, 1998) below the services sector (6%) and above the industrial sector (2.5%). In 2000, the agricultural sector recorded a less than expected growth rate of 2.1 per cent as against the 3.9 per cent achieved in 1999 (National Budget, 2001). The performance of the agricultural sector has not been adequate to ensure food 2 University of Ghana http://ugspace.ug.edu.gh nutrition and food security and to satisfactorily increase real farm income (Ghana-Vision 2020, 1995). A number o f factors and problems have contributed to the poor performance o f the sector. They include low level o f technology used by most agricultural producers, the heavy dependence on climatic conditions especially rainfall and the inadequacy o f farm credit. Nyanteng (1994) has observed that the productivity o f land and labour in small-scale agriculture is generally very low, and for most o f the crops, the average yield for the period 1987-1991 was only about 25-50% o f the yields achieved in isolated cases and elsewhere in Africa and the World (see Table 1). The same is true for average carcass weight of livestock (kg/head). He attributed the above phenomenon to the extensive use of traditional technology and methods o f cultivation. In addition there are problems associated with the type, quantity and quality o f inputs used, farmer education (human capital) improvement, inadequate moisture for plant growth, land tenure and several other factors (Nyanteng and Dapaah, 1997). Table 1.1: Average Yield per Hectare of Major Crops 1987-1996(Mt) Crops Ghana Actual Achievable Africa World Percentage Achievable for Ghana (%) Roots/Tuber Cassava 7.8 28.0 7.7 9.9 28 Yam 6.1 10.0 9.1 9.0 60 Cocoyam 5.6 8.0 4.3 5.6 70 Plantain 7.1 10.0 19.0 25.0 70 Cereals Maize 1.2 5.0 1.6 3.7 24 Sorghum 0.7 2.5 0.8 1.3 28 Millet 0.7 2.0 0.7 0.8 35 Rice 1.0 3.0 2.0 3.5 n o JJ Pulses/Nuts Groundnut 1.4 2.0 na na 70 Cowpea 0.9 2.0 na na 45 Sources: Ministry o f Agriculture (1991); Ministry o f Food and Agriculture (Files) Ghana’s agriculture is characterised by low-skill and low-pay work and most o f the labour force are engaged in less productive smallholder agriculture. The nation can achieve food self­ 3 University of Ghana http://ugspace.ug.edu.gh sufficiency, be secure in food and enjoy a successful agricultural intensification, if it has a healthy and knowledgeable agricultural workforce in crop and livestock production. The individuals in the sector, especially rural farmers, need to have farming skills improved through enhanced educational programmes to adopt modern and sustainable technologies and methods. 1.1.3 Farmer Education and Agricultural Productivity Lele (1990) has observed that an improvement in farm human capital through farmer education is essential for increasing agricultural productivity. Farmers who have acquired high education are likely to adopt new technologies earlier than others with low educational levels, and use inputs that make them more productive. Most Ghanaian crop farmers as shown in the various GLSS reports have had no formal education (never been to school) and farm with relatively unproductive local farming implements like cutlass, hoe, and other traditional farm implements. The average farm size is very small and the farming system that is practised goes to increase land and soil degradation through the depletion o f soil nutrients. The “slash and burn” is still the predominant land preparation method and modern agronomic practices are not common. Also, low earnings from farms are partly the result o f their relatively lower human capital endowment and partly o f labour market discrimination (World Bank, 1995). The illiteracy o f most farmers in Ghana creates communication problems and constrains proper understanding, adoption and also application o f modem and improved farm technologies (ISSER/SOW-VU, 1999; Nyanteng and Dapaah, 1997). Education is critical for economic growth and poverty reduction (World Bank, 1995) and if the nation is to achieve high productivity and incomes from small and medium scale farms and ultimately alleviate poverty, then the relationship between human capital development and 4 University of Ghana http://ugspace.ug.edu.gh productivity in agriculture should be explored fully. The nation will enjoy a successful agricultural intensification envisaged under the First Medium-Term Development Plan (1997-2000), if farmers have all the required skills and knowledge to adopt recommendations on sustainable methods in farming to reduce the adverse impact on the environment that modern farming methods brings. Resulting from the above, a policy framework aimed at generating employment for poverty reduction, needs to focus on increasing substantially productive employment opportunities in agriculture, especially those targeted at the poorer segments o f the rural work force. 1.2 Problem Statement Food security, as an important indicator of human welfare, tells us a good deal about changes taking place in human life (Asenso-Okyere et al, 1997). However, in the last decade, food insecurity has remained a problem in many households in Ghana and other countries in the West African sub-region. This problem has persisted in both dimensions o f food insecurity: availability and accessibility to food (Asenso-Okyere at al, 1997). Not only are many households unable to produce sufficient food for themselves, they also lack the income from cash crops and non-farm activities to purchase food. The whole situation o f food insecurity in the country may better be understood by considering its interconnections with human resource development, since human resources can be seen to have a direct impact on farm productivity (ISSER/SOW-VU, 1999). Agricultural production in Ghana is characterised by low yield due to several of the factors enumerated above. Returns to farm labour and land are low and most o f Ghanaian rural farmers have low incomes from their farm because of several factors among which are the small size of farm-holdings and low educational level and training. Investing in human capital is essential for raising farm productivity, which is a key to the enhancement o f living standards. This is underlined by the fact that such investments constitute effective ways o f increasing the poor farmer’s access to productive resources and to the enhancement of farm level efficiency and increased incomes. The 5 University of Ghana http://ugspace.ug.edu.gh various interrelationships between education (schooling), extension training and farm productivity in Ghana is not well known because of few empirical studies that have been conducted in this direction. This trend can be attributed to the qualitative nature o f most o f the human capital variables that affect agricultural farm productivity. Apart from Jollife (1998) and a few other related human resource-productivity studies, the empirical evidence on the effects o f education on agricultural productivity in Ghana is very scanty. Available evidence shows a low educational level o f farmers in Ghana (GLSS-4 Report, 2000). When both sexes are considered (see Table 2), 42.9% o f the active population who had never been Table 1.2: Educational Levels o f the Active Population (both sexes) by Main Industry (Percent) Education Attainment Main Industry Never been to school Less than MLSC/BECE MLSC/ BECE Secondary or Higher Total Sample Size Agriculture 42.9 24.8 28.3 4.0 100.0 4644 Manufacturing 29.8 24.3 36.3 9.6 100.0 987 Construction 10.3 20.5 44.4 24.8 100.0 117 Financial Services 1.4 4.3 34.8 59.4 100.0 69 T ransportation/Admi nistration 4.9 15.9 55.5 23.6 100.0 182 Source: Ghana Living Standards Survey (GLSS 4) Report, 2000. to school were in the agricultural sector, 24.8% had attained less than MSLC/BECE, 28.3% had attained MSLC/BECE and only 4.0% had secondary education or higher. It must be stressed that among the main industries, agriculture recorded the lowest for secondary school attainment (GLSS 4 Report, 2000). It is evidenced from Table 2 that a significant number of agricultural workers have never been to school formally. This should be a cause for concern for agricultural policy­ 6 University of Ghana http://ugspace.ug.edu.gh makers. There is the need to quantify the effects that education, and exposure to extension training and service have on farm productivity at the household level in Ghana. In Ghana, extension education has largely been limited to agricultural services. The Ministry of Food and Agriculture (MOFA) and COCOBOD (Asenso-Okyere et al, 2000) manage extension education programmes in the country. The ratio o f farmers’ to extension officers in the country is very large and has been increasing, resulting in farmers training needs not being met (Asenso-Okyere et al, 2000). The study therefore seeks to address the following questions; 1. What are the major factors that impact on farm productivity in Ghana? 2. What has been the effect o f exposure to extension contact and educational status o f the farmer on farm productivity in Ghana? 3. To what extent has educational status o f farmers and their exposure to extension contact affected farm income earnings in Ghana? 4. What relevant recommendations can be made to inform policy on the effect o f education on farm productivity and farm income in Ghana and its implication on food security? 1.3 Objectives The main objective o f this study is to estimate and quantify the contributions o f education and exposure to extension service contact on farm productivity and income in Ghana. The specific objectives are the following: 1. To identify factors affecting farm productivity (output per acre) in Ghana. 2. To estimate the impact o f education (schooling) and extension service on farm productivity in Ghana. 3. To estimate the impact o f education (schooling) and extension service on farm household income earnings in Ghana. 4. Finally to provide recommendations that will inform policy on education and farm productivity relationships and their implication for food security. 7 University of Ghana http://ugspace.ug.edu.gh 1.4 Justification of the Study Low agricultural growth recorded in Ghana are mostly attributed to the vagaries o f the weather, costly and low use o f physical inputs (Nyanteng, 1994) but scarcely to the low productivity o f the labour force in the agricultural sector. Ghana, as a developing nation, is aiming at becoming a middle-income country by the year 2020 and therefore the development o f our human resource should be paramount. There is the need for research into the effect that human resource o f the nation has on the productivity o f the people engaged in important sectors o f the economy. The key to food security is sustainability and this calls for increased productivity, which can be achieved, largely through the acquisition o f skills and knowledge (Asenso-Okyere et al, 1997). The identification and analysis o f the relevant factors that determine the productivity level will help decision-makers to formulate appropriate policies that could correct the gross disparities that exist between the rural and urban farming communities in terms of education, health care and nutrition. The nature o f farm productivity-education linkages o f workers on family farms have been largely unexplored, despite the overwhelming importance o f the family farms in developing countries and for that matter Ghana. Many closely related empirical studies have used data from several countries in Asia, South America and other African countries to estimate the effect o f education on agricultural productivity (see Weir, 1999; Lockheed et al, 1980; Phillips, 1994; Moock, 1981). The only notable exception is the work done by Jollife (1998) using data from the GLSS 3. He estimated the effect o f cognitive skills on total household income, farm income and non-farm income. The other studies using data on Ghana mostly centred on the situation o f education in Ghana (Abban, 1986; Asenso-Okyere et al, 2000), impact o f health and nutrition status on productivity (Nube et al, 1999; Croppenstedt and Muller, 1998) and returns to education (Psacharopoulos, 1985,1993; Glewwe and Twum-Baah, 1990). University of Ghana http://ugspace.ug.edu.gh This present study seeks to bridge the gap and contribute to existing knowledge on the relationship between education and farm productivity and incomes in Ghana. In particular, the study establishes the link and the possible effects and interrelationship between education (schooling) farm productivity and farm income in Ghana. Policy recommendations based on the findings will also help policy and decision-makers in the education and agriculture ministries to target specific areas to improve agriculture development and growth in Ghana. The study establishes a baseline data (subset from survey data), method(s) o f analysis and research information on the links between human resource development, farm productivity and farm incomes in Ghana for further research. 1.5 Organisation of the Study The study is organised into six main chapters. The second chapter reviews the relevant literature on the effect of education on farm productivity and farm incomes and the methodologies that have been employed in estimating such effects. The chapter also focuses on issues pertaining to education, economic growth and development relationships. Chapter three provides a detail overview o f the methodology to be used to achieve the study's specific objectives and comprises the theoretical and analytical framework o f the study and empirical models. The results o f the study and the discussion o f these results are embodied in the fourth chapter. The summary, conclusion and policy recommendations from the study form the last chapter. 1.6 Scope and Limitations The present study covered only respondent farmers who are involved mainly in maize and cassava production as sole crops. These crops were chosen because, a high percentage o f respondents whose main occupation was farming, cultivated these crops and also because o f their importance as major staples in the country. The study utilised data from only the fourth round o f the Ghana Living standards Survey (GLSS 4) though it is believed a more detailed work could be done if all the four 9 University of Ghana http://ugspace.ug.edu.gh rounds were used to show policy effects from the late 80’s. Also, due to insufficient information oil cassava input price data, farm income regression analysis was only possible for maize. 10 University of Ghana http://ugspace.ug.edu.gh CHAPTER TWO LITERATURE REVIEW 2.1 Introduction The literature on the effect o f education on productivity and income in specific sectors and enterprises focuses predominantly on the estimation o f social and private rates o f returns to education and the impact human capital has on the productivity of wage earners. Psacharopoulos (1985; 1993) summarised the results from more than 55 such wage studies from Africa, Asia and Latin America. In this study, the available literature comes from studies that use data from all over the world including two1 from Ghana. The literature on the definition o f education and its importance to economic growth and development is outlined. The main issues discussed include education’s contributions to economic growth and poverty reduction, and education as a source o f technical change, especially in agricultural production through increased productivity. Studies on the determinants o f farm productivity and income are also reviewed. The last part o f the chapter concerns itself with some of the competing methodologies used in the estimation o f education effects on productivity especially at the farm household level. 2.2 Education and Development Education is broadly defined as the act o f acquiring knowledge, skills, values, attitudes and best practices (Asenso-Okyere et al, 2000). Education can be divided into two broad categories namely formal and informal education. Formal education has been recognised as the most effective way to 1 see Jo llife (1 9 9 8 ) and V ijv e rb e rg (1 995 ) 11 University of Ghana http://ugspace.ug.edu.gh develop the human potential. Formal education represents all forms o f education that requires people to acquire education or skills through a structured system or institution recognised by the Ministry o f Education (Asenso-Okyere et al, 2000). People who pass through these centres of education are presented with some certificate o f award in recognition o f the successful completion o f the programme. Formal education in Ghana is divided into three divisions and these are: i. Basic Education (Primary and Junior Secondary)2 ii. Secondary Education iii. Tertiary Education Informal education largely deals with the education o f adults or people not through the means that are rigid and do not follow formal classroom education culminating in the award o f certificates or degrees (Asenso-Okyere et al, 2000). To a large extent, informal means o f education are aimed at making people either functionally literate or enable them acquire some skill or vocation. There are three basic types o f informal education and these are: i Adult Education (Non-formal Education) ii Artisanal Training (Apprenticeship) iii Extension Education The functional literacy programme is action-oriented and is designed to address the educational needs of adults, especially the rural poor with emphasis on women and girls. Extension Education programmes are managed by the Ministry o f Food and Agriculture (MOFA) and COCOBOD and have been largely limited to agricultural services. Formal extension education are concerned with the training o f personnel to recognise farmers’ problems, identify farmers’ training needs, design training programmes and, above all, plan intelligent and appreciable extension programmes for the rural people. The various forms o f informal extension education given to farmers include technical 2 T h e d iv is io n cam e abou t a f te r 1987, w hen th e E duca tiona l R efo rm P rog ram m e w as in itia ted . P rev io u s ly w e had B asic E duca tio n (P rim a ry S choo l o f 6 y ears and M idd le schoo l o f 4 years). 12 University of Ghana http://ugspace.ug.edu.gh advice on production, multiplication and supply o f improved planting materials, advice to women farmers on home management and nutrition, among others (Asenso-Okyere el al, 2000). Education, as one of the components of human capital has long been recognised as one o f the major cutting edges o f development and growth o f the economies o f developing countries. Education is a major instrument for economic and social development. According to World Bank (1995), education is central to its strategy for helping countries to reduce poverty and improve their living standards through sustainable growth and investment in people. The two-fold strategy, according to the World Bank (1990) calls for promoting the productive use of labour - the principal asset o f the poor and providing basic social services to the poor. Education is critical for and contributes to economic growth and poverty reduction (Becker, 1964 and 1993; Singh et al, 1986; World Bank, 1986 and 1995; Peaslee, 1965 and 1969), but by itself it will not generate growth (World Bank, 1995). The strongest growth comes about when investment in both human and physical capital takes place in economies with competitive markets (World Bank, 1995). It contributes to growth both through increased individual productivity brought about by the acquisition o f skills and attitudes and through the accumulation o f knowledge. Indeed, education produces the knowledge, skills, values and attitudes that are used in the productive sectors o f the economy o f most countries to earn income and generally for development. Improved attitudes, beliefs and habits may lead to greater willingness to accept risk, adopt innovations, save for investment and generally to embrace productive practices (Appleton and Balihuta, 1996; Cotlear, 1990). A large part o f economic growth stems from improvement in the quality o f the labour force, including increased education and better health, together with technological progress and 13 University of Ghana http://ugspace.ug.edu.gh economies o f scale (Becker, 1993; Schultz, 1964; Denison, 1967; World Bank, 1991). A report by the World Bank (1995) has observed that investment in education leads to the accumulation of human capital, which is key to increased incomes and sustained economic growth. The accumulation o f human capital-knowledge as noted by Romer (1986) facilitates the development of new technologies and is a source o f self-sustaining growth (Lucas, 1988; Barro, 1991). The role o f human capital in the development process has attracted a lot of attention (Schultz, 1964 and 1972; Becker, 1964 and 1993; Welch, 1970). Growth theorists such as Romer (1986,1990), Lucas (1988,1993) and Azariadis and Drazen (1990) have shown that the accumulation of human capital can sustain long-term growth. Economies develop and grow as a result o f increased productivity in all sectors and this can only be achieved if the labour force in these important sectors are well educated. The labour force should as a matter of fact possess all the requisite skills, ideas and attitude that will aid them to increase productivity and achieve high rates o f return to investment (Psacharopoulos, 1984; Becker, 1993; Weale, 1993) by embarking on value for money projects and programmes. The outstanding economic records of Japan, Taiwan, and other Asian economies in recent decades dramatically illustrate the importance o f human capital to growth (Becker, 1993). According to Becker, knowing their limitation and lack o f physical resources, the so-called Asian tigers grew rapidly by relying on a well-trained, educated, hard-working and conscientious labour force. Primary education is the largest single contributor to the economic growth rates o f the high-performing Asian economies (World Bank, 1993). Education is also a source of technical change and source of improvement on technical knowledge and, hence o f labour efficiency (Diwan, 1971). Economic benefits of schooling include the potential to obtain paid employment or to generate income through self-employment, using skills learned in school (Weir, 1999). Colclough (1980) reviewed the relationship between education and 14 University of Ghana http://ugspace.ug.edu.gh economic growth, and concluded that primary schooling increased productivity and thus economic growth. Also education contributes by reducing fertility (Population control) and improving health (Croppenstedt and Muller, 1998; Colclough, 1980) and by equipping people to participate fully in the economy and in society. Also, education is an important determinant o f nutritional status (Thomas, 1989; Strauss, 1990; Behrman and Deolalikar, 1988). Education, especially basic (primary and lower secondary) education contributes to poverty reduction by increasing the productivity o f the poor’s labour (Lockheed et al, 1980; Mooclc, 1994; Philips, 1994; Villaume (1977); World Bank, 1995). As stated by Singh, Squire and Strauss (1986), education should result in an economy’s growing demand for adaptable workers who can readily acquire new skills to support the continued expansion o f knowledge in the nation. Also higher education contributes to self-sustaining growth through the impact o f graduates on the spread o f knowledge (Becker, 1964). In fact more educated workers can and are known to deal more effectively with a rapidly changing environment (Schultz, 1971,1975; Mincer, 1989; World Bank, 1991). Also, schooling may speed the adjustment in use of new technologies (Huffman, 1974). Weir (1999) pointed out that formal schooling is not only useful after new technologies have been adopted but education may also help to determine whether a farmer decides to be an early adopter of innovations and the extent to which new innovation will be used. There are at least three reasons for this phenomenon (Weir, 1999). First, those with schooling tend to be more affluent and are less in danger of starvation if a prospective innovation is unsuccessful. Second, educated farmers may be more likely to be contacted by agricultural extension workers looking for model farmers to test innovations. Lastly, literate farmers are better able to acquire information about potential innovations and to make rational evaluations of the risks involved in trying new inputs, crops or methods. 15 University of Ghana http://ugspace.ug.edu.gh Studies in the early 1990s, notably Azariadis and Drazen, (1990); Lau, Jamison and Louat, (1991) have confirmed the overwhelming importance o f education, especially primary education for growth. Cross-country studies have shown that there exists a threshold level o f human capital accumulation beyond which a country’s growth may accelerate. Most empirical studies (Lockheed, Jamison and Lau, 1980; Moock, 1981; Philips, 1994; Weir, 1999) on the impact o f farmer education on farm productivity have established that primary education is very important in increasing productivity. Agriculture provides a compelling evidence of the link between human capital-education and technology (Becker, 1993). However, he pointed out that education is of little use in traditional agriculture because farming methods and knowledge are readily passed on from parents to children. This is supported by the low social rates of return to education recorded in most studies using data from Africa (Lockheed, Jamison and Lau, 1980; Moock, 1981; Philips, 1994; Weir, 1999). This assertion also receives a lot o f support from African studies reviewed by Appleton and Balihuta (1996). They found that the effect o f schooling on agricultural output is usually not significant, though in some cases it can be large. They suggested several possible reasons for the low and sometimes insignificant effect o f education on farm output, including small farm and sample sizes, errors in measurement o f farm production and wide variation in the actual effects o f education on agricultural output in different areas and under different farming systems. Arrow (1973) postulated that individual farmers with higher ability receive more education and more income and not education itself generating higher incomes. Farmers in developing countries are among the least educated members of the labour force (GLSS 4 report, 2000). By contrast, farmers in developed countries deal with hybrids, breeding methods, fertilisers and other chemicals, complicated equipment, and intricate futures markets for commodities (Becker, 1993) and are therefore well- educated and exposed to more extension contacts to increase productivity on the farm. 16 University of Ghana http://ugspace.ug.edu.gh According to Welch (1970), education is of great value since it helps farmers adapt more quickly to new hybrids and new technologies, and it is the most important to farm production in a rapidly changing technological or economic environment (Schultz, 1964; 1975). Agricultural households are the main forms o f economic organisation in developing countries (Singh, Squire and Strauss, 1986). The agricultural sector plays a major role in economic development in Ghana and since about 60-70% o f the labour force is involved in it, the educational status o f this labour force should be of concern to policy makers. Education may enhance farm productivity by improving the quality o f labour, by increasing the ability o f the farmer to adjust to disequilibria and through its effect upon the propensity to successfully adopt innovation. The importance o f formal schooling to farm production becomes more apparent as technological innovations spread more widely within a country (Weir, 1999). He emphasised on the efficiency advantage for farmers who are better prepared to anticipate and cope with disequilibria. Thus, even in the absence o f innovation, farm productivity may be enhanced by investments in education. Education may have both cognitive and non-cognitive effects upon labour productivity. Cognitive outputs o f schooling include the transmission of specific information as well as the formation of general skills and proficiencies. Education also produces non-cognitive changes in attitudes, beliefs and habits. Increasing literacy and numeracy may help farmers to acquire and understand information and to calculate appropriate input quantities in a modernising or rapidly changing environment (Weir, 1999). In fact schooling enables farmers to learn on the job more efficiently (Rosenzweig, 1995). 17 University of Ghana http://ugspace.ug.edu.gh Becker (1964) established that basic literacy and numeracy (lower occupational levels) and greater logical reasoning capacity, heightened self-expression and sound technical knowledge (higher occupational levels) had a significant impact upon worker effectiveness. Education may either increase prior access to external sources o f information or enhance their ability to acquire information through experience with new technology. That is, it may be a substitute for or a complement to farm experience in agricultural production (Weir, 1999). Education may indirectly increase output through its interaction with other institutional variables. For example, schooling may substitute for access to credit by providing the skills necessary to obtain waged employment, thereby generating cash to finance agricultural investments (Appleton and Balihuta, 1996). 2.3 Determinants of Farm Productivity and Income Reardon et al (1996) uncovered differences in patterns and determinants o f farm productivity over agroclimatic zones, types of technology, degrees o f environmental degradation, and levels of improved inputs. They reported on studies that employed data from Burkina Faso, Rwanda, Senegal and Zimbabwe. Generally the main findings indicated that, rates o f growth in yields and returns per labour-day were low in the four study countries. They also identified productivity determinants in the four study countries to include fertiliser, improved seeds, animal traction, organic inputs and conservation investments. Other determinants were farm size and land tenure, non-cropping income and well-functioning input and output markets. They recommended that, to improve long-term food security in Africa, farmers must be able to pursue sustainable intensification o f farm production by use of improved inputs. They concluded that the use o f fertiliser, organic inputs, animal traction and conservation investments need to rise dramatically. They however failed to examine the role-played by the farmer viz. a viz. his or her educational level and ability to use these inputs effectively. A vast number of available empirical studies include 18 University of Ghana http://ugspace.ug.edu.gh environmental and other human capital variables like health and nutritional status in addition to agricultural inputs as explanatory variables when estimating productivity changes. Strauss (1986) used household level data from Sierra Leone to test whether higher calorie intake enhanced family farm labour productivity. He estimated a farm production function using the Non- Linear Two Stage Least Squares (NL2SLS) developed by Amemiya (1983), accounting for the simultaneity in input and calorie choice. He found a highly significant effect o f calorie intake on labour productivity at the farm level. In all three cases o f estimation procedures,3 both the calorie and calorie squared coefficients were significant at more than the 0.01 level, with calorie consumption contributing positively to output. Nutrition (measured in calorie intake) is found to be a major determinant o f farm productivity. Because nutrition and health variables are known to be collinear with education (Strauss and Thomas, 1995), they were excluded from this study. 2.4 Effects of Education on Farm Productivity: Empirical Evidence The development o f the human capital concept has provided economists with a useful framework for investigating the relationship between education and productivity in the agricultural sector and income (Hansen 1970; Knight, 1987). Human capital analysis assumes that schooling raises earnings and productivity in the agricultural sector mainly by providing knowledge, skills and a way of analysing problems. As stated by Becker (1993) an alternative view, however, denies that schooling does much to improve productivity, and instead it stresses “credentialism"- that degrees and educational qualification convey information about the underlying abilities, persistence, and other valuable traits o f people. Becker admits that credentialism exist but explains that it does not explain most o f the positive association between earnings from agricultural activities and schooling. 3 He u sed th e lin ea r tw o -s tag ed least squ ares e stim ate , the N L2SLS and a rep ea ted N L 2SLS w ith up land and land- u p land in te rac tio n v a riab le s d ropped . 19 University of Ghana http://ugspace.ug.edu.gh Microeconomic evidence on the impact of education on farm productivity is abundant. The conclusions however varied, depending on where the data used in the analysis is taken from. Generally, while it has been established that better-educated workers earn higher wages in the modem sector (Fafchamps and Quisumbing, 1998), it remains a contentious issue whether education raises farm productivity. Studies using survey data from Asia, South America and other developed countries have come out with a positive effect o f education on farm productivity (Jamison and Mooclc, 1984; Lockheed, Jamison and Lau, 1980; Philips, 1994, King, 1980). But evidence from Africa indicate otherwise- low rates o f return to formal schooling and sometimes negative marginal effects (Moock, 1981; Appiah-Kubi et al, 2001; Weir, 1999). 2.4.1 Evidence from Developing Countries Weir (1999) challenged the hypothesis put forth by Croppenstedt and Demeke (1997) that demand for schooling in rural Ethiopia was constrained by the traditional nature o f farm technology and lack of visible benefits of schooling in terms of farmer productivity. He examined the effects o f schooling upon farmer productivity and efficiency by employing data drawn from Ethiopia Rural Household Survey (EHRS) conducted in 1994. Weir (1999), reported of a positive and significant effect o f land, labour, capital and fertiliser when a standard Cobb-Douglas production function is estimated (without education variables). When education attainment variables were introduced, the results showed that there were positive and significant returns to formal schooling in agriculture in rural Ethiopia. In the case o f household heads that farm, the returns were greatest for those who had attained some upper primary schooling (grades four to six) and no more. He argued that given the traditional nature o f farm technology in rural Ethiopia, it wasn’t surprising that secondary schooling added nothing to the productivity of household heads who were farmers. He explained that those who spent more years in school had spent less time in the fields to learn traditional farm methods from their fathers and may have developed negative attitudes toward farm labour. 20 University of Ghana http://ugspace.ug.edu.gh Weir found out that an additional year o f schooling for the household head decreased output by one percent (insignificantly), when site-fixed effects are included in the model specification. Overall, the effect o f one extra year o f schooling for a farmer was to increase output between 1.0 and 2.0 percent and for non-farmers between 3.0 and 4.1 percent. Iiis findings compared well with Lockheed, Jamison and Lau (1980) and Philips (1994), who reported small or negative effects o f an additional year o f schooling. The average farm-specific efficiency estimated from a stochastic frontier production function was 54 percent with an assumed distribution o f one-sided error term and concluded that formal education had no influence upon placement o f a stochastic production frontier (Weir, 1999). Moock (1981) attempted to measure the worker effect o f education in the production o f the staple food crop in an area o f Western Kenya characterised by small farms and considerable off-farm employment. He developed a modified Cobb-Douglas production function, where a measure of education was incorporated into the function as an explanatory variable. The effect o f educational attainment on production was problematic. Men who had completed 4 or more years o f school seemed to produce nearly 2% more maize, ceteris paribus. However, a negative coefficient (-0.111) for the category of respondents with 1-3 years o f schooling, which he described as surprising, was reported in the study. He however reported that exposure to the extension services enhanced technical efficiency, a 10% increase in extension contact was associated, ceteris paribus, with a 0.2% increase in yield of maize. Eisemon, Schwille and Prouty (1990) conducted a survey, which examined the effect o f primary education on the cognitive skills o f farmers in Kenya. They found that farmers who had been to school were able to construct causal models of events in the natural world and to demonstrate how 21 University of Ghana http://ugspace.ug.edu.gh humans could control these events. They were also able to actively observe, diagnose and correct common agricultural problems better than farmers with fewer years o f education. Using data covering 300 farmers of five villages in Irepodun Local Government Area o f Kwara State, Awolola (1998) related three socio-economic factors with the use o f agrochemicals such as herbicides, pesticides, insecticides, and fertilisers. On the basis o f simple cross-tabulations, he found that the educational status, farm size, and income were positively associated with farmers' use o f agrochemicals. The study further reveals that adoption o f agrochemicals is related to the type of education a farmer has. For example, farmers who have secondary and post-secondary education used more agrochemicals than illiterates and adult uneducated farmers. The implication here is that agricultural development involves a sequence o f innovations and adjustments, which increasingly demand a more sophisticated and better-educated farming community. For a successful agricultural development, extension agents were advised in the study to team up with the farmers so as to identify and solve their farm problems. In establishing a relationship between education and technological innovation, Cotlear (1986) stressed the importance o f non-cognitive aspects o f education, such as the receptivity to new ideas, which allowed farmers to employ new technologies. Using survey data from Peru, he estimated the effect o f education on farm output using production functions. He reported that education also affected production by developing analytic modes of problem solving, by allowing farmers to think more abstractly and thus realise the relationship between technology and output. In their study o f the relationship between farmer education and farm efficiency in Nepal vis a vis the role o f schooling, extension services and cognitive skills, Jamison and Moock (1984) found that farmers who could read, write, and understand numbers could allocate inputs more efficiently and thus increased productivity. In general, the study found that direct returns to education were 22 University of Ghana http://ugspace.ug.edu.gh stronger in developing countries than in developed countries. Jamison and Moock (1984) posited that this could simply reflect shortages in minimal skills in developing countries. Lau, Jamison and Louat (1991) studied the effects o f primary education on economic growth in East Asian and Latin American countries. They found that primary education significantly affected economic growth in twenty-two East Asian and Latin American countries. They also found that secondary education affected growth in fifty-four East Asian, Latin American, African and Middle Eastern countries. Lockeed, Jamison and Lau (1980) reviewed and summarised the findings o f 18 studies containing thirty-seven data sets from thirteen developing countries (primarily in Asia) and found that most reported a significant positive effect of education upon output, though the results were mixed. They concluded that four years o f primary education increased the productivity o f farmers 8.7 percent overall and 9.5 percent in countries undergoing modernisation. However, for the group o f studies concerned with the effects o f education in traditional agriculture, the increase in output owing to four years o f schooling was only 1.3 percent on average. Education increased the ability o f farmers to allocate resources efficiently; enabled them to improve their choice o f inputs; and enabled farmers to estimate more accurately the effect of those inputs on their overall productivity. They concluded that as the education level o f an individual increased, so did the individual’s ability to perform more complicated tasks or to adapt to changing conditions or tasks. Phillips (1994) reviewed additional 12 studies using 22 data sets (with more recent data and greater representation o f Latin America), and was able to confirm the general trends noted above. The average increase in output owing to an additional four years of schooling in the studies he considered was 10.5 percent, with the relevant figures for traditional versus modern farming systems at 7.6 and 11.4 percent, respectively. However, his survey was sufficiently geographically 23 University of Ghana http://ugspace.ug.edu.gh diverse to show that (under certain conditions) the effects o f schooling were stronger in Asia than in Latin America, irrespective o f the degree of modernisation. This may have implications for the assumed applicability o f Asian findings to Africa, though too few studies using Africa data were included to draw strong conclusions. Croppenstedt, Demeke and Meschi (1998) found that literate farmers were more likely to adopt and use fertiliser than those who were illiterate, though the quantity o f fertiliser demanded did not depend upon literacy. They used data from a USAID fertiliser marketing survey. In 1997, Croppenstedt and Demeke using data from the Ethiopia Rural Household survey (ERHS), found that literacy had a positive effect upon productivity and that education was weakly correlated with farm efficiency. A study carried out by Colclough (1980) to establish the relationship between primary schooling and economic development asserted that the benefits o f primary schooling arose from cognitive and non-cognitive behavioural changes induced by the schooling experience—even in systems o f very low quality. It also asserted that evidence from fertility and farmer productivity studies suggested that individual behavioural changes that resulted from schooling were stronger when literacy is widespread than when it is concentrated. He postulated that an interactive effect existed between individual and community attitudes and values, which significantly strengthened the economic and social case for universalising access to primary schooling. Croppenstedt and Muller (2000) estimated the impact o f both health and nutritional status on the productivity and efficiency o f cereal-growing Ethiopian peasant farmers. They examined the distribution of residual technical efficiency (TE), after including the various agricultural inputs, including health and nutrition. Analysis was carried out using data from the first round of the Ethiopian Rural Household Survey, conducted in 1994. They estimated a wage equation by 24 University of Ghana http://ugspace.ug.edu.gh including an education variable o f the head and other members. The results showed that generally, the education o f the head and members of the household was very low with an estimated coefficient o f 0.8 from the estimated wage equation. Estimated output elasticity of Weight-for-Height (WFH- wasting) was between 1.90 and 2.26. Returns to investment in nutrition were clearly high in Ethiopia using stochastic production frontier models. An exception to the literature reviewed above is the work done by Diwan (1971). He provided a quantitative estimate o f the impact of education (education defined in terms o f school years) on the efficiency o f labour using time-series data on United States Non-farm Economy for the period 1909-1960. He employed a linear labour efficiency education function and estimated the elasticity of labour with respect to education. The findings showed a positive impact o f education on labour, and the elasticity o f labour efficiency with respect to schooling was around 1.6. 2.4.2 Evidence from Ghana Jolliffe (1998) examined the impact of cognitive skills on the income o f households in Ghana. He used scores on mathematics and English tests to measure cognitive skills and estimated the returns to these skills based on farm profit, off-farm income, and total income. The analysis used data from the Ghana Living Standards Survey (GLSS), which covered 3,200 households between the ages o f 15 and 55. He argued that using test scores in lieu o f years of schooling to measure human capital eliminated the fact that years o f schooling fail to capture any effect that school quality may have on the creation of human capital. He reported that cognitive skills had a positive effect on total and off- farm income but its effect on farm income was statistically insignificant. Basing his arguments on the fact that school quality varied over Ghana and that test scores reflect this variation, the study showed that the returns to skills, as measured in the total income models, were positive and significant. He alluded that the results supplemented the human capital literature because it made it 25 University of Ghana http://ugspace.ug.edu.gh clear that skills were rewarded and also provided evidence against the screening and credentialism theories o f the returns to schooling. In his study of returns to schooling in non-farm self-employment, Vijverberg (1995) examined the relationship between education attainment (schooling) and income from family enterprises. The study also tried to distinguish between productive efficiency and allocative efficiency in studying the contribution o f schooling, using data (a sample of 2,970) derived from the Ghana Living Standards Survey (GLSS). The GLSS data cover September 1987 to August 1988 and give a complete information on 2,983 enterprises. O f these, eight were omitted by imposing a maximum value of 02,000,000 on both Capital and Inventory, in order to avoid problems with outliers. Furthermore, five enterprises were dropped because o f measurement errors associated with outlier’s. Based on a variety o f specifications o f the enterprise income equation and a variety o f estimation strategies, he found that, in Ghana in the late 1980s, schooling o f the entrepreneur increased the income of his enterprise, primarily by raising the firm ’s allocative efficiency and less so through increased productivity efficiency. The study reported o f a rate of return o f 4.1% to the entrepreneur’s own elementary education. The crossover effect of human capital was significant among entrepreneur’s who had no education themselves because family enterprise income rose by 3.3% for each additional year of completed education o f the most educated other family members. He concluded that education contributed positively to firm performance and stated that the estimates could be refined by means of correction for self-selection of the enterprises and for the endogenous choices o f the quantity o f inputs. 26 University of Ghana http://ugspace.ug.edu.gh 2.5 Educational Effects: A Review of Competing Methodologies 2.5.1 Technical and Allocative Efficiency Husain and Byerlee (1995) note two important ways in which education may increase farm output: (1) general skills acquired in school reduce technical and allocative inefficiencies in production; and (2) attitudes acquired in school encourage the adoption o f new technologies, which cause the production frontier to shift outward. When two farmers are observed using similar amounts o f inputs to produce very different quantities of output, it does not necessarily follow that one is more efficient than the other. Both may be equally efficient at using their chosen technology, but one may be using a more effective production technology and operating on a higher production frontier (Jamison and Lau, 1982). Education may cause movement towards the production frontier, in the case o f technical inefficiency (Weir, 1999). A farmer is technically efficient if it is impossible to raise farm output without increasing use o f at least one input (Weir, 1999). Technical inefficiency may arise because of inappropriate timing or method o f input application, which is often caused by a lack o f information but may also reflect problems of input supply (Ali and Byerlee 1991). For allocative inefficiency, education causes movement along the frontier to a profit maximising point. Allocative (or price) efficiency is achieved when the cost o f producing a given output is minimised, as evidenced by the equality o f the marginal products of inputs to the input price ratio (Weir, 1999). Ali and Byerlee (1991) pointed out some o f the causes o f allocative inefficiency. They include lack of information, failure to choose the most cost-efficient combination of inputs, unreliable input supply, sub-optimal tenancy arrangements, risk aversion, and other institutional constraints. 27 University of Ghana http://ugspace.ug.edu.gh Efficiency measures provide an indicator o f performance. Rates of return to education investments and other physical inputs have often provided a good indicator o f measuring efficiency (Psacharoupolus, 1985, 1994; Appiah-Kubi et al, 2001). Other studies (Afari, 1998; Asuming- Brempong, 1987; Darko, 1998) have used the concept of Domestic Resource Cost (DRC) and Pay- Off Matrix to measure efficiency o f resource-use. As farmers gain experience, inefficiency in using the new inputs fall (Weir, 1999). Thus measures o f efficiency at anyone point in time may not reflect the equilibrium situation, but rather indicate deliberate trial and error on the part o f farmers (Welch 1978). Technical progress o f farmers in a particular area may be measured by the proportion o f the population who take up new innovations or by the quantities o f modem inputs, such as fertiliser, which are used. Psacharopoulos and Woodhall (1985) described a framework o f four stages o f agricultural technology adoption and the role education may play in each stage. Stage 1: Traditional farming. Information is passed from father to son, and little or no schooling is needed. Stage 2: Single input adoption (e.g., fertiliser). Basic literacy and numeracy are very useful to farmers for understanding instructions and adjusting quantities of the new input. Stage 3: Adoption of multiple inputs simultaneously. Here, more than literacy and numeracy are necessary. Some basic science knowledge is helpful. Stage 4: Irrigation-based farming. The farmer must make complex calculations o f effect o f changes in crops and weather. More education is needed for efficient production at this stage. 2.5.2 Estimation of the effect of Education on Agricultural production: Production Function Approach The approach o f examining the effect of education on agricultural productivity is divided into two major camps: frontier versus non-frontier (direct) methods for estimating the production function (Weir, 1999). He states that if the researcher is interested mainly in the estimated coefficient on 28 University of Ghana http://ugspace.ug.edu.gh schooling in the production function, non-frontier techniques will suffice. Estimation o f the average (non-frontier) function permits efficiency ranking of firms, in this case farm units or enterprises (Weir, 1999). 2.5.2.1 Non-frontier (Direct) Production Function Approach Estimation o f the effects o f education upon productivity using direct, non-frontier methods dates back to the 1960s (Weir, 1999). Griliches (1964) was the first to use a production function to estimate the effect o f education on agricultural output. Kislev (1965) used non-frontier methods to carry out analysis on the same aggregated data in 1965, but he obtained different result from that obtained by Griliches. Chaudhri (1979) explained that the difference in the results obtained from the two studies (Griliches, 1964 and Kislev, 1965) was due to the level o f aggregation o f the data used. He discussed four possible effects o f schooling: the worker effect, the allocative effect, the innovative effect and the external effect. Based upon his own empirical work, he concluded that an examination o f data at household level would reveal only direct (worker) effects o f schooling on output. As the level o f aggregation is increased up to the state level, more o f the allocative, innovation and external effects are internalised. Welch (1970) estimated a production function with physical output of one farm product as the dependent variable and education and other inputs as explanatory variables. The estimates disclosed only the worker (direct) effect o f education. Using gross output sales o f two or more products as the dependent variable reveals both the worker and allocative effects of schooling. In this way, the role of education in allocating inputs across different outputs can be captured. Estimating a value-added (or profit) production function makes it possible to examine the worker, allocative and input- selection effects of education, if education is included in the specification and other purchased inputs are excluded. 29 University of Ghana http://ugspace.ug.edu.gh Ram and Singh (1988) examined the allocative effect o f schooling in Burkina Faso. They used alternative Mincer/Chiswick-style4 earning (income) function to estimate the effect o f education upon income, one with education plus other farming inputs as the explanatory variables, and one including only education. The coefficient o f education variable in the first equation captures both the worker and the allocative effects because other inputs were no longer held constant. The difference between the coefficients on schooling in the two equations provides an indication of the size o f the allocative effect. To establish causality from education to input choice, Appleton and Balihuta (1996) run regressions with each input as the dependent variable and education and other regressors as explanatory variables in their research on Uganda agricultural productivity. Since they found that education did not have a strong positive effect upon use of capital and other purchased inputs, they concluded that the usual Cobb-Douglas production function which included physical inputs explicitly (i.e., inputs held constant) understated the importance o f education in explaining output. They then estimated the production function without these physical inputs, as suggested by Welch (1970), to capture more fully the effect o f schooling. Weir (1999) observed that when there is adequate information about input and output prices, it is possible to investigate the full effects of schooling without omitting other inputs from the production function. Wu (1977) estimated the worker, allocative and scale effects o f schooling for small-scale farmers in Taiwan using a model of profit maximisation. Farm profits were a function of fixed and variable inputs plus education less cost. Optimal input demand functions were specified as functions o f fixed inputs and education to give an expression for optimal profits. 4 E s t im a t ing re turns to educa t ion and exper ience on earnings (bo th socia l and priva te ra te o f re lu rn) 30 University of Ghana http://ugspace.ug.edu.gh Totally differentiating the actual and optimal profit functions with respect to education provided information on the allocative, worker and scale effects o f education. Jollife (1998) used non-frontier models to estimate the effect o f cognitive skills on farm, non-farm and total income using data from Ghana. He used Powell’s Censored Least Absolute Deviations (CLAD) and Symmetrically Trimmed Least Squares estimators to estimate farm and off-farm income after estimating the income functions by Ordinary Least Squares (OLS). He pointed out that in contrast to Heckman’s two-step or the Tobit estimator, Powell’s estimators are consistent in the presence o f heteroscedasticity and are robust to other violations o f normality. 2.5.2.2 Frontier Production Function Approach A production frontier is estimated based on the most efficient observed use o f inputs to produce each level o f output. The extent to which a production level o f a firm differs from the frontier provides a measure o f technical inefficiency for the sample as a whole or for each individual firm. The causes o f technical inefficiency can be investigated by regressing inefficiency on education and other explanatory variables (Ali and Byerlee 1991). Phillip and Marble (1986) noted that the frontier production function may be more relevant to the study o f the effects o f education on farm productivity than the conventionally used average function because it focused attention on the best- practice farmers in the sample with schooling as an important determinant. Statistical estimation o f production frontiers has had two incarnations: stochastic and non-stochastic (deterministic). The deterministic frontier takes the following general form: Y = f (X )e u, (2.1) where, Y is output, X a vector o f production inputs and u is a non-negative error component representing technical inefficiency (Weir, 1999). 31 University of Ghana http://ugspace.ug.edu.gh Aigner, Lovell and Schmidt (1977), Meeusen and van den Broeck (1977) and Battese and Cora (1977) have developed the stochastic production frontier methodology. Stochastic frontier estimation involves specification o f a two-part error term: Y = f (X)ev'l‘, (2.2) where, v represents random shocks, such as measurement error or factors which are external to the fmn(e.g., weather), and is symmetric and distributed normally. The second component, u, is a one­ sided, strictly non-negative, error representing technical inefficiency. Jondrow et al. (1982) show how to decompose the v-u term from equation 2.2 to provide estimates of technical inefficiency by calculating the expected level o f inefficiency for each farm, E(w,), conditional on the random component, v,- (Ali and Byerlee, 1991). Battese and Coelli (1988) provided a formula to estimate farm-specific efficiency in the case o f a logged dependent variable. One weakness o f the frontier approach is that empirical results are very sensitive to the number of inputs included, the degree o f aggregation o f the data and whether or not environmental factors, such as soil quality, are included (Hussain and Byerlee, 1995). Lovell (1993) noted that the Cobb-Douglas production function is the most commonly used functional form to estimate the production frontier. It has the advantage o f ease o f estimation and interpretation o f the coefficient, u. However, since it assumes constant elasticity o f scale and unitary elasticity o f substitution, variations in elasticity o f scale or substitution may be erroneously attributed to inefficiency. Hence, functional form is a relevant consideration when estimating the production frontier. Results are also highly sensitive to the assumed distribution o f u. the component o f the error term attributed to inefficiency. Commonly assumed distributions include: half-normal, truncated normal, exponential, and gamma (Greene, 1993). 32 University of Ghana http://ugspace.ug.edu.gh Lovell (1993) noted again that the two-stage procedure rested on the assumption that factors such as education affected the efficiency o f the farmer in transforming inputs into output, but did not affect the process by which production occurs. If the variables used in estimating the efficiency were correlated with the variables used to explain the efficiency, the coefficients on the variables used in the first stage to estimate efficiency will be biased. A one-stage model in which all variables, including education, appear in a single equation to estimate efficiency may alternatively be specified. This reduces the problem of omitted variable bias, but may lead to multicollinearity. There is an important difference between the two-stage and one-stage model: in the one-stage model, efficiency is measured controlling for variables such as education; two-stage model, efficiency is measured without controlling for variations in education in the first stage, and regressors such as education are used to explain variations in efficiency in the second stage (Lovell, 1993). Ali and Byerlee (1991) surveyed 12 frontier production function studies, o f which approximately half investigated the sources of measured inefficiency. They noted that variables related to managerial ability o f farmers, including education and technical know-how, were found to be significant in every case, whereas mixed results were found for other potential sources o f inefficiency such as input use and soil quality among others. 2.5.3 Summary 111 summary, we see that the effect o f education on productivity and income at the farm household level has received a lot o f attention with most of the studies employing data from other developing countries. It has been well noted that education (both formal and informal) contributes significantly to economic development and growth, poverty reduction through increased productivity. Also education can be treated as a factor of production together with other fixed and variable inputs and 33 University of Ghana http://ugspace.ug.edu.gh it has a positive effect on output through increased productivity. Some studies though reported a negative effect o f education on output when different levels o f schooling years are considered. The studies reviewed point out the different competing methodologies often used in the estimation procedure. Most o f the studies reviewed used the non-frontier production function by employing the Cobb-Douglas technique with few others using the stochastic frontier approach. The stochastic frontier function approach is not used in the estimation carried out in this study due to data limitation associated with the manner in which answers were solicited from respondents and the difficulty associated with the decomposition o f the error term into its components. The frontier model is used when the measurement error has a symmetric component5 and one-sided component, which captures the effects o f inefficiency relative to the stochastic frontier. The Cobb-Douglas production function is employed because the estimated coefficients can be interpreted easily. 3 R andom shock s ou ts id e th e firm s con tro l, v a ria tio n s and o th e r s ta tis tica l ‘n o is e ’ 34 University of Ghana http://ugspace.ug.edu.gh CHAPTER THREE METHODOLOGY 3.1 Introduction The study employs the theory o f production and profit maximisation at the firm level to derive the empirical models that are used in estimating the factors that affect farm productivity and incomes in Ghana. The traditional Cobb-Douglas production function is adopted in the study to estimate the effect o f factors affecting farm productivity in Ghana. The non-frontier production function is repeated for the second objective with the inclusion o f an education variable as a factor of production. The first and second specific objectives are estimated based on the theory o f production by specifying a model without and with education of the farmer as a variable input. The third objective is estimated based on the theory o f profit maximisation by a farm unit. In estimating the effect o f education o f the farmer on farm income, a restricted net farm income function is specified with education measured in formal years of schooling completed as an explanatory variable. Section 3.2 focuses on the theoretical and analytical framework on production function and profit maximisation. Empirical models used in the estimation are provided in section 3.3. The data to be used in the estimation and its source is outlined in the last section o f the chapter. 35 University of Ghana http://ugspace.ug.edu.gh 3.2 Theoretical and Analytical Framework Available literature on various theories used to explain the effect o f education on productivity has mostly dwelt on the theory o f the firm where education is incorporated into Cobb Douglas production function (see Weir, 1999; Moock, 1981; Lockheed et al, 1980) and the human capital model developed and pioneered by Shultz (1961; 1972), Becker (1964; 1993) and Mincer (1958; 1974). The Human Capital Model (HCM) has been the most widely used model to estimate the rates of return especially in the estimation o f social and private rates o f return to education. According to the theory, given that wage earners are paid their marginal product and that this marginal product rises as more human capital is accumulated, it is possible to estimate rates o f return to additional years o f schooling from earnings data among persons who have different levels o f education. 3.2.1 Production Function The set o f all combinations of inputs and outputs that comprise a technologically feasible way to produce is called a production set. The function describing the upper boundary o f this set is known as the production function and it measures the maximum possible output that can be obtained from a given amount o f inputs applied (Varian, 1999). Inputs for production, which are often called factors of production are often classified into broad categories such as land, labour, capital (financing and physical), raw material among others (Varian, 1999). The firm (in this case the farm unit) is a technical unit whose entrepreneur (owner or manager) decides on how much o f and how one or more commodities will be produced, (Henderson and Quandt, 1980). The entrepreneur (farm manager) transforms inputs into output(s), subject to the 36 University of Ghana http://ugspace.ug.edu.gh technical rales specified by his production function and he either gains or bears the loss which results from his decision (Henderson and Quandt, 1980). The farm production technology or function expresses the relationship between the quantities o f inputs he employs and the quantities of output produced. Following from Varian (1999) and (Henderson and Quandt, 1980), we consider a farm unit which utilizes m variable inputs z =(z], ,zm) and one or more fixed inputs in order to produce a single output, O. The production function states that the quantity o f farm output Q is a function q o f the quantities of inputs: Q = q (p i , .......... ,zm,A) (3.1) Where, A is a vector o f fixed inputs The production function, q shows the maximum output obtainable from various input vectors. We assume that the farm owner or entrepreneur is rational and operating in a competitive market (is a price taker). The function is assumed to be a single valued continuous function with continuous first and second-order partial derivatives. It is also an increasing function o f Zl: Where, i = 1, 2, 3, 4 m. For output maximization or cost minimization, it is assumed that the production function is a regular strictly quasi-concave function. The production function presupposes technical efficiency and states the maximum output obtainable from every possible input combination. When we assume two inputs Z\ and Zj with one of the inputs assigned a fixed value, (say Z?*), the output- produced, Q becomes a variable function of the other (Z i). The average and marginal products for 37 University of Ghana http://ugspace.ug.edu.gh Zi are defined in an analogous manner for particular values o f Z 2 . The average product (AP) o f Z| is its total product divided by its quantity: -I 2) /A (3.2) The marginal product (MP) o f Zj is the rate o f change o f total product (Q) with respect to a unit change in Z i, thus, the partial derivative of (3.1) with respect to Zi is Output elasticity o f Z) is defined as the proportionate rate o f change o f O with respect to Z/ Output elasticities may be expressed as ratio o f marginal and average products and are positive if both MP and AP are positive (AP is always positive). 3.2.2 The Cobb Douglas Production Function The Cobb-Douglas production function has a constant elasticity o f substitution of one and is a member o f the CES class and is homogenous of degree one (Henderson and Quandt, 1980). If a production function has the form q(z/, z2) = A z f& , then it is a Cobb-Douglas production function. The parameter A measures the scale o f production: how much output one would get if one used one unit o f input. The parameter a and b are elasticities and they measure how the amount o f output responds to changes in the input. We generally assume some properties o f farm production 38 (3.4) University of Ghana http://ugspace.ug.edu.gh technology. First, we assume that farm technologies are monotonic: that is, if we increase the amount o f at least one o f the inputs, it should be possible to produce at least much output as you were producing originally. Second, the study assumes a convex farm technology. 3.2.3 Profit Maximization In addressing the third specific objective, the study has profit maximization o f the farm unit as its theoretical underpinning to the estimation o f the effect o f education on farm income (profit). The farm unit chooses a production plan so as to maximize profits (Varian, 1999). The profit, is the difference between his total revenue and his total cost: x = p q ( ) - e (3.5) where, 7t is profit from the farm unit p is output price c is total cost Suppose that the farm unit produces n outputs (q\,...,q„) and uses m inputs (z/, ...,z,„). The profit received by the farm unit can be expressed as: n in n = Z P jV i ~ Z r i z i (3-6) 7 = 1 i -1 The first term on the right-hand side of equation 3.6 is revenue and the second term is cost. The total revenue o f an entrepreneur who sells his output in a perfectly competitive market is given by the number of the units he sells multiplied by the fixed unit price (p) he receives (Henderson and Quandt, 1980). The costs include all the factors of production used by the farm unit, valued at their market prices. The prices of inputs are denoted in equation 3.6 as n. 39 University of Ghana http://ugspace.ug.edu.gh Under certain regularity conditions and the conditions required by the duality theory, efficient production technology is provided by the profit function: 7 c { p \ , . . . .p n ) i \ , . . . . rm ) = m a ^ 2 { p q - r ' z \ f ( z ) > q , z > 0 , q > 0 \ ( 3 .7 ) Equation 3.7 shows the maximum profit available at output pricep and input prices, r (Lovell et al., 1980). 3.2.3.1 Short and Long-run Profit Maximization Let / (Z i Zi) be the production function for the firm with two inputs, p represents the price of output, and /'/ and the prices o f the two inputs Z/ and Z2 respectively. Then the profit- maximization problem facing the farm unit can be written as: max p f { z , z 2) - rxz x - r2z 2 (3 . 8) If Z/ is the profit-maximising choice o f factor Z/, then the product o f the output price and the marginal product o f factor Zy, should be equal to the price o f factor Zy. Assume Z2 fixed for convenience. That is, p . M P 2 i ( z ‘ , z 2 ) = r , ( 3 . 9 ) That is, the value o f the marginal product of a factor should be equal to its price. If the value of marginal product is less than its cost, then decreasing the usage level o f input factor Z/, can increase profit. Other two conditions for changing the level o f input usage are when the value o f marginal product is more than the cost and when they are equal (i.e. value o f marginal product = cost of input). In the long-run the firm is free to choose the level of all inputs. This long-run profit- maximisation problem can be posed as: 40 University of Ghana http://ugspace.ug.edu.gh max SuZ2 p f ( z l z 2) - rxz x - r2z 2 ( 3 . 10 ) Setting the partial derivatives o f equation (3.10) to zero for all inputs we obtain: p.MP , ( z*, z*2 ) = r, ( 3 . 11 ) p.MP 2 (z* ,z*2 )= r2 ( 3 - 1 2 ) The first order condition for profit maximization requires that each input be utilized up to a point at which the value o f its MP equals its price. The farm entrepreneur can increase his profit as long as the addition to his revenue from employment o f an additional unit Z\ exceeds its cost. The marginal value product o f Z\ is the rate at which the entrepreneur’s revenue would increase with further application o f Z\ (Henderson and Quandt, 1980). In the production process the farm household utilizes both fixed and variable farm inputs. The optimum output and profit from the farm is obtained when we have the most efficient production process where all inputs into the process are used to their optimum at the right time, and in the correct quantity. This could only be attained if the farm entrepreneur (who is the decision-maker in the household) has the required skills and knowledge to select the right crop mix and to move into off-farm job if returns from the agriculture enterprise are low. The education or schooling o f the farm household head is hypothesised to be an important input in the production process, helping to achieve high output per unit physical input. The Cobb-Douglas production function is used to estimate the effects o f education, in addition to conventional inputs on agriculture output per unit area cultivated. 41 University of Ghana http://ugspace.ug.edu.gh 3.3 Empirical models 3.3.1 Introduction First, the study employs the Cobb-Douglas production function to estimate the effect o f inputs and household characteristics on the value o f output per acre. To achieve the second objective, educational level o f the household head is introduced as an additional factor o f production to find out if there are positive returns to education for farmers in Ghana. 3.3.2 Farm Technology We estimate a general farm production function denoted as: Qj ~ Q j (Fij> Xjj, Rdj, Z k) (3.13) 1 = labour j = 1 ,2 ,......, J (Crops) i = 1,2,......., I (non-labour variable inputs) d = 1,2,......, D (non-labour fixed inputs) k = 1,2,......,K (household composition variables) From equation 3.13 a Cobb-Douglas production function is stated as Q j = F ^ ' X ffa2R 4 a3Z ka *eao + V j ( 3 . 1 4 ) Finding the natural logarithms o f both sides we obtain a specified double-log function o f farm production. / D k In Qj = cc0+cc, In Fff + Y j h XnX,j + lnRaj + Y f * \ lnZt + (3.15) /'=! d=l k=I 42 University of Ghana http://ugspace.ug.edu.gh Where, Qi = farm production o f jth crop (Farm output per acre) Fij = Labour employed in the production of jth crop (man equivalent person days) X ij = Vector o f non-labour variable farm inputs i used on the jth crop. R cij = A vector o f non-labour fixed inputs d Z|< = A vector o f household characteristics or composition variables a0 = constant a'.v are parameters or coefficients to be estimated u = Error term with zero mean and constant variance 3.3.3 Education as a factor of Production 3.3.3.1 Introduction The study employs a non-frontier technique to estimate the coefficient on years o f schooling in a modified Cobb-Douglas (C-D) production function. We estimate only the “worker effect” o f education. Estimating the “allocative effect” was not possible because o f problems encountered with the aggregation6 o f the two main crops selected. 3.3.3.2 Non-Frontier Empirical Model: Worker Effect of Schooling The worker effect o f schooling refers to the increase in farm output that is attributed directly to education, holding other physical inputs constant (Chaudri, 1979; Welch, 1970). An empirical 6 In m easu rin g the a llo ca tio n e ffec t o f schoo ling , the d ep enden t v a riab le m u s t be the to ta l v a lu e o f fa rm ou tpu t agg reg a ted ov e r a t least tw o c rop s , s in ce no accoun t is taken o f a llo ca tio n o f inpu ts acro ss c om pe tin g uses in th e case o f o n ly one ou tpu t. 43 University of Ghana http://ugspace.ug.edu.gh model is developed to capture the worker effect o f education and extension exposure on farm output per unit land area cultivated using a Cobb-Douglas (C-D) production function. I (3 .16 ) t = 1 i= l,2 ,.........,I (inputs) j-1 ,2 , ,J(outputs) Talcing the natural logarithms o f both sides o f equation 3.15, we obtain / In Qj = Z ln X n + a E f + A + n j (3 .17) i=i Where, Qj is farm output per unit land area cultivated; Xy is a vector o f inputs into production; Bj> is a measure o f education o f household h e ad /in the j th farm and jr, is a stochastic error term. The following hypotheses are tested, where Ho is the null hypothesis and Hi is the alternative hypothesis. Ho: Education exerts no effect on farm productivity Hi: Education exerts a positive effect on farm productivity The hypothesis is repeated same for labour input (L), except the vector o f household head characteristics, Z whose coefficients can be positive or negative. 3.3.3.3 Hypothesis to be tested 44 University of Ghana http://ugspace.ug.edu.gh 3.3.4 Percentage gain per year of education. Following Lockheed et al (1980), the study computed the percentage gain in output value for one additional year o f education o f an individual farmer using equation 3.18a. The percentage increase is obtained by computing the ratio o f output value when the level o f education is Zi year greater than R, to the value when it is V% less, subtracting one (1), and multiplying by 100. Increases in productivity are estimated for two different specifications. In a specification where formal years of education completed is not logged, the percentage increase in output is estimated by using the formula in equation 3.18a: Where, R = Average educational level o f the sample a = Estimated coefficient o f education e - Natural exponential In the other specification where a set o f threshold dummy variables are created to better understand the relative importance o f different levels of schooling, the formula in equation 3.18b can be utilised: Where, N is years o f education in the level specified by the dummy variable indicator. In this calculation it is assumed that the percentage increase due to education can be proportionally attributed to the years o f education. % increase in output l a{R + 0.5) £a( R - 0.5) - 1 x 100 % increase = ------ xlOO N r - 1 45 University of Ghana http://ugspace.ug.edu.gh 3.3.5 Farm Profit (Income) Function From a profit maximization problem of the farmer, we define a restricted profit function: i - 1,2, ,m (3.19) subject to a production constraint Qj < q . (X A, H ;E) where, Qj is of farm output o f crop i in kilogram per acre. In the short-run, the opportunity cost o f the fixed inputs is zero. The farmer needs only to maximise the returns to fixed inputs i.e. the sales value o f output less the cost o f variable inputs used on the farm. The profit from the j-th farm is expressed as a function o f the price o f output and variable input prices and fixed input quantities with education level o f the household head as an input. 7r]=7r] ( p j , r x,r2, rm;A,H;E) (3 .20) For easy computation and estimation, equation (3.20) is expressed in a restricted net farm income function, Yf. The natural logarithm of which is estimated as: Yf = f(A , H; Pf , E) Where, f is increasing and concave in (Xi; A, H) and homogeneous o f degree one in (X(; A, H) meaning it exhibits constant returns to scale. Yf is restricted net farm income in cedis; A is household owned land in acres; H is labour employed for the cultivation in full-time worker 46 University of Ghana http://ugspace.ug.edu.gh equivalents. E is a vector o f variables denoting education and extension training; P r is a vector of input and output prices (Farm input and output prices are cluster averages) and p, is price o f output (cedi/kg). I', is price o f input i (cedis). Since the survey price data on output and inputs are cluster averages, we include them for every farmer found in that particular cluster; therefore the estimated coefficients should be interpreted with caution. 3.3.6.1 Hypothesis to be tested Ho: Education exerts no effect on farm income and Hi: Education exerts a positive effect on farm income. The hypothesis is repeated same for Farm Size (A), Labour input (IT). Output price, pj has a positive effect on farm income but the prices o f variable inputs, r,’s are expected to be inversely related to farm income. Where, Ho is the null hypothesis and Hi is the alternate hypothesis 3.4 The Data The analysis uses data from the Ghana Living Standards Survey (GLSS)7. The GLSS is a nation­ wide household survey carried out by the Ghana Statistical Service with technical assistance from the World Bank. With its focus on the household as a key social and economic unit, the GLSS survey provides valuable insights into living conditions in Ghana. 7 See the G LSS R epo r t o f th e fou rth round fo r de ta il d esc rip tio n o f the sam p ling te chn iqu e and th e se le c tio n o f hou seho ld s 47 University of Ghana http://ugspace.ug.edu.gh The survey administered from October 1998 to September 1999, covers 5,998 households, containing over 25,000 persons, with detailed information on formal and informal labour activities, household farm activities, expenditures, education status o f household members and many other determinants o f household welfare. 48 University of Ghana http://ugspace.ug.edu.gh CHAPTER FOUR RESULTS AND DISCUSSION 4.1 Introduction The analysis o f the effect o f education on farm productivity and farm income is carried out at the level o f two farm enterprise activities. The farm activities are cassava and maize production. Section 4.2 o f the chapter focuses on the definition o f variables and discussion o f the mean values of the variables used in the analysis. Section 4.3 provides descriptive statistics on the head of households selected for the analysis. Descriptive statistics are provided on their age-sex structure, level of educational attainment, and extension contact. Cross-tabulations are also presented to show the distribution o f farmers in the three ecological zones in Ghana by locality; and also educational level completed by locality and sex. Farm productivity regression results obtained from production function without accounting for the education o f the household head or his skill are presented in section 4.4, followed by results o f the regression analysis done with education variables as inputs in the production technology o f the farmer. Section 4.5 focuses and discusses the results obtained for the farm income function. Farm income is estimated for each farm enterprise activity, which is expressed as a function o f a vector o f output and input prices, fixed input variables, formal schooling variables, extension service contact and household composition variables. Linkages between farm productivity, education and food security is provided in the last section. 49 University of Ghana http://ugspace.ug.edu.gh 4.2 Sample selection and variable definition The data for study is drawn from the fourth round o f the Ghana Living Standards Survey (GLSS). The final sample employed for the descriptive analyses are 175 for cassava and 290 for maize farmers. First, the study considered and concentrated on only household heads whose main occupation was crop farming. These farmers were further grouped into cassava and maize crop production with only those who reported as having harvested the crop for sale and were primarily involved in the two activities. Second, farm household heads that had mentioned these crops (cassava and maize) as their main activity but sold other different crops apart from the main activity crop were dropped from the analysis. The number o f observations dropped was small and most had no data on sales and other important indicators and cannot significantly bias the results. Observations with missing data or inconsistent information and negative values were not included in the final analysis. In addition, the following outliers were omitted from the sample: household heads who are above 65 years and those below 15 years; households using less than 1 acre o f land and more than 30 acres for the production of these crops. For maize and cassava production, only two-thirds o f farmers used inputs such as fertiliser (inorganic and organic), pesticides and insecticides but most o f them responded as having spent some money on hired labour, hand tools and other inputs like planting materials, bags and containers. Because o f the difficulty in obtaining actual physical quantities o f inputs from the data, hired labour employed on farm, hand tools and capital inputs are measured in their value terms as determinants o f farm productivity (see Table 4.1). 50 University of Ghana http://ugspace.ug.edu.gh Table 4.1: Variable Definitions and Mean Values ariable N am e D efin ition C assav a M ean S td . D ev M ean M aize S td . D ev ependent v a r iab le ji_C ro p a c re N a tu ra l log o f h a rv e sted ou tpu t p e r acre 6 .67 1.14 5 .52 0 .96 arm and hou seho ld triable j i_ L ab N a tu ra l log o f fam ily labou r (m an eq u iv a len t 6 .02 0.81 5 .76 0 .88 ji_ L an d pe rso n days) N a tu ra l lo g o f c u ltiv a ted land a rea in acres 0 .79 0 .65 0 .98 0 .6 9 j i H lab N a tu ra l lo g o f h ired labou r co st p e r acre 9 .97 1.18 9 .78 1.24 _,n_Cfert N a tu ra l log o f th e v a lu e o f fe r ti lise r and o th e r 9 .04 1.17 8 .65 1.22 ji_H to o I ag ro -ch em ica l N a tu ra l log o f v a lu e o f h and too ls in ced is (lo ca l 8 .64 0 .86 8 .36 0 .88 ji_C ap in p ts hand im po r ted ) p e r acre N a tu ra l log o f c o s t o f c ap ita l in pu t equ ipm en t 8 .96 1.37 8 .66 1.33 j i j i h s i z e p e r acre N a tu ra l log o f h ou seh o ld size 4 .76 2 .5 7 4 .7 6 2 .76 \g ey A ge o f h ou seh o ld head (y ea rs) 4 3 .15 11.36 41 .73 12.16 \g e sq S qu a re o f ag e o f h o u seh o ld h ead (y ea rs ) 1990 .35 992 .78 1888 .78 1058 .58 >ex_l D um m y: 1 i f h ou seh o ld h ead is m a le 0.53 0 .50 0 .77 0 .42 ^work F a th e r’s o c cu p a tio n (fa rm ing = 1, o th e rw ise= 0 ) 1.70 4 .19 1.58 3 .97 ducation V ar iab le s 3chyrs Y ears o f sch o o lin g o f th e h o u seh o ld h ead 5.83 4 .88 5 .12 4.81 Eduprim Dum m y: 1 i f h e ad has com p le ted p rim a ry schoo l 0 .18 6 .39 0 .14 0 .36 Sdumid D um m y: 1 i f h e ad has com p le ted m id d le schoo l 0 .37 0 .49 0.31 0 .46 Sdujss D um m y: 1 i f h ead has com p le ted ju n io r 0.011 0 .10 0 .04 0 .20 Edusec sec o n d a ry schoo l D um m y: 1 i f h e ad has com p le ted seco nd a ry 0 .05 0.21 0 .03 0 .17 EduPsec Extserv educa tio n D um m y: 1 i f h e ad has com p le ted schoo l above s e c o n d a ry level D um m y: 1 i f h e ad has ex ten sio n se rv ice con tac t 0 .10 0 .30 0 .006 0 .09 0 .08 0 .29 Aveschyrs A ve rag e sch oo l y ears o f all o th e r fam ily 4 .06 3 .56 2 .9 7 3.5 ite Dumm ies Rural m em bers a p a r t from the h ead D um m y: 1 i f h ead is lo ca ted in th e ru ra l a rea 0 .86 0 .34 0 .89 0.31 Forest D um m y: 1 i f h ead is loca ted in th e fo res t 0 .76 0 .43 0 .57 0 .49 Coastal e co lo g ica l z on e D um m y :! i f h ead is lo ca ted in th e co asta l 0.11 0.31 0 .10 0 .30 eco lo g ica l zone arm in com e est im a tion ariables Farminc C ro p rev enue less to ta l inpu t c o s t in p roduc tio n Mprice C lu s te r a v e rag e p ric e o f m a ize (p e r kg) 5 .33 6 .08 0 .49 0 .36 Pfert Phtool C lu s te r av e rag e p ric e o f fe rti lise r (av e rag e fo r N .P .K /1 5 -15 -15 and Su lpha te o f A m m on ia ) p e r m in i bag C lu s te r av e rag e p ric e o f h and to o l (a p roxy u s ing th e p ric e o f one s tandard cu tlass) 10.18 8 .80 0 .24 0 .18 Source: Author’s calculations from GLSS 4 Survey Data 51 University of Ghana http://ugspace.ug.edu.gh Fertilizer and other agro-chemicals are included as an aggregated figure for the estimated regression in the cassava and maize production functions. Though individual effects are lost through the aggregation, it was necessary in order to avoid several zero observations found in the data. Land cultivated is measured in acres. Some farmers report unit o f land cultivated under crop in acres while a greater percentage stated units in ropes and poles. All units stated in ropes and poles are converted to acress. The mean farm size cultivated is 0.79 acres for cassava and 0.98 acres for maize. Age and age-squared are used as indicators o f farm work experience. The mean ages are 41.7 for maize farmers, and 43.2 for cassava farmers. Another way o f estimating work experience was to subtract the number o f years o f schooling and the age when the farmer started kindergarten (5 years) from current age to indicate years o f experience on the farm. The regression result did not change much when age squared was replaced with years o f experience. The education o f the head and other adult members o f the household, engaged in cassava and maize farming is generally very low. The average years o f schooling for household heads are 5.83 for cassava farmers and 5.12 for maize farmers (see Table 4.1). This means that most o f the farmers sampled had less than primary level education on the average. The average level o f schooling years completed by other adult household members was 4.06 and 2.97 for households cultivating cassava and maize respectively. To account for the possibility that different levels o f schooling have differing effects upon productivity, a set o f dummy variables representing different levels o f schooling is used as a measure o f education in other specification o f the estimated functions. According to Weir (1999) the coefficient on a 0-1 dummy variable represents the percentage increase in output due to having that level of schooling, as compared with that o f no schooling and other levels completed. 8 See ap p end ix I fo r th e co n v ers io n o f p o le s and ropes to acres u s in g s tand a rd co nvers io n fac to rs 52 University of Ghana http://ugspace.ug.edu.gh A proxy variable was used for extension contact because of the absence o f direct question in the survey questionnaire that required respondents to either indicate whether they had any extension contact or not. Farmers were to indicate the source o f the physical inputs they use, whether from the Ministry o f Food and Agriculture (MoFA), NGO or Cooperative Society. This might not be a good proxy but it is employed on the assumption that farmers who receive their inputs from the above named sources, rather than purchasing from the open market, are assisted, taught or given advice on how to use them. Site dummy variables are included to capture site-specific fixed effects like location and ecological zone effects. Education and/or occupation o f parents are good indicators o f how well a particular individual is doing in his farm enterprise activity. The occupation o f parents (as farmers) is expected to impact positively on farm productivity because children may have learnt a lot of farming skills from them especially if they spent a significant number o f years on their parent’s farm prior to having their own. Dummy variables o f 0-1 were used to indicate the occupation of parents. Together, with the site dummies they serve as control variables to capture some o f the unobserved effects on farm productivity. 4.3 Descriptive statistics and demographic profile of farmers This section provides descriptive statistics on household heads in the two farm enterprise activities selected for the study. The section looks at the age-sex structure o f farmers, their locality and ecological zone, whether they had attended school before or not, their educational attainment by sex and extension service contact. 53 University of Ghana http://ugspace.ug.edu.gh 4.3.1 Farmers’ Locality and Ecological Zone The GLSS 4 survey was carried out in all the three ecological zones o f Ghana. There were four locality classifications but we focus on the first locality classification o f whether farmers were in an urban9 or rural10 location. A cross tabulation is carried out to indicate the distribution o f agricultural household heads into selected farm enterprise activities, their ecological zone and locality. As shown in Table 4.2, cassava and maize production is a rural phenomenon with about 91% of farmers sampled found in the rural area (409). Only 9% o f farm household heads were from the urban locality with more than half o f them involved in maize cultivation. Most o f the farmers were concentrated in the rural forest zone. The percentage o f cassava and maize in the rural forest zone were 75.5 and 58.9 respectively. Table 4.2: Classification of agricultural household heads into selected cassava and maize farmers by ecological zone and locality. C assava M a ize T o ta l A ctiv ity U rban Rura l U rban R ura l U rban R ura l No % No % No % No % Coastal 3 12.5 16 10.6 2 6.3 27 10.5 5 43 Forest 17 70 .8 114 75 .5 13 40 .6 152 58 .9 30 266 Savanna 4 16.7 21 13.9 17 53.1 79 30 .6 21 100 Total 24 100 .0 151 100 .0 32 100 .0 258 100.0 56 409 Source: Author’s calculation from GLSS4 data (2000). The total number o f household heads involved in cassava production was 175 with 151 found in rural area forming about 86.3%. The distribution o f cassava farmers by ecological zone show that 10.8% o f the farmers were found in the coastal zone, 74.9% in the forest zone and 14.3% in the savannah zone o f Ghana. For maize farmers, 10.0% were in the coastal zone, 56.9% in the forest 9 U rban ind ica tes lo ca litie s inhab ited by m o re than 5000 p eo p le (P opu la tio n s ize g re a te r than 5000 ) 10 R ural ind ica tes lo ca litie s inhab ited by less o r equal to 5000 p eo p le (P opu la tio n s ize less than o r equal to 5 0 00 ) 54 University of Ghana http://ugspace.ug.edu.gh zone and 33.1% in the savannah zone. The high percentage of farm household heads in the rural forest zone is not surprising as it supports the view that the production o f major staple crops is mostly done in the forest zone11. 4.3.2 Age-sex structure As shown in table 4.3, male farmers outnumbered their female counterparts in the two different farm activities studied. This might be due to the fact that the sample dwelt mainly on household heads alone excluding their spouses. With the exception o f farmers involved in cassava production, male heads were more than female heads when all age groups are considered. Female heads in cassava production whose ages were in the 41-45 category and between the ages of 56 and 60 were more than their male counterparts. Cassava farmers are concentrated in ages 36-40 and 46-50. The broad age bracket of 36-50 accounts for 46 per cent o f cassava farmers. For maize farmers, their ages mostly fell in the 36-40 bracket. Table 4.3: Age-Sex Distribution of Farmers by Farm activity Age Farm Activity Group Cassava Maize Male Female Total Male Female Total 15-20 1.1 1.1 1.0 1.0 21-25 1.7 1.7 3.4 6.9 0.3 7.2 26-30 6.9 5.7 12.6 12 .1 1.7 13.8 31-35 6.9 4.6 11.4 10.7 2.4 13.1 36-40 11.4 4.6 16.0 12.8 4.1 16.9 41-45 5.1 9.7 14.9 9.3 2.8 12.1 46-50 8.0 8.0 16.0 7.6 3.4 1 1 .0 51-55 5.1 4.0 9.1 5.2 2.1 7.2 56-60 4.0 5.7 9.7 7.6 2.8 10.3 61-65 3.4 2.3 5.7 4.5 2.8 7.2 Total 53.7 46.3 100.0 77.6 22.4 100.0 Source: Author’s calculation from GLSS 4 data (2000). We observe a lower concentration within the age groups o f 15-30. The percentage of farmers in that age group was 17.1 for cassava and 22 for maize. The percentages o f older farmers (51-65) in cassava and maize production were 24.5 and 24.7 respectively. This is not surprising since most 11 See S ijm J. (1 9 9 3 ) 55 University of Ghana http://ugspace.ug.edu.gh studies12 have reported this phenomenon and this should be a major concern for decision-makers. It is thus evident that farming or for that matter production o f major staples crops like cassava and maize have been left to the older folks. It is believed that though they gain experience as they stay on the farms for long they however tend to be late adopters and are more likely to stick to their primitive ways o f farming at the expense o f increasing productivity. 4.3.3 Formal School Attendance Information in Table 4.4 shows the number of farmers in each farm activity distributed over gender who had attended school before and those who had never attended school. The percentage of cassava and maize fanners who responded as having attended formal schooling was 73.7 percent and 63.8 percent respectively. Male household heads that reported as attending school before outnumbered their female counterparts. Table 4.4: Frequency Distribution of Farmers Attending Formal Schooling by Farm Activity _______________________________________ Percentages_____________________________________ Farm Activity Cassava Maize Male Female Total Male Female Total Been to school before 82 47 129 160 25 185 No schooling 12 34 46 65 40 105 Total 94 81 175 225 65 290 Source: Author’s calculation based on GLSS 4 data. Although the agricultural research system in Ghana has generated several improved technologies for food crop production (Obaatanpa, high resistant cassava mosaic virus varieties), the adoption rate o f these innovations among smallholders has generally been slow. This situation has been partly blamed on the low education level or attainment o f the average Ghanaian farmer. 12 See A n ro oy (1 9 9 7 ) and S ijm (1993 ) 56 University of Ghana http://ugspace.ug.edu.gh 4.3.4 Educational Attainment Table 4.5 shows the educational level completed by household heads by activity and sex. Males, who had been to school before, dominated all in the two-farm enterprise activities. Out o f 129 cassava farmers who said they had attended school before, 63.6 percent were males and 36.4 percent were females. For maize farmers, 86.5 percent o f the total (185) were males and 13.5 percent were females. About 90.5 percent o f household heads who had been to school before were males with the rest being females. In relative terms, females had less education than their male counterparts though female numbers in the sample were small. The disparity in educational attainment between male and female household heads for the two farm enterprises is more pronounced at higher levels o f schooling completed with one male maize farmer completing at the university level. As shown in table 4.5, none o f the female heads in the two farm enterprises had completed schooling beyond the senior secondary school level with more than 95 percent completing up to the middle school level. Generally, the educational level completed by cassava and maize farmers were low, with about 90 percent o f them completing up to only basic education. This supports the view that the agricultural workforce in the production o f staple foods and cash crops are in the hands o f mostly illiterate and semi-illiterate older farmers. This has resulted in the production o f these crops suffering from the injection of innovations and increased productivity. 57 University of Ghana http://ugspace.ug.edu.gh Table 4.5: Level of Education completed by Household heads by Sex and Farm Enterprise. Highest CASSAVA MAIZE level completed M A LE FEM A LE TOTAL % M A LE FE M A L E TOTAL % N om e” 4 ( 3 .1 ) 4 ( 3 .1 ) 6.2 13 ( 7 .0 ) 3 ( 1 .6 ) 8.6 K in d e rg a r te n 6 (4 .7 ) 5 ( 3 .9 ) 8.6 9 ( 4 .9 ) 2 (1 .1 ) 6.0 P r im a ry 16 ( 1 2 .4 ) 16 ( 1 2 .4 ) 24.8 33 ( 1 7 .8 ) 10 ( 5 .4 ) 23.2 M id d le 4 4 (3 4 .1 ) 21 ( 1 6 . 3 ) 50.4 81 ( 4 3 .8 ) 9 (4 .9 ) 48.7 JS S 2 (1 .6 ) 1.6 12 ( 6 .7 ) 1 ( 0 .5 ) 7.2 SS S 1 (0 .8 ) 0.8 4 (2 .2 ) 2.2 V o c ./C o m m . 2 (1 .6 ) 1.6 S e c o n d a r y ‘0 ’ 5 (3 .9 ) 3.9 5 ( 2 .7 ) 2.7 T e a c h e r ’s T r a in in g - T e c h n ic a l 1 ( 0 .8 ) - 0.8 1 (0 .5 ) 0.5 N u rs in g 2 ( 1 .6 ) 1.6 P o ly t e c h n ic U n iv e r s i ty 1 ( 0 .5 ) 0.5 K o ra n ic - 1 ( 0 .5 ) 0.5 O th e r - T o ta l 82 (63.6) 47 (36.4) 100.0 160 (86.5) 25 (13.5) 100.0 Source: Author’s estimation from GLSS 4 data (Statistical Service 2000) Figures in parentheses are percentages o f the total. a. No level completed though respondent reported of been to school before The findings compare well with conclusions made by Anrooy (1997), Jollife (1998) and Sijm (1993). 58 University of Ghana http://ugspace.ug.edu.gh 4.3.5: Extension Service Contact Table 4.6 shows the percentages o f farmers who had come into contact with any form o f extension service by farm activity. More than 90 percent o f cassava and maize farmers said they had no form of assistance from MoFA, NGOs and co-operatives. Table 4.6: Extension Service Contact Response Farm Activity Percentage Responding No Yes Total Cassava 90.12 9.88 100 Maize 91.03 8.97 100 Source: Author’s computation from GLSS 4 Data (2000) Only 9.9 percent o f cassava farmers and about 8.97 percent o f maize farmers had their input sources from the MoFA, NGOs and Co-operatives and therefore some form o f extension sendee contact. Extension service contributes to improving farm productivity through informal education of farmers, but if we have just about 10 percent o f major staple crop farmers having access to extension service through input acquisition then this calls for a careful look by policy makers. 4.4 Farm Productivity Regression Functions 4.4.1 Some Econometric issues Careful consideration has been made to omit outliers from the data, in addition to the transformation o f some o f the variables to achieve normality and linearity. We nevertheless expect some o f the variables used in the estimation, to violate some of the assumptions underlying least squares estimation. The principal one is the problem o f non-constant variance (Heteroscedasticity). The two-stage design o f the GLSS sample typically results in the rejection o f the assumption of homoscedasticity because observations drawn from within a cluster are likely to have 59 University of Ghana http://ugspace.ug.edu.gh characteristics that are more similar than observations drawn from different clusters. The difference between intra-cluster correlations will most likely result in heteroscedasticity. The presence of heteroscedasticity causes OLS parameter estimates to be imprecise and their estimated standard deviation to be biased. Heteroscedasticity was tested in all regression estimations by performing the White and Goldfeld - Quandt test13 to find out if the variance o f the disturbance term varies systematically with one or more o f explanatory variables. In all cases we found that heteroscedasticity was a problem. The dependent and other explanatory variables were transformed into logarithmic form and OLS was used to estimate the parameter coefficients. As stated by Jollife (1998) and Vijverberg (1995), the logarithmic transformation lacks flexibility as a method to deal with heteroscedasticity in addition to the skewed spread in farm enterprise size. The study reports the OLS estimates with t-ratios based on heteroscedasticity-consistent standard errors and covariance. One way o f dealing with heteroscedasticity is to use the variable that is causing the heteroscedasticity or has greater variability with the disturbance term as a weight variable in Weighted Least Square (WLS) estimation14. With the exception o f the age and the square o f ages of farmers, correlation coe ffic ien tsestim ated were low and not significant. Multicollinearity was assumed not to be a problem in the estimation since the standard errors were also low in all the regression estimates. See K ou tsoy iann is , A . (1 9 96 ) - rep rin t; W hite , H. (1980 ); M adda la , G .S . (1988 ) See K ou tsoy iann is , A , (1 9 9 6 ) - rep rin t, p ages 189 and 190; N o ra s is, M . J. (1988 ) 15 S ee app end ix 2 fo r e s tim a ted co rre la tio n coeffic ien ts 60 University of Ghana http://ugspace.ug.edu.gh 4.4.2 Basic production function results: Cassava Results o f Ordinary Least Squares (OLS) and Weighted Least Squares (WLS) are shown in Table 4.7. Equation A l and A2 under both OLS and WLS are full estimation results with site and family background variables (see Liu et al, 1999) as control variables. Equations labelled B1 and B2 under both estimators are estimated coefficients when site dummies and highly insignificant explanatory variables are removed. Results shown in the table are estimations based on the production function equation 3.15. Table 4.7: OLS and WLS estimate of the C-D Production Function: Cassava (Without Education) Variable Coefficients Estimates OLS WLS A l B1 A2 B2 Constant 1.155 (0.366) 4.160 (2.600)*** 1.204 (0.748) 1.080 (0.739) Ln_Lab 0.350 (0.945) 0.274 (1.745)* 0.624 (2.952)*** 0.445 (2.174)** Ln_Hlab 0.202 (1.730)* 0.129 (1.425) 0.259 (1.628)* 0.315 (2.013)** Ln C.fert 0.090 (0.734) 0.109 (0.930) Ln_Htool -0.121 (-0.670) -0.042 (-0.283) -0.194 (-1.065) -0.046(-0.243) Ln_Capinpts -0.049 (-0.496) 0.137 (1.248) Ln_FIhsize -0.245 (-0.720) -0.236 (-1.248) Agey 0.094 (1.251) -0.002 (-0.191) 0.004 (0.357) -0.002(-0.267) Agesq -0.001 (-1.223) -0.001 (-1.143) Sex_l 0.172 (0.776) 0.148 (0.793) 0.118 (0.452) Rural 0.182 (0.613) -0.154 (-0.429) Forest 0.194 (0.571) -0.403 (0.760) Coastal -0.009 (-0.023) Pwork 0.408 (1.498) 0.426 (1.862)* 0.476 (1.656)* 0.452 (1.478) R2 0.26 0.24 0.95 0.95 Adj.R2 0.14 0.20 0.94 0.94 F-stats (Prob) 2.25 (0.01) 5.67 (0.000) 184.40 (0.000) 406.79 (0.000) N 107 135 107 135 S o u r c e : A u t h o r ’s c o m p u ta t io n f r o m G L S S 4 d a ta , 2 0 0 0 D e p e n d e n t v a r i a b l e is n a tu ra l lo g a r i th m o f c a s s a v a o u tp u t p e r a c r e . F ig u r e s in p a r e n th e s e s a r e t - r a t io s . S ta r s in d ic a te s ig n i f i c a n c e u s in g a tw o ta i le d t - t e s t a s f o l lo w s : * * * = 0 .0 1 ; ** = 0 .0 5 ; * = 0 .1 0 . N a tu r a l lo g o f fa rm s iz e w a s u s e d a s w e ig h te d v a r ia b le ( P r in c ip a l c a u s e o f h e t e r o s c e d a s t i c i t y ) 61 University of Ghana http://ugspace.ug.edu.gh Most coefficients o f the direct inputs into the production o f cassava were found to be insignificant. Likewise, household composition variables like household size age of household, age-square and gender were found to be insignificant. With the removal of the site dummy variables and highly insignificant variables, the OLS estimates show a 10% level o f significance for family labour, and father’s occupation variable. Output elasticity reduces in the case o f family labour but the dummy variable indicating the main occupation of the household heads’ father had a positive effect on output per acre, all other inputs held constant compared to the case where the father is not a farmer. Output elasticity o f hired labour reduces from 0.20 to 0.13 and becomes insignificant. In equation Al under OLS, the explanatory variables explain only 20% o f the variation in output per acre o f cassava. The R2 reduces to 24% in equation B 1. A test o f heteroscedasticity using White and Goldfeld - Quandts test indicated that land size 16 was the principal cause, in addition to it having a non-zero covariance with the estimated residuals. As a form o f correction, land size was used as a weighted variable in WLS estimation. The results from the WLS estimate show a better F-statistic and R2, with the explanatory variables explaining 95% of the variation in cassava output per acre. The full estimation results o f WLS in equation A2 indicate a positive high output elasticity o f family labour (0.62), which is statistically significant at the 1% level when all other inputs are held constant. Output elasticity o f hired labour cost is statistically significant at 10% level meaning that cassava productivity is very dependent on labour input from the family and amount spent on hired labour o f farmers. Aggregated value of fertiliser and agro-chemicals and the cost of capital inputs have a positive but insignificant effect on cassava output per acre. Apart from the dummy coefficient o f father’s occupation, which remains positive and significant at 10% level, the remaining variables included were found to be not significant with a change in sign for site dummies. With the exclusion o f site 16 T o ta l land s ize used as a sca le v a riab le 62 University of Ghana http://ugspace.ug.edu.gh dummy variables and highly insignificant coefficients, we observe that labour input from the family and cost o f hired labour remains as the most important factors explaining the variations in cassava farm productivity. In addition to hand tools cost, age o f farmer and the dummy signifying the fathers’ occupation, they explain about 95% of the variation in farm productivity (see equation B1 and B2). Low R2 obtained in OLS estimation is explained by the fact that cassava production at the traditional level is influenced by a number of unobserved variables such as access to credit, farm motivation, soil quality, (which are not dealt with in this study). 4.4.3. Basic Production Function Results: Maize Table 4.8 shows the production function estimates obtained by using the OLS and WLS estimators. The dependent variable is the natural log o f maize output divided by the acres o f land farmed. Equation A l incorporates all the basic explanatory variables, including household composition and site dummy variables. Surprisingly, family labour results in less maize output per acre obtained, though the effect is not significant. Among the household composition variables, only household size significantly affects the output o f maize per acre. Output per acre of maize increases by 0.42% as household size goes up by 1% when all other inputs are held constant. The negative coefficient on age of farmer suggests that older farmers are less productive and that productivity declines at a falling rate as the head gets older. This is only true if the head provides the main farm labour. Households headed by males produced more maize per acre (0.41) than female headed households. This may be because males spent more quality time on the farm and do most o f the hard labour intensive activities on the farm. The dummy variables, indicating the locality of farmer and the ecological zone he/she resides had a positive effect upon farm productivity. However, the forest dummy is not significant. As compared with those in savannah areas (omitted variable category), coastal farmers had a reduction in output by 0.41 %. 63 University of Ghana http://ugspace.ug.edu.gh Table 4.8: OLS and WLS estimate of the C-D Production Function: Maize (Without Education) Coefficients Estimates OLS WLS Variable A l B1 A2 B2 Constant 2.566 (1.392) 0.462 (0.430) -0.700(-0.491) 0.538 (0.409) LnJLab -0.158 (-1.388) 0.101 (0.930) 0.166 (1.312) 0.216 (1.814)* Ln_Hlab 0.322 (4.970)*** 0.368 (5.254)*** 0.287 (4.137)*** 0.344 (4.483)*** Ln Cfert 0.115 (1.941)* 0.054 (0.867) 0.108 (1.459) Ln_I [tool -0.085 (-0.790) -0.009 (-0.098) Ln_Inputs 0.161 (2.831)*** 0.086 (1.583) 0.207 (2.995)*** 0.132 (1.950)* Ln_Hhsize 0.420 (3.074)*** 0.021 ( 0.154) 0.045 (0.359) Agey -0.076 (-1.540) -0.033 (-0.845) -0.041 (-0.769) -0.036 (-0.775) Agesq 0.001 (1.403) 0.000 (0.530) 0.000 ( 0.565) 0.443 (2.055)** Sex_l 0.413 (2.021)** 0.322 (4.970)*** 0.389 ( 1.600) 0.000 (0.419) RuralDum 0.472 (2.365)** 0.442 (1.342) ForeDum 0.211 (1.551) 0.399 (2.718)** CoastDum -0.412 (-2.253)** -0.424 (-1.265) Pwork 0.181 (0.904) 0.319 (1.662) 0.139 (0.535) R2 0.46 0.41 0.95 0.95 Adj.R2 0.40 0.38 0.94 0.94 F-stats 8.814 (0.000) 11.008 (0 .000) 218.87 (0.000) 396.60 (0.000) (Prob) N 151 151 151 151 S o u rc e : A u th o r ’s c o m p u ta t io n f r o m G L S S 4 d a ta , 2 0 0 0 D e p e n d e n t v a r i a b l e is n a tu r a l lo g a r i th m o f m a iz e o u tp u t p e r a c r e . F ig u r e s in p a r e n th e s e s a re t - r a t io s . S ta r s in d ic a te s ig n i f i c a n c e u s in g a tw o t a i l e d t - t e s t a s f o l l o w s : * * * = 0 .0 1 ; ** = 0 .0 5 ; * = 0 .1 0 . N a tu ra l lo g o f f a rm s iz e w a s u s e d a s w e ig h te d v a r ia b le ( P r in c ip a l c a u s e o f h e t e r o s c e d a s t i c i t y ) Overall, the included regressors explain 46 percent o f variations in maize output per acre. When site dummies and other highly significant variables are omitted, only hired labour, sex dummy variable and fathers occupation were the only significant variables that affects maize output per acre. While output per acre elasticity of hired labour was 0.37, male heads were likely to increase output by 0.32% compared to their female counterparts holding all other inputs. 64 University of Ghana http://ugspace.ug.edu.gh With the removal o f household size variable the coefficient on natural logarithm o f family labour family turn positive (but insignificant) suggesting that household size may be correlated with fam ily labour. When family labour was instrumented with household size and other variables the coefficient was not significant in all cases with a low coefficient o f determination (R2). To correct for heteroscedasticity, land holdings in acres was used to weight the regressors and the estimated coefficient are presented under WLS in table 4.8. The test for heteroscedasticity under OLS rejected the null hypothesis o f homoscedasticity. Output elasticities o f hired labour cost and value o f other capital inputs were low (0.28 and 0.21 respectively) but statistically significant at 1% level. Male farmers in the forest zone o f Ghana were found to be more productive (0.4% more output on one acre) than those residing in the savannah. The Cobb-Douglas production function estimated by employing WLS are able to explain 95 percent o f the variation in maize output produced per unit land area cultivated. When site dummies and highly insignificant regressors were omitted from equation A2, the coefficient on family labour becomes significant at 10% level with an output elasticity o f 0.22. This compare well with elasticities estimated by Croppenstedt and Muller (2000), Weir (1999) and Chinn (1979). The coefficient for hired labour cost is positive and statistically significant at 1% level. This means that both family labour and hired labour costs are most important factors in increasing productivity o f maize farmers. The value o f other capital inputs has a positive significant effect at least at the 10% level with an output elasticity o f 0.13. Male farmers are found to be more efficient than their female counterparts as they produce 0.4% more output on an acre of farm size compared to female heads. 65 University of Ghana http://ugspace.ug.edu.gh The study tried to correct endogeneity problems by using instrumental variables in piace of potentially endogenous variables. The predicted values o f variables such as family labour and age- squared are used in a second stage o f the two-stage least squares (2SLS), but the results yielded insignificant coefficients. 4.4.4 Effect of Formal Education on Farm Productivity: Cassava The estimated coefficient under WLS estimated for cassava is presented in Table 4.9. Equation 1 in Table 4.9 includes the years o f formal schooling completed by the household head farmer as an input into production and also extension service contact dummy o f 1 -0 , which represents non- formal education. In equation 2 o f Table 4.9, the years o f formal schooling is replaced with dummy variable indicators signifying the level o f schooling completed by the cassava farmer. Both equations are based on equation 3.17 developed earlier with a different variant o f education measure used in each case. Output elasticity of family labour is positive and statistically significant at the 5% level when the specified equation includes the years of schooling completed. The significance level decreases when dummy indicators are introduced in equation 2. A 1 percent increase (decrease) in family labour input induces an increase (decrease) in cassava farm productivity by 0.5% as in equation 1 or only 0.3% in equation 2 with all other inputs held constant. The cost o f hired labour is positively associated with farm productivity and the effect is highly significant. The only possible explanation is that cassava farmers who spent more on hired workers tend to be efficient in production. This assertion may not hold if the hired workers were only recruited during the harvesting period and not at the early stages of land preparation, planting and weeding. 66 University of Ghana http://ugspace.ug.edu.gh Table 4.9: WLS estimates of the C-D Production Function: Cassava (With Education) Coefficients Estimates Equ. 1 Equ. 2 Variable Estimated Coefficient t-ratios Estimated Coefficient t-ratios Constant 0.508 0.313 4.555** 2.000 Ln_Lab 0.501** 2.353 0.344* 1.683 Ln_Hlab 0.379** 2.495 0.270*** 2.661 Lnjrltool -0.135 -0.652 -0.098 -0.578 Agey 0.004 1.480 -0.108 -1.346 Agesq 0.001 1.451 Sex_l - -0.153 -0.584 Schyrs 0.036 1.079 Eduprim 0.585* 1.957 Edumid -0.239 -0.964 Edujss -0.804 -1.710 Edusec 1.433*** 3.348 Edupsec -2.114** -2.026 Aveschyrs 0.010 0.737 Extserv -0.200 -0.768 -0.338* -1.768 Rural Yes (insig) Yes (insig) Forest Yes (insig) Yes (insig) Coastal Yes (insig) Yes (insig) Pwork ■ 0.556* (1.700) 0.454 (1.570) R2 0.94 0.98 Adj.R2 0.93 0.97 F-stats (Prob) 223.740 (0.000) 140 (0.000) N 103 103 S o u rc e : A u t h o r ’s c o m p u ta t io n f r o m G L S S 4 s u r v e y d a ta ( 2 0 0 0 ) D e p e n d e n t v a r i a b l e is n a tu r a l lo g a r i th m o f c a s s a v a o u tp u t p e r a c r e . F i g u r e s i n p a r e n t h e s e s a r e t - r a t i o s . S ta rs in d ic a te s ig n i f i c a n c e u s in g a t w o ta i l e d t - t e s t a s f o l l o w s : * * * = 0 .0 1 ; ** = 0 .0 5 ; * = 0 .1 0 . N a tu ra l lo g o f f a rm s iz e w a s u s e d a s w e ig h te d v a r i a b l e (P r in c ip a l c a u s e o f h e t e r o s c e d a s t i c i t y ) Y e s - v a r ia b le i n c lu d e d in e q u a t io n ; in s ig - n o t s ig n i f ic a n t . All the other regressors were not significant and therefore do not seem to improve farm productivity. The value o f aggregated fertiliser and agro-chemical input, and the value o f other capital inputs were found to be always insignificant with wrong signs and were therefore dropped from the final analysis because they did not contribute to the overall goodness o f fit o f the regression estimates. 67 University of Ghana http://ugspace.ug.edu.gh Household composition variables all had insignificant effects on farm productivity. Estimated coefficient o f formal years o f schooling completed by the household head farmer is positive (0.036) meaning that returns to schooling is positive but insignificant. Additional year o f education has very minimal and insignificant effect in cassava farm productivity. OLS estimates17, which are reported with t-statistics based on heteroscedasticity-consistent standard errors, had a much lower coefficient o f years o f schooling completed (0.002). The results support the view that farmers who employ traditional technology are most likely to be deficient in skills, knowledge and ability to be more efficient. As discussed under section 4.3 and 4.4 many o f the cassava farmers sampled had no or very minimal education and this explains the insignificant effect o f the education variable or captured by the years o f schooling completed by the farmer. In equation specification 2, we find that cassava farm productivity increases by 0.6% when household head farmer has completed primary education as compared to the one who had not received any schooling with all other inputs held constant. The effect o f education dummy variable for cassava farmers with secondary education is positive and statistically significant at the 1% level. We find that cassava farmers who had completed secondary education were more productive and output per acre increase by 1.4% compared to the base case. The generally held view is that those with higher education have negative attitude towards farming and are likely to move into off-farm activities. The results from Table 4.10 however show that farmers with high education up to secondary school level are associated with very high outputs per acre. The negative coefficient on the indicator dummy variable signifying middle school completion is unexpected. It is expected that the completion of middle school by farmers will greatly enhance the farmers’ ability and skill to perceive things differently from those with no schooling. The result is therefore difficult to explain. The OLS estimates however show a positive but insignificant effect. Cassava farmers with junior 17 See a p p end ix 3 fo r e s tim ated resu lts 68 University of Ghana http://ugspace.ug.edu.gh secondary completion and post secondary were found to be less productive though the negative effects were found to be insignificant. Extension service coefficient is negative and significant at the 10% level. Farm productivity in cassava production decreases by -0.3% when farmer has extension service contact than where there is no contact through agricultural input acquisition. This is quite in sharp contrast with the generally held view that technical efficiency increase when farmers are introduced to new and modern inputs and technology and they adopt them. The result may be due to the fact that most o f the farmers sampled obtained their input themselves and therefore responded in the negative when interviewed. Average schooling years completed by other adult household members apart from the head had positive but insignificant effect on productivity o f cassava farms. Fathers’ background occupation dummy was only positive and significant in equation specification 1 , but insignificant in specification 2. Productivity increases by 0.45 to 0.56 percent when the heads father had farming as his main occupation, holding all other inputs constant. Overall, the explanatory variables explain 94% o f the variation in productivity of cassava farms in equation 1 and 98 percent in equation 2. Percentage increase in farm productivity or output per acre due to the addition o f one extra year of schooling is provided in appendix 5. Percentage increase in the case o f equation specification 2 o f Table 4.9 is calculated for only primary and secondary education completion. The results are compared with studies reviewed by Lockheed et al (1980); Phillips (1994) and Weir (1999). The effect o f an additional year o f schooling for the household head who is a cassava farmer is to increase output o f cassava per acre by 3.7%. In general, the results compare quite favourably with studies reviewed by Lockheed et al (1980), which range from -3.3% to 6.5% and those o f Phillip (1994), which ranged from - 3.1% to 8.4%. Weir (1999) however obtained lower response rates of between 1.0% and 2.0% for farmers. For farmers who had completed primary education, the 69 University of Ghana http://ugspace.ug.edu.gh estimated increase in cassava output per acre for one additional year o f education is 13.25% and for secondary education productivity increases by 2 1 .2%. 4.4.5 Effect of Formal Education on Farm Productivity: Maize In the following discussion o f the estimation results, the study focuses primarily on the effect of schooling and extension contact obtained from the Weighted Least Square (WLS) regression and comparison with results obtained by using the Ordinary Least Squares (OLS) estimator. The WLS estimation in table 4.10, show better results with respect to the significant levels o f the estimated coefficients, the R2 values and the F-statistics as compared to the OLS results18 Family labour was found to be insignificant but positive in both equations. Low output elasticities were associated with family labour (0.08 and 0.09). Output elasticities o f hired labour cost, the value o f fertiliser and other agro-chemicals and all other capital inputs were significant (at the 5% level) and were positively associated with maize productivity. Household composition variables were found to affect maize output per acre significantly except in the second specification where the significance o f household size is lost when completed years of formal schooling is replaced by dummy indicators for each level o f schooling completed. While household size is positively associated with maize productivity, we found a negative coefficient on age, which suggests that maize farmers become less productive as they get older. Experience o f the farmer as captured by the square o f their ages has an estimated coefficient o f near zero meaning that experience o f the farmer as captured in this way does not impact on maize farm productivity. Male farmers in maize production were found to be more productive than their female counterparts. 18 See a p p end ix 3 fo r O LS es tim ated re su lts fo r m aize 70 University of Ghana http://ugspace.ug.edu.gh Table 4.10: WLS estimates of the C-D Production Function: Maize (With Education) Variable Coefficients Estimates Equ. 1 Equ.2 Estimated Coefficient T-ratios Estimated Coefficient T-ratios Constant 0.302 0.180 1.445 1.186 Ln_Lab 0.056 0.543 0.088 1.000 LnJHlab 0.371*** 4.792 0.226** 2.602 Ln_Cfert 0.128* 1.928 0.157** 2.472 Ln_Capinpts 0.172** 2.615 0.226*** 3.253 Ln_Iihsize 0.387*** 2.788 0.151 0.941 Agey -0.096* -1.893 -0.092** -1.984** Agesq 0.000* 1.728 0.001 1.635 Sex_l 0.420** 2.2 10 0.398** 2 .2 0 1** Schyrs 0.030 1.194 Eduprim 0.015 0.605 Edumid 0.206 1.573 Edujss -0.427 -0.937 Edusec 0.115 0.346 Edupsec 0.346 0.878 Aveschyrs 0.010 0.493 0.017 0.734 Extserv -0.356 -1.376 -0.514** -2.007** Rural Yes insig Yes (sig) Forest Yes insig Yes (insig) Coastal Yes insig Yes (insig) Pwork 0.402 1.810* 0.414 (1.917)* R2 0.50 0,47 0.96 0.96 Adj.R2 0.43 0.39 0.95 0.95 F-stats (Prob) 6.880 (0 .000) 6.164 180.680 (0 .000) 170.890 (0 .000) (0 .000) N 1 1 0 150 1 10 150 S o u rc e : A u th o r ’ s c o m p u ta t io n f r o m G L S S 4 s u r v e y d a t a ( 2 0 0 0 ) D e p e n d e n t v a r i a b l e is n a tu r a l lo g a r i th m o f m a iz e o u tp u t p e r a c r e . F i g u r e s i n p a r e n t h e s e s a r e t - r a t i o s . S ta rs in d ic a te s ig n i f i c a n c e u s in g a tw o ta i l e d t - t e s t a s f o l l o w s : * * * = 0 .0 1 ; ** = 0 .0 5 ; * = 0 .1 0 . N a tu ra l lo g o f f a rm s iz e w a s u s e d a s w e ig h te d v a r ia b le (P r in c ip a l c a u s e o f h e t e r o s c e d a s t i c i t y ) Maize productivity was found to increase by about 0.4 percent, ceteris paribus among male-headed households than female headed households. The value of hand tools and dummy variables used to represent the locality and ecological zone of maize farmers all had insignificant coefficients and were therefore omitted from the final estimation. Under the WLS estimation we find that an additional year of schooling for household heads that are maize farmers induce an increase in maize productivity by 0.03 percent but this was found not to be 71 University of Ghana http://ugspace.ug.edu.gh significant. The return to formal schooling among maize farmers in Ghana is found to be near zero. The result compares well with several other results that employ data on farmers who are primarily involved in traditional agriculture (Lockheed et al, 1980). The coefficient o f formal education on maize farm productivity does not differ much from that obtained under OLS. Huffman (1974) has attributed small estimated effects o f schooling on crop yields or gross farm output to limited technical-efficiency gains from skills provided by farmers’ education. Taylor and Yunez-Naude (2000) have argued that such studies do not take into account the technological changes and sectoral diversification characterising agricultural transformations in less developed countries (LDCs). Due to the traditional nature o f maize farming in Ghana and the fact that most o f the farmers sampled for the study had only basic education, it is believed that the positive “worker effect” o f formal years o f schooling completed will be more significant if the choice o f or selection into activities is closely studied by looking at the allocative effects. Indicator dummy variables representing different categories or levels o f schooling completed had insignificant coefficients in equation specification 2 o f Table 4.10. While dummy indicators signifying completion o f primary education and middle school had positive coefficients both in the OLS and WLS estimation, that for secondary and post secondary had negative coefficients in the OLS estimation but turned positive except for JSS category in the WLS estimation. The results are characterised by low returns to schooling for the levels though none was significant. Only the coefficient o f middle school completed was significant at the 12% level (not a conventional practice). The percentage increase in productivity of maize farm is 0.20 percent, ceteris paribus for a farmer who had completed middle school as compared with the farmer who had received no education. Extension service contact dummy had a significant negative coefficient at the 5% level. This suggest a reduction in maize productivity levels by 0.5% for a farmers who had extension service 72 University of Ghana http://ugspace.ug.edu.gh contact compared to those who had no contact by the way o f inputs, training or advice from NGOs, MOFA extension agents or any co-operative society. This is surprising and in contrast with the generally held view that farmers who have come into contact with extension agents are expected to be more productive than those whose input sources are private. The only explanation is that most of the farmers sampled reported “no” to having any links with MOFA extension agents, NGOs or co-operatives by way o f acquisition o f relevant inputs. Probably, farmers were not probed enough during the interview as they are likely to report “no” extension contact so they will be included in any list of names o f farmers they erroneously think may be targeted for help. Other studies such as Jamison and Lau (1978), Hopcraft, (1974), Pachico and Ashby (1976) have also reported negative coefficients o f extension on productivity. Evidence o f extension contact and formal education interaction however gave mixed results in most o f the studies reviewed by Lockheed et al (1980) and Philip (1994). We therefore interact extension contact and formal schooling dummies in another specification (see Table A.4b in appendix 4), which is estimated using only the WLS method. The coefficient on the interaction effect is positive but insignificant and turns negative when years o f schooling completed by the head and dummy variable signifying extension service contact are excluded from the production function. The independent effect o f completed years of schooling improves and becomes significant (at the 10% level). We found a significant influence o f maize farmers’ family background dummy variable (fathers occupational background was used as a proxy) on farm productivity. This means that all other things being equal, farmers whose fathers’ main occupation was farming are likely to have an increase in productivity o f 0.4% compared to those whose fathers’ were not farmers. Farmers who are more productive may have spent a greater part of their formative years on the farm and have at least learnt a lot o f skills (at least in traditional way) in making good use o f available inputs at their disposal. The estimated coefficient on average years of formal schooling completed by adult household members is positive but does not play any significant role in raising the productivity of maize farm enterprise o f the household head. 73 University of Ghana http://ugspace.ug.edu.gh Results o f percentage increase in maize output are shown in appendix 5. The estimated increase due to one extra year o f schooling at the mean educational level o f the sample is 3.0%. We estimate the percentage increase only for farmers who have completed middle school. The result show that farmers with one extra year o f schooling above middle school are likely to have increases in maize output per acre by 2.34%. In summary, the WLS estimates provide a better goodness o f fit o f the estimated regression equation model and significance of estimated coefficients. Education as captured by years o f formal schooling completed had no significant effect on farm productivity but there is strong evidence o f threshold effect for schooling. Productivity increases are associated with farmers with primary and secondary education. Extension service contact however was negatively associated with farm productivity. In theory, there is a complementary relation between extension and educational attainment in a production function. As stated by Moock (1981), farmers with a higher educational attainment would have an expanded capacity to understand, interpret in light o f specific conditions, and apply on their own farms the lessons offered by the extension agent(s). WLS Estimation results of interaction effects are provided in Table A.4a under appendix 4. In equation 1 o f Table A.4a, we allow educational attainment to interact with extension service contact without removing their individual effects. The estimated coefficient is positive (0.470) but insignificant. There is no significant improvement in the independent effects o f head’s completed years o f schooling (schyrs) and extension service contact. In equation 2 of Table A.4a, the independent effects o f schooling years completed by the head and extension dummy variables were not significant. There is a negative (-0.646) and significant interaction effect on farm productivity, which is not predicted from theory and therefore difficult to explain. The education (years of schooling completed) of household heads who are maize farmers have no significant effect on productivity in their farms so far as output per land area cultivated is concerned. Only those who have completed middle school 74 University of Ghana http://ugspace.ug.edu.gh were found to have an effect on farm productivity compared with the base case of no schooling (only at 1 2 % level o f significance). 75 University of Ghana http://ugspace.ug.edu.gh 4.5 Full Estimation Results for Farm Net Income Ordinary Least Square (OLS) and Weighted Least Square (WLS) estimation results for maize farm net income are presented in Table 4.11. The estimates are based on equation model 3.20. Due to insufficient information on cassava input price data, farm income regression was only possible for maize. Table 4.11: Results of farm income function estimation: Maize Coefficients Estimates OLS WLS Variable Al B1 A l B2 Constant 5.883 (2.589)** 4.789 (2.697)*** 5.906 (2.676)*** 4.259 (2.392)** Ln_Mprice 0.058 (0.556) 0.052 (0.590) 0.098 (0.864) 0.035 (0.349) Ln Fprice -0.052 (-0.265) -0.039 (-0.254) -0.019 (-0.092) -0.049 (-0.303) Ln_Tprice -0.153 (-0.521) -0.018 (-0.081) -0.268 (-0.782) 0.046 (0.192) Ln_Laba 0.097 (1.462) 0.051 (0.937) 0.135 (1.591) 0.085 (1.223) Ln_Landa 0.394 (4.256)*** 0.309 (4.181)*** 0.471 (3.396)*** 0.338 (3.335)*** Sex_l 0 . 1 1 1 (0.945) 0.112 (1.385) 0.040 (0.275) 0.024 (0.249) Schyrs 0.017 (1.548) - 0.014 (1.212) Educprim - 0.127 (1.338) 0.075 (0.569) Educmid - 0.155 (2.070)** 0.161 (1.965)** Educjss - 0.030 (0.202) - 0.059 (0.361) Educsec 0.241 (1.336) 0.229 (1.169) Educpsec - 0.235 (0.683) 0.279 (2.167)** Aveschyrs 0.003 (0.365) Extserv R2 0.066 (0.421) 0.068 (0.602) 0.169 (1.408) 0.093 (1.142) 0.22 0.20 0.91 0.92 Adj. R2 0.17 0.15 0.90 0.91 F-statistic 4.624 3.829 178.764 182.232 (Prob) (0 .000) (0 .000) (0 .000) (0 .000) N 154 213 154 213 S o u rc e : A u th o r ’s c o m p u ta t io n b a s e d o n G L S S 4 ( 2 0 0 0 ) S u r v e y d a ta . F ig u r e s in p a r e n th e s e s a r e t - r a t io s . S ta r s in d ic a te s ig n i f i c a n c e u s in g a tw o ta i l e d t - t e s t a s f o l l o w s : * * * = 0 .0 1 ; * * = 0 .0 5 ; * = 0 .1 0 . A g e o f f a rm e r w a s u s e d a s a w e ig h t v a r ia b le in W L S e s t im a t io n . It w a s t h e p r in c ip a l c a u s e o f h e te r o s c e d a s t i c i ty a n d th e r e f o r e o m i t t e d f ro m th e r e s u l t s s h o w n in t h e ta b le , a. F ix e d in p u ts i n c lu d e d in t h e f u n c t io n 76 University of Ghana http://ugspace.ug.edu.gh Variable definitions are given in Table 4.1. The estimated farm income equation is specified in two ways. First, the education o f the household head is specified as the years o f schooling completed and the results o f OLS and WLS estimation are provided under equation A l and A2 respectively. Second, dummy variable indicators signifying different levels o f schooling completed by the household head are specified in equation B1 and B2 in Table 4.11. The estimated coefficients o f output price o f maize and input prices were statistically insignificant but had the expected signs in all cases except the coefficient o f hand tool price in equation B2 under WLS. This means that though maize price is positively associated with farm income and input prices have an inverse relationship with farm income, they do not play any significant role in determining the net farm income for maize farmers among the sample. There are two fixed input variables included in the equation specification. Family labour coefficient is positive in all cases but was found not to be statistically significant. However the estimated coefficients and their levels o f significance were better under the WLS estimation. The effect o f land area cultivated under maize was positive and significant at the 1% level. Under the WLS estimation, a 1 percent increase (decrease) in land area cultivated, all other things being equal induces an increase (decrease) in maize farm income between 0.34 to 0.47 percent. Education o f the household head measured as the years o f formal schooling completed by the household head does not significantly affect maize farm income under the WLS estimation but was found to be significant only at the 13% level (not a conventional practice). Even then the estimated coefficient o f formal education was found to be low and almost approaching zero. In equation B1 and B2, where, formal education of the head is captured by employing dummy variable indicators, we found that all the estimated coefficients for all the indicators are positive. All the coefficients were not statistically significant under OLS estimation except the indicator signifying completion of middle school. When all other inputs are held constant, maize farm net income goes up by 0.2 77 University of Ghana http://ugspace.ug.edu.gh percent when the household head has completed middle school compared with those who have not received any education. This means that return to middle school education is better than not receiving any formal education. When the equation is estimated by weighting variables which show much variability, the percentage increase in maize farm income ceteris paribus for the farmer who has completed post secondary education is 0.28 (significant at the 5% level). Also the percentage increase in maize farm income is 0.16, all things being equal for farmers that have completed middle school. The coefficient o f average years o f schooling of other household members was insignificant and therefore dropped from the final analysis. The coefficient o f dummy variable indicator signifying whether the farmer has received any extension service contact or not was positive but insignificant in all specified equations estimated. WLS estimation provides better goodness o f fit for the model specified. The F-Statistic is highly significant meaning that the explanatory variables jointly influence the changes observed in the dependent variable. Overall, 92 percent o f the variation observed in maize farm net income is explained by the explanatory variables included in the model. In summary, we find that the coefficient for middle school completion is significant (OLS and WLS), together with dummy indicator for post secondary education. The evidence thus suggests that middle school and post secondary completion is positively related to farm income. Maize farm income also increases (decreases) as land area cultivated under maize increase (decrease). 4.6 Education, Productivity, and Food security issues Food security is achieved when all people at all times have access to sufficient food for a healthy and productive life (Haddad, 1997). Siamwalla and Valdes (1984), however define food security as the ability o f countries, regions or households to meet their required levels o f food consumption at 78 University of Ghana http://ugspace.ug.edu.gh all times. Food security has three main components: food availability, food access, and food utilisation. Food availability refers to the need to produce sufficient food in a way that generates income for small-scale producers while not depleting the natural resource base, and to the need to get this food into markets for sale at prices that consumers can afford. The second component relates to people’s ability to get economic access to food. Economic access is typically constrained by income: If households cannot generate sufficient income to purchase food, they lack and entitlements to that food. The final component concerns an individual’s ability to use food consumed for growth, nutrition, and health: in an environment lacking clean water, sanitation, child care, and health facilities, the ability to use food to promote health and nutrition will be impaired, particularly for young infants. Many countries aspire to achieve food security, which is not synonymous with food self- sufficiency. Self-sufficiency in food supply refers to the internal capacity o f a country or household to supply all the food it requires, whereas food security does not require self-sufficiency in food supply. Food security can be achieved by importing food or buying food, as long as one has the capacity to import additional food or buying it from others without involving undue supply risks. We have supply-side and demand-side factors determining food security in the country. Supply of food is affected by the national resource endowment, availability o f technology and its dissemination (for food production, storage and preservation), prices, market opportunities and ability to augment own production with external supplies when the need arises. On the demand- side, food security is affected by household incomes and economic assets (including stock o f animals), prices, demographic factors such as number, gender and age composition o f households. 79 University of Ghana http://ugspace.ug.edu.gh and socio-cultural factors like health and sanitation status, educational level, cultural norms and food composition habits. Food Security can be achieved through increased agricultural productivity. But agricultural productivity is low among traditional peasant Ghanaian farmers who apply low productivity- enhancing inputs such as fertilisers and agro-chemicals. Though growth in agricultural production and increases in agricultural productivity have been the main objectives o f the agricultural policies of the past and present governments, sustained success has eluded policy-makers on both counts (Nyanteng and Seini, 2000). Achieving food security would ultimately lead to adequate food consumption and export o f surpluses. Food security in its entire dimension cannot be achieved in a vacuum. It can only be achieved if food crop producers and livestock farmers are technically efficient in their production. But high cost o f inputs, low application o f these inputs and slow rate of adoption among others, constrain productivity. Education has long been an investment good in the economy o f many nations. In several developed countries less than 5% o f the Economically Active Population (EAP) are involved in agriculture but they are able to produce more for home consumption and for export. The level o f education of farmers in these countries plays a major role in agricultural production. The situation is quite different in Ghana, where more than 50% o f the EAP are into agriculture and its related activities but food security attainment is a problem with occasional shortages of food mostly in the northern savannah areas o f the country. This can lead to transitory food insecurity, which could be temporary or transient, arising from temporary shortfalls in food supply relative to requirements, or because o f a temporary loss o f adequate effective purchasing power for food. The factors that lead to temporary food insecurity include seasonal and annual fluctuation in food production, natural catastrophes, temporary loss o f employment etc. It is believed that the more educated the producer is, the more he or she is likely to move into the production of high income earning crops or animals whenever any of these causes are prevalent in an area (Philips et al, 1990). According to Okai (1997) a positive technological intervention is 80 University of Ghana http://ugspace.ug.edu.gh needed to improve the performance o f the agricultural sector through recombination o f resources. The ability to do this can only come from farmers who have more knowledge in modern farming and can shift resources from the production o f one crop to the other. In Ghana, agriculture accounts for about 60-70% o f the labour force as stated earlier in the study and contributes about 40 percent o f GDP but budgetary outlays are inadequate. Smallholder agriculture dominates with most o f the farming population in the rural area. Although technological research has come out with improved modern methods o f farming, adoption rates are low among farm households in Ghana. Even those who have adopted these modern farm methods are unable to expand their holdings due to a number o f reasons. 81 University of Ghana http://ugspace.ug.edu.gh CHAPTER FIVE SUMMARY, CONCLUSION AND RECOMMENDATION 5.1 Summary and Conclusion Education is a key point o f entry for inducing a sequence o f changes, which benefits the rural farmer. A major effect o f education on production is that it enhances the ability o f a producer to acquire or decode information about the productive characteristics of new inputs. Therefore, the more rapid the flow o f new inputs, the greater will be the productivity differential associated with additional education. Low productivity in agriculture in Ghana has been attributed partly to low- skilled labour force, illiterate and semi-illiterate farmers who have persisted in the use o f traditional farm implements and methods in farming. To fully understand the influence that education of an individual farmer has on productivity on his or her farm we have estimated and quantified the contributions o f improved education and exposure to extension training on cassava and maize farm productivity and net farm income from these enterprises. A Cobb-Douglas production function has been specified with education variables as inputs. Due to the problem o f heteroscedasticity, which makes the OLS estimator inefficient, we apply Weighted Least Squares (WLS) method in the estimation. However, OLS estimation results are reported after transformation. For household heads that are cassava farmers, education, as measured by years o f formal schooling completed had no significant effect on farm productivity, though the effect is positive. Increase in productivity is however associated with farmers who have obtained primary or secondary education. Primary and secondary schooling tends to enhance efficiency in cassava production by 82 University of Ghana http://ugspace.ug.edu.gh providing skills, which are useful in terms o f using available inputs at their disposal. Extension service contact, however was negatively associated with farm productivity. This may be due to the fact that most o f the farmers sampled did acquire their own inputs. It must be stated here that the proxy used for extension contact was only used because of the absence o f a direct question in the survey questionnaire that required respondents to either indicate whether they had any extension contact or not. However, percentage increase in cassava output is as much as 3.7 for one year increase in schooling above the mean education level o f the sample. The effect o f an additional year of schooling for the household head who has completed schooling at the primary level is to increase cassava output by 13.3 percent. For secondary schooling, output increases by as much as 21.2 percent. Schooling years completed by maize farmers who are household heads does not significantly improve productivity. Only the coefficient o f middle school completed was significant at the 12% level (not a conventional practice). For these farmers, at least some middle school education is necessary for there to be positive and significant benefits o f schooling. Extension service was found not to enhance productivity in any way. The percentage increase in maize output for one additional year o f education at the mean educational level o f the sample is to increase output by 3.1 percent. For additional year o f schooling above the middle school level with 10 years o f education, maize output is estimated to increase by 2.3 percent. Given the traditional nature o f maize farm technology in Ghana and the fact that most o f the farmers sampled had completed education only up to the basic level, it is not surprising that formal education’s contribution to farm productivity is insignificant. The estimated farm income function indicated that returns to schooling are highest for farmers with middle and post secondary education. Having access to extension service does not significantly result in an increase in maize farm income. 83 University of Ghana http://ugspace.ug.edu.gh 5.2 Policy Recommendation Based on the findings from the study, the following recommendations are made for policymakers in the education and agriculture sector and further research into education and productivity relationships. First, the transient food security problem facing the Sub-Saharan Africa region and especially the northern part o f Ghana demonstrates the need for informed policymaking. The government should pursue macroeconomic policies that do not tax agriculture but provide special incentives for the educated youth. Their attraction into the production o f staple food crops that the nation has comparative advantage in could help the nation in our quest to achieve food security. Second, a strong case is made for government or donor intervention to encourage higher levels o f investment in basic and secondary education in rural Ghana. Increases in budgetary allocation to the education sector would be in the right direction as the rate o f return to investment in education is high compared with other investments. Third, smallholder staple crop farmers should be targeted for special assistance in the area of informal or non-formal education to enable them improve on their efficiency on the farm. Finally, since different levels o f education have different impacts on farm productivity, it is convenient to conclude that the nation can achieve food security if farmers with certain levels o f education are assisted or supported to increase their efficiency in production. Given effective demand of households and their endowments, production decisions need to be taken by knowledgeable heads in the family in order to increase productivity and to acquire income so as to increase their purchasing power. It is said that the key to food security is sustainability therefore there is the need to ensure that, available resource at the disposal o f the farmer is utilised efficiently. To attain food 84 University of Ghana http://ugspace.ug.edu.gh security through increased farm productivity, Ghanaian farmers should have the required skills and knowledge in modern farming methods and be able to comprehend simple instructions on the use of modern farm inputs. 85 University of Ghana http://ugspace.ug.edu.gh REFERENCES: Abban J.B. (1986). Prerequisite o f manpower and educational planning in Ghana. Baffour Educational Enterprises Ltd. Accra. Afari, E. (1998). “Resource-use efficiency in plantain production in Ghana: A case study of Atwima District in the Ashanti Region.” An unpublished dissertation submitted to the Department o f Agricultural Economics and Farm Management, University o f Ghana, Legon. Aigner, D. J„ C. A. Lovell and P. Schmidt. (1977). “Formulation and Estimation o f Stochastic frontier production function models.” Journal o f Econometrics, 6, 21-37 Ali, Mubarak, and Derek Byerlee. (1991). “Economic efficiency o f small farmers in a changing world: a survey of recent evidence.” Journal o f International Development, 3, 1-27 Amemiya, Takeshi. (1983). “Nonlinear Regression Models.” In Handbook o f Econometrics, vol.l edited by Zvi Griliches and Michael Intriligator. Amsterdam: North-Holland. Anrooy van Raymon, (1997). “Commercialisation in Ghana: A comparative study o f maize, cocoa and cassava farmers.” An unpublished MSc thesis submitted to the Department of Economics and Management Wageningen Agricultural University. Appiah-Kubi, K., N.N.N. Nsowah-Nuamah, G.J.M. van den Boom. (2001). “Returns to Education and Experience in Ghana, 1987-1999: Evidence from four rounds o f the Ghana Living Standards survey.” Centre World Food Studies. Staff working paper. Appleton, Simon, and Arsene Balituta. (1996). “Education and Agricultural productivity: evidence from Uganda.” Journal o f International Development, 8, 415-444. Arrow, K.J. (1973). “Higher education as a filter.” Journal o f Public Economics (I) 2:193-216 Asenso-Okyere, W.K., Benneh, G and Tims, W., (1997). “The Status of Food Security in West Africa.” In Sustainable Food Security in West Africa. Kluwer Academic Publishers. Dordrecht, The Netherlands. 86 University of Ghana http://ugspace.ug.edu.gh Asenso-Okyere, N.N.N.Nsowah-Nuamah, FA . Asante. (2000). “Situation o f Education and Training in Ghana.” Human Development Resource, Reseau Ghaneen o f SADAOC FOUNDATION, Institute of Statistical, Social and Economic Research (ISSER). University o f Ghana, Legon. Asuming-Brempong, S. (1987). “Comparative Advantage and Rice Policy in Ghana.” Unpublished MSc. thesis submitted to the Faculty o f the Graduate school, University o f Philippines, Los Banos. Awolola, M. D. (1998). “Use o f Agrochemicals in Nigeria: Farmer's Education, Farm Size, and Income as Determinants”. Unpublished Article pp 62-68. Ibadan, Nigeria Azariadas, C., and A. Drazen. (1990). “Threshold externalities in economic development.” Quarterly Journal o f Economics. 105 (2): 501-526 Barro, R. J. (1991). “Economic Growth in a Cross-Section.” Quarterly Journal o f Economics. 106(2): 407-444 Battese, G.E. and G.S. Corra (1977). “Estimation o f a production frontier model: with application to the pastoral zone o f Eastern Australia, Australian.” Journal o f Agricultural Economies, 21, 169-79 Battese, G. E. and Tim J. Coelli, (1988). “Prediction o f firm-level technical efficiencies with a generalised frontier production function and panel data”, Journal o f Econometrics. 38, 387- 99. Becker, G. (1964). Human Capital: A Theoretical and Empirical Analysis, with Special Reference to Education. New York: Columbia University Press. Becker, G. (1993). Human Capital: A Theoretical and Empirical Analysis, with Special Reference to Education. Third Edition, Chicago. The University o f Chicago Press. Behrman Jere R., Barbara L. Wolfe and David M. Blau (1985) “Human Capital and Earnings Distribution in a Developing Country: The Case of Pre-Revolutionary Nicaragua” In 87 University of Ghana http://ugspace.ug.edu.gh Economic Development and Cultural Change. Volume 34, Number 1. University of Chicago Press. Behrman Jere R., and Anil B. Deolalikar. (1988) Health and Nutrition. In Handbook o f Development Economics, vol. 1. Elsevier Science Publishers B.V. Chaudri, D. P. (1979). “Education, Innovations and Agricultural Development: A study o f Northern India”. A submitted mimeo (London: Croom Helm Ltd. for the International Labour Organisation). Colclough, C. 1980. "Primary Schooling and Economic Development: A Review o f the Evidence." World Bank Staff Working Paper No.399. Washington, DC. Cotlear, D. 1986. "Farmer Education and Farm efficiency in Peru: The Role o f Schooling, Extension Services, and Migration." EDT Discussion Paper 49. World Bank, Population and Human Resources Department. Washington, D.C. Cotlear, Daniel, (1990). “The Effects o f Education on farm productivity” . In Keith Griffin and John Knight, eds., Human Development and International Development Strategy for the 1990s.(London: MacMillan) Croppenstedt Andre and Christophe Muller (2000) “The Impact o f Farmers” Health and Nutritional Status on Their Productivity and Efficiency: Evidence from Ethiopia”. In Economic Development and Cultural Change. Pp. 476-502. University o f Chicago Press. Croppenstedt, Andre and Mulat Demeke. (1997). An empirical study o f cereal crop production and technical efficiency of private farmers in Ethiopia: a mixed fixed-random coefficients approach, Applied Economics, 29, 1217-1226. Croppenstedt, Andre, Mulat Demeke and Meloira M. Meschi. (1998). “Technology adoption in the presence o f constraints: The case of fertiliser demand in Ethiopia”, mimeo (Oxford: Centre for the study o f African Economies). University of Ghana http://ugspace.ug.edu.gh Croppenstedt, Andre and Christophe Muller, (1998). The impact o f health and nutritional status of farmers on their productivity and efficiency: Evidence from Ethiopia, mimeo (Oxford: Centre for the study o f African Economics). Darko David (1998). “The economic Implications o f Agricultural Input Subsidy Removal on Rice Production: A case study o f Dawhenya Irrigation Project” . Unpublished BSc. Agriculture dissertation submitted to the Department o f Agriculture Economics and Farm Management, University o f Ghana, Legon. Denison, Edward F., (1967). “Why Growth Rates Differ: Post-war Experience in Nine Western Countries”. Washington, D.C.: Brookings Institution. Chinn, Dennis L. (1979) “Rural Poverty and the Structure o f Farm Household Income in Developing Countries: Evidence from Taiwan” in Economic Development and Cultural Change. Volume 27, Number 2 University o f Chicago Press. Diwan K. Romesh (1971). “Impact o f Education on Labour Efficiency.” in Applied Economics volume 3, 127-135. Eisemon, J. Schwille and R. Prouty. (1990). “Empirical Results and Conventional Wisdom: Strategies for Increasing Primary School Effectiveness in Burundi”. BRIDGES Research Report Series. April. Cambridge, MA: Harvard University. FAO. (1998). Food and Agriculture Organisation. Trade Year Book. Rome. Fafchamps, Marcel and Agnes R. Quisumbing. (1998). “Human Capita, Productivity, and Labour allocation in rural Pakistan”. Food Consumption and Nutrition Division (FCND) Discussion Paper No. 48. International Food Research Institute. Washington D.C., U.S.A Ghana Vision 2020, (1995). The First Medium-term Development Plan (1997-2000), Accra, Ghana. July 1997 National Development Planning Commission (NDPC). Ghana Living Standards Survey (GLSS 4), 2000. Report of the Fourth Round. Statistical Service of Ghana. Accra. 89 University of Ghana http://ugspace.ug.edu.gh Glewwe, P. and Twum-Baah K.(1990). “Schooling skills and the returns to Education: An econometric exploration using data from Ghana”. LSMS Working Paper 84, World Bank. Greene, William H. (1990). Econometric Analysis. New York: Macmillan Publishing Co. Greene, William H. (1993). “The econometric approach to efficiency analysis.” In Fried, Harold O., C.A. Knox Lovell, and Shelton S. Schmidt (eds.) The Measurement o f Productive Efficiency: Techniques and Applications. (New York: Oxford University Press) Griliches, Zvi. (1964). “Research expenditures, education and the aggregate agricultural production function”, American Economic Review, 54, 961-74. Haddad, L. (1997). Achieving Food Security in Southern Africa: New challenges, New O ppo rtun itie sEds Lawrence Haddad. International Food Policy Research Institute (IFPRI). Washington, D.C. Henderson James M. and Richard E. Quandt. (1980). Microeconomic theory. A Mathematical Approach. 3rd Edition. Me Graw-Hill Book Company. Hopcraft, Peter N. (1974). “Human Resources and Technical Skills in Agricultural Development: An Economic Evaluation o f Educative Investments in Kenya’s Small-Fann Sector.” Huffman, Wallace E. (1974). “Decision-making: the role o f education”, American Journal o f Agricultural Economics, 56, 85-97 Hussain, S. and D. Byerlee, (1995). “Education and farm productivity in post-‘green revolution’ agriculture in Asia.” In G.H. Peters and Douglas D. Hedley, eds., Agricultural competitiveness: Market forces and Policy choice, Proceedings o f the 22nd International conference of Agricultural Economists held in Harare, Zimbabwe (Aldershot: Dartmouth Publishing Company Limited), 554-69. ISSER, (1998). The State o f the Ghanaian Economy. Institute of Statistical, Social and Economic Research (ISSER), University of Ghana, Legon. Accra. 90 University of Ghana http://ugspace.ug.edu.gh ISSER/ SOW-VU (1999). HRD Project Foreign Assistance. Centre for World Food Studies. Stichting Onderzoek Wereldvoedselvoorziening van de Vrije Universiteit (SOW-VU) Activity Report. Jamison, D. and L. Lau. (1982). “Farmer Education and Farm Efficiency”. Baltimore, MD: Johns Hopkins University Press for the World Bank. Jamison, Dean T. and Peter, Moock. (1984). "Farmer Education and Farm Efficiency in Nepal: The Role o f Schooling, Extension Services and Cognitive Skills." World Development 12:67-86. Jin-Tan Liu, James K. Hammitt and Chyongchiou, Jeng Lin, (1999). “Family Background and returns to schooling in Taiwan”. In Economics o f Education Review. Elsevier Science Ltd. Jollife, D., (1998). “ Skills, Schooling, Household Income in Ghana in The World Bank Economic Review. Volume 12, Number 1. Pages 81-104. Jondrow, J., C. A. K. Lovell, I. S. Materov, and P. Schmidt. (1982). “On the estimation o f technical inefficiency in the stochastic frontier production function model”. Journal o f Econometrics, 19, 233-38 King, T., ed. (1980). Education and Income. World Bank Staffing Paper No.402. Washington, DC Knight, J. and R, Sabot. (1987). "Educational Policy and Labour Productivity: An Output Accounting Exercise". The Economic Journal, 97 (March): 199-214. Kislev, Voav. (1967). “Estimating a Production Production Function from U.S. Census of Agriculture Data” . Ph.D. dissertation, University o f Chicago. Koutsoyiannis, A., (1996). Theory o f Econometrics: An Introductory Exposition o f Econometric Methods. Second Edition (re-printed). ELBS, Macmillan Press Limited, Hampshire, Great Britain. Lau, L., D. Jamison and F. Louat. (1991). "Education and Productivity in Developing Countries: An Aggregate Production Function Approach." Policy, Research and External Affairs Working Paper 612. World Bank, Office o f the Vice President Development Economics, and the Population and Human Resources Department. Washington, DC. 91 University of Ghana http://ugspace.ug.edu.gh Lele, U. (1990). “Agricultural Growth and Assistance to Africa: Lessons o f a Quarter Century”. Sector Studies No. 2, International Centre for Economic Growth/ICS Press Publication, San Francisco. Lockheed, M.E; Jamison, D.T; and Lau, L.J; (1980). “Farmer Education and Farm Efficiency: A Survey” in Economic Development and Cultural Change Volume 29, Number 2. University o f Chicago Press. Lovell, C. A. Knox. (1993). “Production frontiers and production efficiency” . In Fried, Harold O., C. A. Knox Lovell, and Shelton S. Schmidt (eds.) The Measurement o f Productive Efficiency: Techniques and Applications (New York: Oxford University Press). Lovell. C. A. Knox, Finn R, Farsund, and Peter Schmidt. (1980). “A survey o f frontier production functions and o f their relationship to efficiency measurement” . In Journal o f Econometrics 13(1980) 5-25. North-Holland Publishing Company. Lucas, R. E. (1963). “Making a miracle.” Econometrica 61(2): 251-272. Lucas, R. E. (1988). “On the mechanics o f economic development.” Journal o f Monetary Economics. 22(1): 3- 22. Maddala, G.G., (1988). Introduction to Econometrics. New York: Macmillan Publishing Co. Meeusen, W. and J. van den Broeck. (1977). “Efficiency estimation from Cobb. Douglas production functions with composed error” . In International Economic Review , 18, 435-44. Mincer, J. (1958). “Investment in Human Capital and Personal Income Distribution.”, Journal o f Political Economy. New York. Mincer, J. ( 1974f_Schooling, Experience and Earnings. New York: NBER. Mincer, J. (1989). “Human capital and the Labour market: a review o f Current Research”. Educational researcher (May): 27-34. 92 University of Ghana http://ugspace.ug.edu.gh Moock, P.R. (1981). “Education and Technical Efficiency in Small-Farm Production” in Economic Development and Cultural Change Volume 29, Number 4. University o f Chicago Press. Moock, P. (1994). "Education and Agricultural Productivity". International Encyclopaedia of Education 1:244-254. Oxford: Pergamon Press. National Budget (2001). Government o f Ghana National Budget for the year 2001. Presented by the Minister o f Finance to Parliament, Accra. Nyanteng, V. K. (1994). “Structural Adjustment and Agriculture in Ghana”. In Structural Adjustment and African Agriculture. Africa Series No.6 . Institute o f Developing Economics. Nyanteng, V. K. and S.K. Dapaah. (1997). “Agricultural Development: Polices and options”. In Policies and Options fo r Ghanaian Economic Development.. Institute o f Statistical, Social and Economic Research. University of Ghana, Legon. Nyanteng, V.K. and A.W. Seini. (2000). “Agricultural Policy and the Impact on Growth and Productivity, 1970-95.” In Economic Reforms in Ghana: The Miracle and the Mirage (Eds) Aryeetey, E., J. Harrigan and M. Nissanke. James Currey, Oxford; Woeli Publishing Services, Accra; Africa World Press, Trenton, NJ. Norusis, M. J. (1988). SPSS-X Advanced Statistics Guide. Second Edition. In association with SPSS incorporated. Chicago, USA Nube', M., G.J.M. van den Boom and Asenso Okyere, W. K., (1999). Food Insecurity and Seasonal Variation in Body Mass Index in the Semi-arid Tropics. Staff Working Paper. Centre for World Food Studies. (SOW-VU). Okai, M. (1997). “Agricultural Production, Food Security and Poverty in West Africa.” In: Asenso- Okyere et al. (eds.), Sustainable Food Security in West Africa, Kluwer Academic Publishers, Dordrecht: The Nertherlands. 93 University of Ghana http://ugspace.ug.edu.gh Pachico, Douglas Ii., and Ashby, Jacquiline A.(1976). “Investments in Human Capital and Farm Productivity: Some Evidence from Brazil.” Unpublished Paper, Cornell University, Ithaca, N.Y. Peaslee, A. (1965). “Elementary Education as a Pre-requisite for Economic Growth.” International Development Review 7. Peaslee, A (1969). “Education's Role in Development.” Economic Development and Cultural Change 11(3): 293-318. University o f Chicago Press. Philips, Joseph M. (1994). “Farmer Education and Farm Efficiency: A Meta-Analysis.” In Economic Development and Cultural Change. Volume 43, Number 1.University o f Chicago Press. Phillips, Joseph M., and Robert P. Marble, (1986). “Farmer education and efficiency: a frontier production function approach.” Economics o f Education Review>, 5, 257-264. Philips P. Truman and Daphne S. Taylor, (1990). “Optimal Control of Food Insecurity: A Conceptual Framework” . In American Journal o f Agricultural Economics Political Economy 88(4): 639-652. PPMED of MOFA (1998). “The agriculture sector in Ghana” : Policy Planning, Monitoring and Evaluation Department (PPMED) o f Ministry o f Food and Agriculture (MoFA) Accra. Psacharopoulos, G. (1984). “The contribution o f education to economic growth: international comparisons”. In International comparisons ofproductivity and causes o f slowdown, ed. J.W. Kendrich. Cambridge, Mass., U. S. A.: Ballinger Publishing Company. Psacharopoulos, G. (1985). "Returns to Education: A Further International Update and Implications." Journal o f Human Resources V o l... (4): 583-597. Psacharopoulos, George and Maureen Woodhall. (1985). “Educational Development: An analysis o f Investment choices” . Oxford University Press. Psacharopoulos, G. (1993). "Returns to Investment in Education: A Global Update". Working Paper Series N o .1067. World Bank, Office of the Director, Latin America and the 94 University of Ghana http://ugspace.ug.edu.gh Caribbean. Washington, DC. Ram, R. and R. Singh (1988). Farm households in rural Burkina Faso: some evidence on allocative and direct returns to schooling and male-female labour productivity differentials. World Development. 16,419-424. Reardon, T. V. Kelly, E. Crawford, T. Jayne, K. Savadogo and D. Clay (1996). “Determinants of Farm productivity in Africa: A system o f four case studies” ; International Development. Page No. 22. Romer, P.M. (1986). “Increasing returns and long-run growth” . Journal o f Political Economy 94 (5): 1002-1037. Romer, P.M. (1990). “Endogenous technological change” . Journal o f Political Economy 98 (5): S71-S102. Rozenweig Mark R. (1995). “Why are there Returns to schooling? American Economic Review: Papers and proceedings o f the American Economics Association 85 (2) 153-58. Sankhayan, P.L. (1988). Introduction to the Economics o f Agricultural Production. Prentice-Hall of India Private Limited. New Delhi. Schultz, Theodore W. (1964). “Transforming Traditional Agriculture” (New Haven: Yale University Press) Schultz. T.W. (1971). "Investment in Human Capital." In M. Blaug, ed., Economics o f Education. Middlesex, England: Penguin Books. Schultz, T.W. (1972). “A 'Guide' to Investors in Education with Special Reference to Developing Countries” . In K. Thompson and F. Champion, eds., Education and Development Reconsidered. Volume 2. Bellagio, Italy: Rockefeller Foundation and Ford Foundation. Schultz, T. W. (1975). ‘The ability to deal with disequilibria’. Journal o f Economic Literature. 13: 827-846. Siamwalla, A. and Valdes, A. (1984). Food Security in Developing countries’ International Issues: 95 University of Ghana http://ugspace.ug.edu.gh In: Etcher, C.K. and Staaz, J.M. (eds), Agricultural development in the third vmrld. Baltimore: Johns Hopkins University Press Sijm J. (1993). Food Security and Policy Interventions in Ghana. Tinbergen Instituut, Amsterdam. Singh, I.. L. Squire, and J. Strauss, eds. (1986). Agricultural household models. Extensions, Applications, and Policy. Baltimore, M.d. U.S.A.: John Hopkins University Press for the World Bank. Strauss, John, and Duncan Thomas. (1995). “Human Resources: Empirical Modelling o f Household and Family Decisions”. In Jere Behrman and T. N. Srinivasan, eds., Handbook o f Development Economics, Volume 3. Amsterdam: North-Holland. Strauss, J. (1990). “Household, Communities, and Pre-school Children’s Nutritional Outcomes: Evidence from Rural Cote D ’Ivoire.” Economic Development and Cultural Change. University of Chicago Press. Taylor J. E. and Yunez-Naude A. (2000). “The Returns from Schooling in a Diversified Rural Economy’. American Journal o f Agricultural Economics. 82: 287-297. Thomas, D. (1989). “Intra-Household Resource Allocation: an Inferential Approach.” New Haven: Yale University, mimeo. Varian, Hal R. (1999). Intermediate Microeconomics. A Modern Approach. 5Ul Edition. W. W. Norton and Company Inc., New York, London. Villaume, J. (1977). “Literacy and Agricultural Innovation in Kenya.” Ph.D. thesis. Harvard University, Cambridge, MA. Vijverberg Wim P.M. (1995) “ Returns to Schooling in Non-Farm self-employment: An Econometric Case Study o f Ghana” in World Development. Vol. 23, No 7 pp 1215-1227 Elsevier Science Ltd Weale, Martin. (1993). “A critical Evaluation of Rate o f Return analysis”. Economic Journal University of Ghana http://ugspace.ug.edu.gh 103 (418): 729-37 Weir, S. (1999). “The effects o f Education on Farmer productivity in Rural Ethiopia” . CSAE Working Paper series 99-7 (Oxford: Centre for the study of African Economies.) Welch, F. (1970). “Education in Production”. Journal o f Political Economy 78 (January-February) : 35-39. Welch, F. (1978). “Minimum wages: Issues and evidence” . Washington, D.C: American Enterprise Institute. White, Halbert (1980). “ A heteroscedasticity-consistent covariance matrix estimator and a direct test for heteroscedasticity.” Economelrica. Vol. 48, Number 4, pp 817-838. World Bank. (1986). Financing Education in Developing Countries: An Exploration o f Policy Options. Washington, DC. World Bank. (1990). World Development Report: Poverty. New York: Oxford University Press. World Bank. (1991). “ Republic of Ghana: Community Secondary Construction Project.” Staff Appraisal Report 9556-GH. African Region, Country Department IV, Washington D.C. World Bank. (1993). “The East Asian Miracle: Economic Growth and Public policy” A World Bank Policy Research Report. New York: Oxford University Press. World Bank. (1995). Development in Practice. Priorities and Strategies fo r Education. A World Bank Review. The World Bank, Washington, D. C. Wu, Craig C. (1977). “Education in farm production: the case o f Taiwan.” American Journal o f Agricultural Economics, 59, 699-709. 97 University of Ghana http://ugspace.ug.edu.gh APPENDICES Appendix 1 Conversion Factors The GLSS 4 survey data recorded the units o f land area cultivated (farm size) and weight o f farm products in local units. Following Anrooy (1997), the different units are converted into international units o f measurement. Variation in the recorded local units o f area in the data is converted into acres The conversion factors used are as follows: 1 Rope = 0.125 acre 1 Pole =0.225 acre Variation in recorded units o f weight in the data is converted into kilograms. Cassava has the following conversion factors: 1 Minibag = 91.00 kg 1 Maxibag = 182.00 kg 1 Basket 1 Bowl 1 Stick 22.75 kg 2.00 kg 0.50 kg Maize has the following conversion factors: 1 Minibag 1 Maxibag 1 Basket 1 Bowl 1 American Tin = 50.00 kg = 100.00 kg = 12.50 kg = 2.70 kg = 0.50 kg 98 University of Ghana http://ugspace.ug.edu.gh Table A2a: Correlation Coefficients between Variables: Cassava Appendix 2 OTP LAB IILB 1NP IITO CF LND HSIZ AGE AGSQ SEX SCH ASCII PRIM JSS MID SEC PSEC EXT FB 1.00 0.03 0.02 0.03 0.06 0.17 -0.15 0.01 0.17 0.14 0.01 0.03 0.02 0.08 -0.16 0.15 0.23 -0.42 -0.18 -0.04 0.03 1.00 0.15 0.25 0.65 0.39 -0.61 0.49 -0.02 -0.03 0.08 0.20 0.09 -0.23 0,11 0.30 -0.13 0.08 -0.06 -0.07 0.02 0.15 1.00 0.46 0.51 0.53 -0.10 0.21 0.25 0.25 0.41 0.14 -0.13 0.00 -0.03 -0.04 0.13 0.12 0.30 -0.12 0.03 0.25 0.46 1.00 0.52 0.47 -0.41 0.06 0.16 0.14 0.08 -0.07 0.17 0.18 -0.08 0.01 -0.23 0.03 0.15 -0.04 0.06 0.65 0.51 0.52 1.00 0.63 -0.53 0.23 0.14 0.12 0.18 0.09 0.00 -0.02 0.03 0.13 -0.19 0.13 0.07 0.01 0.17 0.39 0.53 0.47 0.63 1.00 -0.32 0.31 0.18 0.17 0.19 0.29 0.22 -0.02 -0.03 0.19 -0.07 0.21 0.30 -0.16 LND -0.15 -0.61 -0.10 -0.41 -0.53 -0.32 1.00 0.07 -0.01 -0.01 -0.01 0.00 -0.05 0.18 0.08 -0.33 0.30 0.02 0.15 0.07 J HSIZ 0.01 0.49 0.21 0.06 0.23 0.31 0.07 1.00 0.03 0.01 0.04 0.31 0.20 0.03 0.21 0.04 0.12 0.12 0.03 -0.20 AGE 0.17 -0.02 0.25 0.16 0.14 0.18 -0.01 0.03 1.00 0.99 -0.02 0.11 -0.01 -0.13 -0.27 -0.05 0.13 0.29 -0.11 0.13 AGSQ 0.14 -0.03 0.25 0.14 0.12 0.17 -0.01 0.01 0.99 1.00 -0.01 0.12 -0.01 -0.16 -0.22 -0.04 0.12 0.31 -0.12 0.12 SEX 0.01 0.08 0.41 0.08 0.18 0.19 -0.01 0.04 -0.02 -0.01 1.00 0.21 -0.19 -0.04 0.08 0.00 0.15 0.12 0.22 -0.02 SCH 0.03 0.20 0.14 -0.07 0.09 0.29 0.00 0.31 0.11 0.12 0.21 1.00 0.15 -0.23 0.04 0.46 0.38 0.40 -0.01 -0.11 AS CH 0.02 0.09 -0.13 0.17 0.00 0.22 -0.05 0.20 -0.01 -0.01 -0.19 0.15 1.00 -0.05 -0.01 0.21 -0.03 -0.04 -0.08 -0.22 PRIM 0.08 -0.23 0.00 0.18 -0.02 -0.02 0.18 0.03 -0.13 -0.16 -0.04 -0.23 -0.05 1.00 -0.08 -0.51 -0.15 -0.12 0.09 0.15 JSS -0.16 0.11 -0.03 -0.08 0.03 -0.03 0.08 0.21 -0.27 -0.22 0.08 0.04 -0.01 -0.08 1.00 -0.14 -0.04 -0.03 -0.06 -0.30 MID 0.15 0.30 -0.04 0.01 0.13 0.19 -0.33 0.04 -0.05 -0.04 0.00 0.46 0.21 -0.51 -0.14 1.00 -0.24 -0.20 -0.10 -0.10 SEC 0.23 -0.13 0.13 -0.23 -0.19 -0.07 0.30 0.12 0.13 0.12 0.15 0.38 -0.03 -0.15 -0.04 -0.24 1.00 -0.06 0.16 -0.09 PSEC -0.42 0.08 0.12 0.03 0.13 0.21 0.02 0.12 0.29 0.31 0.12 0.40 -0.04 -0.12 -0.03 -0.20 -0.06 1.00 -0.08 0.11 EXT -0.18 -0.06 0.30 0.15 0.07 0.30 0.15 0.03 -0.11 -0.12 0.22 -0.01 -0.08 0.09 -0.06 -0.10 0.16 -0.08 1.00 0.03 FB -0.04 -0.07 -0.12 -0.04 0.01 -0.16 -0.07 -0.20 0.13 0.12 -0.02 -0.11 -0.22 0.15 -0.30 -0.10 -0.09 0.11 0.03 1.00 99 University of Ghana http://ugspace.ug.edu.gh Table A2b: Correlation Coefficients between Variables: Maize O TP LAB HLB INP H TO C F LND H SIZ AGE AGSQ SEX SCH ASCH PR IM JSS M ID SEC PSEC EXT FB i . OTP 1.00 0.23 0.58 0.39 0.31 0.41 -0.19 0.24 -0.14 -0.14 0.26 0.06 0.05 -0.02 -0.10 0.06 0.07 -0.06 -0.05 0.00 ii- LAB f 0.23 1.00 0.31 0.22 0.50 0.39 -0.65 0.45 0.24 0.22 -0.02 -0.04 0.14 -0.08 -0.09 0.13 -0.13 -0.04 -0.07 -0.10 IpHLB 0.58 0.31 1.00 0.48 0.43 0.54 -0.33 0.07 0.03 0.02 0.05 -0.07 0.06 0.06 -0.06 0.00 0.02 -0.13 -0.07 -0.08 fJN P 0.39 0.22 0.48 1.00 0.46 0.45 -0.27 0.08 0.01 0.01 0.12 0.14 -0.17 -0.09 -0.03 0.09 0.07 0.05 0.10 -0.10 HTO 0.31 0.50 0.43 0.46 1.00 0.46 -0.68 0.01 -0.09 -0.08 0.06 -0.01 -0.01 -0.01 -0.01 0.04 -0.09 -0.05 0.14 -0.20 CF 0.41 0.39 0.54 0.45 0.46 1.00 -0.50 0.05 -0.04 -0.04 0.07 -0.01 -0.14 -0.15 0.04 -0.06 0.09 0.08 0.10 -0.16 LND -0.19 -0.65 -0.33 -0.27 -0.68 -0.50 1.00 0.17 0.03 0.00 0.16 0.02 -0.04 -0.01 -0.11 0.04 -0.09 0.21 0.12 0.22 H S IZ 0.24 0.45 0.07 0.08 0.01 0.05 0.17 1.00 0.35 0.29 0.20 0.11 0.21 -0.22 -0.27 0.38 -0.23 0.11 0.02 0.14 AGE -0.14 0.24 0.03 0.01 -0.09 -0.04 0.03 0.35 1.00 0.99 0.00 0.05 0.02 -0.02 -0.34 0.27 -0.20 0.12 -0.04 0.15 AGSQ -0.14 0.22 0.02 0.01 -0.08 -0.04 0.00 0.29 0.99 1.00 0.01 0.04 0.00 -0.01 -0.29 0.22 -0.18 0.12 -0.04 0.13 SEX 0.26 -0.02 0.05 0.12 0.06 0.07 0.16 0.20 0.00 0.01 1.00 0.10 0.13 -0.28 -0.01 0.13 0.09 0.05 0.12 -0.16 SC H 0.06 -0.04 -0.07 0.14 -0.01 -0.01 0.02 0.11 0.05 0.04 0.10 1.00 0.17 -0.29 0.05 0.51 0.31 0.30 0.14 -0.17 ASCH 0.05 0.14 0.06 -0.17 -0.01 -0.14 -0.04 0.21 0.02 0.00 0.13 0.17 1.00 -0.03 -0.03 0.06 0.20 -0.01 0.06 -0.12 PR IM -0.02 -0.08 0.06 -0.09 -0.01 -0.15 -0.01 -0.22 -0.02 -0.01 -0.28 -0.29 -0.03 1.00 -0.13 -0.51 -0.11 -0.06 -0.07 -0.06 JSS -0.10 -0.09 -0.06 -0.03 -0.01 0.04 -0.11 -0.27 -0.34 -0.29 -0.01 0.05 -0.03 -0.13 1.00 -0.31 -0.07 -0.04 0.02 -0.16 M ID 0.06 0.13 0.00 0.09 0.04 -0.06 0.04 0.38 0.27 0.22 0.13 0.51 0.06 -0.51 -0.31 1.00 -0.27 -0.15 -0.01 0.10 SEC 0.07 -0.13 0.02 0.07 -0.09 0.09 -0.09 -0.23 -0.20 -0.18 0.09 0.31 0.20 -0.11 -0.07 -0.27 1.00 -0.03 -0.08 -0.11 PSEC -0.06 -0.04 -0.13 0.05 -0.05 0.08 0.21 0.11 0.12 0.12 0.05 0.30 -0.01 -0.06 -0.04 -0.15 -0.03 1.00 0.41 -0.12 EXT -0.05 -0.07 -0.07 0.10 0.14 0.10 0.12 0.02 -0.04 -0.04 0.12 0.14 0.06 -0.07, 0.02 -0.01 -0.08 0.41 1.00 -0.10 FB 0.00 -0.10 -0.08 -0.10 -0.20 -0.16 0.22 -0.14 0.15 0.13 -0.16 -0.17 -0.12 -0.06 -0.16 0.10 -0.11 -0.12 -0.10 1.00 100 University of Ghana http://ugspace.ug.edu.gh Where: O T P .................................................................. Output LA B ...................................................................Labour H LB ...................................................................Hired labour IN P .................................................................... Other inputs H TO .................................................................. Hand Tools C F ...................................................................... Chemical Fertilizer LN D .................................................................. Land HSIZ ..................................................................Household size AGE.................................................................. Age o f household head AGSQ ............................................................... Square o f the age o f household head SEX ................................................................... Sex o f household head SCH....................................................................Years o f schooling ASCH.............................................................. Average school years o f other family members PRIM.................................................................Primary education JS S .....................................................................Junior Secondary School education M ID ................................................................ Middle School education SEC .................................................................. Secondary education PSEC ............................................................... Post- Secondary education EX T ..................................................................Extension education F B ..................................................................... Family Background 101 University of Ghana http://ugspace.ug.edu.gh Appendix 3 Table A3a: OLS regression results: Cassava OLS Coefficients Estimates Equ.l_________________ Equ.2 Variable Est. Coefficients T-ratios Est. Coefficients T-ratios Constant 3.164 (1.156) 1.482 (0.670) LnJLab 0.035 (0.190) 0.452 (2.178)** Ln_Hlab 0.171 (1.330) 0.142 (1.234) Ln_Cfert 0.238 (1.865)* 0.269 (1.761)* LnJHtool -0.283 (-1.581) Agey 0.086 (1.467) 0.028 (0.312) Agesq - 0 .0 0 1 (-1.327) - 0.000 (-0.125) Sex_l 0.092 (0.281) 0.096 (0.372) Schyrs 0.002 (0.054) Educprim 0.289 (0.878) Educmid 0.024 (0.863) Educjss -0.651 (-1.067) Educsec 1.313 (2.052)** Educpsec -2.525 (-2 .222 )** Aveschyrs -0.007 (-0.253) Extserv -0.443 (-1.641) -0.680 (_9 RuralDum Yes (insig) Yes (insig) ForeDum Yes (insig) Yes (insig) CoastDum Yes (insig) Yes (insig) Pwork 0.496 (1.800)* 0.332 (1.164) R2’ 0.26 0.39 Adj.R2 0.14 0.26 F-stats 2.158 (0.000) 3.010 (0.000) (Prob) N 86 86 S o u rc e : A u th o r ’s c o m p u ta t io n f r o m G L S S 4 d a ta , 2 0 0 0 D e p e n d e n t v a r i a b l e is n a tu r a l lo g a r i th m o f c a s s a v a o u tp u t p e r a c r e . F ig u r e s in p a r e n th e s e s a r e t - r a t io s . S ta r s in d ic a te s ig n i f i c a n c e u s in g a tw o t a i l e d t - t e s t a s f o l l o w s : * * * = 0 .0 1 ; ** = 0 .0 5 ; * = 0 .1 0 . a. In c lu d e d a s a s c a le v a r ia b le Y e s - in c lu d e d in t h e m o d e l ; in s ig - e s t im a te d c o e f f i c i e n t n o t s ig n i f i c a n t E q u a t io n s l a b e l le d 1 a n d 2 a r e b a s e d o n e q u a t io n 3 .1 7 u s in g d i f f e r e n t v a r ia n t s o f e d u c a t io n m e a s u r e . 102 University of Ghana http://ugspace.ug.edu.gh Table A 3b: OLS regression results: Maize OLS Coefficients Estimates Equ. 1 Equ. 2 Variable Est. Coefficients T-ratios Est. Coefficients T-ratios Constant -1.935 (-1.542) 0.106 (0.083) Ln_Lab 0.201 (1.527) 0.078 (0.699) Ln_Hlab 0.385 (4.536)*** 0.288 (3.842)*** Ln_Cfert 0.107 (1.318) 0.147 (2.119)** Ln_Inputs 0.136 (1.803)* 0.147 (2.352)** Agey -0.045 (-1 .0 2 1 ) -0.057 (-1.321) Agesq 0.000 (0.724) 0.000 (1.096) Sex_l 0.580 (2.270)** 0.453 (2.651)*** Schyrs 0.025 (1.139) Educprim 0.088 (0.415) Educmid - 0.031 (0 .2 10 ) Educjss -0.446 (-1.346) Educsec -0.052 (-0.160) Educpsec -0.172 (-0.285) Aveschyrs 0.004 (0.171) 0.022 (1.160) Extserv -0.568 (-2.630)*** -0.158 (-0.804) RuralDum Yes (insig) Yes (sig) ForeDum Yes (insig) Yes (insig) CoastDum Yes (insig) Yes (insig) Pwork 0.436 (1.954)* 0.312 (1.571) R2 0.50 0.47 Adj.R2 0.43 0.39 F-stats 6.880 (0 .000) 6.164 (Prob) N 150 150 S o u rc e : A u th o r ’s c o m p u ta t io n f r o m G L S S 4 d a ta , 2 0 0 0 D e p e n d e n t v a r i a b l e is n a tu r a l lo g a r i th m o f m a iz e o u tp u t p e r a c r e . F ig u r e s in p a r e n th e s e s a r e t - r a t io s . S ta r s in d ic a te s ig n i f i c a n c e u s in g a tw o t a i l e d t - t e s t a s f o l l o w s : * * * = 0 .0 1 ; ** = 0 .0 5 ; * = 0 .1 0 . a . In c lu d e d a s a s c a le v a r i a b l e Y e s - V a r i a b l e in c lu d e d in t h e m o d e l ; in s ig - e s t im a te d c o e f f i c i e n t is n o t s i g n i f i c a n t ; s ig - c o e f f i c i e n t is s ig n i f ic a n t E q u a t io n s la b e l le d 1 a n d 2 a r e b a s e d o n e q u a t io n 3 .1 7 u s in g d i f f e r e n t v a r ia n t s o f e d u c a t io n m e a s u r e . 103 University of Ghana http://ugspace.ug.edu.gh Appendix 4 Table A 4a: WLS estimation results showing interaction effect for Extension Service and Years of schooling completed : Cassava Variable WLS Coefficients Estimates Equ. 1 Equ. 2 Est. Coefficients T-ratios Est. Coefficients T-ratios Constant 0.540 0.172 0.586 0.026 LnJLab 0.357 1.113 0.486* 1.953 Ln_Hlab 0.300 1.276 0.243 1.460 Ln_Cfert 0.116 0.616 0.086 0.788 Ln_Htool -0.193 -0.871 -0.210 - 1 . 1 1 2 Ln_Inputs 0.060 0.376 0.135 1 .0 12 Agey 0.062 0.618 0.044 0.532 Agesq -0.000 -0.654 -0.000 -0.544 Sex_l -0.128 -0.288 0.001 0.004 Schyrs 0.012 0.213 Extserv*Schyrs 0.470 0.685 -0.646** -2.201 Extserv -1.009 -1.519 RuralDum -0.394 -0.644 -0.277 -0.767 ForeDum -0.411 -0.689 -0.533 -1.082 CoastDum 0.012 0.018 -0.333 -0.577 Pwork 0.696* 1.781 0.582** 1.856 R2 0.93 0.95 Adj.R2 0.92 0.94 F-stats (Prob) 69.476 (0.000) 137.009 (0.000) N 107 107 S o u rc e : A u th o r ’s c o m p u ta t io n f r o m G L S S 4 d a ta , 2 0 0 0 D e p e n d e n t v a r i a b l e is n a tu r a l lo g a r i th m o f c a s s a v a o u tp u t p e r a c r e . F ig u r e s in p a r e n th e s e s a r e t - r a t io s . S ta r s in d ic a te s ig n i f i c a n c e u s in g a tw o ta i l e d t - t e s t a s f o l l o w s : * * * = 0 .0 1 ; ** = 0 .0 5 ; * = 0 .1 0 . L an d a r e a c u l t iv a te d w a s u s e d a s a s c a le o r w e ig h te d v a r ia b le E q u a t io n s la b e l le d 1 a n d 2 a r e b a s e d o n e q u a t io n 3 .1 7 u s in g d i f f e r e n t v a r ia n t s o f e d u c a t io n m e a s u r e . 104 University of Ghana http://ugspace.ug.edu.gh Table A 4b: WLS estimation results showing interaction effect for Extension Serv ice and Years of schooling completed : Maize Variable WLS Coefficients Estimates Equ. 1 Equ. 2 Est. Coefficients T-ratios Est. Coefficients T-ratios Constant -1.656 -0.993 -0.678 -0.503 LnJLab 0.198 1.292 0,162 1.343 Ln_Hlab 0.368*** 4.244 0.262*** 3.355 Ln_Cfert 0.119* 1.724 0.116 1.484 LnJHtool 0.026 0.241 -0.010 -0.010 Ln_Inputs 0.155** 2.031 0.216*** 3.056 Agey -0.049 -0.928 -0.034 -0.717 Agesq 0.000 0.711 0.000 0.489 Sex_l 0.447 1.497 0.395 1.634 Schyrs 0.040* 1.710 - Extserv*Schyrs 0.296 0.547 -0.191 -0.918 Extserv -0.374 -0.795 RuralDum 0.299 0.778 0.473 1.522 ForeDum 0.303* 1.682 0 4 1 7 *** 2.774 CoastDum -0.507 -1.390 -0.437 -1.292 Pwork 0.292 0.917 0.107 0.400 R2 0.96 0.95 Adj.R2 0,95 0.94 F-stats (Prob) 147.364 (0,000) 219.779 (0.000) N 151 151 S o u rc e : A u th o r ’s c o m p u ta t io n f r o m G L S S 4 d a ta , 2 0 0 0 D e p e n d e n t v a r i a b l e is n a tu r a l lo g a r i th m o f m a iz e o u tp u t p e r a c r e . F ig u r e s in p a r e n th e s e s a r e t - r a t io s . S ta r s in d ic a te s ig n i f i c a n c e u s in g a tw o t a i l e d t - t e s t a s f o l l o w s : * * * = 0 .0 1 ; * * = 0 .0 5 ; * = 0 .1 0 . N a tu ra l lo g o f f a rm s iz e w a s u s e d a s a w e ig h te d v a r i a b l e ( P r in c ip a l c a u s e o f h e t e r o s c e d a s t i c i t y ) E q u a t io n s l a b e l le d 1 a n d 2 a r e b a s e d o n e q u a t io n 3 ,1 7 u s in g d i f f e r e n t v a r ia n t s o f e d u c a t io n m e a s u r e . 105 University of Ghana http://ugspace.ug.edu.gh Appendix 5 Table A 5: Percentage increase in output associated with an extra year of schooling of head farmer* Farm Enterprise Education Variable Percentage increase Activity in output Cassava Years o f Schooling 3.66 Dummy Indicator Primary 13.25 Secondary 21.19 Maize Years o f Schooling 3.05 Dummy Indicator Primary 0.25 Middle 2.34 Secondary 0.81 Post Secondary 2.29 Source: Author’s computation from GLSS 4 data (2000), using the formula developed by Lockheed, Jamison and Lau (1980) * Results based on equations 3.18a and 3.18b 106 University of Ghana http://ugspace.ug.edu.gh