DEMAND, SUPPLY RESPONSE AND PREFERENCE SWITCH FOR RICE IN NIGERIA AbiodunOlayinkaSamsideen AYANWALE MATRIC. NO: 108521 B. Agric. (Agricultural Economics) (UNAAB), Abeokuta M.Sc. Agric. Economics (UI), Ibadan A Thesis in the Department of Agricultural Economics submitted to the College of Agriculture and Forestry in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY, UNIVERSITY OF IBADAN February, 2014 DEDICATION This research is heartily dedicated to ‘my father’, Professor Fatai Lekan Ayanwale, former Vice President, World Veterinary Association, and lecturer, Tuskegee University, Alabama, U.S.A., for taking full responsibility of my postgraduate studies; and all those who showed concern in the pursuit of my Ph.D programme. ii ABSTRACT The phenomenon of increasing rice importation defying several policy interventions has been of great concern in Nigeria. This rising importation is however driven by increasing demand, shortage in domestic supply and consumers’ preference for imported rice. Yet, comprehensive national studies on determinants of demand, supply response and preference switch for rice are scarce. Thus, the determinants of demand, supply response and preference switch for rice were investigated. Secondary data from the Nigeria Living Standard Survey (NLSS) of 2004 conducted by the National Bureau of Statistics (NBS) and time series data from the official records of International Rice Research Institute (IRRI), 1960-2008 were used. Due to elimination of households with missing values on variables of interest, a total of 18,861 out of 21,900 households were used in the NLSS. Variables used in NLSS included Household Size (HS), Non-Food Total Expenditure (NFTE), Years of Education (YE), sector (urban/rural), occupation (farming/non-farming) and Membership of Association (MA) which were hypothesized to influence household expenditures on Imported Rice (IR), Improved Domestic Rice (IDR) and Local Rice (LR). Data on area cultivated, level of import, fertilizer consumption and prices were used in IRRI rice statistics and these variables were also hypothesized to influence supply (output) of rice. Data were analysed using descriptive statistics, Tobit regression model, vector error correction model and generalised least square regression at p= 0.05. The HS and YE were 4.9±2.9 and 6.8±6.3 years, respectively. Rural dwellers, farmers and members of association constituted 76.1%, 82.7% and 54.2%, respectively. Monthly rice expenditure was N2, 712.40, representing 25.0% of total monthly food expenditure. The expenditure share of IR (45.0%) was higher than IDR (30.0%) and LR (25.0%). Urban sector, -03 -04 -03 YE, HS and NFTE increased the demand for IR by 4.0×10 , 2.0×10 , 1.0×10 and -09 -03 1.0×10 , respectively, while Farming Occupation (FO) reduced it by 9.0×10 . Also, FO -03 increased IDR demand by 8.0×10 . Conversely, HS, NFTE, and MA reduced IDR demand -04 -08 -09 by 9.0×10 , 2.0×10 and 1.0×10 , respectively. Also, NFTE and MA, respectively, -09 -03 increased LR demand by 6.0×10 and 4.0×10 . Price elasticities of IR, IDR and LR which -03 -04 -03 were -3.0×10 , -7.0×10 and -2.0×10 , respectively implied that rice was price inelastic. -08 -07 Also, income elasticities of IR, IDR and LR which were, respectively, 7.0×10 , 2.0×10 -07 and 1.0×10 classified rice as ‘necessities’ and ‘normal’ good. In the long-run, area iii cultivated and fertilizer consumption increased rice output by 2.8 and 2.3 respectively. Rural Sector (RS), HS, FO, and price of IR increased consumers’ switch from IR to IDR by 55.1, 6.6, 130.4, and 30.7, respectively, while price of IDR reduced it by 19.4. Price of IR and RS positively influenced switch from IR to LR by 2.0 and 70.2, respectively, while price of LR reduced it by 16.3. Education and urban livelihood increased demand for imported rice. Increasing rice area cultivated and usage of fertilizer may boost domestic rice supply. Price reduction will be a veritable tool in switching consumers’ preference from imported to improved domestic and local rice. Keywords: Rice demand, Supply response, Preference switch, Imported rice Word count: 486 iv ACKNOWLEDGEMENTS All praises, gratitude and adoration belong first to the Almighty Allah for his immeasurable favour on me throughout the course of this project. I glorify Him equal to the number of His creatures; I thank Him as much as it pleases him; my gratitude to Him is as enormous as the adornment of His throne and equal to the number of His words. My hearty and sincere appreciation goes to my able and capable supervisor, Professor V.O. Akinyosoye, for his inspiration, fatherly concern, intellectual guidance, data gathering assistance, psychological and moral support throughout this research. I cannot thank him enough for his patience and understanding despite his tight schedule. He is just more than a supervisor. I am greatly indebted to Dr. S.A. Yusuf, my co-supervisor, who doubled as the chairman departmental postgraduate committee, for providing a viable complementary supervisorial role by providing the necessary guidance and direction at every stage of the research; he was indeed more than a co-supervisor. Also, the efforts of Dr. O.A. Oni as the second co-supervisor can never be forgotten. He created time out of none to assist in perusing my write-ups and served as motivator along the course of the research. I will forever be grateful to the Head of Department, Professor V.O. Okoruwa for his support, encouragement and official roles played throughout the Ph.D programme. The Departmental Postgraduate Co-ordinator, Dr. K.K. Salimonu, is highly appreciated for his formal and informal roles in facilitating the completion of this programme, especially from the abstract stage. Other faculty members, such as Professor J.K. Olayemi (Rtd.), Professor A.O. Falusi (Rtd.), Professor M.A.Y. Rahji, Dr. T.T. Awoyemi, Dr. B.T. Omonona, Dr. (Mrs) Adenegan, Dr (Mrs) Adeoti, Dr. A.S Oyekale, Dr (Mrs) Obayelu, Dr (Mrs) Adepoju, late Dr (Mrs) Ajani and others are acknowledged for their support, encouragement and statutory roles in seeing this work to completion. Mr Alabi of the GIS unit, International Institute of Tropical Agriculture is acknowledged for assisting me in providing the needed climate data. I equally extend my gratitude to the Director and staff of National Food Reserve Agency (NFRA) and the head and staff of Planning Research and Statistics Department of OGADEP, Abeokuta for providing me with needed data. My academic benefactors and mentors: Dr. Taofiq Azeez, of University of Abuja; Professor L.O. Sanni, Dr. I.A. Ayinde and Dr. K. Adebayo, of the University of Agriculture, Abeokuta; Prof. A.A. Tijani of Obafemi Awolowo University, Ile Ife; Dr. v Ibrahim Uthman, of University of Ibadan; Alhaji S.A. Akinwunmi, of Olabisi Onabanjo University Teaching Hospital and Dr. GBK Ajayi, of Ojulowo Eye Clinic, Ibadan, are immeasurably appreciated for their academic, financial, moral and spiritual support. I sincerely appreciate the following friends, colleagues, classmates and associates for the assistance rendered during the conception of the research work, data mining, data analysis, write-up reviews, financial and moral support: Dr F. Ogundele (NISER); Dr. Dontsop (IITA); Dr. Idowu and Dr. O.C. Ologbon (OOU); Dr. W. Ashagidigbi; Dr. Balogun and Dr. F. Sowunmi (U.I); Dr. T. Dauda (IAR&T); Mr. Omoniyi (Redeemers College); Dr. S. Karim , Dr. Dele Akinbode, Dr. R. Adebowale, Dr. S.A. Adewuyi, Dr. K. Bello, Dr. R. Bello and Dr. S. Iposu, all of UNAAB; Dr. R. Karim (Crescent University); Dr. L. Akinola (Fountain University); Dr. (Mrs) G. Adeshina-Uthman (NOUN); Dr. Wale Olayide, Dr Wole Ogunyemi, Mr Monsuru Muritala and Mr Saheed Olatunji (U.I); Mr AbdulRauf Oyedele; Engineer and Mrs Oketokun; Mr. M. Zakariyah; Mr. D. Oladunni; Mr. Y. Oke; Mr T. Olaiwola; and others too numerous to mention. I am indeed grateful to the departmental secretary, Mrs F.M Oladejo, Mrs Atanda, Mr Lawrence, Mr Lekan Animasaun, Mrs Nike Olawande and other administrative staff of the Department of Agricultural Economics for the roles they played in the course of my Ph.D programme. This acknowledgment will be incomplete without appreciating my lovely family: my darling, caring and enduring wife, Azeezat Adesola Amope, who gave all support and encouragement in pursuing this programme; my children- Ghalib Adesina, Mujibah Adeola and Mazeedah Adebola; my dear mother, Alhaja Aduke Ayanwale; and my late father, Alhaji Raimi Ayanwale of blessed memory, for their patience and perseverance throughout my academic pursuit. I am greatly indebted to my elder brothers and sisters: Alhaji Rasaq Ayanwale, Late Alhaji Ganiu Ayanwale, Alhaja Taibat Ologunebi, Professor Lekan Ayanwale, Mr. Musibau Ayanwale, Alhaji Osuolale Ayanwale, Engineer Leye Ayanwale and Dr Jibola Ayanwale; and my siblings- Mr. Akeem, Mr. Moruf and Alhaji Bashir Ayanwale- for their perseverance, moral, spiritual and financial support throughout my protracted educational career. I equally thank my bosses- Mr. Moses Ebikwo, Engnr. A. Akinola and Mr I.E Sotiyo and other colleagues at work for granting me the needed time and support to pursue this PhD programme. May Allah grant all of the mentioned and the unmentioned their heartfelt desires. vi CERTIFICATION I certify that the research culminating in this thesis was carried out by Abiodun Olayinka Samsideen Ayanwale under my supervision in the Department of Agricultural Economics, University of Ibadan, Nigeria. Professor V.O. Akinyosoye BSc. (Agric.) Ibadan MSc. (Agric. Economics) Ibadan M.A. Wisconsin, U.S.A Ph.D (Agricultural Economics) Wisconsin, USA Professor of Applied Economics and Data Management vii TABLE OF CONTENTS Title page i Dedication ii Abstract iii Acknowledgements v Certification vii Table of contents viii List of tables xi List of figures xii List of appendices xiii Abbreviation and acronyms xiv CHAPTER ONE: INTRODUCTION 1.1. Background to the study 1 1.2. Statement of the problem 3 1.3. Objectives of the study 5 1.4. Hypotheses of the study 5 1.5. Justification of the study 6 1.6. Organization of the thesis 8 CHAPTER TWO: THEORETICAL/CONCEPTUAL FRAMEWORK AND LITERATURE REVIEW 2.1 Theoretical and conceptual reviews 9 2.1.1 Concept and theory of food demand 9 2.1.2 Supply response theory 12 2.1.3 Consumption preference theory 14 2.2 Methodological review 15 2.2.1 Estimation of demand 16 2.2.2 Methodological issues in time series modeling 17 2.2.3 Methodological issues in measurement of supply response 23 2.3 Empirical reviews 26 2.3.1 Trends in rice demand and preference in Nigeria 26 2.3.2 Rice production systems and processing 30 2.3.3 Rice production trend in Nigeria 31 viii 2.3.4 Rice import trend in Nigeria 37 2.3.5 Nigerian rice policy environment 40 2.3.6 Nigeria and the quest for food self-sufficiency 44 2.3.7 The Agricultural Transformation Agenda (ATA) 44 2.3.8 Determinants of food demand and preference 47 2.3.9 Review of supply response Studies 49 2.4 Conceptual linkage 51 CHAPTER THREE: RESEARCH METHODOLOGY 3.1 The study area 53 3.2 Sources and type of data 57 3.3 Analytical tools and models 58 3.3.1 Descriptive statistics 58 3.3.2 Tobit Regression Model 59 3.3.3 Linearised AIDS Model 62 3.3.4 Cointegration-ECM Analysis 66 3.3.5 Paired Sample t-Test 72 3.3.6 Generalised Least Square Regression 72 3.4 Limitations of the study 73 CHAPTER FOUR: RESULTS AND DISCUSSION 4.1 Household expenditure pattern on rice and socioeconomic characteristics of respondents 75 4.1.1 Distribution of expenditure of households on rice 75 4.1.2 Share of rice in total food expenditure 84 4.1.3 Socioeconomic characteristics of the respondents 86 4.1.4 Distribution of share of rice expenditure and socioeconomic characteristics 88 4.2 Rice self-sufficiency in Nigeria 97 4.3 Determinants of rice demand in Nigeria 100 4.3.1 Tobit Model Estimate for rice demand 100 4.3.2 Almost Ideal Demand System Estimate for rice demand 107 4.3.3 The Tobit Model compared with the AIDS Model 111 4.4 Supply response analysis 111 4.4.1 Unit Root Test 111 ix 4.4.2 Pairwise Granger Causality Test 113 4.4.3 Tests for cointegration (Johansen Test) 115 4.4.4 The Vector Error Correction Model 118 4.5 Preference Switch Analysis 122 4.5.1 Preference direction 122 4.5.2 Determinants of preference switch of rice commodities 125 CHAPTER FIVE: SUMMARY, CONCLUSION AND RECOMMENDATIONS 5.1: Summary 129 5.2 Policy Implications of the findings 132 5.3 Conclusion 133 5.4 Policy recommendations 133 5.5 Contributions to knowledge 135 5.6 Suggestions for further research 136 References 137 Appendices 153 x LIST OF TABLES Table 1: Nigeria Geopolitical Zones Table 2: A priori Expectation for Demand Variables Table 3: A priori Expectation for Supply Response Variables Table 4: Distribution of the Respondents by Total Monthly Expenditure on Rice Table 5: Distribution of the Respondents by Expenditure on Imported Rice Table 6: Distribution of the Respondents by Expenditure on Improved Domestic (Agric.) Rice Table 7: Distribution of the Respondents by Expenditure on Local Rice Table 8: Description of Household Expenditure on Rice Commodities (National) Table 9: Distribution of the Respondents by Share of Rice in Total Food Expenditure (National) Table 10: Socioeconomic Characteristics of the Respondents Table 11: Distribution of Share of Rice expenditure by Household size Table 12: Distribution of Share of Rice Expenditure by Education Table 13: Distribution of Share of Rice Expenditure by Membership of Community Society Table 14: Distribution of Share of Rice Expenditure by Marital Status Table 15: Distribution of Share of Rice Expenditure by Primary Occupation Table 16: Distribution of Share of Rice Expenditure by Location Table 17: Distribution of Share of Rice Expenditure by Age Table 18: Tobit Regression Result for Rice Demand in Nigeria (Marginal Values) Table 19: Tobit Elasticity Estimates for Rice Demand in Nigeria Table 20: AIDS Regression Result for Rice Demand in Nigeria Table 21: AIDS Elasticity Estimates for Rice Demand in Nigeria Table 22: Result of ADF Unit Root Test of Variables Table 23: Pairwise Granger Causality Tests Table 24: Cointegration Test for all Specifications Table 25: Co-integration Test for Intercept and No deterministic trend in the data Table 26: Rice Supply Response- Long Run Model Table 27: Short Run Equilibrium Model (VECM) Table 28: Paired Sample t-test for Preference Switching Table 29: Determinants of Preference Switch A xi LIST OF FIGURES Fig. 1: Rice Consumption Trend in Nigeria (1995-2009) Fig. 2: Rice Area Trend in Nigeria (1995-2011) Fig. 3: Rice Production Trend in Nigeria (1995-2011) Fig. 4: Rice Yield Trend in Nigeria (1995-2011) Fig. 5: Trend in Rice Import in Nigeria (1995-2011) Fig. 6: Conceptual Linkage Fig. 7: Map of Nigeria Showing States and Geopolitical zone Fig. 8: Map of Nigeria Showing Coarse Grain Crop Zones Fig. 9: Relationship between Rice Supply and Demand (1960-2008) Fig. 10: Self-Sufficiency Ratio of Rice in Nigeria xii LIST OF APPENDICES Appendix 1: Nigerian Rice Production Systems Appendix 2: Analysis of Objectives Appendix 3: Taxonomy of Trade Policy on Rice in Nigeria Appendix 4: NBS Rice Production Statistics Appendix 5: FAO Trend in Rice Production, Consumption and Import in Nigeria (1995- 2011) Appendix 6: Estimates of Aggregate Agricultural Supply Response Appendix 7: Description of Household Expenditure on Rice Commodities (Zones) Appendix 8: Distribution of Respondents by Share of Rice in Total Food Expenditure(Zones) Appendix 9: Nigeria Rice Self- Sufficiency Ratio Appendix 10: Farm harvest Price of Rice in Nigeria (1960-2008) Appendix 11: Fertilizer Consumption in Nigeria (1960-2008) Appendix 12: Mean Annual Rainfall in Nigeria (1960-2008) Appendix 13: Determinants of Preference Switch B xiii xiii ABREVIATIONS AND ACRONYMS ADF–Augmented Dickey-Fuller AIDS–Almost Ideal Demand System AR–Autoregressive ARIMA–Autoregressive Integrated Moving Average ATA–Agricultural Transformation Agenda CBN–Central Bank of Nigeria CE–Cointegrating Equation DF–Dickey-Fuller DW– Durbin-Watson DADTCO–Dutch Agricultural Development and Trading Company FAO–Food and Agricultural Organisation FAOSTAT–Food and Agricultural Organisation Statistics FIML–Full Information Maximum Likelihood FMWA–Federal Ministry of Women Affairs FMAWRD–Federal Ministry of Agriculture, Water and Rural Development GNP– Gross National Product IRRI–International Rice Research Institute IDR–Improved Domestic Rice IR–Imported Rice LAIDS–Linearized Almost Ideal Demand System LES–Linear Expenditure System LR–Likelihood Ratio LR–Local Rice NAMIS–National Agricultural Market Information System NCRI– National Cereals Research Institute NBS–National Bureau of Statistics NERICA– New Rice for Africa NFRA–National Food Reserve Agency NISER– National Institute of Social and Economic Research NLSS–Nigeria Living Standards Survey PCU–Project Coordinating Unit QUES– Quadratic Expenditure Demand System RIFAN–Rice Farmers Association of Nigeria xiv SCPZs–Staple Crop Processing Zones SURE–Seemingly Unrelated Regression TIN–Trade Investment Nigeria TDS–Translog Demand System UNEP–United Nations Environment Programme UNIDO–United Nation Industrial Development Organisation USDA–United State Development Agency VAR–Vector Autoregressive VECM–Vector Error Correction model (VECM) WARDA–West African Rice Development Agency xv CHAPTER ONE INTRODUCTION 1.1 Background to the study The agricultural food sub-sector in Nigeria parades a large array of staple crops,made possible by the diversity of agro-ecological production systems. The major food crops are cereals, such as sorghum, maize, millet, rice, wheat; tubers, such as yam, cassava; legumes, like groundnut, cowpeas; and others such as vegetables (Akande, 2007). These are the commodities that are of considerable importance for food security and incomes of households. Rice is one of the leading staple food crops in Nigeria. It is cultivated in virtually all the agro-ecological zones of Nigeria, from the mangrove and swamps environment of the coastal areas, to the dry zones of the Sahel in the North (Akande, 2007). In 2007, about 1.7 million hectares were under rice cultivation in Nigeria with estimated national production of 3.4 million metric tons, representing 22% increase over 2006 level and a growth rate of 0.6 percent from 1999 (National Food Reserve Agency, NFRA, 2008). NFRA (2008) also notes that rice yield in the same year was estimated at 2 metric tons per hectare, a negligible decrease of 0.03 percent over 2006 and 1 percent annual growth rate from 1999. Also, National Bureau of Statistics (NBS) claims that 4. 5 million metric tons of rice was produced in 2010 as against the 3.2 million metric tons (from 1.84ha) and 3.1 million metric tons (from 1.77ha) reported by Food and Agricultural Organisation (2014) for 2010 and 2011, respectively. IRRI (2014) equally reported a contrastingly lower output of 2.85 million metric tons for rice in Nigeria in 2012. The demand for rice has been soaring over the years. Since the mid-1970s, rice consumption in Nigeria has risen tremendously, growing by 10.3% per annum, a result of accelerating population growth rate of 2.8% per annum, increasing per capita consumption of 7.3% per annum, rapid urbanisation, increased income levels, and associated changes in family occupational structures (Akpokodje et al., 2001; UNEP,2005; Akande, 2007, Bashorun, 2010; Oyinbo et al., 2013). The ease with which rice is prepared fits into urban lifestyle, where households rush up daily to catch up with career demands (Oguntona and Akinyosoye, 1986; Bamidele et al., 2010; Oyinbo et al., 2013). The average rice consumption expenditure represents 60% of the total expenditure on cereals and 17% of expenditure share on food commodities (NLSS; NBS 2004). The demand for rice in Nigeria amounts to 4.1 million 1 metric tons in 2002 (Akande, 2007) and has risen astronomically to 5.2 million metric tons and 5.9 million metric tons in 2011 and 2012, respectively (IRRI, 2014). The foregoing notwithstanding, the production increase has been unable to match the consumption increase (Okoruwa et al., 2006; Rahji et al., 2008) and domestic production capacity is below the national requirements for rice (Rahji and Adewumi, 2008). Nigeria is the largest producer of rice in West Africa, but the country with a population of over 150 million people still relies on massive rice importation (Bello, 2004; Okoruwa et al., 2006; Rahji et al., 2008). Bello (2004) states that Nigeria imports US$700 million worth of rice in 2003. It also accounts for 20% of sub-Saharan Africa‘s rice imports (Omotola and Ikechukwu, 2006). Similarly, Workman (2008) reports that Nigeria imported 1.4 million tons of rice, equivalent to 4.8 percent of global rice imports and therefore tops the list of rice importers in the year 2007. Nigeria also expends US$1.3 billion every year to import 2.2 billion kg of rice in order to fulfil its domestic requirements (Akosile, 2009). It also spends whooping N630 billion annually on the importation of agricultural products, of which rice gulps N75 billion, following wheat and fish as the most imported agricultural products (Sanusi 2011). These imports represent a substantial foreign exchange outlay for the Nigerian economy. Given the size and value of the imports, there is considerable policy interest in reducing rice imports by promoting domestic rice production and consumption (Sanusi, 2003; Omotola and Ikechukwu, 2006). Thus, rice has become a strategic commodity in the Nigerian economy. The Nigerian government has the objective of self-sufficiency in rice high on its agenda as epitomised by the intermittent import bans, government‘s attention on varietal improvement, seed multiplication, varying tariff regimes on imported rice in the past (Akpokodje et al., 2001; Erenstein et al., 2004; Akande, 2007), more recent special rice projects, import substitution policies, presidential initiative on rice (NFRA, 2008; Akosile, 2009) as well as Agricultural Transformation Agenda (ATA) that targets 2018 for self-sufficiency in rice production (Akinwunmi, 2012; Adeyeye, 2012). The Nigerian government has intervened in the rice sector in the past few decades, yet domestic production has been unable to catch up with demand, resulting in continuous importation of milled rice. Given this scenario, self- sufficiency (a balanced ratio of domestic supply to demand) in rice remains a proximate objective of the Nigerian government. 2 1.2 Statement of the Problem In the 1960s, Nigeria was 99 percent self-sufficient in the rice consumed by its citizens. In the following two decades (1970s and 1980s), self-sufficiency declined to 38 percent, resulting from demand outstripping supply (Akande, 2007). The 360,000 tons of rice produced in the 1960s was enough to meet local demand, but the 1.45 million tons produced in the 1990s was not (IRRI, 1991; IRRI, 1995). Thus, importation of rice rose from an annual average of 7,000 tons in the 1960s to 657,000 tons in the 1990s (IRRI, 1995). In1999, the value of import was US$259 million, partly leading to a drain on Nigeria's foreign exchange reserve, which stood at US$407.5 million in the 1960s but dropped to US$58 million in the 1990s (IRRI, 1999). Also, between 1961 and 1999,Nigeria had spent $4 billion on rice importation alone, an average annual import value of US$102 million (RIFAN, 2006). The Central Bank of Nigeria (2002) also notes that US$578 million worth of rice was imported in 2002. Nigeria expends N250 billion annually on agricultural products, rice alone gulps N60 billion (NAMIS, 2004). The rice importation bill has risen to US$1.3 billion annually (Akosile, 2009). Worse still, Nigeria's rice import burden was predicted to swell, as demand was estimated to double supply growth in 2013 (TIN, 2010). Given the precarious balance of payment position of the country, rice import has become a major source of concern. According to RIFAN (2006), as at 2003, demand for rice was estimated at 5 million metric tons while production was 3 million metric tons of rice; a short-fall of 2 million metric tons, which was augmented by importation. The 2 million metric tons importation out of 5 million metric tons demand at a cost of US$300 million dampened the hope of possible improvement in the level of domestic rice production.The target during the first national conference on harmonisation for sustainability of self-sufficiency in rice productionheld in Abuja, Nigeria‘s capital, in 2003, was that, by November 2005, locally produced rice would be 4.2 million metric tons leaving a gap of 800,000 metric tons to fill the vacuum created by domestic demand, but this target was not attained (RIFAN, 2006). Nigeria, which is about one-fifth of the population in the sub-Saharan Africa, consumes the highest volume of rice within the region (Momoh, 2007). Rice is now a staple food for over 60 percent Nigerians (RIFAN, 2006).The inadequate level of cereal production to match demand in Nigeria is manifested in high prices and an annual increasing expenditure on importation (CBN, 2000; Bashorun, 2013). This rising demand for rice is a function of several price and non-price factors that need to be identified and managed. 3 Nigeria has suitable ecologies and a potential land area for rice production.The potential of riceyield has not been fully realised inspite of increasing area of cultivation (Akpokodje et al., 2001; Akande, 2007). The risk and uncertainty faced by agricultural firms is much higher than that faced by other standard firms. The agricultural sector is characterised by high imperfection in price and other information.As a result, the production behaviour of agricultural firms greatly differs from that of other firms, yet risk factors are often neglected in the analysis of supply response and dynamic modelling has not been employed in most cases (McKay et al., 1999; Muchapondwa, 2008). Similarly, little has been done in Nigeria to capture the impact of non -price factors(such as, climate, area and import level) on supply (Rahji and Adewumi, 2008).This is a gap that needs to be filled. Numerous general and specific agricultural research, policies and programmes in Nigeria such as, previous import bans, government‘s attention on varietal improvement, seed multiplication, varying tariff regimes on imported rice, special rice projects, multinational NERICA rice dissemination project, import substitution policies, presidential initiative on rice and Agricultural Transformation Agenda (Akpokodje et al., 2001; Erenstein et al., 2004; Akande, 2007; NFRA, 2008; Tiamiyu, 2009; Akosile, 2009; Adeyeye, 2012; Akinwumi, 2012) have been executed in Nigeria over time.However, local rice production has not kept up with the domestic demands of the Nigerian populace and, consequently, rice is still massively imported (Rahji and Adewumi, 2008). This calls for the reexamination of the effectiveness of these policies. Also, with increased production of local rice, it is still not certain that consumers will purchase local rice if there is preference for imported rice (Sanusi, 2003), thus, defeating the goal of self-sufficiency in the face of increased production. Socioeconomic factors play a key role in determining the direction of preference for various rice commodities. Researches in the area of consumer preference and switching factors for rice, especially at the national level, have received little attention, a gap filled by this study. Similarly, recent development in the fast food industry that involves the promotion of consumption of local foods including local rice, and the inclusion of local rice in ceremonial delicacies point to the fact that the Nigerian consumers have tendency to switch to local food if it is made acceptable and competitive enough through improved processing and quality enhancement. This further necessitates a study on the determinants of preference switch from foreign to local rice and vice versa. 4 In line with these facts, the following research questions become fundamental in this study. 1. What is the pattern of rice consumption in Nigeria? 2. Which trend does rice self-sufficiency ratio follow in Nigeria? 3. What are the factors that determine the demand for rice in Nigeria? 4. Does domestic rice supply respond to price and non-price factors in the short run and the long run? 5. Are there socioeconomic factors that switch consumers‘ preference from foreign to domestically produced rice or vice versa? 1.3 Objectives of the study The main objective of this study was to determine the factors that influence the demand, supply response and preference switch for rice in Nigeria. The specific objectives were to: 1. describe the expenditure pattern of rice in Nigeria. 2. examine the self-sufficiency ratio of rice in Nigeria. 3. estimate a demand model for rice in Nigeria. 4. analyse the supply response of rice in Nigeria. 5. isolate the determinants of preference switch from foreign to domestically produced rice and vice versa. 1.4 Hypotheses of the study The following hypotheses were tested: Ho: There is no significant relationship between the demand for rice and socioeconomic factors in Nigeria. Ho:Rice supply does not respond to price and non-price factors in the short run and the long run. Ho:Socioeconomic factors do not determine the preference switch from foreign to local rice and vice versa. 5 1.5 Justification of the study In view of the ever-increasing demand for rice and the inability of local rice supply to meet demand that necessitates continuous importation of milled rice at a cost that drains Nigeria‘s foreign exchange earnings(Okoruwa and Ogundele, 2006; Momoh, 2007; TIN, 2010), a study of this nature, that estimates the determinants of demand for rice in a bid to stimulate domestic rice consumption and increase the market share of domestically processed rice as stipulated in the Agricultural Transformation Agenda, is relevant to save the nation‘sforeign exchange reserve. Estimation of demand functions is also useful as they provide us with income and price elasticities which are required for the design of different policies; for example, policy design for indirect taxation and subsidies requires knowledge of these elasticities for tradable commodities and services (Deaton, 1988). Most past Nigerian studies on rice like Imolehin and Wada (2000), Akpokodje et al. (2001), Kebbeh et al.(2003), Ajetomobi (2005),Okoruwa and Ogundele (2006), Okoruwa et al. (2006), Tijaniet al. (2006), Bamidele et al. (2010), Bamba et al. (2010), Adeyeye et al. (2012) and Oyinbo et al. (2013)dealt with either the supply or the demand side of the rice sector, only very few studies that combine demand and supply response have been carried out in Nigeria (Rahji and Adewumi, 2008). This study combined the two sides of the rice market to bring about a holistic view of the Nigerian rice economy, thus contributing to the literature in this regard and assisting in designing policies on the attainment of rice self-sufficiency from a comprehensive perspective. Furthermore, most supply response studies in Nigeria, such as Rahji et al.(2008), limited their analysis of supply response to price factors, yet rice supply has been hypothesized to be a function of several price and non-price factors.This study specified a supply response function for rice, inclusive of price and non-price factors in both the short run and the long run, thereby exposing the response of rice to non-price factors. This will provide additional information on policy design for rice production, especially when rice is non-responsive to price. Unlike previous studies on supply response in Nigeria (such as Abalu, 1974; Ajakaye, 1987; Oyejide, 1990; Yunus, 1993; Koc, 1998 and Rahji et al., 2008), which paid less attention to the impact of risk and uncertainty on agricultural production, this study captured the influence of rainfall- a critical climate variable (non-price factor) on rice supply in Nigeria. 6 It has also been established that there are variations in the demand and supply of rice in different regions and geopolitical zones in Nigeria (NLSS, 2004; Erenstein et al., 2004; NFRA, 2008; Bashorun, 2013; Adeyeye, 2012). This study analysed demand from the perspective of geopolitical zone. This will enable the problem of disequilibrium in supply and demand of rice to be examined on regional basis rather than the conventional approach of focusing on the national level on the assumption that there are no ecological and cultural variations among the people of Nigeria. In addition, this study will also provide insight into the extent to which government policies (trade liberalisation and importation) are effective in impacting rice supply in Nigeria. Agricultural response in the form of increased food production could assist in moderating inflation and thus contribute to the process of internal adjustment (Muchapondwa, 2008). Furthermore, as long as consumers continue to exhibit preference for foreign rice above local rice, the goal of rice self-sufficiency may not be totally met. Thus, consumer preferences and switching factors which are analysed in this study are very vital. This is an area which has receivedvery littleattention in Nigerian studies at the national level (Nwachukwu et al., 2008; Agwu, et al., 2009, Adeyeye, 2012). Methodologically, the strength of this study lies in the use of co-integration and error correction procedure in modelling supply response in Nigeria as against the traditional Nerlove‘s model. McKay et al.(1998) asserts that the advantage of using Error Correction Model (ECM) include the fact that spurious regression problems are bypassed, and that ECM offers a means to incorporate the levels of the variables x and y alongside their differences. The ECM also conveys information on both short-run and long-run dynamics. Nickel (1985) demonstrates that the ECM specification represents forward-looking behaviour, such that the solution of a dynamic optimisation problem can be represented by an Error Correction Model. The ECM can, thus, be interpreted as describing farmers reacting to ‗moving‘ targets and optimising their objective function under dynamic conditions. In summary, this research isolated the factors that affect demand, supply response as well as preference switches of rice and generated policy measures to effectively manage demand and boost domestic supply, as contributions to the goal of attaining rice self-sufficiency in Nigeria.This is in consonance with the broad agricultural policy objectives of thevarious tiers of government, which include the attainment of self-sufficiency in food and 7 fibre,improvement in the socio-economic welfare of the people, reduction in the rate of food price inflation, diversification of the country‘s sources of foreign exchange earnings through the rejuvenation of agricultural export commodities and the production of raw materials for local agro-based industries.It is also in tune with the ongoing Agricultural Transformation Agenda in the country(Sanusi, 2003; Bello, 2004, Akinwumi, 2012). 1.6 Organisation of the study The thesis is structured into five chapters.The first chapter is the introductory chapter and contains the background, statement of the problem, objectives, hypotheses and justification of the study. The second chapter reviews concepts, theories and existing literature relating to demand, supply and preference switch of rice. Chapter three presents the research methodology consisting of the study area, data and sampling procedure as well as analytical techniques employed. In the fourth chapter, the results of the analysis on socioeconomics, determinants of demand, supply response and preference switch are presented and discussed in detail. The final chapter is the inferential part, where summary, conclusion, policy implications, recommendations and area of further research were rendered. 8 CHAPTER TWO THEORETICAL/CONCEPTUAL FRAMEWORK AND LITERATURE REVIEW This chapter presents the basic concepts and theories of demand, supply and preference for commodities. It further reviews the methodologies of demand analysis and issues in supply response analysis. The chapter also covers empirical review on rice demand, supply and import trends as well as determinants of rice demand, preference switch and supply response. 2.1 Theoretical and conceptual reviews This section reviews theories and concepts of demand, supply and preference for food. 2.1.1 Concept and theory of food demand Seale et al. (2003) described demand analysis as a science of consumer choice or preferences among different goods and services. Since the demand for any good or group of goods is dependent on the price and availability of other products, analysing consumer demand is essentially the act of analysing consumer preferences, that is, how the consumer chooses to allocate his income among different products. Economists often use the concept of utility to define the level of satisfaction or welfare that comes from a specific allocation of income among different products. Deaton and Muellbauer (1980) states that the basis of demand analysis is the problem of how to maximise utility subject to a given level of income, the latter also being known as budget constraint. This can be expressed as: 𝑀𝑎𝑥𝑖𝑚𝑖𝑧𝑒 ∪= ʋ 𝑞1 , 𝑞2 ……………𝑞𝑛 … . ………………………………………………… . . (1) Subject to ∑𝑝𝑘 , 𝑞𝑘−𝑥 where ∪ is a utility function of the quantities of goods consumed, x is total income, and p and q are prices and quantities,respectively. Solving this maximisation problem by setting up the Langrangean function will lead to a set of demand equations that expresses the quantity demanded for each good as a function of the price and total income: 𝑞𝑖− 𝑔𝑖 𝑥, 𝑃 ………………… . ……………………………… . …………………………………… 2 where 𝑃 is the vector of commodity prices 9 This type of demand function, based on utility maximisation, is known as a Marshallian or uncompensated demand function. For a logarithmic utility function, both income and price elasticity can be calculated by taking the derivative of the Langrangean function, resulting in the following equation: 𝑛 𝑑𝑙𝑜𝑔𝑞𝑖 − 𝜂𝑖𝑑𝑙𝑜𝑔𝑥 = 𝜇𝑖𝑗 𝑑𝑙𝑜𝑔𝑝𝑖 ……………………… . . ………………………………… (3) 𝑗 =1 where ηi is the income elasticity and μij are the uncompensated price elasticities. So that changes in prices and total expenditure do not violate the budget constraint in the demand function, the following conditions on the elasticities must hold, 𝑛 𝑛 𝑤𝑗η𝑖 − 1 𝑎𝑛𝑑 𝑤𝑖μ𝑖𝑗 + 𝑤𝑗 − 0 …………… . …………………………………………… (4) 𝑗=𝑖 𝑗=𝑖 where w is the budget share. These two conditions are known as Engel and Cournot aggregation, respectively, and together are sometimes referred to as the adding up restriction. The Marshallian demand function is the solution to the consumer‘s problem of maximising utility subject to the budget constraint. However, the consumer‘s problem can also be expressed as one of minimising total expenditures or costs subject to a predetermined utility level or, 𝑀𝑖𝑛𝑖𝑚𝑖𝑧𝑒 𝑥 − 𝑝𝑘𝑞𝑘𝑠𝑢𝑏𝑗𝑒𝑐𝑡 𝑡𝑜 𝑣(𝑞1 , 𝑞2 ……………𝑞𝑛) − 𝑢 …………………………… (5) The solution to the problem is the Hicksian demand function, which is equivalent to the Marshallian demand function when evaluated at the optimal utility level 𝑞𝑖 − 𝑕𝑖 𝑢, 𝑝 − 𝑔𝑖 𝑥, 𝑝 ……………………… . ……………………………… . . ……………… (6) The Hicksian demand function is also known as the compensated demand function, since it represents demand when utility is held constant. Price elasticities derived from the Hicksian demand function are called ―compensated‖ or ― Slutsky‖ price elasticities and are equal to the uncompensated price elasticity (also called ‗‘Cournot‘‘ price elasticities) plus the product of the income elasticity and the budget share. This is stated thus: 𝜀𝑖𝑗 = 𝜇𝑖𝑗 + η𝑖w𝑗 ……………………………… . ………………………………………………… (7) where εij is the Slutsky price elasticity In this study, we follow the assertion of Attanasio (1999) and Olayemi (2004b), that effective food demand is equal to food consumption. Food consumption is a component of the food system at which people‘s nutritional needs are met at individual or household level. 10 Familiarity with modern consumption research requires understanding of three fundamental modules: Keynes‘s Absolute Income Hypothesis (AIH), Friedman‘s Permanent Income Hypothesis (PIH) and Modigliani‘s Life Cycle Hypothesis (LCH). Modern consumption research is however based, to varying degrees, on at least one of these approaches. The concept of consumer demand refers to the variations in the quantities of a commodity that a consumer is expected to buy at specified (different) prices and time period, assuming that his income, prices of other (substitute) commodities, tastes and preferences, and all other pertinent factors remain constant. In mathematical form: 𝑄𝑑 = 𝑓 𝑝, 𝑦, 𝑝 ∗ , 𝑎, 𝑧 …………………………………………… . . …………………………… . (8) where: Qd = quantity of commodity demanded, p = price of commodity, y = consumer‘s income, a = taste and preferences p* = prices of related commodities (substitute or complement), z = other factors. According to Olayemi (2004b), demand theory suggests an inverse (negative) relationship between the quantities demanded of that product and its (own) price. The relationship between the quantity demanded of one commodity and price of other commodities may be positive, negative or zero. This is called a cross-price effect. Relationships are expected to be positive for substitute products. For complementary products, the relationship is expected to be negative. That is, an increase in the price of one commodity may lead to a decrease in demand for the other. The relationship is expected to be zero for independent products, meaning that the price of one product does not affect the demand for the other. Also predicted by economic theory is a direct relationship between the consumer‘s income and the quantity demanded of a product at any given price, that is, as consumer‘s income (y) increases, quantity demanded (q) is expected to increase. But based on Engel‘s surveys of families‘ budget and expenditure patterns, Engel (1974) notes that, with rising incomes, the share of expenditures for food products declines. The resulting shift in expenditure affects demand patterns and employment structures but Engel‘s law does not suggest that the 11 consumption of food products remains unchanged as income increases. It suggests that the consumers increase their expenditures for food products (in percentage terms) less than their increased income. The poorer a family is, according to Engel‘s law, the greater the proportion of the total amount of money that must be used for food. Within a country, the poor spend a higher proportion of their income on food than the rich in the same society, and at the aggregate level, poor countries spend more of their Gross National Product (GNP) on food than wealthy ones. Several factors may affect a product‘s elasticity of demand, but, generally, it is true to say that essential goods have inelastic demand, while luxury goods have elastic demand. Since food is regarded as an essential good, human beings need food in order to live. Once humans have enough food to satisfy their needs, they do not generally buy more food. So, consumers demand for food is income-inelastic. As consumers‘ incomes increase, households spend their money on luxuries (such as manufactured goods, holidays, and so on). The producers of these products, in turn, receive higher incomes (one person‘s spending is another person‘s income). Another noticeable economic theory suggests that as output or supply of a product increases, its price will fall. As the price of a product falls, normally consumers will demand more of it, but the demand for food is price inelastic. In fact, any fall in the price of food effectively increases consumers‘ disposable income, and they are likely to spend that money on more luxuries (Olayemi, 2004b). Consumers may continue to make purchases on the basis of habit if prices have changed. Tastes and preferences of individual consumers may change for a variety of reasons such as age, education and social status. For example, consumer education about health and nutrition may influence the type of foods purchased. Similarly nutritional knowledge of household head plays an important role in the consumption of food items(Agwu et al., 2009). In line with the foregoing, this study considered the effect of price, income and other socioeconomic factors on rice demand and preference switch.Relevant elasticities were also computed in this regard. Since it has also been established that demand is also a function of tastes and preferences borne out of socio-cultural affiliations, a geopolitical zone analysis of factors affecting demand is relevant in Nigeria, as considered in this study. 2.1.2 Supply response theory As stated by Muchapondwa (2008), the modelling of the aggregate supply response has its foundations in the theory of the firm. The interest is on the output supply function, and not on 12 input demand functions. Hence, the commonly used approach of expressing the firm‘s problem in an output perspective is usually employed. Such an approach assumes that optimisation has already been achieved in the input space and that the firm uses the least cost combinations for the production of any output level. This least cost approach is conceptually plausible because producers would just want to produce a given output with the minimum cost outlay rather than try to directly optimise in the input space by equating marginal factor productivity to marginal factor cost. Producers are only aware of the costs they pay for inputs and do not generally have an idea of the input marginal productivities. A profit maximising firm produces output up to the point where it equates marginal revenue to its marginal cost. When producers are price-takers, as the general case for farmers, profit maximisation behaviour equates the marginal cost to price. As such, the firm‘s supply function is simply its marginal cost function. The supply function is defined only in the range where price is greater or equal to the minimum of the average variable cost. Hence, the quantity of a product produced and supplied depends on its own price, the prices of substitute and complementary products, and the prices of inputs(McKay et al., 1999). Supply can, thus, be expressed as the inverse of the marginal cost function. The fundamental result from the theory of the firm is that price is the most important determinant of supply (Rahji, 1999; Begum et al., 2002). Therefore, the price of rice is included in the analysis. Muchapondwa (2008) further claims that the analysis underlying the theory of the firm assumes instantaneous response between inputs and outputs, which is not applicable to agriculture. Firstly, the agricultural sector is characterised by biological lags between input application and output production. Secondly, for the agricultural firm, the technical rules imply that the production function may actually change during the course of the production process. Thirdly, for agricultural firms, there exist technological and institutional factors which prevent intended production decisions from being fully realised during some periods. Fourthly, the assumption of perfect knowledge and foresight is not valid for the majority of agricultural firms.The agricultural sector is characterised by high imperfections in price and other information. Finally, the risk and uncertainty faced by agricultural firms is much higher than that faced by other standard firms. As a result, the production behaviour of agricultural firms might be expected to divert from what the theory of the firm stipulates. For example, as a result of the presence of risk and uncertainty, farmers might not have the profit maximisation goal, but, rather, they might seek to minimise risks and maintain food security. Modifications and extensions to the theory of the firm would thus be needed to capture the 13 realistic production processes of the agricultural firms in any attempt to model agricultural supply response. All the above problems have been dealt with in the literature in a number of ways. The generic solution to these problems has been the use of dynamic models in modelling agricultural supply response (Deb, 2003). This was adopted in this work. Risk, technical and institutional factors were also included in the analysis to reflect the peculiarity of the agricultural firm. Cobweb theorem The prices of agricultural goods fluctuate over time because of unplanned variations in supply and the difficulty of altering supply in the short run. This fluctuation in prices is explained by the Cobweb theorem, which represents a dynamic model, that farmers base their production decisions for next year (Qt+1) on the current price (Pt). Generally, the higher the current price, the more they will be willing to produce next year. This implies that the quantity to be supplied next year is a function of the current price. This means that current supply quantity (Qt) is a function of last year`s price (Pt-1) and that current supply is not a function of current price. However, the current demand for the commodity is affected by and is a function of the current price. Over all, fluctuations in the price from one year to the other may steadily approach the equilibrium price, resulting in convergent cobweb model; or the fluctuations may become wider and wider over successive periods, leading to a divergent cobweb model (Gujarati, 2003; Olayemi, 2004a; Gujarati and Sangheeta, 2007; Rahji and Adewumi, 2008). In line with the cobweb theorem, lag values of prices and output were included in the supply response analysis specification for this study.The coefficients of adjustment to equilibrium in our supply model were also estimated. 2.1.3 Consumption preference theory The theory of household consumer behaviour is based on the concept of consumer preference and assumed existence of consumer utility functions. The theory assumes that, when a consumer is faced with alternative bundles or ‗baskets‖ of commodities each of which has some amounts of utility content, he/she will prefer a bundle with higher utility content to one with lower utility content. The consumer utility function defines the satisfaction which can be derived from various commodity bundles within his or her choice range, provided every 14 consumer is a utility maximiser and there exists a perfect competition in the market (Olayemi, 2004b; Nwachukwu et al., 2008). Utility was originally based on the theory of cardinal utility, which assumes that utility is measurable on cardinal scale, that is, consumers are assumed to be able to assign numerical utility value to alternative bundles of commodities. This later developed into a simpler concept of ordinal utility and indifference curve. The concept of ordinal utility proposes that the consumer is assumed to be rational enough to, at least, be able to rank commodities in order of preference (Atanasio, 1999).The Indifference curve describes the locus of combination of commodities that gives the same level of satisfaction. This concept of utility is relevant to this study as preference switch for different types of rice- ‗imported‘, ‗improved domestic‘ and ‗local‘-were analysed based on the utility theory. For the purpose of this study, the utility function is defined as follows: u1  f (q1,q2 ,q3 ,q4 ) ...............................................................................................................(9) where U1= utility derived from rice consumption q1= Imported rice q2 = Agric. rice q3 = Local rice q4 = Other Food Commodities This could be transformed as: u2  f kq1,kq2,kq3,kq4 for k > 0 without changing the preference ranking…...………....(10) Above all, the overriding assumption is still that the consumer is rational and has the objective of utility maximisation. However, consumption preferences have been hypothesized to be a function of many factors. Hence, utility maximization, in line with consumption preference, is constrained by income, price and other demographic/socioeconomic factors in this study. 2.2 Methodological review In this section, various methods of analysing demand are discussed. Also, issues in time series modelling and methods of estimating supply response are explained in detail. 15 2.2.1 Estimation of demand Estimation of demand for goods and services has attracted the attention of both theoreticians and empiricists, and very dense literature is available on this subject. Some of the past studies have ignored the required connections between the theory and empirical analysis, and concentrated on the estimation of single demand equations. Food demand analysis is dominated by the econometric estimation of demand systems based on aggregate market data and steady progress that have been made in analytical techniques (FAO, 2007). The main approaches are the use of simple Engel curves and the use of demand systems. Demand is empirically measured with the use of mathematical symbols involving the estimation of functional forms and it is the most straightforward and convenient way in demand estimation. Some of the existing models in use include the probability models, such as the Probit, and Logit models for qualitative dependent variables, Heckman two-stage model and the Tobit model applicable to continuous dependent variables of truncated or censored data(Tobin, 1958; Akinyosoye, 2007; FAO, 2007). Others are flexible functional forms that estimate completedemand systems,such as the Translog Demand System (TDS) of Christensen et al.(1975); Generalised Leontief Demand System and Linear Expenditure System (LES) proposed by Stone (1954); Quadratic Expenditure Demand System (QUES), Rotterdam Model of Theil (1965) and the Almost Ideal Demand System (AIDS) of Deaton and Muellbauer (1980). This new modelling has attracted a great deal of attention, and has been used extensively in empirical works. Moreover, extensions of the standard AIDS has been developed to make this modelling as rich as possible. Among these are Inverse AIDS, by Moschini and Vissa (1992), Quadratic AIDS, by Banks et al. (1997) and more recently Semi- Flexible Almost Ideal Demand System, by Moschini (1998). In order to take care of the censored and truncated nature of the NLSS data, this study utilisedthe Tobit model to estimate the demand for rice for the entire data set and further tried the AIDS model on non–zero observations for flexibility of estimation. These will be fully discussed in the chapter three. 16 2.2.2 Methodological issues in time series modelling (1) Serial correlation theory A common finding in time series regressions is that the residuals are correlated with their own lagged values. This serial correlation violates the standard assumption of regression theory, that disturbances are not correlated with other disturbances. As a result, we find that OLS is no longer efficient among linear estimators; standard errors computed using classical OLS procedures are not correct, and are generally understated; and if there are lagged dependent variables on the right-hand side of the equation, OLS estimates are biased and inconsistent (Greene, 1997, Gujarati and Sangheeta, 2007). In view of the problems of serial correlation in time series regression, it is important to examine the residuals of a regression equation for evidence of serial correlation before using an estimated equation for statistical inference (for example, hypothesis tests and forecasting). The general specification of a regression with an autoregressive process of order p, AR(p) error is given by: yt  X  t ...……………………………………………………………………...............(11) t  1t1 2t2 .............ptp et ………………………..…...............................(12) The autocorrelations (i) of a stationary AR(p) process gradually die out to zero, while the partial autocorrelations for lags larger than p are zero. The most widely used approach for testing for existence of serial correlation is the Durbin-Watson (DW) statistic. Other diagnostic tools include the Q-statistic and the Breusch-Godfrey LM test. The DW statistics tests for first-order serial correlation, AR(1), or linear association between adjacent residuals from a regression model. It is a test of the hypothesis: =0 in the specification: t  t1  et , where the t is the vector of OLS residuals in the regression. yt  X  t . N 2 N 2 The test statistic is: d et  et1 /et . t2 t1 If there is no serial correlation, the DW statistic will be around 2. The DW statistic will fall below 2 if there is positive serial correlation (in the worst case, it will be near zero). If there is negative correlation, the statistic will lie somewhere between 2 and 4 (Greene, 1997) 17 Several techniques are available for estimating models containing autoregressive residual terms (AR models). The most widely used are the Cochrane-Orcutt and Prais-Winsten iterative procedures, which are multi-step approaches designed so that estimation can be performed using standard linear regression. However, as pointed out by Greene (1997), these approaches suffer from important drawbacks which occur when working with models containing lagged dependent variables as regressors, or models using higher-order AR specifications. Such problems are usually overcome by estimating the higher order AR (q) process by maximum likelihood techniques and, better still, by specification and estimation of autoregressive integrated moving average (ARIMA) model (Gujarati and Sangeetha, 2007). The fact that assumption of stationarity does not hold in this work and that ARIMA model is not based on proven theories consistent with the agricultural industry will not make the model applicable in this study. (2) Problems with non-stationary time series A number of approaches for modelling serially correlated time series, such as the ARIMA, are based on the assumption that the time series are stationary. A series is said to be (weakly or covariance) stationary if its mean, variance and covariance do not vary with time. Any series with time varying mean, variance and auto covariance is said to be non-stationary.The canonical example of a non-stationary series is found in the random walk: yt  yt1  t …………………………………………………………..….........................(13) where t is a stationary random disturbance term, and the AR(1) process is characterised by a unit root, that is, 1. The series y has a constant forecast value, conditional on t, and the variance is increasing over time. The random walk is a difference stationary series since the first difference of y is stationary: yt  yt1  t ………………………………………………………………..…………….(14) A difference stationary series is said to be integrated and is denoted as I(d), where d is the order of integration. The order of integration is the number of unit roots contained in the series, or the number of differencing operations it takes to make the series stationary. For the random walk above, there is one unit root, so it is an I(1) series. Similarly, a stationary series is I(0). 18 Standard inference procedures do not apply to regressions, which contain an integrated dependent variable or integrated regressors. This is because the application of least square regression to equations containing non-stationary series results in spurious regression 2 (Granger and Newbold, 1974). It is a situation where such regression produces high R and t- ratios that are biased towards rejection of the null hypothesis of no relationship even when there is relationship between the variables.Against this background, it has become necessary to check whether a series is stationary or not before using it in a regression. The formal method for testing the stationarity of a series is the unit root test: that is, a test that: H0:  1against Ha:  1 in regressions yt  yt1 t involving a series yt where  t is a stationary random disturbance term (Dickey and Fuller, 1981). (3) Approaches to unit root testing There are three widely used unit root tests: the Dickey-Fuller (DF), Augmented Dickey- Fuller (ADF) and Phillips-Perron (PP) tests. Others are Kwiatkowski-Philips-Schmidt-Shin, Elliot- Rothenberg-Stock point-optimal, Ng-Perron (Gujarati and Sangeetha, 2007; EViews, 2009). For a time series Yt, two forms of the ADF test, which are based on t-test of significance of the coefficient associated with the lagged value of the series (Yt-1) in any of the following two forms of ADF regression equations: p Yt 1Yt1  jYt j  t …...…………………………………………………………(15) j1 p Yt 0 1Yt1 2t  jYt j t ...……….……………………………...…………(16) j1 where  t for t = 1,…., N is assumed to be Gaussian white noise. Equation (15) is with no constant and trend while (16) is with both constant and trend. The number of lagged term p is chosen to ensure that the errors are uncorrelated. When1  0 , the time series is non- stationary so that standard asymptotic analysis cannot be used to obtain the distribution of the test statistics (Gujarati and Sangeetha, 2007). Various researchers (Fuller, 1976; Dickey and Fuller, 1981, Guilkey and Schmidt, 1989) have designed Monte Carlo experiments to generate critical values that can be used for testing purposes. As an alternative to the 19 inclusion of lag terms to allow for serial correlation, Phillips and Perron (1988), cited in Gujarrati (2003)proposed a nonparametric method of controlling for higher-order serial correlation in a series. The test regression for the Phillips-Perron (PP) test is the AR(1) process: yt  yt1 t ………………………………..……...……………………………..(17) while the ADF test corrects for higher order serial correlation by adding lagged differenced terms on the right-hand side, the PP test makes a correction to the t-statistic of the coefficient of the lagged residual term in the AR(1) regression to account for the serial correlation in  . A consortium of tests for stationarity as obtainable in EViews 5.0 version was employed in this study. They gave similar results. The ADF test was reported in this study because it caters for AR(n) process in case the model did not follow AR(1) and because of its wide usage and acceptability. (4) Cointegration and error correction theory The need to include non-stationary series in econometric models spurred the development of cointegration techniques pioneered by Granger (1969) and Engle and Granger (1987). Engle and Granger (1987), observe that a linear combination of two or more non-stationary series may, in fact, be stationary. Such a stationary linear combination of I(1) non-stationary time series are said to be “cointegrated”. The stationary linear combination is called the “cointegration equation‖ and may be interpreted as a long-run equilibrium relationship between the variables. Two or more variables would cointegrate, that is, exhibit long-run equilibrium relationship(s), if they share common trend(s) (Gujarati and Sangheeta, 2007), such that, even though the I(1) series may drift apart in the short run, a stable long-run relations is guaranteed between them. Further evidence from cointegration theory suggests that, if two variables are cointegrated, the finding of no causality in the relationship between themis ruled out (Granger, 1969; Granger and Newbold1974), just as the possibility of the estimated relationship being spurious is also ruled out (Gujarati and Sangheeta, 2007). As pointed out by Engle and Granger (1987), the relationship between two or more non- stationary economic variables that are co-integrated may be given an error correction representation. To illustrate, suppose we seek to model a simple econometric relationship of the form: Yt   Xt ut …………………………………..……………………………………….(18) 20 and we know that both X and Y are non-stationary I(1) series that are cointegrated. The idea behind ECM is that short-term ―shocks‖, like those occasioned by changes in policy environments, weather conditions and random factors disturb the long-term equilibrium relationship which exists between X and Y, after which the two variables returns to their equilibrium. Therefore, ECM tries to capture the short-term and long-term dynamics in the relationship, and a simplified representation for the cointegrating equation in (18) may be written as: Yt  Xt1  (Yt1  Xt1) ………….…..………………………………………….(19) In this simple ECM, ̂ captures the short-term relationship between X and Y; ̂ and ̂ capture the long-term relationship between X and Y; while ̂ gives the rate at which the model ―re-equilibrates‖, that is, returns to its equilibrium. Formally, ̂ tells us the proportion of the dis-equilibrium, which is corrected with each passing period. Noting that, the second term in the right hand side of (19) is simply ut1 , where ut is the stochastic error term in the cointegrating equation (18), the Engle and Granger (1987) approach to ECM consists of three steps: 1. Estimation of the cointegrating regression Yt  Xt  et 2. From these estimate, the residual term eˆ Ytˆt ˆXt are generated; and 3. The residual term is included in the short-term equation Yt  Xt1  eˆt1 as an ―error correction term‖. The coefficients ̂ , ̂ , ̂ and ̂ obtained in this process are then interpreted and used as earlier illustrated with the simplified ECM representation in (19). The Engle and Granger (1987) approach to testing for co-integration is to test for stationarity of the stochastic residuals generated in the second stage of the three stages ECM procedure. (5) Vector autoregression and vector error correction modelling According to Gujarati and Sangheeta (2007), the need to analyse the dynamics of simultaneous relations that often exists among economic time series led to the analysis of vector autoregressive model by incorporating co-integration and error correction mechanism into Vector Auto Regression (VAR). Cointegration and Vector Error Correction Modelling (VECM) have emerged as the currently dominant approach to econometric analysis of time series.VAR is commonly used for forecasting systems of interrelated time series and for 21 analysing the dynamic impact of random disturbances on the system of variables (EViews, 1998, 2009). The VAR approach sidesteps the need for structural modelling, which is often complicated by the fact that endogenous variables may appear on both the left and right side of the equation, by modelling every endogenous variable in the system as a function of the lagged values of all the endogenous variables in the system. The mathematical form of a VAR is: yt  A1yt1 ....... Ap ytp Bxt et .……………………………………………...………(20) where yt is a k vector of endogenous variables, xt is a d vector of exogenous variables, A1.....Ap and B are matrices of coefficients to be estimated, and et is a vector of innovations (stochastic residual terms) that may be contemporaneously correlated with each other but are uncorrelated with their own lagged values, and uncorrelated with all of the right-hand side variables. Since only lagged values of the endogenous variables appear on the right-hand side of each equation, there is no issue of simultaneity, and OLS is the appropriate estimation technique. Note that the assumption that the disturbances are not serially correlated is not restrictive because any serial correlation could be absorbed by adding more lagged y‘s. A Vector Error Correction model (VECM) is a restricted VAR that has cointegration restrictions built into the specification, so that it is designed for use with non-stationary series that are cointegrated. The VECM specification restricts the long-run behaviour of the endogenous variables to converge to their cointegrating relationships while allowing a wide range of short-run dynamics (EViews, 1998; EViews, 2009). The general form of a VECM is obtained by rewriting the VAR in (20) as follows: p1 yt yt1 iyt1 Bxt et ……………………………………….…………..…(21) i1 p p where  Ai  I and i  Aj i1 ji1 In this special case, yt is a k-vector of non-stationary I(1) variables, xt is a d vector of deterministic variables, such as a constant and/or trend terms, and et is a vector of innovations. Granger‘s representation theorem (E-views, 2009) asserts that, if the coefficient matrix  in (21) has reduced rank r −𝜀𝑖 …………… . ………………………………………………………… . . (28) 𝑌𝑖> 0 if 𝑋𝑖𝛽 < −𝜀𝑖 The model was estimated over the entire selected survey sample using Maximum Likelihood estimator routine in LIMDEP (Vogelvang, 2005; Long and Freese, 2006). In the estimation of 2 Tobit model, the conventional coefficient of determination R is an inappropriate measure of goodness of fit (Vogelvang, 2005). To test the specification of such models, an LR-test is used by obtaining Lu, which is the value of the log-likelihood function of the unrestrictedly estimated model and LR, the value of the log-likelihood function of the restricted estimated equation that has only the intercept as regresssor. Then, the LR-test statistic is LR = -2{ln (LR 2 – ln (Lʋ)}. LR has an asymptotic X (K-1)distribution under the null hypothesis of zero-slope coefficients. The LR-test statistic is usually a component of LIMDEP output when Tobit 2 models are estimated with constant terms. The Pseudo-R is also an accompanying result from LIMDEP output and its value indicates the robustness of the Tobit model estimates as it gets closer to unity. The marginal values directly generated by the Tobit model are estimates of elasticities. Four separate Tobit regression models were estimated for aggregate and various rice commodities. 59 Variables description The variables for the purpose of this study are specified as follows: Yi = F (P, Y, S, L, U)…...………………………,,…………………………………………(29) Where Yi = Expenditure share on rice commodities P = Vector of price Variables Y= Vector of Income Variable S = Vector of Socioeconomic Variables L= Locational Dummies U= Stochastic Term Definitions of dependent variables Y1 = Aggregate expenditure share ofrice (N) Y2 = Expenditure share ofimported rice (N) Y3 = Expenditure share ofimproved domestic (agric.) rice (N) Y4 = Expenditure share of local rice (N) Note: within the context of this study, the followings are defined thus: Imported rice: Rice varieties produced outside Nigeria, especially from Thailand and other Asian countries, polished, packaged and imported into the Nigerian market. Improved Domestic Rice (Agric Rice): Rice varieties that are products of domestic varietal improvement, such as FARO series, Tox, ITA series and NERICA Local Rice:Indigenous rice commodities grown domestically in Nigeria, such as Ofada, Gboko, Abakaliki Definitions of explanatory variables Socioeconomic variables X1–Household Head‘s Age (years) X2–Primary Occupation (D=1 farming, 0 if otherwise) X3–Household Size (number-Adult equivalent) X4–Education (No of years of education) X5–Marital Status (1-Married, 0 if otherwise) X6–Membership of Community Organization (1= Member, 0 if otherwise) X7–Total Asset Value (N) X8–Non-Food Total Expenditure share (N) 60 Locational variables X9–North-East Region (1, 0 otherwise) X10–North-West Region (1, 0 otherwise) X11–South-East Region (1, 0 otherwise) X12–South-South Region (1, 0 otherwise) X13–South-West Region (1, 0 otherwise) X14–Location dummy (D=1 Rural, 0 if otherwise) Note: North–Central is chosen as the base and hence not included in the model Income and price variables X15–Household total expenditure adjusted for regional cost difference (as a proxy for income) (N/month) X16–Price of Imported Rice (N/kg) X17–Price of Agric. Rice (N/kg) X18–Price of Local Rice (N/kg) X19–Price of Yellow Garri (N/kg) X20–Price of White Garri (N/kg) X21–Price of Yam Tuber (N/kg) X22–Price of Brown Beans (N/kg) X23–Price of White Beans (N/kg) X24–Price of Millet (N/kg) X25–Price of Guinea Corn (N/kg) X26–Price of White Maize (N/kg) X27–Price of Yellow Maize (N/kg) Note: *Major food commodities consumed in Nigeria include root and tuber as well as cereals, thus the prices of major root and tuber products as well as other cereals within the limit imposed by the available prices in the NBS, 2004 price data were included in the demand analysis. Beans has often been traditionally associated with rice consumption, hence the need to include beans price in the demand model. *A cross–sectional analysis is involved in this study; hence, there was no need for price deflation, trending or addition of seasonality variable in the demand analysis. 61 3.3.3 Linearized AIDS Model In order to ensure a more flexible analysis of demand, this study equally tried theAlmost Ideal Demand System (AIDS) on non-zero observations of the NLSS data. AIDS model is a flexible functional form that is based on duality theory and a two-stage budgeting procedure. This model is quite useful for providing insight into how consumers allocate expenditure among disaggregated food commodities and how they make decisions concerning food purchases (Akbay and Boz, 2001). Some important advantages of the AIDS model are that the expenditure function from which the AIDS model is derived is flexible. The model also allows for testing and imposition of homogeneity and symmetry restrictions, thus conserving degree of freedom. Furthermore, the model gives an arbitrary first-order approximation to any demand system, which satisfies the axioms of choice exactly and lastly the underlying class of preferences contains desirable aggregation properties, and largely avoids the need for nonlinear estimation (Deaton and Muellbauer, 1980). The stochastic version of the AIDS budget share demand function can be written as: 𝑀 e𝑤𝑖 = 𝛼𝑖 + ∑ 𝑛 𝑗=1 𝛾𝑖𝑗 𝑙𝑛𝑝𝑗 + 𝛽𝑖 ln + 𝑒𝑖 …… . ………………………………… . ……… . (30) 𝑃 th where 𝑤𝑖 is the budget share of the i good, M is the total consumption expenditure, Pj is the th price of the j good, P is a properly defined price aggregator. The AIDS model is based on the consumer‘s expenditure function, as seen clearly in equation 30. The equation expresses the budget share of a given commodity as a function of total expenditure and prices. The open form of the price aggregator is given by: 𝑛 𝑛 𝑛 1 ln 𝑝 = 𝛼0 + 𝛼𝑖𝑙𝑛 𝑝𝑖 + 𝛾𝑖𝑗 𝑙𝑛𝑝𝑖 𝑙𝑛𝑝𝑗 …………………………………………… (31) 2 𝑖=1 𝑖=1 𝑗 =𝑖 where the coefficients are coming from the expenditure function of an individual household. Because of the existence of non-linear parameters and difficulties in the estimation of constant term in the price index expressed in the preceding equation, it is difficult to achieve convergence. To circumvent these difficulties, the linear approximation AIDS (LAAIDS) model was substituted for the original model in many applied studies. This model involves the replacement of log P with simpler index used by Stone (1954) and Akbay and Boz (2001). 62 𝑙𝑛𝑝 = ∑𝑛𝑖=1 𝛼𝑖𝑙𝑛 𝑝𝑖 …………………………………………………………………………………………………………....(32) With the following parameter restrictions, equation (31) satisfies the adding up, homogeneity and symmetry properties derived from the standard demand theory. Ʃ 𝛼𝑖 = 1, Ʃ βi = 0, Ʃ 𝛾𝑖𝑗 = 0, Ʃ 𝛾𝑗𝑖 = 0 and 𝛾𝑖𝑗 = 𝛾𝑗𝑖 Expenditure and price elasticities then can be easily derived as follows: ɳ𝑖 = 1 + 𝛽𝑖/𝜔𝑖……………………………………………………………………………………………………………………………………………… (33) Ԑii = -1 + (𝛾𝑖𝑗 /𝑤𝑖) – 𝛽𝑖 ...........................................................................................................(34) Ԑij = (𝛾𝑖𝑗 /𝑤𝑖) – 𝛽𝑖𝑤𝑗 ………………………………………………...……………………...(35) where ɳi is the expenditure elasticity, 𝑤𝑖 is the budget share of good i, Ԑii is the own price elasticity and Ԑij represents the cross-price elasticity, in Marshalian terms (uncompensated). Compensated (Hicksian) price elasticities, eij, can be derived easily by using ɳi,Ԑii and Ԑij and the following relation: eij = Ԑij + 𝜂𝑖* 𝑤𝑗 ………………………………………………………..…………………...(36) A system of share equations based on equation (30) and subject to the restrictions (adding-up, homogeneity, and symmetry) is estimated using iterative Seemingly Unrelated Regression (SURE) method of Zellner. This method is equivalent to Full Information Maximum Likelihood (FIML) estimation. The adding-up property of demand causes the error covariance matrix of system to be singular. So, one of the expenditure share equations is dropped from the system to avoid singularity problems. The estimates are invariant of which equation is deleted from the system. The coefficients pertaining to the expenditure share equation of local rice which is dropped from the system in the estimation stage are obtained by using the adding-up property. Symmetry is imposed during the estimation of the system of equations. The AIDS model in equation (30) is modified by the inclusion of some household variables, namely: X1–Household Head‘s Age (years) X2–Primary Occupation (D=1 farming, 0 if otherwise) X3–Household Size (number- Adult equivalent) X4–Education (No of years of education) 63 X5–Marital Status (1-Married, 0 if otherwise) X6–Membership of Community Association (1= Member, 0 if otherwise) X7–Total Asset Value (N) X8–Non-Food Total Expenditure share (N) X9–Location dummy (D=1 Rural, 0 if otherwise) X10–Household total expenditure adjusted for regional cost difference (as a proxy for income) (N/month) X11–Price of imported rice (N/kg) X12–Price of agric. rice (N/kg) X13–Price of local rice (N/kg) 64 Table 2: Apriori Expectation for Demand Variables Variables Unit Expected Signs Authorities Age Years -ve Heilig(1999),Choi and Lee(2000) Agwu et al. (2009), Adeyeye (2012) Primary Occupation Dichotomous +ve/-ve Household size Country Adult Eqiv. +ve Abdulai et al. (1999), Choi and Lee(2000), Agwu et al. (2009), Bamidele et al. (2012), Oyinbo et al. Education Years of education +ve (2013) Jenson(1995); Babatunde et al. (2007), Nwachukwu et al. (2008), Agwu et al. Marital Status Dichotomous +ve (2009), Adeyeye (2012) Membership of Comm. Dichotomous +ve Adeyeye (2012) Total asset N +ve Abdulai et al. (1999) Non Food Total Exp. N -ve Zonal Dummies/ Dichotomous +ve/-ve Location Bamba et al. (2010), Adeyeye (2012) Own Price N -ve Rahji and Adewumi (2008) Nwachukwu et al. (2008), Agwu et al. (2009),Akbay,and Boz (2001), Okoruwa et al. (2008), Odusina (2008), Other Comm. Price N +ve/-ve Adeyeye (2012), , Oyinbo et al. (2013) Odusola(1997), Babatunde et al. (2008); Okoruwa et al. (2008), Adeyeye Per Capita Expenditure N +ve (2012) (income) Agwu et al.(2009), Bamidele et al. (2010) Note:All price and income Elasticities for rice are expected to be less than unity (i.e inelastic) in conformity with Engels law and past studies such as,Nwachukwu et al.(2008), Okoruwa et al. (2008), Agwu et al.(2009). Source:Composed by the author from literature review 65 3.3.4 Cointegration-ECM Analysis This study estimated the responsiveness of rice supply to price and non- price factors by applying recent time series techniques and using data spanning different pricing regimes (pre- SAP and post-SAP regimes).This study improved upon the methodology ofMcKay et al. (1999) by making use of a more recent cointegration technique, the Vector Autoregressive Error Correction Model. The most widely known single equation approach to cointegration is the Engle-Granger two-step procedure. This approach has some limitations. Firstly, it ignores short-run dynamics when estimating the cointegrating vector. When short-run dynamics are complex, this biases the estimate of the long-run relationship in finite samples. To counter this, a test based on the coefficient of the lagged dependent variable in an autoregressive distributed lag framework has been proposed (Banerjee et al., 1998). However, the parameter estimates are only asymptotically efficient on the assumption of weak exogeneity of the regressors. McKay et al. (1999) adopted this approach but there are reasons to believe that agricultural prices may not be weakly exogenous, thus shading doubt on the asymptotic efficiency and consequently validity of their estimates. Secondly, the procedure only assumes that one cointegrating vector exists leading to inefficiency in estimation in the event that more than one cointegrating vector actually exists. The Johansen estimation procedure deals with this problem but, like the Engle-Granger procedure, it presupposes that the order of integration of all the variables is the same and known with certainty. In this study, Johansen method nested in vector error correction modelling was employed since there may be more than one cointegrating relationship and it is an improvement over the Engle-Granger traditional procedure. 1. Test of stationerity The development in time series modelling points to the need to exercise some caution, by first examining the statistical properties of the series and incorporating these in the final model where necessary so as to guarantee non-spurious regression (Granger and Newbold,1974). The first step in the analysis is to identify the order of integration of the variables. The Dickey-Fuller (DF) approach and the Augmented Dickey-Fuller (ADF) can be applied to test the null hypothesis that a series contains a unit root (is non-stationary). In case of Dickey-Fuller, it involves estimating the equation below for each variable yt and testing the null hypothesis approach: 66 Ho: = 1 against the alternative H1: < 1. yt =  + t + (yt-1 + t.......………………………………………………………...(37) If the variable does not follow an AR(1) process but is AR (n), then the Augmented Dickey Fuller (ADF) test should be used; and in place of (37), we estimate: yt =  + t + (yt-1 + i liyt-i + t ………………………....................................(38) If Ho cannot be rejected, then yt contains a unit root and hence is not stationary. If its first difference is then tested and found stationary, yt is I(1). If not, yt needs to be differenced further. In this study, the Augmented Dickey Fuller test was estimated and differenced further and until stationarity was attained in the variables. 2. Vector Auto Regressive Error Correction (VECM) Having ascertained that most of the series in the economic model are non-stationary in their level, but stationary in their first difference and bearing in mind the need to accommodate the interdependence of relationships between most economic variables, the economic model was re-conceptualised as a vector autoregressive system, allowing for the possibility of cointegration among the endogenous variables. 4 yt  Bxt iyt1 yt1  et ….………………………………………………………(39) i1 where x is vector of deterministic variables, constant (C) and/or trend; y is vector of I(1) endogenous variables – Output, Area, Price, Import, Fertilizer consumption, Rainfall, Policy B,  and  are matrices of coefficients to be estimated, while e is a vector of stochastic residuals. Terms in B give the influence of the associated deterministic variables, while  represents short-term elasticities of response. And, where evidence of r<5 Cointegrating relations exist, by Granger causality theorem,   , in which  is the cointegrating vector (containing the long-run elasticities), while elements of  are the adjustment parameters in the vector error correction model. 3. Test for Cointegration This test was implemented in EViews using procedures for Johansen‘s (1992, 1995b) system- based techniques. The test utilizes a trace statistic-based likelihood-ratio (LR) test for the 67 number of cointegrating vectors in the system. In implementing the Johansen technique,however, two main issues have to be addressed. The first is the choice of the optimal lag length in the VAR system. Noting that the lag length ought to be set long enough to ensure that the residuals are white noise (EViews, 1998, EViews, 2009), and considering limitations imposed by the data (consumption of too much degree of freedom; and the result of the performance of additional lag from the Granger causality test), this study stuck to the use of one lag in the VAR. A second issue that has to be addressed is whether deterministic variables, such as a constant and trend, should enter into the long-run cointegrating space or the short-run model. Gujarati and Sangeetha (2007) observe that there are, in general, three possible ways of incorporating these deterministic components into an analysis: (a) That, if there are no linear trends in the levels of the data, a most restrictive specification would be to restrict the constant to lie in the cointegration space only, simply in order to account for the units of measurement of the variables. (b) That, a less restrictive option would be to permit a constant in both the cointegration space and the short-run model in situations where linear trends are present in the levels of the data. (c) That, with respect to the trend term, if quadratic deterministic trends are absent in the levels of the variables (which is not usually a possible long-run outcome), the least restrictive specification would be to force the trend term to lie in the cointegration space so that any long-run linear growth is captured by a linear deterministic trend in levels. EViews provides facilities for conducting and comparing cointegration tests based on five scenarios that accommodatethe suggestions above. These may be listed, from the most restrictive to the least restrictive options as follows: Option A: Assumes no deterministic trend in the data, and allows no intercept nor trend in the cointegrating equation (CE) or test VAR; Option B: Also assumes no deterministic trend in the data, and allows intercept (no trend) in the CE and no intercept in the VAR; Option C: Allows for linear deterministic trend in the data, with intercept (no trend) in the CE and test VAR; 68 Option D: Allows for linear deterministic trend in the data, with intercept and trend in the CE but no trend in the VAR; Option E: Allows for quadratic deterministic trend in the data, with intercept and trend in the CE and linear trend in the VAR. Because significant trends were not found in series in the model, the final choice among the options was based on application of the Pantula principle (Johansen, 1992), which permits joint test of the rank order of the long-run matrix and the presence of deterministic components. This involved estimating all the possible specifications, and conducting Johansen's likelihood-ratio tests for the rank order of the long-run matrix sequentially from the most restrictive to the least restrictive specification. The first time the null hypothesis is not rejected indicates both the rank order of the long-run matrix and the appropriate specification for the deterministic components (Gujarati and Sangeetha, 2007). The final stage of the analyses, having established that one cointegrating vector existed in the data, is to estimate the restricted VAR in (39) using VECM facility in EViews. Model Specification Following McKay et al. (1999) and Muchapondwa (2008), the output function adopted in this study is specified as follows: Δ𝑅𝑡 = 𝛽0 + 𝛽1 Δ𝑝𝑡−𝑖 + 𝛽2 Δ𝑅𝑡−𝑖 + 𝛽3 Δ𝑍𝑡 + 𝛽4𝑃 + λ𝐸𝐶𝑇𝑡 + 𝜀𝑖 𝑖 𝑖 𝑡 ……………… (40) where Rt is the supply in year t Pt-i are the lagged value of producer prices, Rt-i are the lagged value of supply, Zt are values of other determinants of rice supply, P is the policy variable, ECT is the error correction term 𝜀𝑡 is the stochastic disturbance. β‘s and λ are parameters to be estimated. Following Mc Kay et al. (1999), Nayaran (2005), Akmal (2007), Rahji and Adewumi (2008), Rahji et al., (2008) and Muchapondwa (2008), the variables are defined as follows: 69 Dependent variable Rt = Rice Supply in year t, Proxied by Rice Output (tons) Explanatory Variables Prt = Price of Rice in year t (N/tonne) Prt-1 = Lagged value of Price of Rice in year t (N/tonne) Rt-1 = Lagged value of Rice Supply in year t (tons) Zt‘s are: Wt = Amount of Rainfall in year t (mm) It = Rice Import Level in year t (tons) Ft = Fertilizer consumption in year t (tonne) At = Area of rice cultivated in year t (Ha) P = Policy Variable (1-Policy intervention era, 0- Non- policy intervention era) • The period before 1986 has been classified as non-policy (liberal policy) intervention era on rice, while the period from 1986 has been classified as policy intervention era on rice in Nigeria owing to ban on rice and the commencement of trade liberalization in this period. Introduction of SAP and the abolition of Commodity Boards to provide production incentives to farmers through increased producer prices started from 1986 (Ogundele, 2007; Rahji et al 2008) Typically, agricultural economists have modelled expected output as being determined by past prices (cobweb behaviour, distributed lags and adaptive expectation models). Farmers are supposed to react to recent past information and there is no use of current information. In addition to this, a study by Lopez and Ramos (1998) considered the cobweb model appropriate for basic grains and that the price farmers expect is the price they received in the preceeding period.In line with Nerlove (1956), the models portraying the structural relationship in the production of local rice can be postulated as output response. Following the partial adjustment model, the price of substitute is never considered (Gafar, 1997). Similarly, several studies like McKay et al. (1999), Rahji et al. (2008), Rahji and Adewumi (2008) equally omitted the price of substitute in their analysis owing to the consumption of degree of freedom because of the limited data points. Hence, the non-inclusion of price of substitutes in this study becomes justified. 70 Table 3: Apriori Expectation for Supply Response Variables Variables Unit Expected Signs Authorities Area ‗000 hectares +ve Mc Kay et al.(1999);Nayaran(2005), Muchapondwa(2008) Price N /tonne +ve(less than Rahji and Adewumi unity) (2008),Rahji et al. (2008),Muchapondwa(2008) Import ‗000 Tons -ve Ogundele (2007) Fertilizer Consm. ‗000 Tons -ve Muchapondwa(2008) Rainfall MM +ve Begum et al. (2002) Policy Dichotomous +ve/-ve Rahji et al. (1999), Ogundele (2007) Source: Author‘s compilation from literature review 71 3.3.5 Paired sample t-test The paired sample t-test statistics was used to estimate the direction of preference switch between various pairs of rice commodities. The paired means were checked for statistical significance. The individual means of paired rice commodities were then compared while the one with higher mean was the preferred.Thus, a switch was observed towards it. The sign of the t-statistics further confirms the direction of the switch (Straus, 1982; Nwachukwu et al., 2008 and Agwu et al., 2009). 3.3.6 Generalized Least Square Regression In estimating the socioeconomic determinants of preference switch from one rice commodity to another, a multiple regression model of three functional forms (Linear, semi-log and double-log) were employed. The fittest of the model, based on economic and statistical criteria was selected as the primary model. Four separate models were estimated. The implicit form of the model is expressed as: Yi = f ( Xa, Xb, Xc, Xd ……………………….Xq)…………………………………………(41) Following Nwachukwu et al.(2008) and Agwu et al. (2009), the variables were defined thus: Dependent variable Yi = Switch from one rice commodity to the other Y1 = Index of Preference switch from Imported to Agric. rice (Expenditure share of imported rice- Expenditure share of agric. rice) Y2 = Index of Preference switch from Imported to local rice (Expenditure share of imported rice- Expenditure share of local rice) Y3 = Index of Preference switch from Agric. to Imported rice (Expenditure share of agric. rice – Expenditure share of imported rice) Y4 = Index of Preference switch from local to Imported rice (Expenditure share of local rice – Expenditure share of imported rice) Explanatory variables Xa- Household Head‘s Age (years) Xb-Primary Occupation (D=1 farming, 0 if otherwise) Xc- Household Size (number- adult equivalent) Xd- Education (No of years of education) 72 Xe- Marital Status (1-Married, 0 if otherwise) Xf- Membership of Community Organization (1= Member, 0 if otherwise) Xg- Total Asset Value (N) Xh- North-East Region (1, 0 otherwise) Xi- North-West Region (1, 0 otherwise) Xj- South-East Region (1, 0 otherwise) Xk- South-South Region (1, 0 otherwise) Xl- South-West Region (1, 0 otherwise) Xm- Location dummy (D=1 Rural, 0 if otherwise) Xn Household total expenditure adjusted for regional cost difference (as a proxy for income) (N/month) X0- Price of imported rice (N/kg) Xp- Price of agric. rice (N/kg) Xq- Price of local rice (N/kg) However, following the detection of strong positive spatial autocorrelation with Durbin- Watson values of 0.025, 0.020 0.025 and 0.020 for the estimated equation Y1- Y4,respectively,in the ordinary least square regression model, the chosen model (semi-log) was modified using General Least Square(GLS) of first difference. We can afford to lose the first data point without necessarily transforming through Cochrane-Orcutt, Prais-Winsten or any other iterative procedures since we are dealing with relatively large samples with high degree of freedom (17, 18844) (Gujarati, 2006; Gujarati and Sangheeta, 2008). 3.4 Limitations of the Study The major limitations encountered are discussed below: 1. The study was mostly constrained by data availability. The Nigerian Bureau of Statistics (NBS) collects and releases data at distant periodic intervals, a minimum of 5 years. The 2004 NLSS data (which came after the 1999 data set) was the most current national data as at the time of analysis of this work in 2009-2011. The new set of data was released in 2012/2013 after this research work had been completed. However, except for the inclusion of few other variables and the panel nature of the data, the new sets of data were very similar to the 2004 NLSS data. The socioeconomic characteristics which are of interest in our analysis have remained relatively stable over the period of time. Hence, analysis in this study is as relevant as possible in the present time. The results from these studies are comparable with a 73 similar national study by Nigeria Institute of Social and Economic Research (Adeyeye, 2012) and other localised studies, such as Oyinbo et al. (2013). However, analysing with the NLSS data set released in 2013 may improve the relevance of the study in current time. 2. We could not incorporate factors like grain characteristics and processing in the demand and preference switch models owing to the unavailability of such qualitative data in the NLSS, 2004. 74 CHAPTER FOUR RESULTS AND DISCUSSION This chapter presents and discusses the results of various analyses. The discussion focuses on descriptive statistics of expenditure patterns and socioeconomic characteristics, inferential statistics of Tobit and Almost Ideal Demand System (AIDS), Vector Error Correction Model (VECM), Generalised Least Square Regression for demand, supply response and preference switch analysis. 4.1 Household expenditure pattern on rice and socioeconomic characteristics of the respondents In this section, the distribution of households according to their expenditure on total rice consumption as well as individual rice commodities (imported, agric., and local) were fully examined and analysed both at the national and individual geo-political zones. The section further analyses the expenditure share of rice in relation to the total food expenditure and socioeconomic characteristics. 4.1.1 Distribution of expenditure of households on rice Tables 4 to 7below classify the respondents according to their monthly expenditure on aggregate rice and individual rice commodities (imported, agric. and local rice) for different geopolitical zones and the national aggregate. Table 8compares statistical distribution of expenditure of various rice commodities. At the national level, Table 4 shows that, over 40 percent of the respondents spent less or equal to N2, 000 on rice monthly, a little more of them expended between N2, 001 and N4,000 on rice while the remaining few spent more thanN4,000 on rice monthly.As seen in Table5, about half of the respondents spent between N501 andN1, 000 on imported rice, while the rest were sparsely distributed across various categories of expenditure. Tables 6 reveal that the majority of the respondents spent between N501 and N1, 000 on agric. rice.This group was followed by expenditure group of ≤ N500 with a marginal number expending above N1, 500. The respondents fell into three main categories for local rice expenditure nationally (Table 7).The modal expenditure group was those that spent between N501 and N1, 000, while the expenditure group of less or equal to N500 followed. 75 Table 4: Distribution of the Respondents by Total Monthly Expenditure on Rice Cumulative Zone Category Frequency Percent Percent South-South <=2000 2300 80.6 80.6 2001-4000 554 19.4 100.0 Total 2854 100.0 South-East <=2000 1577 58.8 58.8 2001-4000 1104 41.2 100.0 Total 2681 100.0 South-West <=2000 2560 85.5 85.5 2001-4000 433 14.5 100.0 Total 2993 100.0 North Central <=2000 1019 30.6 30.6 2001-4000 1813 54.4 85.0 >=8001 499 15.0 100.0 Total 3331 100.0 North-East <=2000 545 17.0 17.0 2001-4000 2657 83.0 100.0 Total 3202 100.0 North-West 2001-4000 2182 57.4 57.4 4001-6000 1158 30.5 87.9 6001-8000 460 12.1 100.0 Total 3800 100.0 National <=2000 8001 42.4 42.4 2001-4000 8743 46.4 88.8 4001-6000 1158 6.1 94.9 6001-8000 460 2.4 97.3 >=8001 499 2.7 100 Total 18861 100 Source: Computed from NLSS Data (2004) 76 The result of the distribution of respondents according to expenditure on aggregate rice invarious geopolitical zones,presented in Table 4 revealed that the South-South zone recorded only two categories of expenditure, wherein the expenditure category of less or equal to N2,000 was in the majority over the N2001- N4000 category. The aforementioned two groups existed in the South-East zone as well; the households in the former category slightly outnumbered those in the latter. A clear tilt was observed towards the expenditure category of less or equal toN2, 000 against theN2, 001- N4, 000 expenditure category in the South-West. In the North-Central zone, a higher percentage was in favour of respondents that spentbetweenN2, 001- N4, 000. The North Eastern region followeda similar trend with that of the North-Central zone. Three groups featured in the North-West geopolitical zone.Respondents that fell within the expenditure bracket of N2, 001 and N4, 000 recorded the highest frequency. The zonal report for expenditure on imported rice, in Table 5 revealed that all respondents in the South-South region spent between N501 and N1, 000 on imported rice. South-Eastern region had three valid groupings with respect to imported rice; close to 60 percent expended between N501- N1,000, while the remaining were split between the expenditure group of N1,501- N2,000 and N2, 001- N2, 500. Unlike the South-East, two valid groups existed in the South-West, with the larger percentage tending towards expenditure bracket of N501 andN1, 000. The North-Central region was well dispersed in consumption of imported rice.The modal group (N501- N1, 000) constituted one–third of the respondents.The majority of the respondent in the North-Eastern zone recordedexpenditure between N501 and N1000, while two expenditure groups,N501-N1, 000 and N1, 001-N1, 500, dominate the North- Western zone. Considering the zonal distribution of agric. rice expenditure (Table 6), more than half of the respondents expendedbetween N501 and N1, 000 on agric. rice. All the respondents in the South-Eastern region fell within the Agric. rice expenditure category of N501 and N1, 000, while South-West had expenditure group of N501- N1, 000 in the majority. The North- Central zone featured three different expenditure groups for agric. rice;N1, 001- N1, 500 category being the modal group. The same three categories featured in the North-East, with the N501- N1, 000 group outnumbering the other two groups. The result obtained was quite different in the North-Western region; four different groups were represented. N501- N1, 000 constituted the largest category of rice consumers in the zone. 77 Table 5: Distribution of the Respondents by Expenditure on Imported Rice Cumulative Zone Category Frequency Percent Percent South-South 501-1000 2854 100.0 100.0 South-East 501-1000 1577 58.8 58.8 1501-2000 540 20.1 79.0 2001-2500 564 21.0 100.0 Total 2681 100.0 South-West 501-1000 2560 85.5 85.5 1501-2000 433 14.5 100.0 Total 2993 100.0 North-Central <=500 470 14.1 14.1 501-1000 1036 31.1 45.2 1001-1500 555 16.7 61.9 1501-2000 254 7.6 69.5 2001-2500 517 15.5 85.0 2501-3000 499 15.0 100.0 Total 3331 100.0 North-East <=500 545 17.0 17.0 501-1000 589 18.4 35.4 1001-1500 2068 64.6 100.0 Total 3202 100.0 North-West 501-1000 1062 27.9 27.9 1001-1500 1120 29.5 57.4 2001-2500 575 15.1 72.6 3001-3500 460 12.1 84.7 >=3501 583 15.3 100.0 Total 3800 100.0 National <=500 105 5.4 5.4 501-1000 9678 51.3 56.7 1001-1500 3743 19.8 76.5 1501-2000 1227 6.5 83.0 2001-2500 1656 8.8 91.8 2501-3000 499 2.6 94.5 3001-3500 460 2.4 96.9 >= 3501 583 3.1 100.0 Total 18861 100.0 Source: Computed from NLSS Data (2004) 78 Table 6: Distribution of the Respondents by Expenditure on Improved Domestic (Agric.)Rice Cumulative Zone Category Frequency Percent Percent South-South <=500 1019 35.7 35.7 501-1000 1835 64.3 100.0 Total 2854 100.0 South-East 501-1000 2681 100.0 100.0 South-West <=500 1052 35.1 35.1 501-1000 1941 64.9 100.0 Total 2993 100.0 North-Central <=500 1019 30.6 30.6 501-1000 741 22.2 52.8 1001-1500 1571 47.2 100.0 Total 3331 100.0 North-East <=500 545 17.0 17.0 501-1000 2150 67.1 84.2 1001-1500 507 15.8 100.0 Total 3202 100.0 North-West 501-1000 1663 43.8 43.8 1001-1500 1149 30.2 74.0 1501-2000 528 13.9 87.9 3001-3500 460 12.1 100.0 Total 3800 100.0 National <= 500 3635 19.3 19.3 501- 1000 11011 58.4 77.7 1001-1500 3227 17.1 94.8 1501-2000 528 2.8 97.6 3001-3500 460 2.4 100 Total 18861 100 Source: Computed from NLSS Data (2004) 79 As presented in Table 7, the gross population of the respondents in the South-South expended less or equal to N500 on local rice against the N501- N1, 000 category. The same pattern was obtained in the South-East and the South-West region. In the North-East and the North-West zones, the higher frequency was in favour of the N501- N1, 000 local rice expenditure category. The North-Central region added an additional category of expenditure ≥ N3,001, yet the N501- N1, 000 expenditure group for local rice remained the modal group for the region. As a whole, the expenditure on imported, agric. and local rice was relatively higher in the northern region than the southern region. This agrees with the findings of Adeyeye(2012). The results discussed so far are a pointer towards sociocultural diversity in rice consumption in Nigeria. Hence, the importance of inclusion of locational factor in rice demand analysis is further buttressed. Also, a comparisonof the results from the rice expenditure tables above, shows that Nigerians spend more on foreign (imported) rice than agric. and local rice (domestically produced rice). This is no doubt a threat to food self - sufficiency if the trend is not reversed. 80 Table 7: Distribution of the Respondents by Expenditure on Local Rice Cumulative Zone Category Frequency Percent Percent South-South <=500 2445 85.7 85.7 501-1000 409 14.3 100.0 Total 2854 100.0 South-East <=500 1610 60.1 60.1 501-1000 1071 39.9 100.0 Total 2681 100.0 South-West <=500 1992 66.6 66.6 501-1000 1001 33.4 100.0 Total 2993 100.0 North-Central <=500 517 15.5 15.5 501-1000 2315 69.5 85.0 >=3501 499 15.0 100.0 Total 3331 100.0 North-East <=500 999 31.2 31.2 501-1000 2203 68.8 100.0 Total 3202 100.0 North-West <=500 534 14.1 14.1 501-1000 3266 85.9 100.0 Total 3800 100.0 National <=500 8097 42.9 42.9 501-1000 10265 54.4 97.4 >=3501 499 2.6 100.0 Total 18861 100.0 Source: Computed from NLSS Data (2004) 81 The national statistics of various expenditures and expenditure shares of rice are presented in Table 8. In most cases, the skewness value greater or less than zero as well as kurtosis values greater than 3 indicate that the expenditures of various rice commodities are not normally distributed. The mean expenditure for imported rice (N1, 256.545) was higher than that of agric. rice (N797.748) and local rice (N658.110). Consequently, the share of expenditure of various rice commodities follows the same trend since imported rice is on the top with mean share of total rice expenditure of 0.451. Rice averagely constituted about 25 percent of total food expenditure. The zonal variation is presented in Appendix 7. Appendix 7further reveals that, on the average, the North-West zone ranked highest in total mean expenditure and expenditure share on almost all rice commodities. This zone was closely followed by the North–Central region. The South-East and North-East region were at close range, while the South-South zone ranked lowest in terms of overall rice consumption expenditure. For imported rice, a comparison of the mean expenditures at various zones showed that it followed the same ranking order as the total rice expenditure. North-West led, followed by North-Central; South-East, North-East, South-West and South-South followed in that sequence. With regard to the mean expenditure of agric. rice, North-East took the usual lead, followed by North-Central; the North-East overtook the South-East for agric. rice expenditure but with minimal difference. As also presented in Appendix 7, the South-West and the South-South maintained their normal ranking. In terms of expenditure on local rice, the North-Central took an exceptional lead. This is not, in any way, surprising; it could be attributed to the fact that the zone is the Nigerian basket for the production of local rice. The zone was followed by North-West, then North-East. The mean expenditure in the southern zones for local ricewas generally low; South-East followed the northern region, while the South-West and the South–South zones were at close range in the consumption of local rice. In all, it is clear that the northern region consumed more of rice commodities than the southern region simply because of the wide varieties of dishes prepared from rice and relatively higher production in the northern zones. 82 Table 8: Description of Household Expenditure on Rice Commodities (National) Expenditure/Share Mean S. D Skewness Kurtosis Imported Rice (IR) 1256.545 6.007 1.646 2.120 Agric. Rice (IDR) 797.748 3.611 3.171 13.113 Local Rice (LR) 658.110 4.453 5.226 27.695 Total Rice (TR) 2712.404 11.375 2.225 4.913 Share of IR 0.451 0.001 -0.500 -0.222 Share of IDR 0.301 0.075 -0.232 -0.169 Share of LR 0.248 0.091 1.397 2.522 Share of TR 0.254 0.307 9.486 180.920 Source: Computed from NLSS (2004) 83 4.1.2 Share of rice in total food expenditure The national expenditure share on rice is presented in Table 9. The expenditure share of 0-10 and 11-20 percent was at close range and in the majority, followed by 21-30 percent share. Other expenditure share categories were thinly distributed at the national level. The fact that expenditure on rice represents a tangible share of total food consumption expenditure is an exposure of the potential of rice to solve the problem of food self-sufficiency through increased production. Therefore, unavailability of the commodity constitutes a threat to food security and self-sufficiency in Nigeria. As shown in Appendix 8, almost half of the respondents in the South-South zone expended0- 10 percent of their total food expenditure on rice.Another substantial percentage spent up to 30 percent, while the rest were thinly distributed across various expenditure share groups. The situation was a bit different in the South-East, as closer percentages fell within the expenditure share group of 0-10 and 11- 20 percent. Yet, about 16 percent spent between 21 and 30 percent of their monthly food expenditure on rice. A similar trend to that of the South- East zone was obtained in the South-West, North-Central and North-East zones. The expenditure share group of 11-20 percent represented the modal group for North-Central and North-East unlike in the Southern region where 0-10 percent expenditure share were in the majority. The 21-30 percent expenditure share category took an exceptional lead in the North-West, with other expenditure categories closely distributed, save the 60-70 percent expenditure share. 84 Table 9: Distribution of the Respondents by Share of Rice in Total Food Expenditure (National) Expenditure Frequency Percent Valid Cumulative Percent Percent National 1-10 3926 20.8 28.3 28.3 11-20 3810 20.2 27.5 55.9 21-30 2458 13.0 17.7 73.6 31-40 1571 8.3 11.3 85.0 41-50 963 5.1 7.0 91.9 51-60 671 3.6 4.8 96.8 61-70 450 2.4 3.2 100.0 Total 13849 73.4 100.0 Missing value 5012 26.6 Total 18861 100.0 Source: Computed from NLSS (2004) 85 4.1.3 Socioeconomic characteristics of the respondents Table 10 presents the summary statistics of socioeconomic characteristics of selected respondents in the Nigerian Living Standards Survey. The frequency distribution and percentages by categories are stated, while the mean, mode, median, median, standard deviation, skewness and kurtosis are also presented where applicable. As captured in Table 10, about one-third of the selected sample had no formal education; even the educated category was heavily skewed to the lower primary and secondary school. It is believed that education has a role to play in consumption pattern of individuals. Most of the respondents had less than 10 household members, averaging about 5 household members. The household size reduces down the category line; larger households are expected to consume more food than the lower household size. A gross number of the sampled respondents were farmers, mostly resident in the rural area. Rural dwellers who are essentially farmers are expected to have more access to domestically produced rice, especially in areas where they are produced. Most of the respondents were married while a little above half of them belonged to one community society or the other. The respondents were mainly middle-aged, with the modal group being the age bracket of 41 and 50. The mean per capita expenditure of household was N14, 873. 86 Table 10: Distribution of the Respondents by Socioeconomic Characteristics Category Freq. Percentage Mean Mode Median S.D Skewness Kurtosis Per Capita Exp. 14873 5315.190 23345.510 31622 16.740 801.740 < 5,000 9027 47.9 5001- 10000 3957 21.0 10001-15000 1627 8.6 15001-20000 934 5.0 20001-25000 611 3.2 > 25001 2702 14.3 Total 18861 100.0 Education 6.782 0.000 7.000 6.336 0.552 -0.540 No formal 6457 34.2 Primary 2867 15.7 Secondary 5581 29.6 ND/NCE 2889 15.3 BSc./HND 248 1.3 Post Grad. 819 4.3 Total 18861 100.0 Household Size 4.847 4.000 4.000 2.905 1.121 2.059 <5 12316 65.3 6-10 5683 30.1 11-15 815 4.3 16-20 38 0.2 21-25 8 0.0 >25 1 0.0 Total 18861 100.0 Primary Occupation Non farming 3267 17.3 Farming 15594 82.7 Total 18861 100.0 Membership of Society Non Member 8640 45.8 Member 10221 54.2 Total 18861 100.0 Marital Status Single 4171 22.1 Married 14690 77.9 Total 18861 100.0 Location Rural 14361 76.1 Urban 4500 23.9 Age 47.399 45.000 40.000 14.531 0.510 -0.133 < 18 500 2.65 19-30 2040 10.8 31-40 4653 24.7 41-50 4833 25.6 51-60 3517 18.6 > =61 3318 17.6 Total 18861 100.0 Source: Computed from NLSS (2004) 87 4.1.4Distribution of share of rice expenditure and socioeconomic characteristics Here the total expenditure share on rice is cross tabulated with socioeconomic characteristics of the respondents to give a casual insight into the relationship between expenditure share of rice in total food expenditure and socioeconomic characteristics. The analyses are presented in tables 11–17. As seen in Table 11, the percentage of respondents in ≤ 5 household size increased with increasing share of rice up to a point and then declines. The reverse was the case for all other household size categories except the last two groups that follow irregular pattern. It means that variation seems to exist in rice expenditure among households of varying sizes. The cross-examination of expenditure share with education in Table 12 presents an irregular pattern. It could be noted that the respondents with higher educational level spent less on rice consumption. Also, for all categories of expenditure share, the majority falls within the ‗no formal‘ education group. 88 Table 11: Distribution of Share of Rice expenditure by Household size Household Size Distribution Share of Rice Expenditure <=5 6-10 11-15 16-20 21-25 >=25 Total 1-10% Frequency 2258 1400 246 16 5 1 3926 % within grouping of 57.5% 35.7% 6.3% 0.4% 0.1% 0.0% 100.0% percent share of totalrice 11-20% Frequency 2470 1142 185 12 1 0 3810 % within grouping of 64.8% 30.0% 4.9% 0.3% 0.0% 0.0% 100.0% percent share of totalrice 21-30% Frequency 1643 700 111 3 1 0 2458 % within grouping of 66.8% 28.5% 4.5% 0.1% 0.0% 0.0% 100.0% percent share of totalrice 31-40% Frequency 1094 415 59 3 0 0 1571 % within grouping of 69.6% 26.4% 3.8% 0.2% 0.0% 0.0% 100.0% percent share of totalrice 41-50% Frequency 671 267 24 0 1 0 963 % within grouping of 69.7% 27.7% 2.5% 0.0% 0.1% 0.0% 100.0% percent share of totalrice 51-60% Frequency 467 186 17 1 0 0 671 % within grouping of 69.6% 27.7% 2.5% 0.1% 0.0% 0.0% 100.0% percent share of totalrice 61-70% Frequency 306 131 14 0 0 0 451 % within grouping of 67.8% 29.0% 3.1% 0.0% 0.0% 0.0% 100.0% percent share of totalrice Total Frequency 8909 4241 656 35 8 1 13850 % within grouping of 64.3% 30.6% 4.7% 0.3% 0.1% 0.0% 100.0% percent share of totalrice Source: Computed from NLSS Data (2004) 89 Table 12: Distribution of Share of Rice Expenditure by Education Distribution of education Share of Rice Expenditure No formal Primary Secondary ND/NCE BSc./HND Post Grad Total 1-10% Frequency 1252 619 1204 655 21 175 3926 % within grouping of 31.9% 15.8% 30.7% 16.7% 0.5% 4.5% 100.0% percent share of totalrice 11-20% Frequency 1128 602 1244 620 17 199 3810 % within grouping of 29.6% 15.8% 32.7% 16.3% 0.4% 5.2% 100.0% percent share of totalrice 21-30% Frequency 839 378 705 389 10 137 2458 % within grouping of 34.1% 15.4% 28.7% 15.8% 0.4% 5.6% 100.0% percent share of totalrice 31-40% Frequency 521 216 480 264 5 85 1571 % within grouping of 33.2% 13.7% 30.6% 16.8% 0.3% 5.4% 100.0% percent share of totalrice 41-50% Frequency 333 141 282 150 5 52 963 % within grouping of 34.6% 14.6% 29.3% 15.6% 0.5% 5.4% 100.0% percent share of totalrice 51-60% Frequency 249 83 203 81 5 50 671 % within grouping of 37.1% 12.4% 30.3% 12.1% 0.7% 7.5% 100.0% percent share of totalrice 61-70% Frequency 161 82 127 49 2 30 451 % within grouping of 35.7% 18.2% 28.2% 10.9% 0.4% 6.7% 100.0% percent share of totalrice Total Frequency 4483 2121 4245 2208 65 728 13850 % within grouping of 32.4% 15.3% 30.6% 15.9% 0.5% 5.3% 100.0% percent share of totalrice Source: Computed from NLSS Data (2004) 90 As shown in Table 13, at lower share of rice expenditure (1-40 percent) membership of community society outnumbered non-members, while at higher rice expenditure shares the reverse was the case. Also in Table 14, the result revealed that married people took the larger percentage in all categories of share of expenditure on total rice.In this case, marriage can have some influence on expenditure share of households on rice. Farmers generally expended more on total rice consumption while the rural populace was more in all categories of rice expenditure share (Table 15). Table 17 also shows that there was high level of variation in cross-examining age with expenditure share of rice.In most cases, the age bracket of 31-40 or 41-50 represented the modal group in all categories of expenditure shares. 91 Table 13: Distribution of Share of Rice Expenditure by Membership of Community Society Distribution by Membership of Community Society Comm. Share of Rice expenditure Non Member Member Total 1-10% Frequency 1654 2272 3926 % within grouping of 42.1% 57.9% 100.0% percent share of totalrice 11-20% Frequency 1633 2177 3810 % within grouping of 42.9% 57.1% 100.0% percent share of totalrice 21-30% Frequency 1092 1366 2458 % within grouping of 44.4% 55.6% 100.0% percent share of totalrice 31-40% Frequency 714 857 1571 % within grouping of 45.4% 54.6% 100.0% percent share of totalrice 41-50% Frequency 519 444 963 % within grouping of 53.9% 46.1% 100.0% percent share of totalrice 51-60% Frequency 345 326 671 % within grouping of 51.4% 48.6% 100.0% percent share of totalrice 61-70% Frequency 248 203 451 % within grouping of 55.0% 45.0% 100.0% percent share of totalrice Total Frequency 6205 7645 13850 % within grouping of 44.8% 55.2% 100.0% percent share of totalrice Source: Computed from NLSS Data (2004) 92 Table 14: Distribution of Share of Rice Expenditure by Marital Status Distribution of Respondents by Marital status Share of Rice Expenditure Single Married Total 1-10% Frequency 727 3199 3926 % within grouping of 18.5% 81.5% 100.0% percent share of totalrice 11-20% Frequency 890 2920 3810 % within grouping of 23.4% 76.6% 100.0% percent share of totalrice 21-30% Frequency 584 1874 2458 % within grouping of 23.8% 76.2% 100.0% percent share of totalrice 31-40% Frequency 376 1195 1571 % within grouping of 23.9% 76.1% 100.0% percent share of totalrice 41-50% Frequency 237 726 963 % within grouping of 24.6% 75.4% 100.0% percent share of totalrice 51-60% Frequency 139 532 671 % within grouping of 20.7% 79.3% 100.0% percent share of totalrice 61-70% Frequency 74 377 451 % within grouping of 16.4% 83.6% 100.0% percent share of totalrice Total 3027 10823 13850 Frequency % within grouping of 21.9% 78.1% 100.0% percent share of totalrice Source: Computed from NLSS Data (2004) 93 Table 15: Distribution of Share of Rice Expenditure by Primary occupation Distribution by Primary Occupation Total Non- Share of Rice Expenditure Non Farming Farming Farming 1-10% Frequency 1111 2815 3926 % within grouping of 28.3% 71.7% 100.0% percent share of totalrice 11-20% Frequency 645 3165 3810 % within grouping of 16.9% 83.1% 100.0% percent share of totalrice 21-30% Frequency 339 2119 2458 % within grouping of 13.8% 86.2% 100.0% percent share of totalrice 31-40% Frequency 165 1406 1571 % within grouping of 10.5% 89.5% 100.0% percent share of totalrice 41-50% Frequency 100 863 963 % within grouping of 10.4% 89.6% 100.0% percent share of totalrice 51-60% Frequency 59 612 671 % within grouping of 8.8% 91.2% 100.0% percent share of totalrice 61-70% Frequency 28 423 451 % within grouping of 6.2% 93.8% 100.0% percent share of totalrice Total 2447 11403 13850 Frequency % within grouping of 17.7% 82.3% 100.0% percent share of totalrice Source: Computed from NLSS Data (2004) 94 Table 16: Distribution of Share of Rice Expenditure by Location Distribution by Location Share of Rice Expenditure Rural Urban Total 1-10% Frequency 2607 1319 3926 % within grouping of 66.4% 33.6% 100.0% percent share of totalrice 11-20% Frequency 2785 1025 3810 % within grouping of 73.1% 26.9% 100.0% percent share of totalrice 21-30% Frequency 1957 501 2458 % within grouping of 79.6% 20.4% 100.0% percent share of totalrice 31-40% Frequency 1328 243 1571 % within grouping of 84.5% 15.5% 100.0% percent share of totalrice 41-50% Frequency 834 129 963 % within grouping of 86.6% 13.4% 100.0% percent share of totalrice 51-60% Frequency 582 89 671 % within grouping of 86.7% 13.3% 100.0% percent share of totalrice 61-70% Frequency 405 46 451 % within grouping of 89.8% 10.2% 100.0% percent share of totalrice Total Frequency 10498 3352 13850 % within grouping of 75.8% 24.2% 100.0% percent share of totalrice Source: Computed from NLSS Data (2004) 95 Table 17: Distribution of Share of Rice Expenditure by Age Distribution of Respondents by Age Share of Rice <=30 31-40 41-50 51-60 >=61 Total 1-10% Frequency 351 989 1203 769 614 3926 % within grouping of 8.9% 25.2% 30.6% 19.6% 15.6% 100.0% percent share of totalrice 11-20% Frequency 508 890 963 720 729 3810 % within grouping of 13.3% 23.4% 25.3% 18.9% 19.1% 100.0% percent share of totalrice 21-30% Frequency 362 588 610 426 472 2458 % within grouping of 14.7% 23.9% 24.8% 17.3% 19.2% 100.0% percent share of totalrice 31-40% Frequency 240 411 332 286 302 1571 % within grouping of 15.3% 26.2% 21.1% 18.2% 19.2% 100.0% percent share of totalrice 41-50% Frequency 145 221 232 182 183 963 % within grouping of 15.1% 22.9% 24.1% 18.9% 19.0% 100.0% percent share of totalrice 51-60% Frequency 103 180 156 115 117 671 % within grouping of 15.4% 26.8% 23.2% 17.1% 17.4% 100.0% percent share of totalrice 61-70% Frequency 82 119 106 68 76 451 % within grouping of 18.2% 26.4% 23.5% 15.1% 16.9% 100.0% percent share of totalrice Total Frequency 1791 3398 3602 2566 2493 13850 % within grouping of 12.9% 24.5% 26.0% 18.5% 18.0% 100.0% percent share of totalrice Source: Computed from NLSS Data (2004) 96 4.2 Rice self-sufficiency in Nigeria As graphically depicted in Figures9 and 10, Nigeria was self-sufficient in rice from 1960 to 1975, as the self- sufficiency ratio was approximately unity during this period. From 1975 the self-sufficiency ratio began to decline up to 1987. Rice farmers responded positively through production increase from 1987 when a ban was placed on importation of rice. The self–sufficiency ratio was around unity in 1987 and was even greater than unity in 1989. Perhaps, owingto lifting of the ban on importation of rice, demand increased astronomically, outstripping supply again from 1990 leaving Nigeria insufficient in rice from that period till 2012. Although there were some fluctuations and improvement in self-sufficiency ratio in some years (such as 1985-1987), the unity status witnessed in the 1960s up to mid-1970s has never been restored. Appendix 9 givesdetails of yearly self-sufficiency ratio of rice in Nigeria between 1960 and 2012. Even in the face of improved rice self-sufficiency ratio in some years, it is surprising that rice is still massively imported and the importation figure is on the rise with time. The fact is that, not all farm outputs constitute marketed output (real supply), bearing in mind loss in transition owing to poor storage facilities, transportation, intermediate consumption and other constraints that can lead to over-estimation of output and consequently rice self-sufficiency. As a matter of standard, 30 percent reduction in output is usually made for cereals for losses in transition in the Central and Eastern European nations (Hallam, 2000). Sanni (2000) equally reported 20-40 percent losses in cereals output before marketing owing to poor storage in Nigeria. Similarly, IRRI;FAO (2014) rice statistics reported 277, 000 and 263, 000 tons loss in rice output from farm to processing in 2008 and 2009, respectively. If we impose such 30 percent average losses on estimated farm level rice output, the self-sufficiency ratio will definitely decline more for all the years. Besides that, other factors such as preference for imported rice against local rice can also lead to increasing rice importation in the face of increased rice production and improved self-sufficiency ratio. Hence, a dynamic analysis of demand in the light of the preference switchis justified by this study. 97 7000 6000 5000 4000 Production 3000 Consumption 2000 1000 0 NOTE: Points of Intersection between supply and demand represents rice self- sufficient years Figure 9: Relationship between Rice Supply and Demand (1960-2012) Source: Computed from IRRI; USDA Rice Statistics (2014) 98 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 2012 Rice Self-Sufficiency Ratio 1.4 1.2 1 0.8 0.6 Rice Self Sufficiency Ratio 0.4 0.2 0 Figure 10: Self-Sufficiency Ratio of Rice in Nigeria (1960-2012) Source: Computed from IRRI; USDA Rice Statistics (2014) 99 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 2012 4.3 Determinants of rice demand in Nigeria This section discusses the factors that influence rice consumption in Nigeria. It also presents interpretation of relevant elasticity estimates derived from the Tobit and AIDS models. 4.3.1 Tobit model estimate for rice demand As presented in Table 18, the diagnostic test; Log Likelihood for all the four Tobit regressions indicates the fitness of the model. The F-statistics was significant at 1 percent probability level. The Akaike Information Criterion was equally low enough to confirm the fitness of the model. The relative low R-square value is typical of consumption studies owing to extraneous and qualitative variables not usually totally captured in such studiesand this is in line with the finding of authors of similar studies, such as,Abdulai et al. (1999), Akbay and Boz (2001), Okoruwa et al.(2008.)and Oyinbo et al. (2013). The R-square value of 0.46, 0.47, 0.27 and 0.49 were obtained, respectively for aggregate, imported, agric. and local rice. Socioeconomics factors As shown in Table 18, household size was significant in determining aggregate rice demand at 5 percent level (P ≤ 0.05). Similarly, the demand for imported rice and agric. rice were significantly influenced by household size at 1 percent probability level. An additional membership increase in household size increased the total rice and imported rice -03 -03 consumption by a factor of 3.449×10 and 1.071×10 ,respectively. This exposes the tendencies of populated household to consume generally more of rice and specifically more of imported rice which is readily available, easier to prepare and relatively cheaper. This is consistent with the findings of Abdulai et al. (1999), Bamidele et al., (2010) and Oyinbo et al., (2013). However, increase in household size reduced the consumption of agric. rice by -04 9.232×10 perhaps because of relatively high price of agric. rice and its availability. Although household food security is guaranteed with increasing demand for imported rice,this constitutes a great threat to food self- sufficiency in Nigeria. The age of the respondents was found to be statistically significant at 1 percent in explaining total rice consumption. An inverse relationship was equally observed, implying that older populace generally consumed less of rice commodity.The same trend was observed for the agric. rice consumption as inverse relationship significant at 1 percent resulted from the Tobit regression estimate.In contrast, an increase in age by a year leads to increasein local rice consumption -04 by a factor of 1.568×10 .This result is expected as older generations are known for conservativeness in terms of local food consumption. Heilig (1999), Choi and Lee (2000) and 100 Babatunde et al. (2007) equally found age to significantly affect rice consumption. The result also conforms to the findings of Agwu et al.(2009) and Adeyeye (2012). A sustained consumption of domestically produced rice not only by older generation, but also by people of all age categories will certainly boost rice self- sufficiency in Nigeria. As for the influence of educational factor, only imported rice was significantly influenced by education at 5 percent level of probability. An increase in year of educational attainment -04 resulted in an increase in imported rice consumption by 1.557×10 . This depicts the preference of educated people for imported foods as against local food. This corroborates the findings of Abdulai et al. (1999),Jenson(1995),Babatunde et al.(2007), Nwachukwu et al.(2008), Agwu et al.(2009), Bamidele et al. (2010) and Adeyeye (2012). If this trend continues, it has a great implication for rice self-sufficiency in Nigeria. -03 Marital life increased agric. rice consumption by 7.346×10 , while local rice consumption -03 decreased by 6.832×10 . The marital status variable was significant at 1 percent probability level. Most often, marriage produces children, which increase the household membership. The relatively higher price of local rice will, no doubt, make it unaffordable for larger families in a poverty-ridden community like Nigeria.Thus, its consumption decreases with marital status. This corroborates the findings of Adeyeye (2012). Also, taking farming as an occupation reduced consumption of imported rice by a factor of -03 9.108×10 ; this factor was significant at 1 percent level of probability. A direct relationship was observed between occupational factor and agric. rice consumption (significant at P ≤ 0.01),implying that agric. rice farmers set aside some proportion of their harvest for home consumption purposes. Most farmers are rural dwellers and hence consume more of domestically produced rice based on availability and/ or conservativeness. This development is a positive step towards increasing rice self-sufficiency in Nigeria. At 1 percent level, an inverse relationship occurredbetween membership of community society and agric. rice consumption but a positive relationship was obtained with reference to local rice. This is in agreement with the findings of Abdulai et al. (1999).This implies that group influence could stimulate the consumption or otherwise of rice commodities. 101 Table 18: Tobit Regression Result for Rice Demand in Nigeria (Marginal Values) Variables Aggregate Rice Imported Rice Agric. Rice local Rice -03 -03 -04 -03 HHSIZ 3.449x10 ** 1.071x10 *** -9.323x10 *** -1.383e -02 -04 -04 -04 (1.259x10 ) (3.041x10 ) (2.602x10 ) (2.628x10 ) -7 -09 -08 -09 NFDTOT -2.794x10 * 9.912x10 *** -1.610x10 *** 6.191x10 ** -7 -09 -09 -09 (1.454x10 ) (3.563x10 ) (3.048x10 ) (3.079x10 ) -3 -05 -04 -04 AGE -1.260x10 *** -2.367x10 -1.331x10 *** 1.568x10 *** -3 -05 -05 -05 (1.684x10 ) (4.055x10 ) (3.470x10 ) (3.505x10 ) -3 -04 -04 -05 EDUC 3.920x10 1.557x10 ** -1.204x10 -3.532x10 -03 -05 -05 -05 (0.3669x10 ) (8.865x10 ) (7.587x10 ) (7.662x10 ) -03 -04 -03 -03 MARST -3.141x10 -5.134x10 7.346x10 *** -6.832x10 *** -02 -03 -03 -03 (7.516x10 ) (1.780x10 ) 1.540x10 (1.556x10 ) -02 -03 -03 -03 PROCC 8.263x10 -9.108x10 *** 7.719x10 ***1.389x10 -02 -03 -03 -03 (6.670x10 ) (1.595x10 ) (1.365x10 ) (1.378x10 ) -09 -10 -11 -10 TASSET -1.373x10 -2.452x10 -2.768x10 2.728x10 -08 -10 -10) -10 (1.007x10 ) (2.480x10 ) (2.122x10 (2.143x10 ) -02 -04 -03 -03 COMEM 4.293x10 7.396x10 -4.903x10 *** 4.163x10 *** -02 -03 -04 -4 (4.681x10 ) (1.129x10 ) (9.658x10 ) (9.754x10 ) -01 -03 -03 -3 LOCAT -1.976x10 *** 3.693x10 ** -1.602x10 -2.091x10 -02 -03 -03 -3 (6.243x10 ) (1.495x10 ) (1.280x10 ) (1.293x10 ) -01 -01 -2 SS 1.901*** -1.841x10 *** 2.493x10 *** -6.523x10 *** -01 -03 -03 -3 (2.501x10 ) (5.913x10 ) (5.060x10 ) (5.110x10 ) -01 -02 -1 SE 1.533*** 1.458x10 *** -1.791x10 *** -1.278x10 *** -01 -03 -03 -3 (1.433x10 ) (3.377x10 ) (2.890x10 ) (2.921x10 ) -02 -02 -1 SW 1.472*** 2.146x10 *** 9.754x10 *** -1.190x10 *** -01 -03 -03 -3 (1.575x10 ) (3.764x10 ) (3.221x10 ) (3.254x10 ) -01 -02 -01 -1 NE 5.049x10 *** - 3.669x10 *** 1.005x10 *** -6.381x10 *** -01 -03 -03 -3) (1.116x10 ) (2.716x10 ) (2.325x10 )(2.348x10 -01 -02 -01 -1 NW 8.864x10 *** 4.011x10 *** 1.075x10 *** -1.477x10 *** -01 -03 -03 -3 (1.170x10 ) (2.840x10 ) (2.431x10 ) (2.455x10 ) -01 -1 CONST. -4.245*** -1.538*** 5.463x10 *** 6.075x10 *** -01 -02 -02 -1 (7.048x10 ) (1.701x10 ) (1.456x10 ) (1.470x10 ) 2 MacFaden R 0.461 0.468 0.273 0.489 LogLikelihood-56804.990 22109.460 25045.040 24858.480 Akaike Info Crt.13.468 -2.341 -2.652 -2.633 F(30, 18830) 536.220*** 552.270*** 235.400*** 600.49*** ***Values significant at 1%; **Values significant at 5%, *Values significant at 10% Source: Computed from NLSS Data (2004) 102 The non-food total expenditure decreased with increase in total rice expenditure by a factor of -07 2.794×10 . The same trend was observed for agric. rice commodities, while a direct relationship ensued for imported and local rice commodities. The beta coefficient of non-food total expenditure was significant at 5 percent level for local rice commodity, while others were significant at 1 percent probability level. Expenditure on rice could generally reduce the amount spent on non-food commodities, especially for people with low income in agreement with Engel‘s law.This does not hold for expenditure on imported rice and local rice, as obtained in this study. Locational factors As for the rural-urban dichotomy, urban livelihood partly explains the variation in the total rice consumption and imported rice at 1 percent and 5 percent probability level, respectively. A positive relationship was observed between urban livelihood and imported rice consumption as a priori expected. This is a further confirmation of the fitness of imported rice into urban lifestyle as they often desire easy to prepare food because of their career demand. On the other hand, an inverse relationship holds between urban livelihood and total rice consumption. This effect of urbanisation on rice consumption has been earlier isolated by Bashorun (2013). The estimate of the coefficient was -0.198. All the geopolitical zone dummies were found to be statistically significant at 1 percent level (P≤ 0.01). In relation to the basal North-Central zone, South-South zone increased aggregate rice consumption the more by a factor of 1.901. On the other hand, residing in the South- South zone led to decreased consumption of imported rice relative to the North-Central zone by a factor of 0.184. Furthermore, more agric. rice was consumed in the South-South zone relative to the North-Central. Also, consumption of aggregate and imported rice increased in the South-East zonemore than the North-Central by a factor of 0.143 and 0.146,respectively. In case of agric. and local rice, less was consumed in the South-East relative to the North- Central zone. This is expected because North-Central zone produces more local rice than any other zone in Nigeria. In the South-West zone, more of total, imported and agric. rice was consumed than the North-Central. The cosmopolitan nature of the South-West accounted for wide variety of rice consumption. The high production of rice in the North-Central zone also comes into play here, as the South-West zone consumed less of local rice than the North- Central. 103 In the North-Eastern zone, at the aggregate level, more rice was consumed than the North- Central zone. On individual rice commodities,more of agric. rice, and less of imported and local rice was consumed in the North-East relative to the North-Central. Lastly, the North- Westzone recorded more consumption of total rice, imported rice and agric. rice relative to the North-Central. As usual, no region, North-West inclusive, superseded the North-Central geopolitical zone in the consumption of local rice. Abdulai et al. (1999), Choi and Lee (2000), Adeyeye (2012) and Bashorun (2013) equally found location factors significantly influencing food demand in India, Korea and Nigeria. Therefore, policy on rice production, either through increasing production or stimulation of consumption, should be location based. Elasticity estimates The estimated income elasticities from the Tobit model (Table 19) show that imported, agric. -08 and local rice were non-income elastic, as their respective values of elasticities-7.266×10 , -07 -07 1.727×10 and 1.001×10 ,were less than unity. In this case, the various commodities of rice could be conveniently classified as ‗necessities‘ and because the elasticity values are greater than zero, they are equally classified as ‗normal good‘. This result clearly supports the assertion of the Engel‘s curve, which states that, at higher income, families spend lesser proportion of their income on food. Once the food requirement is satisfied, additional income is rather expended on luxuries (Olayemi, 2004). In all cases, positive relationship existed between income and consumption of various rice commodities, thus confirming the traditional direct relationship between income and demand, a finding consistent with that of Odusola(1997), Miller (2002), Nwachukwu et al. (2008), Agwu et al. (2009), Bamidele et al. (2010) and Oyinbo et al. (2013). This implies that, as income increases, the demand for -08 -07 - imported, agric. and local rice increases by 7.266×10 , 1.727×10 and 1.001×10 07 ,respectively, but in a less than proportionate magnitude to increase in income due to inelasticity. On the own price elasticities, all the resulting coefficients for imported, agric. and local rice commodities displayed the expected negative signs showing the usual inverse relationship between price and quantity demanded.This confirms the traditional law of demand. More importantly, all the values were less than unity, confirming the inelasticity of rice demand to price,like many other food commodities. This is in line with the findings of Nwachukwu et al.(2008), Agwu et al.(2009), Rahji et al. (2008) and Jimoh et al. (2010). Necessities, unlike luxuries, are often associated with lower price elasticities, as further confirmed in this study. 104 A unit decrease in price leads to increase in the demand for imported, agric. and local rice by -08 -04 -03 a factor of 2.923×10 , 7.392×10 and 1.825×10 ,respectively. The own price coefficients were statistically significant at 1 percent probability levels. The overall implication of these income and own price elasticities is that a change (increase or decrease) in income and price results in a less than proportionate change in demand for all rice commodities.Hence, price and income instrument have limited effect on demand for various rice commodities. The results of the cross-price elasticities for imported rice revealed that local rice, white garri, yam and brown beans were substitutes to imported rice, as the coefficients displayed positive signs. Other food commodities, such as agric. rice, yellow garri, white beans, millet, guinea corn, white maize and yellow maize displayed inverse relationship,thus indicating complementarity with imported rice.For the agric. rice commodity, yellow garri, guinea corn, millet, white maize and yellow maize were found to be substitute products, while imported rice, local rice, white garri, yam, brown beans and white beans were found to be complementary to agric. rice. In the same vein, the competitive products to local rice included agric. rice, yellow garri, yam, millet and guinea corn, while imported rice, white garri, brown beans, white beans, white and yellow maize emerged as complementary products to local rice. All the cross price coefficients were highly significant at 1 percent (P ≤ 0.01). The significance of all price coefficients in determining the demand for rice is consistent with the findings of earlier researchers, notablyRahji and Adewumi (2008), Nwachukwu etal. (2008), Odusina (2008), Agwu et al.(2009), Oyinbo et al. (2013) and Adeyeye (2012).However, the degree of substitution of a particular rice commodity for other rice or food commodities was minimal because of the cross price inelasticity of demand for various rice commodities. 105 Table 19: Tobit Elasticity Estimates for Rice Demand in Nigeria Imported Rice Agric. Rice Local Rice -08*** -07*** -07*** Income 7.266x10 1.727x10 1.000x10 -03*** -04*** Price of Imported Rice -03***-2.923x10 -2.189x10 -7.367x10 -03*** -04*** -03*** Price of Agric. Rice -3.296x10 -7.392x10 2.556x10 -03*** -03*** -03*** Price of Local Rice 3.928x10 -2.102x10 -1.825x10 -03*** -03*** -04*** Price of Yellow Garri -3.633x10 3.131x10 5.020 x10 -03*** -03*** -03*** Price of white Garri 8.263x10 -6.585x10 -1.677x10 -04*** -03*** -04*** Price of Yam Tuber 7.005x10 -1.330x10 6.233x10 -03*** -04*** -03*** Price of Brown Beans 4.589x10 -3.431x10 -4.246x10 -03*** -03*** -04*** Price of White Beans -3.925x10 4.700x10 -7.749x10 -03*** -04*** -03*** Price of Millet -1.642x10 5.168x10 1.125x10 -03*** -04*** -03*** Price of Guinea Corn -6.608x10 6.070x10 6.000x10 -03*** -03*** -03*** Price of White Maize -1.209x10 5.190x10 -3.981x10 -03*** -03*** -03*** Price of Yellow Maize -1.322x10 8.528x10 -7.960x10 Bold Values are own price elasticities; otherprice values are cross price elasticities ***Values significant at 1% **Values significant at 5% *Values significant at 10% Source: Computed from NLSS data (2004) 106 4.3.2 Almost Ideal Demand System (AIDS)Estimate for rice demand The result of the AIDS regression is presented in Table 20, while the elasticity estimates are shown in Table 21. As presented in Table 20, all categories of rice commodities were significantly affected by household size. A direct relationship occured between household -03 size and imported rice (β = 1.350×10 ), while an inverse relationship was observed with -03 -04 agric. (β = -0.746×10 ) and local rice (β = -5.990×10 ). The beta coefficients for imported, agric. and local rice were significant at 1, 5 and 10 percent probability levels respectively. The reason that could be adduced for this fact is the relative cheapness, ease of cooking and availability of imported rice in most part of Nigeria that propel the larger households to demand more of it. In most cases the cost of local rice is unbearable for larger household size and is not readily available as much as imported rice coupled with cooking difficulties. Non-food total expenditure negatively influenced the consumption of imported, agric. and local rice. A unit rise in non-food expenditure reduces the consumption of imported, agric. -08 -08 -08 and local rice by 1.124×10 , 1.429×10 and 2.264×10 ,respectively. Age positively -04 affected imported rice demand at 1 percent level by a factor of 4.690×10 but had an indirect relationship with agric. and local rice at the same level of statistical significance.In agreement with the previous result from the Tobit model, education was a determinant of -04 imported rice as it exerted a positive influence on demand by a factor of 9.430×10 , while it -04 -03 reduced the consumption of agric. and local rice by a factor 4.157×10 and 5.278×10 , respectively. Urban livelihood similarly had a significant positive influence on imported rice consumption. This was significant at 1 percent probability level. The consumption of agric. and local rice was however, favoured by rurality at 5 and 1 percent levels, perhaps because many of the rural people were farmers and were engaged in the production of agric. and local rice. As seen in Table 20, farmers expectedly demanded more of agric. rice and less of imported rice when compared with people of other primary occupation other than farming.The occupational variable was significant at 1 percent and 5 percent, respectively, for imported and agric. rice. Marital status also showed an inverse relationship with imported and local rice but a direct relationship with agric. rice. Also, demand for local rice significantly appreciated with belongingto a community group but rather depreciated in case of agric. rice. 107 Table 20: AIDS Regression Result for Rice Demand in Nigeria Variables Imported Rice Agric. Rice local Rice -02 -03 -03 HHSIZ 0.135e *** -0.746e ** -0.599e * -03 -03 -03 (0.382e ) (0.291e ) (0342e ) -07 -07 -08 NFDTOT -0.124e *** -0.143e *** 0.266e *** -08 -08 -08 (0.456e ) (0.348e ) (0.408e ) -03 -03 -03 AGE -0.470e *** -0.241e *** -0.229e *** -04 -04 -04 (0.506e ) (0.387e ) (0.453e ) -03 -03 -03 EDUC 0.943e *** -0.416e *** -0.528e *** -03 -04 -04 (0.111e ) (0.850e ) (0.997e ) -02 - -01 -02 MARST -0.613e *** 0.112e *** -0.502e ** -02 -02 -02 (0.231e ) (0.176e ) (0.207e ) -02 -02 - -04 PROCC -0.339e * 0.346e ** 0.708e -02 -02 -02 (0.198e ) (0.151e ) (0.177e ) -09 -09 -10 TASSET 0.190e -0.126e -0.632e -09 -09 -09 (0.319e ) (0.244e ) (0.286e ) - -03 -02 -02 COMEM 0.535e -0.803e *** 0.857e *** -02 -02 -02 (0.113e ) (0.109e ) (0.128e ) -01 -02 -01 SECTOR 0.130e *** -0.309e ** -0.161e *** -02 -02 -02 (0.176e ) (0.134e ) (0.157e ) CONST. -0.539*** 0.272*** 0.189*** -02 -02 -02 (0.461e ) (0.352e ) (0.413e ) 2 R 0.119 0.414 0.917 2 Adj R 0.118 0.407 0.906 F(15,18845) 169.200*** 54.310*** 126.810*** Log Likelihood 304728.140 304728.140 304728.140 Akaike Info Crt. -2.060 -2.379 -2.060 ***Values significant at 1% **Values significant at 5% *Values significant at 10% Source: Computed from NLSS Data (2004) 108 Elasticity estimates For the AIDS model in Table 21, the demand for various rice commodities was found to be income-inelastic. The expected direct relationship was observed in all cases. The coefficients for imported, agric. and local rice were found to be significant at 1 percent probability level. The result is equally in conformity with the Engel‘s law. Hence, rice is a ‗necessity‘. The price variable coefficients revealed that all categories of rice,except imported rice, were price inelastic. This is typical of necessities like rice. According to the result of the AIDS model, it also follows that price instrument could be used in manipulating imported rice demand but to a lesser extent for local and agric. rice. The expected demand-price inverse relationship holds in all cases of imported, agric. and local rice, implying that all rice commodities obeyed the traditional law of demand. In addition, all price coefficients were statistically significant at 1 percent (P ≤ 0.01). The R-square value implies that the factors considered jointly explained about 12 percent variation in the demand for imported rice, 41 percent for agric. rice and 92 percent for local rice consumption. The low R-squared obtained for imported rice is typical of a consumption study, especially the fact that importation is a function of many macroeconomic variables probably unaccounted for in modelling the demand for imported rice in this study. 109 Table 21: AIDS Elasticity Estimates for Rice Demand in Nigeria Imported Agric Local Price -1.804*** -0.975*** -0.945*** Income 0.999** 0.999** 0.999** ***Values significant at 1% **Values significant at 5% *Values significant at 10% Source: Computed from NLSS Data (2004) 110 4.3.3 The Tobit Model compared with the AIDS Model The results from the two models showed considerable similarities in the significance of variables as well as the direction of relationship. Even the non-significant variables were similar in the two models. However, the Tobit model had the edge of utilizing all the NLSS data.Also, considering the numbers of variables that were accommodated in the Tobit model as well as the theoretical plausibility of the model results,the Tobit model gave a better estimate of demand for rice commodities in Nigeria. Unlike the AIDS model, the Tobit model in this study permittedthe estimation of various cross-price elasticities as well as showing the effect of variables on demand across various geopolitical zones. 4.4 Supplyresponseanalysis This section presents and discusses the result on supply response of rice to price and non- price factors applying the Vector Autoregressive Error Correction Model (VECM). As a matter of convention, it begins with the unit root test applying the ADF, and the cointegration test using the Johansen test. The estimation of the long-run and the short-run model was completed in a vector error correction model. 4.4.1 Unit Root Test The summary of the results of Augmented Dickey Fuller (ADF) unit root analysis is presented in Table 22. The result of the ADF unit root test revealed that output, area, price, import and fertilizer consumption had a unit root. At their various levels, the null hypotheses of the presence of unit root in the variables (ρ=1) were accepted at one percent (P≤ 0.01). The variables, however, became stationary at first difference implying that they were all integrated of the order of 1(that is, they were I(1)). This is further confirmation of the fact that most macroeconomic variables are first difference stationary (Tijani and Ajetomobi, 2006; Gujarati and Sangeetha, 2007).The yield and rainfall variables were stationary at their levels, the unit root null hypotheses (ρ=1) was, therefore, rejected at their levels. The unit root hypothesis for rainfall and yield was rejected at one percent and five percent significant levels, respectively. Tijani and Ajetomobi (2006) equally found rainfall to be level stationary in their supply response analysis for cocoa export. In addition, rice price was also found to be trend stationary at five percent probability level. From the result above, it follows that the output of rice can exhibit a long-run relationship with Area Cultivated, Own Price, Fertilizer Consumption and Quantity of Rice Imported. Yield could also co-integrate with rainfall in the long run. 111 Table 22: Result of ADF Unit Root Test of Variables Variables Level First Difference Order of Untrended Trended Untrended Trended Integration Outp 0.672 -2.043 -8.437 8.883* I(1) Area 0.742 -2.540 -9.450* 9.649* I(1) Yield -3.059** 3.742** I(0) Pric -2.433 3.865** -7.947* 7.856* I(1) FCon -1.398 -1.383 -5.997* -5.968* 1(1) Impt -0.790 -2.297 -6.398* -6.392* 1(1) Rain -5.959* 5.943* I(0) ***Values significant at 1% **Values significant at 5% *Values significant at 10% Source: Computed from IRRI Rice Statistics (2011) 112 4.4.2 Pairwise Granger Causality Test The Granger test of causality to determine the appropriate lag length and see the causal effect andrelative importance of variables is presented in Table 23. From the result of the pairwise Granger causality test, at the first lag, it was observed that output was Granger-caused by one variable (policy), area was Granger-caused by one variable (Policy), price was Granger-caused by four variables, namely: output,area, fertilizer consumption and policy.Fertilizer consumption and rainfall were not Granger-caused by any variable while policy was Granger-caused by only one variable(price). This suggests that output was most affected byprice, followed by import, area and policy as well as fertilizer consumption and rainfall. However, increasing the lag length up to the fourth lag did not significantly improve the significance of the variable; hence, a one lag model was supportedfrom the result of the Granger causality test. 113 Table 23: Pairwise Granger Causality Tests Causal Outp Area Pric Fcon Impt Rain Poly variables Output — N Y N Y N N Area N — Y N Y N N Pric N N — N N N Y Fcon N N Y _ N N N Impt N N N N — N N Rain N N N N Y — N Poly Y Y Y N N N — 1 1 4 0 3 0 1 * Y-Granger-caused *N-Not Granger-caused Source: Computed from IRRI Rice Statistics (2011) 114 4.4.3 Tests for Cointegration (Johansen Test) Table 24 shows the results of the cointegration test for all possible specifications of the vector error correction model using the Johansen test.As for the summary of various cointegration tests, the trace and Maximum Eigen tests reported rank 1 for all possible specification of cointegration except trace test that reported rank 3 for the specification with no intercept and trend in the CE and VAR. The Pantula principle states that the lower the rank of the specification, the better the model. The data for this study supported the use of a vector error correction model (VECM) with all specifications except the one with no intercept or trend in the CE and VAR,since it has a trace value higher than others specifications. We estimated the cointegration specification with intercept and no trend since it has value significant at 5 percent (P<0.05) level. The Trace and Maximum- Eigen value presented in Table 24 below indicates the rejection of the null hypothesis of no cointegration equation (CE) at 5% level of significance. The null hypothesis of at most one CE was thus accepted at 5% level of significance. In conformity with the specification stated by the Pantula principle, the Johansen(1992, 1995a) trace and max-Eigen value revealed that one cointegrating equation exists among the variables in the economic model. 115 Table 24: Cointegration Test for all Specifications Lags interval: 1 to 1 Data Trend: None None Linear Linear Quadratic Rank or No Intercept Intercept Intercept Intercept Intercept No. of CEs No Trend No Trend No Trend Trend Trend Log Likelihood by Model and Rank 0 -1455.410 -1455.410 -1449.609 -1449.609 -1444.725 1 -1436.538 -1434.135 -1428.338 -1427.478 -1422.995 2 -1425.719 -1422.769 -1417.023 -1415.610 -1411.437 3 -1416.056 -1412.055 -1410.049 -1405.354 -1402.598 4 -1412.901 -1408.876 -1407.313 -1400.823 -1398.241 5 -1412.901 -1407.163 -1407.163 -1398.089 -1398.089 Akaike Information Criteria by Model and Rank 0 64.36565 64.36565 64.33083 64.33083 64.33586 1 63.97991 63.91891 63.84077 63.84687 63.82587 2 63.94432 63.90298 63.78361 63.80912 63.75813* 3 63.95897 63.91544 63.91519 63.84150 63.80860 4 64.25656 64.25547 64.23101 64.12274 64.05395 5 64.69134 64.65925 64.65925 64.48213 64.48213 Schwarz Criteria by Model and Rank 0 65.35947 65.35947 65.52342 65.52342 65.72722 1 65.37127 65.35002* 65.43089 65.47675 65.61476 2 65.73321 65.77138 65.77126 65.87628 65.94455 3 66.14538 66.22112 66.30037 66.34594 66.39255 4 66.84051 66.99843 67.01373 67.06446 67.03543 5 67.67282 67.83949 67.83949 67.86114 67.86114 Trace test Rank = 3 Rank = 1 Rank = 1 Rank = 1 Rank = 1 Max-Eig Rank = 1 Rank = 1 Rank = 1 Rank = 1 Rank = 1 Source: Computed from IRRI Rice Statistics (2011) 116 Table 25: Co-integration Test for Intercept and No deterministic trend in the data Series: OUTPUT AREA PRIC FCON IMPORT Hypothesized Trace 0.05 No. of CE(s) Eigenvalue Statistic Critical Value Prob.** None * 0.603467 96.49433 76.97277 0.0008 At most 1 0.389936 53.94447** 54.07904 0.0514 At most 2 0.372366 31.21170 35.19275 0.1263 At most 3 0.129109 9.785010 20.26184 0.6609 At most 4 0.071773 3.426018 9.164546 0.5043 Trace test indi cates 1 cointeg rating equation at the 0.05 lev el * denotes rejection of the hypothesis at the 0.05 level ** Value Significant at 5 percent Source: Computed from IRRI Rice Statistics (2011) 117 4.4.4 The Vector Error Correction Model Following the evidence from the cointegration tests in the previous section, the vector error correction model was estimated using EViews, with one cointegrating restrictions imposed. The normalisation adopted was in respect of the output of rice. This permits assessment of the long-run influence of area cultivated, price of rice, fertilizer consumption and the quantity of import on the output variable. Tables 26 and 27 present the normalised cointegrating vectors in the VECM for the long-run and the short-run equilibrium models, respectively. 1. The Long-Run Model As shown in Table 26, the estimated coefficients in the long-run equilibrium performance model are quite plausible, as all variables elasticities are consistent with economic theory and previous findings. Lagged values of area cultivated, fertilizer consumption and import quantities significantly influenced the supply of domestic rice in Nigeria at 1, 10 and 5 percent level of probability, respectively. Area cultivated remains the most critical factor that affected rice supply (output) in Nigeria. A one percent change in area cultivated in the previous year caused the output quantity to increase by about 3 percent. This is expected, as farmers cultivate more hectares, the volume of output (supply) increases ceteris paribus. Policies that make more farm land available to farmers for rice cultivation will surely go a long way in increasing production to meet up with demand. Fertilizer consumption improves output by 2.3 percent when increased by 1 percent. Fertilizer consumption was also significant at 5 percent level.By implication availability of improved input,such as fertilizer,could stimulate production and increase the output of domestic rice. A one percent rise in import resulted in a rise in output by 0.3 percent contrary to apriori expectation. In the longrun, some level of rice importation could stimulate domestic production through competitiveness.In view of the estimated value of 0.27 for price coefficient, rice output was found to be price-inelastic in the long run. This is consistent with the findings of Rahji (1999), Rahji et al. (2008) and Muchapondwa (2008) among others. Therefore, a pricing policy in the long run may not yield any significant result in stimulating production to meet demand. 118 Table 26: Rice Supply Response Long-Run Model Variable Coefficients Standard Error t- Statistics Output (-1) 1.000 Area (-1) 2.809*** 0.109 26.077 Price (-1) 0.273 0.169 1.640 Fertilizer Consumption 2.327* 0.364 6.483 Import (-1) 0.279** 0.279 3.290 Constant 151.405 59.556 Log-likelihood -1428.338*** ***Values significant at 1% **Values significant at 5% *Values significant at 10% Source: Computed from IRRI Rice Statistics (2011) 119 1. The Short Run Model As reported in Table 27, in the short run, supply of rice responded to one-year lagged value of output, area, fertilizer consumption and import. An increase in output in the preceeding year resulted in decrease in output in the current year by 0.264. Hence, rice output in Nigeria exhibited cobweb behaviour in the short run. Also, an increase in area was also associated with increasing rice output by 0.240. Rice output positively responded to increasing fertilizer consumption as well by a factor of 0.044. An inverse relationship existed between rice supply and import quantity. This implies that a reduction or ban on importation could assist farmers to increase output in the short run. As earlier observed, this assertion does not hold in the long run as farmers are seen to compete favourably by increasing their output with increasing importation of rice.The result of the short-run VECM model indicated an adjustment coefficient of 0.26 for rice output (that is, the error correction term).This implies that the adjustment to any disequilibrium caused by shocks in all factors affecting rice supply will be corrected within 12/0.26 (46) months. The adjustment coefficient of area relative to output was 0.24, thus shocks due to area will be corrected within 12/0.24 (50) months. Table 27 also reveals the coefficient of adjustment of fertilizer consumption in relation to output as 0.04, implying an adjustment speed of about 4 percent of output to any variation caused by fertilizer consumption. Imported quantity recorded adjustment coefficient of 0.37, thus farmers adjusted to short run fluctuation in import within 12/0.37 (33) months. In all, the speeds of adjustment with respect to all the variables were sluggish, as the highest speed was 37 percent. Yet, rice output was not price-elastic or significant in the short run. Pricing policy in this regard cannot be effective in stimulating rice supply in Nigeria. This corroborates the findings of Rahji et al. (2008) and Muchapondwa (2008). The explanatory factors considered jointly accounted for about 25 percent variation in rice output, as seen from the R-squared value. The statistical significance of the Log-likelihood ratio (-1653.236) and the lower value Akaike information and Schwarz criteria confirms the fitness of the vector error correction model. 120 Table 27: Short-Run Equilibrium Model VECM Error Correction: D(OUTPUT) D(ACREAGE) D(PRICE) D(FERT_CONS) D(IMPORT) D(POLICY) D(RAINFAL) Coin tEq1 -0.26 3537 0.23 9486 0.09 3049 0.043 648 -0.36 9250 4.13 E-06 0.093 898 (0.15731) (0.05347) (0.10192) (0.01900) (0.08399) (7.6E-05) (0.06279) [ -1.67524] [ 4.47912] [ 0.91294] [ 2.29666] [-4.39654] [ 0.05407] [ 1.49538] D(OUTPUT(-1)) -0.400339 -0.176539 0.114588 0.055390 0.240491 2.77E-05 -0.025409 (0.21965) (0.07465) (0.14231) (0.02654) (0.11727) (0.00011) (0.08767) [-1.82264] [-2.36479] [ 0.80521] [ 2.08741] [ 2.05082] [ 0.25957] [-0.28982] D(ACREAGE(-1)) 0.553426 0.025678 -0.154127 -0.040300 -0.713216 -4.51E-05 -0.020270 (0.54011) (0.18357) (0.34993) (0.06525) (0.28836) (0.00026) (0.21559) [ 1.02465] [ 0.13988] [-0.44045] [-0.61761] [-2.47339] [-0.17211] [-0.09402] D(PRICE(-1)) 0.273709 0.122483 -0.014202 -0.008179 -0.209293 8.35E-05 -0.031028 (0.25917) (0.08808) (0.16791) (0.03131) (0.13836) (0.00013) (0.10345) [ 1.05612] [ 1.39052] [-0.08458] [-0.26122] [-1.51263] [ 0.66332] [-0.29994] D(FERT_CONS(-1)) -2.176345 -0.653792 -1.755982 -0.168226 2.474122 -0.000651 -0.336631 (1.43053) (0.48621) (0.92683) (0.17282) (0.76374) (0.00069) (0.57100) [-1.52135] [-1.34468] [-1.89461] [-0.97341] [ 3.23950] [-0.93691] [-0.58955] D(IMPORT(-1)) 0.152642 -0.007875 0.056673 0.026685 -0.077921 -5.69E-05 -0.083339 (0.24157) (0.08211) (0.15651) (0.02918) (0.12897) (0.00012) (0.09642) [ 0.63186] [-0.09591] [ 0.36210] [ 0.91436] [-0.60417] [-0.48509] [-0.86429] D(POLICY(-1)) 571.4313 -180.0903 -45.82642 -64.58049 -91.28526 0.023531 -162.5472 (367.242) (124.817) (237.934) (44.3663) (196.064) (0.17829) (146.585) [ 1.55601] [-1.44283] [-0.19260] [-1.45562] [-0.46559] [ 0.13198] [-1.10889] D(RAINFAL(-1)) 0.587497 0.328950 0.614194 0.107130 0.013489 5.96E-05 -0.464862 (0.39857) (0.13546) (0.25823) (0.04815) (0.21279) (0.00019) (0.15909) [ 1.47402] [ 2.42831] [ 2.37848] [ 2.22488] [ 0.06339] [ 0.30776] [-2.92202] C 110.9417 67.36111 18.59008 2.246679 37.84380 0.023999 6.721778 (52.4340) (17.8211) (33.9716) (6.33452) (27.9935) (0.02546) (20.9291) [ 2.11584] [ 3.77985] [ 0.54722] [ 0.35467] [ 1.35188] [ 0.94275] [ 0.32117] R-squared 0.25 0510 0.50 2306 0.22 8687 0.337 920 0.453 232 0.05 1964 0.34 6150 Adj. R-squared 0.088458 0.394697 0.061917 0.194768 0.335012 -0.153017 0.204777 Sum sq. resids 3934757. 454529.0 1651672. 57427.44 1121520. 0.927427 626891.9 S.E. equation 326.1055 110.8358 211.2813 39.39661 174.1016 0.158321 130.1654 F-statistic 1.545863 4.667864 1.371270 2.360560 3.833797 0.253507 2.448491 Log likelihood -326.4757 -276.8338 -306.5103 -229.2528 -297.6069 24.52044 -284.2286 Akaike AIC 14.58590 12.42756 13.71784 10.35882 13.33074 -0.674802 12.74907 Schwarz SC 14.94368 12.78533 14.07562 10.71660 13.68851 -0.317024 13.10685 Mean dependent 107.8913 46.10870 17.40652 4.956522 34.73913 0.021739 -6.461087 S.D. dependent 341.5622 142.4602 218.1425 43.90341 213.4989 0.147442 145.9658 121 4.5 Preference Switch Analysis This section presents and discusses the direction of switch of various commodities of rice when paired. The results of influence of socioeconomic factors on preference switch from one rice commodity to the other are equally presented. 4.5.1 Preference direction The result in Table 28 shows the preference direction for various commodities of rice both at the national and zonal levels. At the national level, a preference is observed towards imported rice when compared with agric. rice. This is evident fromthe relatively higher mean value recorded for imported rice and a positive t-value. Similarly, when imported rice is paired with local rice, a preference is observed towards imported rice. The pair sample t-statistics also indicates a preference for agric. rice as against the local rice with a higher mean value of 797.75. All the t-values for the three pairs of rice commodities were statistically significant at 1 percent (P ≤ 0.01). The zonal result for the paired mean sample t-test shows that, in the South-South, the preference was observed towards imported rice when paired with agric. and local rice, while the preference was towards agric. rice when paired with local rice. The t- values for the three rice commodity pairs for the South-South geopolitical zonewere significant at 1 percent level. The result of the South-East and the South-West zones followed the same preference pattern as the South-South. All the t-values were significant at 1 percent probability level. The result of rice preference in the North-Central geopolitical zone revealed that there was a usual tilt towards imported rice when paired with agric. and local rice. However, in this zone, preferencewas observed towards local rice when paired with agric. rice, as seen from the higher mean value for local rice and a negative t-value. All the coefficients t-values were significant at 1 percent probability levels. Thehigh production of local rice in theNorth- Central region was equally reflected here. In the North-East zone, there was preference for imported rice in relation to agric. and local rice. All t-values for the three pairs were statistically significant at 1 percent probability level. The rice preference results from the North-West Nigeria trended the same pattern as other zones, as preference was observed towards imported rice when compared with agric. and local rice and towards agric.rice in relation to local rice. All the t-values were significant at 1percent (P ≤ 0.01). 122 The general preferencetowards imported rice, in Table 28, is in agreement with the studies of Nwachukwu et al. (2008) and Agwu et al. (2009), Bamba et al. (2010) and Oyinbo et al. (2013) in separate localised studies of foreign and local rice preference in Nigeria. As mentioned earlier, the high preference for imported rice might be due to its lower price and availability when compared with agric. and local rice. This has a negative implication for rice self-sufficiency in Nigeria. The determinants of preference switch from one commodity to another is fully presented and discussed in the next section. 123 Table 28: Paired Sample T-test for Preference Switching Individual Paired Mean 95% C. Intv. of t-stat Mean Difference Lower Upper National Imported 1256.55 Agric 797.75 458.80 449.55 468.04 97.25*** Imported 1256.55 Local 598.44 587.29 609.58 105.25*** 658.11 Agric 797.75 Local 658.11 139.64 130.27 149.01 29.20*** South-South Imported 700.63 168.65 163.20 174.10 60.67*** Agric. 531.98 Imported 700.63 265.33 260.23 270.43 102.01*** Local 435.30 Agric. 531.98 96.68 93.62 99.75 61.78*** Local 435.30 South-East Imported 1327.74 694.63 672.01 717.24 60.23*** Agric. 633.12 Imported 1327.74 828.12 807.21 849.04 77.65*** Local 499.62 Agric. 633.12 133.50 127.89 139.11 46.66*** Local 499.62 South-West Imported 869.82 289.71 282.97 296.45 84.31*** Agric. 580.11 Imported 869.82 456.57 445.50 467.64 80.87*** Local 413.25 Agric. 580.11 166.86 161.13 172.59 57.10*** Local 413.25 North-Central Imported 1372.26 168.65 163.20 174.10 48.03*** Agric. 852.46 Imported 1372.26 265.33 260.23 270.43 10.53*** Local 1209.95 Agric. 852.46 96.68 93.62 99.75 -18.19*** Local 1209.95 North-East Imported 1044.11 320.09 304.41 330.76 58.81*** Agric. 724.03 Imported 1044.11 495.06 480.90 509.22 68.55*** Local 549.05 Agric. 724.03 174.98 168.93 183.02 42.66*** Local 549.05 North-West Imported 2006.00 706.92 672.66 741.18 40.46*** Agric. 1299.09 Imported 2006.00 1267.71 1233.931301.50 73.57*** Local 738.30 Agric. 1299.09 560.79 537.62 583.97 47.44*** Local 738.30 Source: Computed from NLSS Data (2004) 124 4.5.2 Determinants of preference switch of rice commodities The result of the determinants of switch from one type of rice commodity to the other is presented in Table 29. As shown in the table, preference switch from imported to agric. rice was determined by a number of factors, particularlyhousehold size, per capita expenditure, primary occupation, sector, zonal factors and price of imported and agric. rice. All the aforementioned factors were highly significant at 1 percent probability level.Other factors were marital status and education which were significant at 5 and10 percent, respectively.Household size, age, education, per capita expenditure,primary occupation, price of imported and local rice as well as locational factors significantly influenced switch from imported to local rice. Similarly, switch from agric. to local rice was significantly influenced by household size, age, education, per capita expenditure, marital status, primary occupation, community membership, price of agric. rice and all zonal variables. All these factors were significant at 1 percent probability level. These findings are in tandem with those of Bamidele et al. (2010), Adeyeye et al. (2012) and Oyinbo et al. (2013) in their studies on determinants of preference for foreign and local rice. An increase in household membership reduced switch from imported to agric. and local rice by 6.585 and 5.161 respectively, and also reduces the switch from agric. to local rice by 11.746. This is because larger households often consume more of imported rice due to availability, ease of cooking and cost saving. Age significantly increased switch from imported to local rice, and also increased switch from agric. to local rice. Older population often has tendencies to consume more of locally produced food than imported foods owing to conservativeness or desire for nutritional value. Education increased switch from imported to agric. rice by a margin of 1.056,and reduced switch from imported to local rice and from agric. to local rice by a factor of -1.623 and -2.678, respectively. Nwachukwu et al. (2008) and Agwu et al.(2010) found education to influence consumer‘s switch from foreign to local rice in Abia State. This implies that, through increasing educational awareness of nutritional value of domestically produced rice, educated people could switch from imported to agric. rice. However, this does not hold for imported to local rice or from agric. to local rice perhaps because of the poor quality of processing of local rice that makes it unattractive to educated people. Increasing per capita expenditure was associated with increasing switch from -03 imported to agric. rice by 0.922×10 and a reduction in switch from imported to local rice and from agric. to local rice. Hence, the consumption of agric. rice is income-driven owing to the higher price of agric. rice. 125 Being married significantly reduced switch from imported to agric. rice and from imported to local rice. Marriage is usually accompanied with increasing household size and has implication for cost if consumers have to switch to a rice commodity of higher price. Expectedly, taking farming as primary occupation switched consumers from imported to agric. and local rice as well as from agric. to local rice. Farmers often plant agric. and local rice. Even when they do not plant it, they have bias for food produced within their locality because of availability and sustenance of production. Community membership also reduced consumers switch from agric. to local rice. On the factor of location, rural dwelling positively influenced switch from imported rice to agric. and local rice by 55.128 and 70.194 respectively. Similarly, rural living switched consumers from agric. to local rice. This was because the domestically produced rice commodities were more available in their immediate community and they were equally conservative with respect to consumption of locally produced food. Residing in the South- South influenced switch from imported rice to agric. and local rice less than the North- Central, whereas switch from agric. to local rice increased more than the North-Central. Residence of the South-East geopolitical zone switched more from imported to agric. and local rice and from agric. to local rice relative to the North-Central zone. In the South-West zone, consumers switched less from imported to agric. rice but switched more from imported to local rice in relation to the basal North-Central, perhaps because of the increasing awareness of the nutritional value of local rice. More switches from agric. to local rice were observed in the South-West zone relative to North-Central. Rice consumers in the North-East switched less from imported to agric. rice but switched more from imported to local rice relative to North-Central. They equally switched more from agric. to local rice. The effect of location on rice consumption, as evident in this study, attests to the findings of Adeyeye (2012) and Bashorun (2013). An increase in the price of imported rice resulted in consumers‘ drift from imported to agric. and local rice by 30.680 and 1.972. An increase in the price of agric. and local rice reduced switch from imported to agric. and local rice by a factor of 19.385 and 16.309 respectively. Increase in the price of agric. rice switched consumers‘preference from agric. to local rice. Nwachukwu et al. (2008) and Agwu et al.(2010) equally found price of rice to influence consumer‘s switch from foreign to local rice in Abia State. Hence, pricing policy in terms of price reduction in agric. rice or increasing price of imported rice (for example, through high 126 tariff) could serve as a veritable tool in stimulating demand for agric. rice as against imported rice. Generally, the result of preference switch from imported to agric. rice, imported to local rice, and agric. to local rice is diametrically opposed to the results of agric. to imported rice, local to imported rice,and local to agric. rice, respectively(Appendix 13). This scenario is true for all socioeconomic factors considered. The coefficients (magnitude) and diagnostic statistics were same for the two pairs except for the difference in the signs. In essence, all factors that influenced switch from imported to agric. rice were the same factors that influenced switch from agric. to imported rice but in a reverse direction. This same holds for imported to local rice as against local to imported rice and similarly for agric. to local rice as against local to agric. rice. For instance, as earlier noted, an increase in price of imported rice leads to a switch from imported to agric. rice. Conversely, an increase in the price of agric. rice (in the agric. to imported rice switch model) resulted in consumers‘ switch from agric. to imported rice. The same held for all significant socioeconomic variables. 127 Table 29: Determinants of Preference Switch A Variables Imp-Agr Imp-Loc Agr-Loc HHSIZ -6.585*** -5.161** -11.746*** (2.071) (2.150) (2.149) AGE 0.286 0.687** 0.973*** (0.276) (0.287) (0.287) EDUC 1.056* -1.623** -2.678*** (0.604) (0.627) (0.627) PCEXP -0.922e-03*** -0.287e-03** 0.635e-03*** (0.130e-03) (0.135e-03) (0.135e-03) MARST -30.560** -14.227 44.787*** (12.259) (12.727) (12.718) PROCC 130.425*** 93.622*** 36.802*** (10.863) (11.278) (11.270) TASSET -0.987e-06 -0.213e-05-0.114e-05 (0.169e-05) (0.175e-05) (0.175e-05) COMEM 35.916 -6.409 -42.325*** (7.687) (7.981) (7.975) SECTOR 55.128*** 70.194*** 15.067 (10.186) (10.575) (10.568) SS -1605.268*** -612.241*** 993.027 *** (40.275) (41.814) (41.784) SE 772.490*** 1081.11*** 308.624 *** (23.002) (23.881) (23.863) SW -263.829*** 183.316*** 447.145*** (25.642) (26.623) (26.603) NE -576.549*** 314.775*** 891.324*** (18.503) (19.210) (19.196) NW 69.910*** 1293.064*** 1223.153*** (19.344) (20.084) (20.070) PRIMP 30.680*** 1.972** (0.940) (0.976) PRAGR -19.385*** 2.932*** - (0.533) (0.553) PRLOC - -16.309*** -0.546 (0.733) (0.732) CONST. 3920.572*** -1650.722*** 2269.85*** (115.872) (120.301) (120.214) R2 0.380 0.540 0.351 Adj R2 0.379 0.540 0.350 F (17, 18843) 385.32*** 737.47*** 339.63*** ***Values significant at 1%; **Values significant at 5%, *Values significant at 10% Note: Imp-Agr means switch from imported to agric. rice Imp-Loc means switch from imported to local rice Agr-Loc means switch from agric to local rice Source: Computed from NLSS Data (2004) 128 CHAPTER FIVE SUMMARY, CONCLUSION AND RECCOMMENDATION This is the concluding chapter of the thesis. It consists of the summary of the research with greater emphasis on the major findings, the conclusion and policy implication of the findings. The contributions of the research to knowledge are documented, appropriate policy recommendations are made from the findings and areas of further research are equally suggested. 5.1 Summary The inability of domestic supply to match rising demand for rice alongside preference for imported rice resulting in increasing importation of milled rice in Nigeria was the focus of this study. Against this background, the study examined the determinants of demand, supply response and preference switch for rice in Nigeria. Data from the Nigeria Living Standard Survey (NLSS) of 2004 conducted by the National Bureau of Statistics (NBS) and time series data from the official records of International Rice Research Institute (IRRI), 1960-2008 were used in this study. Data were analysed using descriptive statistics, Tobit regression model, AIDS model, vector error correction model and generalised least square regression. The results of the analysis in this research are summarised thus: 1. Rice averagely constituted about 25 percent of total food expenditure in Nigeria. The mean expenditure for imported rice (N1, 256.545) was higher than that of improved domestic (agric.) rice (N797.748) and local rice (N658.110). Consequently, the share of expenditure of various rice commodities followed the same trend with imported rice on the top with mean share of total rice expenditure of 0.451. 2. At the zonal level, the North-West region ranked highest in total expenditure on all rice commodities. This region was closely followed by the North–Central region. The South East and North-East region were at close range; the South-West followed while the South-South region ranked lowest in terms of overall rice consumption expenditure.North-West zone led in terms of mean expenditure on imported and agric. rice, while North-Central ranked highest in expenditure on local rice. 3. Nigeria was self-sufficient in rice from 1960 to 1975, the self–sufficiency ratio began to decline thereafter. From 1987, there was remarkable increase in self-sufficiency ratio owing to ban on importation but still less than unity till now. 129 4. Total rice demand was significantly influenced by household size, non-food total expenditure, age, sectoral and zonal factors. All factors were significant at 1 percent probability level, with the exception of non-food total expenditure, that was significant at 10 percent level and household sizethat was significant at 5 percent. Non-food total expenditure and age showed inverse relationship with total rice expenditure, while other significant factors displayed positive signs. 5. The determinants of imported rice demand were: household size, non- food total expenditure, education, primary occupation, sectoral and zonal factors. All the factors were significant at 1 percent probability level, with the exception of educational level, that was significant at 10 percent and the dummy for sectorthat was significant at 5 percent. Non-food total expenditure, primary occupation and two of the zonal dummies were negatively related to imported rice demand, while others were directly related. 6. For improved domestic (agric.) rice demand, the determinants were household size, non-food total expenditure, age, marital status, primary occupation, community membership and all the zonal dummies. A positive relationship existed between agric. rice demand and factors like marital status, primary occupation and all zonal (except South-East) variables. Others were inversely related to agric. rice demand. All the mentioned factors were highly significant at 1 percent probability level. 7. Local rice demand was found to be statistically influenced by the following variables: non-food total expenditure, age, marital status, community membership and all zonal variables. Non-food total expenditure, marital status and all zonal dummies showed negative signs, while other factors were positively related to local rice demand. 8. The estimated elasticities from the Tobit model showed that imported, agric. and local -08 rice were non-income-elastic as their various values of elasticities: 7.266×10 , -07 -07 1.727×10 and 1.001×10 ,were less than unity but greater than zero. Hence, all rice commodities were classified as ‗necessities‘ and ‗normal good‘. They were equally non-price-elastic as all price elasticities were less than unity. However, the traditional inverse demand-price and direct demand-income relationship was maintained in all the coefficients. 9. The results of the cross price elasticities for imported rice revealed that local rice, white garri, yam and brown beans were substitute products to imported rice; while agric. rice, yellow garri, white beans, millet, guinea corn, white maize and yellow maize were classified as complements. 130 10. For the agric. rice commodity, yellow garri, guinea corn, millet, white maize and yellow maize were found to be substitute products; while imported rice, local rice, white garri, yam, brown beans and white beans, were found to be complementary to agric. rice. 11. Similarly, the competitive products to local rice includedimported rice,agric. rice, yellow garri, yam, millet and guinea corn; while, white garri, brown beans, white beans, white maize and yellow maize emerged as complementary products to local rice. 12. Area cultivated, fertilizer consumption and import quantities were found to significantly influence the supply of domestic rice in Nigeria in the Vector ECM short-run and long-run model. 13. The result of the short-run VECM model stated that the adjustment to any disequilibrium caused by shocks in output of rice was corrected within 12/0.26 (46) months. Shocks due to area were corrected within 12/0.24 (50) months. The adjustment speed of fertilizer consumption was 0.04. Farmers adjusted to short-run fluctuation in import within 12/0.37 (33) months. 14. Local rice output was not price elastic in the long run and short run in the error correction model. 15. Household size, per capita expenditure, education, marital status, primary occupation, prices of imported and agric. rice and all locational factors exercised significant influence on switch from imported to agric. rice. All the factors were significant at 1 percent probability level with due exception to education and marital status, which were significant at 10 and 5 percent, respectively. Per capita expenditure, marital status, and price of agric. and two zonal dummies showed inverse relationship, while other factors were directly related. All the factors that were significant above equally influenced switch from agric. to imported rice but in reverse direction. 16. Preference switch from imported to agric. rice was determined by a number of factors, particularly household size, per capita expenditure, primary occupation, sector, zonal factors and price of imported and agric. rice. All the aforementioned factors were highly significant at 1 percent probability level. Other factors were marital status and education, which were significant at 5 and10 percent, respectively. 17. Household size, age, education, per capita expenditure,primary occupation, price of imported and local rice as well as locational factors significantly influenced switch from imported to local rice at various probability levels. 131 18. Similarly, switch from agric. to local rice was significantly influenced by household size, age, education, per capita expenditure, marital status, primary occupation, community membership, price of agric. rice and all zonal variables. All these factors were significant at 1 percent probability level. 5.2 Policy implications of the findings 1. Rice is consumed in all geopolitical zones of Nigeria and constitutes a significant share of total food expenditure, hence, attention on increasing local supply and stimulation of consumption of local rice will certainly contribute to the attainment of overall food self-sufficiency and food security in Nigeria. 2. Although price and income significantly influenced the demand for various rice commodities, the inelasticity of price and income make them influence demand in a less than proportionate trend. It also follows that there is limit to the use of pricing and income policy in stimulating rice demand. Similarly, the degree of substitution of one rice commodities type for another and other food commodities is largely limited. 3. Variation exists among different sectors and geopolitical zones in terms of rice demand. Urbanization and cosmopolitan nature of zones seem to favour the consumption of imported rice as against local rice probably owing to availability, acceptability (resulting from quality processing and good packaging) and ease of cooking. Therefore, improvement in availability and processing of local rice has become necessary to stimulate its consumption 4. Pricing policy is rather a blunt instrument in motivating farmers for rice supply, as supply is price-inelastic in Nigeria. 5. Increase in area cultivated and fertilizer consumption could increase productivity and boost domestic rice supply in the short run and long run in Nigeria. 6. Import restriction could also stimulate domestic rice production in the short run but importation allowance will ensure competitiveness and stimulate local rice production in the long run. 7. Since price is inversely related to switch from imported to domestically produced rice, a switch of such could be driven by price reduction in domestically produced rice (cost reduction production technologies are essential to farmers in this regard). 8. Switch from imported to agric. rice consumption has equally been found to be income driven, hence policies that increase the income of the populaceare of relevance in increasing agric. rice demand 132 9. Education has also been observed to positively influence switch from imported to agric. rice, therefore, educational awareness on nutritional value of agric. rice will be of great importance. 5.3 Conclusion Against the background of rising importation bills and consequent drains on foreign exchange earnings necessitated by the ever increasing demand, shortage in supply of domestically produced rice and preference switch towards imported rice, this study examined the determinants of demand for rice, estimated a supply response model for rice alongside analysing the determinants of preference switch from one rice commodity to the other. The analysis isolated the determinants of demand for various rice commodities, estimated the supply response of rice and examined the determinants of preference switch from one rice commodity to another in Nigeria. Demand for rice was largely influenced by household characteristics, which included: non-food total expenditure, age, education, price and location among other factors. The price and income elasticities show that demand for rice is price and income-inelastic. In addition, complements and substitutes of various rice commodities were identified, although with limited degree of substitution owing to inelasticity. On the other hand, supply of rice responded to area cultivated, fertilizer consumption and quantity of imports but was non-price or climate responsive. In addition to income, education and few other socioeconomic variables, the price of rice commodities could switch consumers from imported to domestically produced rice. To this end, the following recommendations are made: 5.4 Policy Recommendations (a) Enhancing demand/consumption of domestically produced rice 1. Since education largely favours the consumption of imported rice as against local rice, re- educating literates and students on the nutritional value of domestically produced rice as well as local rice baiting (feeding students with local rice in schools) will assist in stimulating local rice demand. 2. Residing in the North-Central zone relatively stimulated local rice consumption more than other zones,therefore,efforts at stimulation of consumption of local rice should be more intensified in other geopolitical zones than the North-central zone. 133 3. Imported rice was found to increase with urban livelihood (as a result of availability, improved processing, packaging and ease of cooking). This calls for making domestically produced rice available in urban centres. Improved processing and packaging of domestically produced rice will also increase acceptance and consumption by urban dwellers.This can be achieved through the followings of which most are components of the Agricultural Transformation Agenda: a. Public and private sector investment in rice mill; b. Training on improved parboiling and drying methods; c. Dissemination of improved small-scale milling technology; d. Dissemination of destoners and their use, and e. Establishment of rice development fund within existing financial system establish fund to facilitate investments in new rice processing technology/ equipment. (b) Increasing output/supply of domestically produced rice 1. In view of the responsiveness of rice supply to area cultivated and fertilizer use, increasing farmers‘ holdings and encouraging land transfer to rice farmers will be effective in boosting domestic rice supply. Also, increased use of improved input, especially fertilizer (through effective extension delivery, subsidy programme, private participation to increase avialability) is a veritable means of increasing domestic rice productivity and supply. One of the key elements of the transformation agenda is the sanitisation of the government subsidy programme and subsequent liberalisation of the fertilizer industry. This is expected to improve the availability of fertilizers to farmers and consequently boost productivity if properly implemented. 2. Quantity imported has been found to vary negatively and positively with rice supply in the short run and long run, respectively.It follows that policy on importation control could confer a short-run advantage on domestic rice producer. Allowance of some level of importation to ensure competitiveness may also be beneficial in boosting rice supply in the long run. Hence, government should not close its gate totally to rice importation in the long run as planned in the transformation agenda 134 (c) Switching Consumers‟ Preference from Imported to domestically produced rice 1. Since consumers are seen to switch from imported to local rice upon price reduction, cost reducing measures (such as development and adoption of improved varieties and technologies) are essential in reducing domestic rice price and increasing switch towards domestically produced rice. 2. Similarly, income-increasing policies could drive consumers‘ towards agric. rice consumption because switch from imported to agric. rice was found to increase with increasing income. 3. Education has been associated with increasing switch from imported to agric. rice, hence education and reorientation on the nutritional value of domestically produced rice has the potential to switch consumer from imported to agric. rice consumption. 5.5 Contributions to knowledge This study has contributed to knowledge in the following ways: 1. Through disaggregation of determinants of rice demand, the study empirically showed that household size, non-food total expenditure, education and urban livelihood increased imported rice demand; marital status and farming occupation increased agric. rice demand, while non-food total expenditure, age and membership of community organization favour local rice consumption among other socioeconomic characteristics.Factors that reduced demand for various rice commodities were equally isolated. Hitherto, determinants of demand for domestically produced rice have never been disaggregated into agric. and local rice components. 2. By factoring in the effect of geopolitical zone into demand analysis, the study revealed that the residents of North-Central relatively consumed more of domestically produced rice than virtually all other zones while residing in the South- East and South-West stimulated the consumption of imported rice relative to North- Central. 3. The study showed that consumers‘ preference could be switched from imported to domestically produced rice through education, income and price reduction in domestically produced rice among other factors. 135 4. Finally, it was revealed that import restriction could stimulate domestic production of rice in the short run while some allowance of importation could assist farmers in competing favourably through production cost reduction in the long run. 5.6 Suggestions for further research 1. Due to data limitation, this study could not incorporate organoleptic characteristics of rice (size, colour, texture, taste, aroma, rising capacity and so on) as well as value addition characteristics (processing and packaging) in the demand and preference switch models. Future research could incorporate thesequalitative variables to capture better the determinants of demand and preference switch of various rice commodities. 2. Rice marketed output rather than the farm level output data will be better in analysing rice self-sufficiency and supply response. Such data should be sourced to provide a better estimate of rice self-sufficiency and supply response in future research. 3. Studies on determinants of demand, supply response and preference switch should be extended to other important staple foods in Nigeria. 4. There is a need for further research on the determinant of more consumption of local rice in the North central relative to other zones 5. Further researchers should update their data for both demand and supply response analysis to elicit more current results on demand and supply of rice in Nigeria. 136 References Abalu, G.O.I 1974. Supply response to producer prices; acase study of groundnut supply to theNorthern States Marketing Board.Savannah 4(1): 33-40. Abdulai, A. and P. Rieder 1995. The impacts of agricultural price policy on cocoa supply in Ghana: an Error Correction Estimation.Journal of African Economies 4 (3): 315-35. Abdulai, A; D.K Janin and A.K. Sharma 1999. Household food demand analysis in India. India Journal of Agricultural Economics 50(2): 316-327. Adeyeye, V.A. 2012. Strategies for enhancing consumption of locally produced rice in Nigeria. The Nigerian Institute of Social and Economic Research (NISER), Ibadan Agwu, N.M.; I.N. Nwachukwu and B.C. Ezekwem 2009. Preference switching among rice consumers in Abia State.Journal of Sustainable Development 6(1): 63-69. Ajakaye D.O. 1987. The SAP for Nigeria; its impact on price and income. In: Philips A.O. and E.C. Ndekwu eds. SAP for the developing economy: the case study of Nigeria. Ibadan: NISER. Ajetomobi, J.O. 2005. Supply response, risk and institutional change in Nigerianagriculture. African Economic Research Consortium (AERC), Nairobi Akande, S. O. and G. Akpokodje 2003. Rice prices and market integration in selected areas in Nigeria, project report - The Nigerian rice economy in a competitive world: constraints, opportunities and strategic choices. Ibadan: NISER.ii-20 pp. Akande, S.O. 2007.An overview of the Nigerian rice economy, Ibadan: NISER, Available at http://www.unep.ch/etu/etp/events/Agriculture/nigeria.pdf.Accessed January 8, 2009 Akbay, C. and I. Boz 2001.Food consumption patterns of socioeconomic groups: an application of censored systems of equation. Paper presented at the ERC/METU Conference in Economics Meeting in Ankara, Turkey, September 10-13, 2001 137 Akinwumi A. 2012. Agricultural Transformation Agenda: Repositioning agriculture to drive Nigeria‘s economy.Abuja: Federal Ministry of Agriculture and Rural Development Akinyosoye, V.O. 2009. Demand for Dairy Products in Nigeria: Evidence from the Nigerian Living Standard Survey.Journal of Rural Economics and Development16(1): 13-26. Akmal S. M. 2007. Stock returns and inflation: An ARDL econometric investigation utilizing Pakistani data.Pakistan Economic and Social Review 45(1): 89-105. Akosile, A. 2009.Nigeria: rice Demandpushes upglobal price, Available at: All African.com.Accessed 05/05/2009 Akpokodje, G., F. Lançon, and O. Erenstein 2001. Nigeria's rice economy: state of the art. Project report - The Nigerian rice economy in acompetitive world: constraints, opportunities and strategic choices. Bouake: WARDA. ii-55 pp. Attanasio O.P. 1999.Consumption.In: J.B Taylor and M. Woodford eds., Vol.1B, Amsterdam, North Holland. Pp741-812 Babatunde, R.O., A.O. Omotesho and O.S. Sholotan 2007. Socioeconomic characteristics and food security status of farming households in Kwara State, North Central Nigeria.Pakistan Journal of Nutrition 6 (10): 49-58. Bamba, I., A. Diagne, J. Manful and A. Ajayi 2010. Historic opportunities for rice growers in Nigeria. Grain de sel. No. 51, July – September 2010. Bamidele, F.S., O.O. Abayomi and O.A. Esther 2010. Economic analysis of rice consumption patterns in Nigeria.Journal of Agricultural Science Technology12: 1-11. Banerjee, A., J. Dolado, J. Glabraith and D.Hendry 1993.Co-integration, error correction and the econometric analysis of non-stationary data, Oxford: Oxford University Press. Banerjee A, J.J. Dolado, and R. Mestre 1998. Error correction mechanism tests for cointegration in a Single-Equation Framework, Journal of Time Series Analysis 19(3): 267-83. 138 Bangkok business News 2008.Rice price dips on slow orders, Available at: www. bangkokpost.com. Accessed march 2009 Banks, J., R. Blundell, and A. Lewbel 1997.Quadratic Engel Curves and consumer demand, Review of Economics and Statistics4: 527-539. Bashorun, J.O. 2013.Expository analysis of rice processing in Igbemo, rural Nigeria. American Journal of Social Issues and humanities 3(2): 78-85. Begum M.A.A.; S.M. Fakhrul Islam; M. Kamruzzaman; M, Jahangir Kabir and S. M.A. Shiblee 2002. Supply response of wheat in Bangladesh; an application of Partial Adjustment Model, Pakistan Journal of Biological Science5 (2): 225–229. Bello, A. 2004.Nigeria imported $US700 million rice in 2003: Federal Minister of Agric. and Rural Development, Tribune, July 7, 2004. CBN 2000.Central Bank of Nigeria, Annual Report, Statement of Account. Central Bank of Nigeria (CBN) 2006.CBN Statistical Bulletin CBN/NISER 1992. Central Bank of Nigeria/Nigerian Institute of Social and Economic Research.Rice Policy in Nigeria. Choi J.H. and K.I. Lee 2000.Food consumption analysis using Engel Equation.Korea Journal of Rural Development 23(2): 209-227. Christensen, L.R., D.W. Jorgensen, L.J. Lau, 1975. Transcendental logarithmic utility functions.American Economic Review (65): 367-83. Clements, W.K., A. Selvanathan, S. Selvanathan 1996.Applied demand analysis: a survey. Economic Record (72): 63-81. DADTCO 2009.Rice import;a survey conducted by the Dutch Agricultural Development and Trading Company.Available at: All Africa.com.Accessed 05/05/2009 139 Deaton, A., 1988.Quality, quantity, and spatial variation of price.American Economic Review. 78: 418-430. Deaton, A. 1990.Price elasticities from survey data: extensions and Indonesian results, Journal of Econometrics 44(3): 281-309. Deaton, A. and J. Muellbauer 1980a.An Almost Ideal Demand System.TheAmerican Economic Review70 (3): 312-33. Deaton, A., and J. Muellbauer 1980b. Economics and consumer behavior, Cambridge: Cambridge University Press. Deb S. 2003. Terms of trade and supply response of Indian agriculture: analysis in cointegration framework. Centre for Development Economics, Working Paper No. 115. Dickey, D.A. and W.A. Fuller 1979.Distribution of the estimators for autoregressive time series with a unit root.Journal of the American Statistical Association74(366): 427- 431. Dickey, D. A. and W. A. Fuller1981. Likelihood ratio statistics for Autoregressive Time Series with a unit root. Econometrica 49 (4): 1057-1072. Economic Commission of Africa (ECA) 1998.Food security and food self-sufficiency in Africa, issues on African economy and telematics, African Study Center, University of Pennsylvania. Available at www.africa.upenn.edu/ECA/Food Engle, R.F. and C.W.J. Granger 1987.Co-integration and error correction: representation, estimation, and testing,Econometrica 55 (2): 251-276. Erenstein, O., F. Lancon, S.O. Akande, S.O. Titilola, G. Akpokodje, O.O. Ogundele 2003. The Nigerian rice economy in a competitive world: constraints, opportunities and strategic choices: rice production systems in Nigeria, a survey, Abidjan: WARDA - The Africa Rice Centre, Cote D‘voire. 140 Erenstein, O., F. Lançon, O. Osiname, and M. Kebbeh 2004. Operationalising the Strategic Framework for Rice Sector Revitalisation in Nigeria. Project Report - The Nigerian rice economy in a competitive world: constraints, opportunities and strategic choices. Abidjan: WARDA -The Africa Rice Centre. ii-38 pp. EViews 1998.EViews Version 3.1 Help Topics.EViews, Quantitative Micro Software. EViews 2009.EViews Version 5.0 Help Topics. EViews, Quantitative Micro Software. Fagbemi, A. 2012. Government boosts rice production, gets FAO recognition. Available TH at:http://www.ngrguardiannews.com/index.php?. Accessed 20 February 2014. Food and Agriculture Organization (FAO) 2002. Population: Annual time Series. Nigeria FAO 2007. Econometrics model for consumption analysis, agriculture and consumer protection, FAO Corporate Document Repository, pp.1-16, Available at: http://www. Fao.org/DOREP/005/Y4475e/y4475e07.htm. Accessed 22/04 2010 FAOSTAT, 2006. FAO (Food and Agriculture Organizaton) Database, Rome, downloaded November 10, 2007 from http://www.beta.irri.org/statistics FAOSTAT, 2008. FAO (Food and Agriculture Organizaton) Database, Rome.Downloaded January 12, 2009 from http://www.beta.irri.org/statistics FAO Rice Web 2001. Rice and Nigeria Agriculture Information http://www.riceweb.org/countries/nigeria.htm FAO and UNIDO 2008.Agricultural mechanisation in Africa: time for action, report of an expert group meeting, Vienna, Australia, January, 2008. Federal Ministry of Agriculture, Water Resources and Rural Development FMAWRRD (1988).Agricultural Policy for Nigeria, Abuja: FMAWRRD 141 Federal Ministry of Women Affairs (FMWA) 2004.Implementation of the Beijing Platform for Action and Commonwealth Plan of Action, April, 2004 Fuller, W.A. 1976.Introduction to Statistical Time Series.New York: Wiley international Fuller, F.H., J.C. Bashin and S. Roselle 2004.Urban demand for dairy products in China: evidence from new survey data. Working paper 04- WP 380, Centre for Agricultural and Rural Development, IOWA State University, Ames, USA. Gafar, J. 1980. Price responsiveness of tropical agricultural exports;a case study of Jamaica 1954-1972, The Developing Economies18(3): 288-297. Ghatak, S and J. Seale (Jr.) 2001a. Supply response risk and institutional change in Chinese agriculture. Journal of Development Studies37(5): 141-150. Gill and A. Gracia 2001.The demand for alcoholic beverages in Spain, Agricultural Economics, 26 (1): 71-83. Granger, C.W.J. 1969. Investigating causal relations by econometric models and Cross- Spectral methods.Econometrica 37(3): 424-438. Granger, C.W.J. and P. Newbold 1974.Spurious regressions in econometrics.Journal of Econometrics2: 11-120. Granger, C.W.J. and P. Newbold 1986.Forecasting Economic Time Series.2/e Academic Press. Green, R. and J.M. Alston 1991. Elasticities in AIDS Models: a clarification and extension. American Journal of Agricultural Economics 73: 874-75. Greene, William H. 1997.Econometric Analysis, 3rd edition, Prentice-Hall. Griliches, Z. 1960.Estimates of the aggregate US farm supply functions, Journal of Farm Economics 42 (2): 282-293. Guilkey, D.K. and P. Schmidt 1989.Extended Tabulations for Dickey-Fuller Tests. Economic Letters. 142 Gujarati D.N. and Sangheeta 2007.Basic Econometrics.New Dehli, New York: Tata McGraw-Hill, Hallam, D 2000. Analysing agricultural supply response in economies in transition, CIHEAM, option mediterranennes Hallam, D. and R. Zanoli 1993.Error correction models and agricultural supply response.European Review of Agricultural Economics 20: 151-156. Hansen, P.R. 2002.Generalised reduced rankregression. Economics Working Paper 2002- 02. Brown University, http://www.econ.brown.edu/fac/Peter_Hansen/ Papers/grrr.pdf. Heilig G.K. 1999.China food: Can China feed itself? HASA, Lasenburg Cd Rom Vers. 1 Houthakker, H.S. 1952. Compensated change in quantities and qualities consumed,Review of Economic Studies 19: 155-164. Imolehin, E. D. and A. C. Wada 2000. Meeting the rice production and consumption demandsof Nigeria with improved technologies, International Rice Commission Newsletter. FAO 49: 33-41. International Rice Research Institute (IRRI) 1991.World Rice Statistics.Pp. 34-42. International Rice Research Institute (IRRI) 1995.World Rice Statistics. Pp 34-42. International Rice Research Institute 2004.World Rice Statistics. International Rice Research Institute, Manilla, Phillipines International Rice Research Institute (IRRI) 2014.World Rice Statistics. Available at: http://beta.irri.org/solutions/index.php? 143 Jensen, K. 1995. Fluid milk purchase patterns in the south: effects of use of nutritional information and household characteristics.Journal of Agricultural and Applied Economics. 27(2): 644-657. Jimoh, W.O., A.O.S. Ayanwale, R.O. Kareem and T.A. Akionsho2010.Consumption preference for cultured, captured and frozen fish in Abeokuta metropolis,Nigerian Journal of Farm Management.5(2): 72-79. Johansen, S. 1988.Statistical analysis of cointegrating vectors.Journal of Economic Dynamics and Control, 12:231-254. Johansen, S. 1991.Estimation and hypothesis testing of cointegrating vectors in Gaussian Vector Autoregressive Models.Econometrica 59 (6): 1551- 1580. Johansen, S. 1992.Determination of cointegration rank in the presence of a linear trend, Oxford Bulletin of Economics and Statistics 54 (3): 383-397. Johansen, S. 1995a.Likelihood-based Inference in Cointegrated Vector Autoregressive Models.Oxford; Oxford University Press. Johansen, S. 1995b. Identifying restrictions of linearequations –with applications to simultaneousequations and cointegration.Journal of Econometrics 69: 111-132 Johansen, S. and K. Juselius 1990a. Some structural hypotheses in a multivariate cointegrationanalysis of the purchasing power parity and the uncovered interest parity for UK, Discussion Papers90/05. University of Copenhagen, Department of Economics. Johansen, S. and K. Juselius 1990b.Maximum likelihood estimation and inferences on cointegration—with applications to the demand for money.Oxford Bulletin of Economics and Statistics 52 (2): 169-210. Karfoot. R. 2010. Self -sufficiency in food, agriculture in the United Kingdom, Department of Environment, Food and Rural Affairs, United Kingdom 144 Kebbeh, M., S. Haefele, and S. O. Fagade 2003. Challenges and opportunities for improving irrigated rice productivity in Nigeria. Project report - The Nigerian Rice Economy in a Competitive World: Constraints, Opportunities and Strategic Choices. Abidjan: WARDA. ii-24pp. Koc A. and S. Alpay 1988. Household demand in Turkey; an application of Almost Ideal Demand System with spatial cost index, SESRTCIC Working paper 0226 Koc, A.A.1998. Acreage allocation model estimation and policy evaluations for crops in Turkey.Technical Report 98 – TR42 November. Kormawa P.M., V.O. Okorowa, A. O. Adejobi and A.O.S Ayanwale 2004.Profitability and determinants of fertilizer usage by crop farmers in the drier savannah zone of Nigeria.Nigerian Journal of Science 38 (1):145-153. Lluch, C., A. Powell, R. Williams 1977.Patterns in household demand and savings.Oxford: Oxford University Press. Long, J.S. and J. Freese 2006.Regression models for categorical and limited dependent variables using Stata. 2. Ed. Collegestation, TX: stata Lopez, R.A. and H.H. Ramos 1998.Supply response and demand for basic grains in El- Salvador.Agribusiness 14(6): 475-481. McKay A., O. Morrissey and C. Vaillant 1998.Aggregate export and food cropsupply response in Tanzania.DFID-TERP: Credit Discussion Paper 98/4 (CDP04),Centre for Research in Economic Development and International Trade,University of Nottingham. McKay A, Morrisey O, and Vaillant C. 1999. Aggregate supply response in Tanzanian agriculture,The Journal of International Trade and Economic Development8(1):107- 123. 145 Miller, D. 2002. Crop prices and supply response, Economic and Agricultural issues. #5 Momoh, S. 2007.Nigeria faces rice shortage as world supply dwindles. Available at: http://www.tradenet.biz//groups.Accessed 05/03/2009 Moschini, G. 1998.The Semiflexible Almost Ideal Demand System, European Economic Review.42: 349-364. Moschini, G. and A. Vissa 1992.A linear inverse Demand System.Journal of Agricultural Resource Economics. 17: 294-302. Muchapondwa E. 2008. Estimation of the aggregate agricultural supply response in Zimbabwe: The ARDL approach to cointegration, Working Paper Number 90, School of Economics, University of Cape Town, August 20, 2008 NAMIS 2004. Nigeria Agricultural Marketing News Bulletin No. 4.Available at: www.afmin.net. Accessed 06/12/2008. Narayan K. P. 2005. The saving and investment nexus for China: evidence from cointegration tests. Applied Economics 37: 1979-1990 National Bureau of Statistics (NBS) 2004.A report on expenditure on food items, National Living Standard Survey (NLSS) National Bureau of Statistics (NBS) 2004.Nigeria Living Standard Survey, NLSS.Available at: www.nigerianstat.gov.ng National Bureau of Statistics (NBS) 2009.Agric. data.Statistical Bulletin. Available at: www.nigerianstat.gov.ng National Bureau of Statistics (NBS) 2010.Area and Climate.Statistical Buletin.Available at: www.nigerianstat.gov.ng 146 National Bureau of Statistics (NBS) 2012a.The review of Nigerian economy.Available at: www.nigerianstat.gov.ng National Bureau of statistics (NBS) 2012b.Consumption pattern in Nigeria 2009/2010 Preliminary Report, Available at: www.nigrianstat.gov.ng National Food Reserve Agency (NFRA) 2008.Report of 2007 Agricultural Production Survey(APS), Federal Ministry of Agriculture and Water Resources National Population Commission (NPC) 2013.Facts and Figures.www.population.gov.ng Nerlove 1956.Estimates of supply of selected agricultural commodities, Journal of Farm Economics38: 496-509. Nerlove, M. 1958. The dynamics of supply: estimation of farmer‘s response to price. Baltimore, Johns Hopkins Press. Nerlove, M. and K. L. Bachman 1960.The analysis of changes in agriculture supply problems approaches.Journal of Farm Economics 42: 531-554. Nickell S. (1985). Error correction, partial adjustment, and all that: an expository note. Oxford Bulletin of Economics and Statistics 47(2): 119-29. Nwachukwu, I.N., N.M. Agwu, and C.I. Ezeh 2008.Comparative study of consumer purchase attitude of local and foreign rice in Abia State, A paper presented at the nd 42 Annual Conference of the Agricultural Society of Nigeria at Ebonyi State th rd University, Abakaliki, 19 -23 October.Pp. 764-767 Odusina, O.A. 2008. Urban rice demand analysis: A case Study of Ijebu Ode township. Middle-East Journal of Scientific Research 3 (2): 62–66. Odusola, A.F. 1997. Poverty in Nigeria,an electric appraisal, NES 1998 Annual Conference Proceeding 147 Ogundele, F. 2007.Trade liberalisation and import demand for rice in Nigeria: A dynamicmodeling.Journal of Rural Economic and Development. 16(1): 35-46 Oguntona, T. and V.O. Akinyosoye 1986. Manifestation and cause of the Nigerian food crisis.Quarterly Journal of International Agriculture. 25(5): 258-270 Okoruwa, V.O., O.O. Ogundele 2006.Technical efficiency differentials in rice production technologies in Nigeria. Available at: http.//www.csae.ox.ac.uk/conferences/2006- E01-RPI/papers/case/okoruwa.pdf Okoruwa, V.O., O.O. Ogundele and S.O. Oyewusi 2006.Efficiency and productivity of farmers in Nigeria: A study of rice farmers in North Central Nigeria. Poster Paper Prepared for Presentation at the International Association of Agricultural Economists Conference, Gold Coast, Australia, August 12-18, 2006. Okoruwa, V.O., B.C. Onwurah and J.O. Saka 2008. Food demand among HIV Households in North CentralNigeria.European Journal of Science 5(4): 91-98. Olayemi, J.K. 2004a.Elements of applied econometrics.Ibadan: El Shaddai Global Ventures Olayemi J.K. 2004b.Principle of economics for applied economics analysis, Ibadan: Fayoson Printers.383pp. Omotola and Ikechukwu 2006. Rice milling in Nigeria, A Report on Development of Agriculture in sub-Saharan Africa. Available at: www.ricenigeria.com Oyejide, T.A. 1990. Supply response in the context of Structural Adjustment in sub-Saharan th Africa, AERC Special paper 1, Agricultural policies proceedings of the 12 Annual Symposium of the International Association of research Scholars and fellows held at IITA, Ibadan. Pp. 31 Oyinbo O., R.A. Omolehin and Z. Abdulsalam 2013. Household consumption preference for imported and domestic rice in Kaduna, Nigeria: implication for rice quality improvement.PAT9(1): 29-37. 148 Pesaran, M. H., Shin Y., and Smith J. R. 2001. Bounds testing approaches to the analysis of levelrelationships, Journal of Applied Econometrics 16(3): 289-326. Philippines News Agency 2009.Nigeria: rice demandpushes upglobal price.Available at: All Africa.com.Accessed 05/05/2009 Rahji, M.A.Y. 1999.Dimension and rural poverty and the food self-sufficiency gap in Nigeria. In: poverty alleviation and food security in Nigeria, Fabiyi, Y.L. and E.O. Idowu Eds. Pp33-37. Rahji, M.A.Y., O.O. Ilemobayo and S.B. Fakayode 2008.Rice supply response in Nigeria; an application of the Nerlovian Model, Medwell Agricultural Journal 3(3):229-234. Rahji, M.A.Y. and Adewumi, M.O. (2008).Market supply response and demand for local rice in Nigeria: implications for self-sufficiency.Central European Agriculture Journal. 9(3): 567-574. Raji A. (2013). Making Agricultural Transformation Agenda a reality, Proceeding of the th 47 Annual Conference of the Agricultural Society of Nigeria held at Federal College of Animal Health and Production Technology, Moore Plantation Ibadan, 5-8 November 2013.Daily Independent Wednesday, February 19, 2014 Rice Farmers Association of Nigeria (RIFAN) 2006.Rice Production in Nigeria. Conference Paper presented by the Secretary Rice Farmers Association of Nigeria th Saka I. and Kolawole, Y. 2008.Rice importation dangerous,This Day, 12 May, 2008, Available at: All Africa.com. Accessed 05/05/2009 Sams N. 2010. Nigeria spends N106B to import rice from Thailand, Available at www.allAfrican.com Sanni L. O. 2000. Agricultural development without post-harvest system: any hope for success?, University of Agriculture, Abeokuta Alumni Association Lecture Series No. 2.,23pp. 149 Sanusi J. O. 2003.Special Remarks delivered at the Seminar on Sustainable Rice Production in Nigeria, held at the Hamdala Hotel, Kaduna, 14th & 15th, January 2003. Sanusi, L. 2011.Nigeria Import Burden, The Punch,Wednesday, July 6, 2011, Pp.20. Available at: www.punch.com Schiff, M. and E.C. Montenegro 1997: Aggregate agriculture supply response in developingCountries: a survey of selected issues. Economic Development and Cultural Change 45(2) available at: http//www.hartford-hwp.com/archives/55a/index-dcd.html Schvedler, W. 2005.Likelihood Estimation for Censored Random Vectors.Econometric Review24(2): 195-217. Seale, J., A. Regni and J. Bernstein 2003.International evidence of food consumption patterns, USDA Technical Bulletin No. 1904 Available at: http//www.ers.usdagov/data/International food demand Accessed December, 2006 Shimbun, M. 2004.Food self-sufficiency. Editorial, Mainichi Newspaper Co. Stone, J. 1954. Linear expenditure systems and demand analysis—an application to the pattern of Britishdemand.Economic Journal64: 511-527. Strauss, J. 1982.Determinants of food consumption in rural Sierra Leone: application of Quadratic Expenditure System to the Household Firm Model.Journal of Development Economics. II. Pp327-353. Theil, H. (1965).The information approach to demand analysis.Econometrica 33: 67-87. Thiele, R. 2000. Estimating aggregate agriculture supply response: a survey of techniques and results for developing countries, Keil Working Paper No. 1016, Keil Institute of World Economics, Germany 150 Tiamiyu 2011.Efficiency and technology use among growers of NERICA rice varieties in the Savanna Zone of Nigeria, Unpublished Ph.D Thesis, Department of Agricultural Economics, University of Ibadan, Ibadan Tijani, A.A. 2006. Analysis of technical efficiency of rice farms in Ijesha Land of Osun State, Nigeria.Agrekon 45(2): 126-135. Tijani, A.A., J.O. Ajetomobi, O. Ajobo 1999. A cointegration analysis of Nigeria cocoa exportsupply.Journal of Rural Economics13(1): 45-56. Tobin, J. 1958.Estimation of Relationships for Limited Department Variables.Econometrica 26(1): 24-36. Townsend, R. 1996. Price liberalization, technology and food self-sufficiency: an analysis of summer grains in South Africa‘, mimeo. Townsend, R. and C. Thirtle 1995.Dynamic acreage response: an Error Correction Model for maize and tobacco in Zimbabwe, University of Reading, Discussion Papers in Development Economics, Series G, 2(20). Trade Invest Nigeria (TIN) 2009.Nigeria's rice import burden predicted to swell TIN, Wednesday, 26 August, 2009 United Nations Environment Programme(UNEP) 2005.Integrated assessment of the impact oftrade liberalization; a country study of the Nigerian rice sector. Genera Switzerland USDA/FAS, 2001a.Attaché reports. http://www.fas.usda.gov/ainfiles/200112/135682904.pdf USDA/FAS, 2001b.Attaché reports. http://www.fas.usda.gov/gainfiles/200104/80680426.pdf USDA/FAS, 2003. Nigeria product brief, rice 2003 GRAIN Report, USDA/FAS, Washington. 151 Vogelvang. B. 2005.Econometrics: Theory and application with EViews, FT Prentice-Hall Publishers Workman, D. 2008.Rice import dependent countries. Available at: http://internationaltradecommodities.suite101.com/article.cfm/rice_import_dependent countries.Accessed January, 2010 World Bank 1996. Poverty in the midst of plenty: the challenge of growth with inclusion, Washington DC. World Bank 2008.World Report Agriculture for Development. Available at: http://website/external/topics/exttpoverty Yunus, M., (1993).Farmers response to price in Bangladesh, The Bangladesh developments Studies21(3): 101-109 152 Appendix 1: Nigerian Rice Production Systems Production Major States Covered Estimated share of Ave. Yield System national rice area (ton/ha) Rainfed Ogun, Ondo, Osun, Ekiti, Oyo, Edo, 30% 1.7 Upland Delta, Niger, Kwara, Kogi, Sokoto, Kebbi, Kaduna and Benue states Rainfed Ondo, Ekiti, Delta, Edo, Rivers, Bayelsa, 47% 2.2 Lowland Cross River, Akwa Ibom, Lagos, all major river valleys, e.g., shallow swamps of Niger basin, Kaduna basin and inland swamps of Abakaliki and Ogoja areas Irrigated Niger, Sokoto, Kebbi, Borno, Benue, 16% 3.5 Kogi, Anambra, Enugu, Ebonyi and Cross River states Deep water Flooded areas of Rima valley-Kebbi state 5% 1.3 /Floating and deep flood areas of Ilushi, Delta state Mangrove Ondo, Ekiti, Delta, Edo, Rivers, Bayelsa, 1% 2.0 Swamp Cross River, Akwa Ibom Lagos Source: Akpokodje et al. (2001) 153 153 Appendix 2: Analysis of Objectives OBJECTIVE FOCUS DATA REQUIREMENTS ANALYTICAL TOOLS 1. To describe the This objective described Data on consumption Descriptive Statistics: expenditure pattern the Expenditure on expenditures on rice, other frequency tables, mean of rice in Nigeria various types of rice and food commodities and median, mode, standard also related it to other socioeconomic characteristics deviation, percentages, food commodities and skewness and kurtosis. socioeconomic characteristics 2. To examine the self- In this objective we Data on production, Descriptive Statistics: sufficiency ratio of examine the ratio of consumption and import frequency tables, mean, rice in Nigeria domestic rice supply to quantities of rice in Nigeria median, mode, standard demand in Nigeria in deviation, percentages. relation to the quantity imported annually 3. To estimate a Here, the determinants Data on household Tobit Regression Model, demand model for of rice demand, their expenditure on various rice Almost Ideal Demand rice in Nigeria signs, magnitude and commodities, Price, income System (AIDS) corresponding and other socioeconomic elasticities were variables, such as age, estimated household size, education, etc. 4. To analyse the Here, the response of Data on rice output, area and Cointegration and Error supply response of rice output to price and prices, weather variable, Correction Model (Vector rice in Nigeria non-price factors were fertilizer consumption etc. autoregressive procedure) tested,the elasticities and coefficientof adjustment determined in the short run and long run. 5. To isolate the This objective analysed Data on expenditure share on Paired sample t-test and determinants of the direction and local, agric. and imported rice; Generalized least square preference switch determinants of socioeconomic variables: age, (GLS) multiple regression from foreign to local preference switch income, education, price etc. rice or vice versa between various pairs of rice commodities. Source: Author‘s Compilation Appendix 3: Taxonomy of Trade Policy on Rice in Nigeria 154 PERIOD POLICY MEASURES Before april 1974 60% tariff April 1974-April 1975 20% April 1975- April 1978 10% April1978 – June 1978 20% June 1978 –October 1978 19% October 1978 – April 1979 Import in container under 50kg were banned April 1979 Import under restricted license only Government agencies September 1979 6 months ban on all rice imports January 1980 Import license issued for 200,000 tonnes of rice October 1980 Rice under general import license with no quantitative restrictions December 1980 Presidential task force (PTF) on rice was created and it used the Nigeria National Supply Company to issue allocations to customers and traders May 1982 PTF commenced issuing of allocations directly to customers and traders in addition to those issued by NNSC January 1984 PTF disbanded, rice importation placed under general license restriction October 1985 Importation of rice banned. July 1986 Introduction of SAP and the abolition of Commodity Boards to provide production incentives to farmers through increased producer prices 1995 100% 1996 50% 1998 50% 1999 50% 2000 50% 2001 85% Source: Akpokodje et al. (2001); Akande et al.(2007) Appendix 4: NBS Rice Production Statistics 155 a. Rice Area, Output and Yield (2002-2007) 2002/03 2003/04 2004/05 2005/06 2006/07 AREA(„000Ha) 1,361.17 1,389.13 1,454.57 1,590.37 1,526.00 OUTPUT(„000MT) 2,737.61 2,980.57 3,183.39 3,247.52 3,333.00 YIELD (Kg / ha) 2,011.2 2,145.7 2,188.5 2,042.0 2,184.0 Source: National Bureau of Statistics (NBS) (2010) b. Rice Output (2006-2010) 2006 2007 2008 2009 2010 OUTPUT(„000MT) 3333.00 3561.55 3369.70 3402.59 4468.04 Source: National Bureau of Statistics (2012b) Appendix 5: Trend in Rice Production, Consumption and Import in Nigeria (1995 - 2011) 156 Year Productn Area Yield Consumption Import („000) („000ha) (tonne/ha) 1995 2920 1796 1.63 2249 300 1996 3122 1784 1.75 2429 346 1997 3268 2048 1.60 2880 679 1998 3275 2044 1.60 2781 594 1999 3277 2191 1.50 3000 813 2000 3298 2199 1.50 2994 786 2001 2752 2117 1.30 3402 1710 2002 2928 2185 1.34 3146 1236 2003 3116 2210 1.41 3284 1601 2004 3334 2348 1.42 3251 1398 2005 3567 2494 1.46 3182 1188 2006 4042 2725 1.65 3372 976 2007 3186 3000 1.30 3601 1217 2008 4179 2382 1.75 3324 972 2009 3403 1837 1.93 3545 1164 2010 3219 2433 1.84 NA 1885 2011 3045 2580 1.77 NA NA Source: IRRI; FAOSTAT World Rice Statistics (2014) Available at: http://www.beta.irri.org/statistics Appendix 6: Estimates of Aggregate Agricultural Supply Response 157 Studies Country Supply Response Short-run Long-run Griliches (1959) N US 0.28 - 0.30 1.20 - 1.32 Griliches (1960)G US 0.10 - 0.20 0.15 Tweeten and Quance (1968)G US 0.25 1.79 Rayner (1970)G UK 0.34 0.42 Pandey et al (1982)N, G Australia 0.30 0.6 - 1.00 Reca (1980)N Argentina 0.21 - 0.35 0.42 - 0.78 Bapna (1980)N India (Ajmer) 0.24 na Krishna (1982)N India 0.20 - 0.30 na Chhibber (1989) N India 0.20 - 0.30 0.40 - 0.50 Bond (1983)N: Ghana 0.20 0.34 Kenya 0.10 0.16 Côte d‘Ivoire 0.13 0.13 Liberia 0.10 0.11 Madagascar 0.10 0.14 Senegal 0.54 0.54 Tanzania 0.15 0.15 Uganda 0.05 0.07 Burkina Faso 0.22 0.24 SSA (ave.) 0.18 0.21 Those studies indicated by N use the Nerlove Model, those indicated by G use the Griliches approach. Source: McKay et al. (1999) Appendix 7: Description of Household Expenditure on Rice Commodities Zone/ Rice type Mean S. D Skewness Kurtosis North Central 158 Imported Rice (IR) 1372.255 869.648 0.621 -0.887 Agric. Rice (IDR) 852.455 408.138 -0.553 -1.243 Local Rice (LR) 1209.948 1273.898 1.913 1.752 Total Rice (TR) 3434.657 2273.684 1.454 0.860 Share of IR 0.4000 0.122 40.149 -0.985 Share of IDR 0.264 0.093 40.272 -1.463 Share of LR 0.336 0.143 0.157 -1.179 Share of TR 0.253 0.168 0.888 0.241 North East Imported Rice (IR) 1044.116 358.860 -0.654 -1.279 Agric. Rice (IDR) 724.029 204.561 0.855 -0.109 Local Rice (LR) 549.054 63.118 0.767 -0.621 Total Rice (TR) 2317.199 456.163 -0.453 -0.377 Share of IR 0.437 0.095 -0.404 -1.520 Share of IDR 0.312 0.048 -0.188 -1.244 Share of LR 0.251 0.076 0.602 -1.055 Share of TR 0.249 0.167 0.917 0.409 North West Imported Rice (IR) 2006.004 1132.540 0.597 -1.296 Agric. Rice (IDR) 1299.087 805.944 1.670 1.755 Local Rice (LR) 738.293 135.475 -0.772 -0.556 Total Rice (TR) 4043.383 1740.027 0.889 -0.506 Share of IR 0.478 0.124 0.076 -1.254 Share of IDR 0.321 0.107 -0.008 -1.660 Share of LR 0.202 0.048 -0.811 -1.084 Share of TR 0.440 0.575 6.211 63.223 South-South Imported Rice (IR) 700.629 150.365 0.976 -0.606 Agric. Rice (IDR) 531.982 123.980 -0.003 -1.062 Local Rice (LR) 435.299 58.271 -0.020 -1.614 Total Rice (TR) 1667.909 275.274 -0.720 -1.357 Share of IR 0.420 0.047 -0.466 -1.166 Share of IDR 0.317 0.039 0.831 -0.182 Share of LR 0.264 0.025 -0.102 -1.212 Share of TR 0.146 0.123 1.799 4.088 South East Imported Rice (IR) 1327.744 631.793 0.621 -1.320 Agric. Rice (IDR) 633.117 80.114 -0.213 -1.675 Local Rice (LR) 499.620 133.781 0.635 -1.164 Total Rice (TR) 2460.480 771.019 0.721 -1.099 Share of IR 0.515 0.090 0.124 -1.698 Share of IDR 0.275 0.063 -0.018 -1.058 Share of LR 0.211 0.050 1.051 -0.120 Share of TR 0.208 0.161 1.273 1.229 South-West Imported Rice (IR) 860.821 274.675 1.621 1.329 Agric. Rice (IDR) 580.111 97.519 1.157 0.137 Local Rice (LR) 413.253 119.192 0.541 -1.453 Total Rice (TR) 1863.185 375.772 1.091 0.376 Share of IR 0.460 0.054 0.798 -0.635 Share of IDR 0.314 0.028 -0.432 -0.861 Share of LR 0.226 0.061 0.101 -1.254 Share of TR 0.194 0.140 1.258 1.631 Source: Computed from NLSS (2004) Appendix 8: Distribution of Respondents by Share of Rice in Total Food Expenditure (Zones) Expenditure Frequency Percent Valid Percent North Central 159 1-10 359 10.8 20.6 11-20 494 14.8 28.3 21-30 342 10.3 19.6 31-40 236 7.1 13.5 41-50 140 4.2 8.0 51-60 109 3.3 6.3 61-70 64 1.9 3.7 Total 1744 52.4 100.0 Missing value 1587 47.6 Total 3331 100.0 North East 1-10 548 17.1 21.5 11-20 689 21.5 27.0 21-30 505 15.8 19.8 31-40 368 11.5 14.4 41-50 216 6.7 8.5 51-60 117 3.7 4.6 61-70 106 3.3 4.2 Total 2549 79.6 100.0 Missing value 653 20.4 Total 3202 100.0 North West 1-10 338 8.9 13.3 11-20 489 12.9 19.3 21-30 552 14.5 21.7 31-40 404 10.6 15.9 41-50 311 8.2 12.2 51-60 267 7.0 10.5 61-70 179 4.7 7.0 Total 2540 66.8 100.0 Missing value 1260 33.2 Total 3800 100.0 South-South 1-10 1246 43.7 48.9 11-20 752 26.3 29.5 21-30 293 10.3 11.5 31-40 136 4.8 5.3 41-50 68 2.4 2.7 51-60 37 1.3 1.5 61-70 16 0.6 0.6 Total 2548 89.3 100.0 Missing value 306 10.7 Total 2854 100.0 South-East 1-10 704 26.3 32.4 11-20 660 24.6 30.4 21-30 338 12.6 15.6 31-40 194 7.2 8.9 41-50 135 5.0 6.2 51-60 82 3.1 3.8 61-70 58 2.2 2.7 Total 2171 81.0 100.0 Missing value 510 19.0 Total 2681 100.0 South-West 1-10 731 24.4 31.8 11-20 726 24.3 31.6 21-30 428 14.3 18.6 31-40 233 7.8 10.1 41-50 93 3.1 4.0 51-60 59 2.0 2.6 61-70 27 0.9 1.2 Total 2297 76.7 100.0 Missing value 696 23.3 Total 2993 100.0 Source: Computed from NLSS (2004) Appendix 9:Nigeria Rice Self- Sufficiency Ratio Year Consumption Production Import Self-sufficiency Ratio 1960 239 240 1 0.996 160 1961 229 230 1 0.996 1962 246 247 2 0.996 1963 202 204 1 0.990 1964 269 270 1 0.996 1965 236 232 1 0.996 1966 270 271 1 0.996 1967 260 261 1 0.996 1968 249 250 1 0.996 1969 257 258 1 0.996 1970 284 285 6 0.997 1971 307 313 11 0.981 1972 310 321 2 0.966 1973 342 344 4 0.994 1974 348 352 6 0.989 1975 390 396 94 0.985 1976 406 500 446 0.812 1977 412 750 789 0.549 1978 394 950 242 0.415 1979 372 854 394 0.440 1980 523 850 686 0.615 1981 579 1227 666 0.472 1982 648 1337 903 0.485 1983 607 1648 629 0.368 1984 579 1220 569 0.474 1985 680 1249 462 0.544 1986 630 1042 642 0.605 1987 1184 1152 344 1.028 1988 1249 1350 164 0.925 1989 1982 1550 224 1.279 1990 1500 2757 290 0.544 1991 1911 2207 440 0.866 1992 1956 2436 382 0.803 1993 1839 2221 300 0.828 1994 1456 2136 300 0.682 1995 1752 2200 350 0.674 1996 1873 2175 731 0.861 1997 1961 2712 900 0.723 1998 1965 2815 850 0.698 1999 1966 2866 1250 0.686 2000 1979 3029 1906 0.653 2001 1651 3051 1897 0.541 2002 1757 3307 1448 0.531 2003 1870 3670 1369 0.510 2004 2000 3750 1777 0.533 2005 2140 3800 1600 0.563 2006 2546 4040 1600 0.630 2007 2008 4000 1550 0.502 2008 2632 4200 1800 0.627 2009 2730 4580 2000 0.596 2010 2615 5030 2000 0.520 2011 2709 5200 2550 0.521 2012 2850 5950 2700 0.479 Computed from IRRI (2014); USDA Rice Statistics Data Available at: http://www.beta.irri.org/statistics Appendix 10: Farm Harvest Price of Rice in Nigeria (1960-2008) 161 Farm Harvest Price of Rice(N/ton) 120000 100000 80000 60000 Farm Harvest Price of 40000 Rice(N/ton) 20000 0 Source: Computed from IRRI Statistics (2011) Appendix 11: Fertilizer Consumption in Nigeria (1960-2008) 162 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 fertilizer consumption(tons) 500 450 400 350 300 250 200 fertilizer 150 consumption(tons) 100 50 0 Source: Computed from IRRI; FAO Rice Statistics (2011) Appendix 12: Mean Annual Rainfall in Nigeria (1960-2008) 163 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 Mean Annual Rainfall(mm) 1600 1400 1200 1000 800 Mean Annual 600 Rainfal(mm) 400 200 0 Source: Computed from IITA Rainfall Data (2010) Appendix 13: Determinants of Preference Switch B 164 1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 Variables Agr-Imp Loc-Imp Loc-Agr HHSIZ 6.585*** 5.161** 11.746*** (2.071) (2.150) (2.149) AGE -0.286 -0.687** -0.973*** (0.276) (0.287) (0.287) EDUC -1.056* 1.623** 2.678*** (0.604) (0.627) (0.6267) PCEXP 0.922e-03*** 0.287e-03** -0.635e-03*** (0.130e-03) (0.135e-03) (0.135e-03) MARST 30.560** 14.227 -44.787*** (12.259) (12.727) (12.718) PROCC -130.425*** 93.622*** -36.802*** (10.863) (11.278) (11.270) TASSET 0.987e-06 0.213e-05 0.114e-05 (0.169e-05) (0.175e-05)(0.175e-05) COMEM -35.916 6.409 42.325 (7.687) (7.981) (7.975) SECTOR -55.128*** -70.194*** -15.067 (10.186) (10.575) (10.568) SS 1605.268*** 612.241*** -993.027*** (40.275) (41.814) (41.784) SE -772.490*** -1081.11*** -308.624*** (23.002) (23.881) (23.863) SW 263.829*** -183.316*** -447.145*** (25.642) (26.623) (26.603) NE 576.549*** -314.775*** -891.324 (18.503) (19.210) (19.196) NW -69.910*** -1293.064*** -1223.153 (19.344) (20.084) (20.069) PRIMP -30.680*** -1.972** (0.940) (0.976) PRAGR 19.385** -2.932 (0.533) (0.553) PRLOC - 16.309*** 0.546 (0.733) (0.732) CONST. -3920.572*** 1650.722*** -2269.85 (115.872) (120.301) 120.214 R2 0.380 0.540 Adj R2 0.379 0.540 F (30, 18830) 385.32*** 737.47*** ***Values significant at 1%; **Values significant at 5%, *Values significant at 10% Note: Agr-Imp means switch from agric. to imported rice Loc-Imp means switch from local to imported rice Loc-Agr means switch from local to agric. rice Source: Computed from NLSS Data (2004) 165