The results found in this study have implications for local Nigerian food producers, retailers, other participants of the food sector, and government food policy makers.
In this thesis the demand analysis for onion, peppers, fresh okra and tomato in Nigeria was conducted using General Household Survey data collected by the World Bank and the Nigeria National Bureau of Statistics. The two stage estimation procedure and Linear Approximation Almost Ideal Demand System addressing censoring were used to analyze the demand system. The analyses are based on the assumption that every household is maximizing its utility subject to a budget constraint. Standard errors on both stages of the estimation as well as for the calculated elasticities were adjusted using a bootstrap procedure.
Most of the demographic characteristics determining consumption were significant. Marshallian cross price elasticities suggest that the products are a mix of gross substitutes and complements, whereas positive values of Hicksian cross-price elasticities indicate that all vegetables are net substitutes. According to expenditure elasticities, not all of the vegetables appear to be normal goods. Negative expenditure elasticity for fresh okra indicates that the vegetable is an inferior good.
A combination of policies that increase purchasing power of population, and fosters food supply would benefit a developing country, like Nigeria, the most. Increased supply would trigger an increase in quantity demanded, improving the livelihood of agricultural producers, poor households and potentially creating more jobs in agricultural and related industries.
TABLE OF CONTENTS
CHAPTER 1: INTRODUCTION - 1
1.1. Problem Identification and Explanation - 1
1.1.1. Research Objectives - 10
CHAPTER 2: LITERATURE REVIEW - 11
2.1. Demand analysis - 11
CHAPTER 3: CONCEPTUAL FRAMEWORK - 23
3.1 Utility maximization - 23
3.1.1. Separability - 25
3.1.2. Homogeneity - 27
3.1.3. Slutsky symmetry - 28
3.1.4. Adding up restriction - 29
3.2. Empirical specification - 30
CHAPTER 4: METHODS AND PROCEDURES - 32
4.1. Data - 32
4.1.1. Data cleaning and modification - 33
4.2. Sample selection bias. Two-stage estimation procedure - 41
4.2.1. First stage: Probit regressions - 46
4.2.2. Multicollinearity - 49
4.2.3. Second stage: Iterated Seemingly Unrelated Regression (ITSUR) - 49
4.3. Model - 51
4.4. Bootstrap estimation of the standard errors - 56
4.5. Demand Elasticities for the censored LA/AIDS model - 58
CHAPTER 5: RESULTS - 62
5.1. Sample structure - 62
5.2. Multicollinearity Diagnostics - 65
5.3. Results of Probit Regressions - 65
5.4. Marginal effects - 67
5.4.1. Onion - 67
5.4.2. Peppers - 68
5.4.3. Fresh Okra - 69
5.4.4. Tomato - 70
5.5. LA/AIDS model - 71
5.6. Elasticities - 71
5.6.1. Marshallian Own and Cross Price Elasticities - 72
5.6.2. Expenditure Elasticities - 73
5.6.3. Hicksian Own and Cross Price Elasticities - 73
CHAPTER 6: CONCLUSIONS - 75
REFERENCES - 78
APPENDICES - 109
APPENDIX A: How to attain copies of the data - 110
APPENDIX B: The GHS Sample Design - 111
APPENDIX C: Tables 1 – 17 - 112
APPENDIX D: List of figures - 132
VITA - 134
LIST OF TABLES
Table - Page
Table 1. Units of Measure………………………………………………………………………113
Table 2. Correspondence of certain units of measurement to the commodity items…………... 113
Table 3. Kcal consumption…………………………………………………………………….. 114
Table 4. Components of the Wellbeing Index…………………………………………………. 115
Table 5. Definitions of the Wellbeing Index components …...………………………………... 117
Table 6. Calculating the index for the household 10001………………………………………. 117
Table 7. Definitions of the dependent and explanatory variables related to the models………. 118
Table 8. Distribution of the final sample data by Geographic Location………………………..120
Table 9. Sample statistics of expenditures, quantities, prices and expenditure shares, N = 3033 households………………………………………………………………………………………122
Table 10. Diagnostics of Multicollinearity in first stage: Probit models………………………. 123
Table 11. Diagnostics of Multicollinearity in second stage: SUR……………………………... 124
Table 12. Maximum Likelihood Estimates of Probit Models…………………………………..125
Table 13. The estimated parameters of the LA/AIDS model………………………………….. 128
Table 14. The results of Likelihood Ratio test………………………………………………….128
Table 15. Uncompensated (Marshallian) Price and Expenditure Elasticities………………….. 129
Table 16. Compensated (Hicksian) Price Elasticities………………………………………….. 130
Table 17. Marginal effects at variables means. Results from Probit models…………………...131
LIST OF FIGURES
Figure - Page
Figure 1. Graphical representation of per capita quantities consumed of different products in different geographic regions of Nigeria. - 133