CHAPTER ONE INTRODUCTION
Background
In developing countries, detailed knowledge of the relationship between expenditure shares and total household expenditure plays a significant role in household welfare analysis. Food and nutritional programs, poverty reduction strategies as well as tax reforms, all require information on how expenditure share respond to changes in household total expenditure for effective formulation and implementation. Knowledge of how expenditure shares respond to changes in household total expenditure, however, is often ascertained through Engel curve analysis. Engel curves as noted in Girma and Kedir (2007) describe the relationship between household expenditure on particular items and household total expenditure. The most studied Engel curve is food Engel curve which describes how food expenditure increases as total expenditure increases. Such studies often show that household food consumption increases less than proportionally with income, and food expenditure as a share of total expenditure decreases as income increases (Girma and Kedir, 2007; Banks et al., 1997; Deaton and Muellbauer, 1980; Engel, 1857). This relationship is known as the Engel’s law.
For a developing country such as Ghana, knowledge of food Engel curve is important for a number of reasons. First, food Engel curve is important in welfare analysis. For poorer household food expenditure share is high but it decreases as the household income increases. Therefore, high food expenditure share can be used as indicator of poverty levels while decreases in food share can be associated with rising incomes. Second, since food Engel curve shows the relationship between food share and income, it can be used to predict how food
expenditure will change as income changes. These predictions can be used to simulate the impact of tax policies on consumers as well as the local economy. The food Engel curve reveals important information about aspects of consumer behavior which has relevant policy implications (Deaton and Muelbauer 1980). For instance, food Engel curve may indicate proportionately the burden of tax imposed on food items consumed by the poor. Third, aside from it (food Engel curve) showing how food share responds to changes in income, the food Engel curve also reveals the income elasticity of food and thus, shows if food is a normal good, inferior good or luxury good.
Despite the above benefits, estimating food Engel curves (particularly in developing countries) is often fraught with a number of challenges. One of such is the appropriate functional specification to assume for the relationship between food expenditure share and total income, for instance whether to use a linear or non-linear specification. Earlier studies such as Deaton and Muellbauer (1980) and Leser (1963) used the Working-Leser specification in which expenditure shares are assumed to be linear functions of total income. However, studies such as Girma and Kedir (2007), Banks et al (1997), Hausman et al. (1995) and Gozalo (1997) have shown that food expenditure share can be non-linearly related to total income. Given the benefits of food Engel curves, it is important that the right specification is used in food Engel curves analysis hence the objective of this thesis is to estimate both linear and non-linear food Engel curve for Ghana in order to ascertain the right specification that best fits for Ghana. It uses four rounds of the Ghana Living Standard Survey (GLSS 3, 4, 5 and 6) and compares the relationship across survey periods.
Another challenge with food Engel curves analysis is that household expenditure captured in most expenditure surveys is often associated with measurement error. This is because
expenditure on an item captured in an expenditure survey consists of both cash expenditure and own consumption. While cash expenditure is already monetized, own consumption is often imputed using different set of prices. This imputation in the calculation of expenditures adds an additional layer of error to standard reporting errors made by respondents and interviewers. This type of error has serious consequences in the estimation of food Engel curves particularly when estimation employs share form equations (Lewbel 1996). To correct for this measurement error, the thesis uses the augmented regression approach of Hausman (1978) and Blundell & Robin (1999).
Statement of problem
As noted earlier, food Engel curve is important tool in understanding household welfare analysis because they provide an insight into the relationship between food consumption and household total expenditure. However, getting the expenditure share of food and total expenditure relationship right is contingent on using the right specification in estimation. Different specifications have been employed in many studies, but the result for the linear and non-linear estimations have not been the same in most cases. Although a number of studies on the food Engel curve have been done for other African countries, studies on Ghana is, however, limited. The few studies on Ghana have assumed either linear relationship or non- linear relationship between expenditure share of food and total expenditure. This approach may lead to bias estimates particularly when the data does not support the assumed specification. As far as this study is concerned no study on Ghana has compared both the linear and non-linear specification to establish which best fit the Ghanaian data.
To fill this gap in the literature the study estimates the two most common functional specifications employed in the literature: the Working-Leser form popularised by Deaton and
Muellbauer (1980) and the quadratic version of this model introduced by Banks et al. (1997). This will afford us the opportunity to compare the two functional specifications and determine the appropriate functional specification for food Engel curve for Ghana. Such information will play a significant role in policy formulation. This study is unique in two ways. First, the study does not only estimate functional specification for food Engel curve, but it also compares food Engel curve across periods to ascertain whether Food Engel curves have been changing over time. As far as we are concern this is the first study in Ghana to have done that. Second, the study explicitly correct the measurement error associated with total expenditure; most studies are silent on it.
 Research Questions
In line with the above problem statement, the thesis seeks to answer three main research questions.
- Which of the two functional specifications of food Engel curve (that is, the Working- Leser model or the quadratic model) best fit the Ghana Living Standard Survey (GLSS) data?
- Is this functional specification the same for all survey periods? If not, what explains the differences in specification across survey periods?
- What has been the nature of food consumption patterns in Ghana between 1990 and 2013?
 Research Objectives
The study seeks to broadly determine the nature of food Engel curve for Ghana from 1990- 2013 using the GLSS. Specifically, the objectives of the study are:
- To determine the functional specification of food Engel curve for Ghana using the Working-Leser and the quadratic functional forms and determining the best functional form for Ghana using the last four rounds of the GLSS.
- To analyze food consumption patterns in Ghana using the last four rounds of the GLSS.
 Significance of the study
The study seeks to contribute to ongoing discussions on the relevance of Engel curve estimation by focusing on a country-specific study which is the case of Ghana. The determination of the appropriate functional form and the correct nature of the food Engel curve for Ghana will guide the formulation of policies to promote the welfare of the poor and also help in the attainment of the Sustainable Development Goals one and two of no poverty and zero hunger respectively.. It will enable welfare assessments of households which is an important tool for tax policy. Thus, the study will help to improve the design of public policies that aim to shield the poor from the adverse effects of income shocks.
Data
The study will employ the Ghana Living Standard Survey (GLSS) in its analysis. GLSS is a nationally representative household survey, conducted by the Ghana Statistical Service (GSS), which provides reliable, disaggregated and internationally comparable welfare and living conditions statistics in Ghana. The survey provides detailed information on demographic characteristics of the population, education, health, employment, migration, housing conditions, household incomes, expenditure and agriculture. The variables of focus for this study will be the expenditure on food by households and the total expenditure by these households. The study will employ the last four rounds of the GLSS i.e. GLSS 3, 4, 5 and 6.
Organization of the study
The study will be in six (6) chapters. Chapter one deals with the introduction of the study, focusing on the background of the study, the statement of the problem, objective of the study and significance of the study. Chapter two reviews both theoretical and empirical literature. Chapter three covers the overview and trends of food consumption and total consumption of households in Ghana. Chapter four presents the methodology used for the study. Chapter five involves the estimation and empirical analysis of the results whilst chapter six summarises, concludes and makes policy recommendations from the study.
