Effect of income on nutrient intake case of households in the southeast region of vietnam 2000

Therefore, this thesis will examine the effects of income change on calorie intake by using a large household-level survey on food consumption, incomes and other household characteristic

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES

Vietnamese-Dutch Project for M.A Program in

Nguyen Phuong Chi

In Partial Fulfillment of the Requirements for Obtaining the Degree of

MASTER OF ARTS IN ECONOMICS OF DEVELOPMENT

Supervisors

M.Sc Nguyen Hun Dung

Dr Le Bach Mai

THU V1L~l"~

ACKNOWLEDGEMENTS

First of all, I would like to express my acknowledgement to Dutch Lady Vietnam Food and Beverage Company for their sponsorship to my trip to the National Institution of Nutrition in Hanoi to finish this thesis I also convey my gratitude to the University of Economics - Ho Chi Minh City, the Institute of Social Studies -The Hague and the National Institution ofNutrition for creating such good chance for me to do the research

I am deeply indebted to my supervisors, Mr Kees Voogd, the Innovation Manager and Mr Nguyen Huu Thiet, the Human Resource Manager of Dutch Lady Vietnam Food and Beverage Company; Mr Nguyen Huu Dung, the coordinator of the MDE program; Dr Nguyen Cong Khan, the director of the National Institution of Nutrition, Dr Le Bach Mai, the manager of the Public Nutrition Department and

Dr Vu Quynh Hoa, staff of the National Institution of Nutrition for their valuable instructions, comments and suggestions during the process of writing the report

Finally, I wish to express my thanks to my parents, relatives and friends for their spiritual encouragement and material support Without those sources, the thesis would have been made impossible

I alone bear full responsibility for any errors remaining in the report

Nguyen Phuong Chi

Ho Chi Minh City, August 2003

ABSTRACT

It is a fact that 75 percent undernourished children lives in India, Bangladesh, Pakistan, Indonesia, Nigeria and Vietnam (Foster and Leathers, 1999) Still, developed countries like European and the United States give a lot of attention to this problem They increase incomes of the poor and food production per person to avoid malnutrition (Wolfe and Behrman, 1983) Unfortunately, Engel's law stated that as incomes grow, people would reduce the food spending proportion in their budget and increase the budget share for non-food commodities Besides, people often consume foods to suit their tastes and habits rather than to get more calories

in their choice

The thesis examines the effect of income on nutrient intake of households in the Southeast region of Vietnam in 2000 All the data is taken from The National Food and Nutrition Survey of the National Institution of Nutrition (NIN) in 2000 Regression result analyzes show that there is a positive effect of expenditure on calorie intake but the effect is rather low People have shifted their spending to animal foods such as meat, eggs, milk and milk products Gender does not have any correlation with calorie intake

TABLE OF CONTENTS

CERTIFICATION 1

ACKNOWLEDGEMENTS ~ !

ABSTRACT 2

TABLE OF CONTENTS 3

LIST OF TABLES 6

LIST OF FIGURES 7

ACRONYMS AND ABBREVIATIONS 8

CHAPTER 1: INTRODUCTION 9

1.1 PROBLEM STATEMENT 9 1.2 OBJECTIVES OF THE THESIS 11 1.3 RESEARCH QUESTIONS 11 1.4 RESEARCH HYPOTHESIS 12 1.5 RESEARCH METHODOLOGY 12 1.6 THE ORGANIZATION OF THE THESIS 13 CHAPTER 2: LITERATURE REVIEW AND EMPIRICAL STUDIES 15

2.1 CONCEPTS OF NUTRIENT INTAKE AND THE HOUSEHOLD 15 2.1.1 2.1.2 2.1.3 2.1.4 NUTRIENT INTAKE 15

HOUSEHOLD 16

HOUSEHOLD INCOME 16

l-IOUSEHOLD HEADSHIP 17

2.2 LITERATURE REVIEW 17 2.2.1 THE THEORY OF CONSUMER CHOICE 18

2 2.1.1 The Traditional Model of Consumer Behaviour 18

2.2.1.1.1 Assumptions of the model 18

2.2.1.1.2 The Budget Constraint and The Budget Line 19

2.2.1.1.3 Indifference Curves 19

2.2.1.1.4 Income effect on Consumption 20

2.2.1.2 Criticisms on The Theory ofConsumer Choice 22

2.2.1.2.1 Advantages of The Theory of Consumer Choice 22

2.2.1.2.2 Disadvantages ofthe Theory of Consumer Choice 23

2.2.2 MODELS OF INTRA-HOUSEHOLD ALLOCATION 24

2.2.2.1 The Unitary Model ·.24

2.2.2.2 The Collective Model 25

2.2.3 EQUIVALENT LAWS OF ECONOMICS 26

2.2.3.1 Engel's Law 27

2.2.3.2 Bennett's Law 27

2 2 3 3 ·Slutsky's Law : 28

2.2.4 MEASUREMENT ERRORS OF NUTRIENT INTAKE 29

2.3 EMPIRICAL STUDIES 30 2.3.1 NUTRIENT INTAKE AND INCOME 30

2 3.1.1 Studies with low calorie income elasticities 3 0 2.3.1.1.1 Study ofBehrman and Wolfe (1983) 30

2.3.1.1.2 The study ofBehrman and Deolalikar (1987) 31

2.3.1.1.3 The study ofTimmer and Alderman (1979) 32

2 3.1 2 Studies with high calorie income elasticities 3 3 2.3.1.2.1 The study of Pitt (1983) 33

2.3.1.2.2 The study ofPinstrup-Andersen and Caicedo (1978) 34

2.3.2 NUTRIENT INTAKE AND HOUSEHOLD CHARACTERISTICS 35

2.3.2.1 Nutrient intake and household size and demographic structure 36

2 3 2 2 Calorie intake and household head 3 7 2 3 2 3 Calorie intake and female education 38

2.4 CONCLUSION 38 CHAPTER 3: SOCIAL AND ECONOMICAL SITUATION, FOOD CONSUMPTION AND NUTRIENT INTAKE IN THE SOUTHEAST REGION OF VIETNAM: GENERAL PICTURE 40

3.1 SOCIAL AND ECONOMICAL SITUATION IN THE SOUTHEAST REGION OF VIETNAM 40 3.2 FOOD AND NUTRITION SITUATION IN THE SOUTHEAST REGION OF VIETNAM 41 3.3 SOME NUTRITIONAL PROGRAMS IMPLEMENTED TO SOLVE MALNUTRITION PROBLEMS IN VIETNAM 43 3.4 CONCLUSIONS 45 CHAPTER 4: ANALYTICAL FRAMEWORK AND RESEARCH METHOD ••.••••••.•.•••••.••••••••••••.••••••••••••••••.•.•.••••••••.•.••.••••••.•••••.•••••••••••••••••••••.•••••••••••••••••••••••••••• 46 4.1 ANALYTICAL FRAMEWORK 46 4.1.1 TRANSFORMING CONCEPTS INTO VARIABLES .46

4.1.1.1 The nutrient intake 46

4.1.1.2 Household income 47

4.1.1.3 Household characteristics 48

4.1.2 MODEL SPECIFICATION · 51

4.2 ~ESEARCH METHOD 52 4.2.1 DATA REQUIREMENT 52

4.2.2 DATA SOURCE : .'53

4.2.3 SAMPLE SELECTION , 53

4.2.3.1 The surveysites 53

4.2.3.2 Sample size and sampling , 53

4.2.4 METHOD OF ANALYZING DATA 54

CHAPTER 5: EFFECTS OF INCOME ON NUTRIENT INTAKE OF HOUSEHOLDS IN THE SOUTHEAST REGION OF VIETNAM IN 2000 55

5.1 FOOD CONSUMPTION PATTERN, NUTRIENT INTAKE AND HOUSEHOLD CHARACTERISTICS 55 5.1.1 FOOD CONSUMPTION PATTERN , 55

5.1.2 NUTRIENT INTAKE 58

5 1.3 HOUSEHOLD EXPENDITURE 60

5.1.4 HOUSEHOLD CHARACTERISTICS 61

5.1.4.1 Household size 61

5.1.4.2 Household head 62

5.1 4 3 Education 63

5.2 REGRESSION RESULTS 64 5.2.1 DESCRIPTION OF VARIABLES 64

5.2.2 RELATIONSHIP BETWEEN CALORIE INTAKE AND TOTAL EXPENDITURE, CALORIE INTAKE AND ENERGY ORIGINATED FROM ANIMAL PRODUCTS 67

5 2 2.1 Calories per capita and total expenditure of household 67 ·

5 2 2 2 Calories per capita and calories originated from animal foods 68

5.2.3 REGRESSION RESULTS 70

CHAPTER 6: CONCLUSIONS AND POLICY RECOMMENDATIONS 77

APPENDIX 1 81

APPENDIX2 83

BI~LIOGRAPHY ~87

LIST OF TABLES

Table 2.1 : Income and expenditure elasticities of calorie demand 3 6 Table 3.1: Mean share (%) of food groups in total calories consumed by a

household by regions ····:···: 43

Table 4.1: List of variables and expected signs in relation with calorie intake 51

Table 4.2: Four forms of commonly used Engel functions 52

Table 4.3: Number of households by city/ provinces in the Southeast region 55

Table 5.1: Major food group- average intake (grams/household/day) 57

Table 5.2: Food consumption pattern by household expenditure quintiles in the Southeast region (grams/household/day) 59

Table 5.3: Energy composition in the Southeast region and the whole country 60

Table 5.4: Highest diploma obtained by gender in the Southeast region in 2000 64

Table 5.5: Percentage of female per household who gets the highest diploma by expenditure quintile in the Southeast regionin 2000 65

Table 5.6: Summary statistics 66

Table 5 7: Distribution of quantitative variables 667

Table 5.8: Calorie groups by mean total expenditure per household 68

Table 5.9: Daily calories per capita originated from animal foods by calorie groups and expenditure quintiles 70

Table A2.1: Multicollinearity test 84

Table A2.2: Heterokedasticity test 85

Table A2.3: Functional form test 85

Table A2.4: Autocorrelation test 86

LIST OF FIGURES

Figure 2.1: Matching Consumer Desires with Available Consumer Choices 21 Figure 2.2: Income-Consumption curve and its form 22 Figure 3.1 : Percentage of children under five suffering from mild underweitght and stunting (malnutrition level 1) 44

·Figure 5.1: Average consumption level of other foods in the Southeast region (grams/household/day) 58 Figure 5.2: Total expenditure and food expenditure by quintiles in the Southeast region (1 000 dongs/person/month) 62 Figure 5.3: Histogram of quantitative variables with normal curve 667 Figure 5.4: The relationship between daily calories per capita and total expenditure per household 69 Figure 5.5: The relationship between calories per capita and calories from animal foods 70 Figure A2.1: Histogram of residual from the regression on linear function with unstandardized residual 87 Figure A2.2: Histogram of residual from the regression on linear function with standardized residual 87

ACRONYMS AND ABBREVIATIONS

Asian Development Bank Body Mass Index

BaRia- Vung Tau Centimeters

Food and Agriculture Organization Gross Domestic Product

General Statistical Office

Ho Chi Minh City Kilocalories Non Governmental Organization National Institution ofNutrition Ordinary Least Squares

Protein Energy Malnutrition Statistical Package for the Social Sciences United Nations

United States United Nations International Children's Fund Vietnam Economic Review

Vietnam's Socio-Economic Development World Bank

Chapter 1: INTRODUCTION

1.1 PROBLEM STATEMENT

When human beings become the subject of development in any certain country, nutrition then becomes a great concern According to the Asian Development Bank (ADB) (2001: 26), 'Nutrition plays a very important role in welfare, and good nutrition is essential for good, long-term health for both children and adults' A person with better nutrition and health has more capabilities to live, study, work and enjoy his life than ill people (Ingham, 1993) Socially, a country with a nourished population may yield a good chance to improve anthropometrical as well

as intellectual indicators of its present and future generations (Behrman and Deolalikar, 1988), to increase output due to faster productivity (Ray, 1998) and to meet its developmental goals such as higher income and life expectancy (Ingham, 1993)

Once a person's income increases, his nutritional status may be improved accordingly (Ray, 1998) Statistics of FAO (1998) show a substantial gap between developed and developing countries in average daily kilocalorie (kcal) availability per capita An average daily energy supply per capita of developed countries is 3,340 kcal, while that indicator of the poorest countries is just 2,060 kcal This seems consistent with the idea of the WB in its World Development Report in 1980 that 'the most efficient long-term policies are those that raise the incomes of the poor, and those that raise food production per person' (cited in Wolfe and Behrman, 1983: 525) However, income itself is not enough to reflect the availability of each good in the market and how much of each good that a person can afford and prefer

to consume (Dasgupta, 1995) As a result, whether income is an important determinant of nutrition is still in debate (Wolfe and Behrman, 1983)

According to Engel's law, when people earn additional income, they tend to spend more on non-food commodities and less on food items (Gillis et al., 1996)

Moreover, as the income of people goes up, they often prefer foods with higher quality rather than higher calorie content The effects of income in the hands of men and in those of women are also different Women tend to care more about health and nutrition of all members in the household, especially for children, compare to men (King and Hill, ed., 1993)

Over the last decade, Vietnam has made good results in human development such

as income growth and poverty reduction (WB, 2001 b) From 1993 to 1998, GDP per capita grew on average 8.9 percent per year Poverty rate, which is measured by 2,100 calories per day as stated by the WB, reduced from 55 to 37 percent during the same period

According to reports from UNICEF in 1996 (stated in Foster and Leathers, 1999), Vietnam is on the list of 50 most seriously undernourished countries by three different measures, i.e daily per capita calorie supplies as a percent of requirements, under-five mortality rate per 1000 live births and percent of children underweight In 1999, the percentage of undernourished people in total population

in Vietnam is 19 percent

Although the living standards of Vietnamese households have been improved over the last decade, their nutritional status is still in problem According to ADB (200 1 ), Vietnamese households may spend their additional income on food without referring to richer calorie content or on other non-food commodities Besides, some recent empirical studies show that income elasticities of nutrient intake are low in poor countries (Ravallion, 1990) Therefore, this thesis will examine the effects of income change on calorie intake by using a large household-level survey on food consumption, incomes and other household characteristics of Vietnam in 2000 conducted by the NIN In particular, data from the Southeast region will be focused for analyzing

1.2 OBJECTIVES OF THE THESIS

Health and nutrition have been of great concern m Vietnam for a long time Together with better living standards and higher income, nutritional status of people should be improved Therefore, it is important to kt1ow the impact of income change on nutrient intake of people so that the government can apply efficient policies to solve this problem The study will focus on this situation in the Southeast region that achieved the highest speed of income growth in Vietnam

People tend to care about their nutrition based on their preferences over its characteristics (Dasgupta, 1995) Even though people are undernourished, they would not spend the whole income for food consumption Also, food patterns of people depend on many non-food factors such as taste, cultural or household characteristics that may outweigh interests on nutritional contents (Deaton, 1997) Besides, malnutrition often damages welfare of people more seriously than ~hey used to think There is a trap: weak people cannot work and without working, they lack ability to improve their malnutrition (Ray, 1998) Thus, nutritional value is the 'first necessity' that may be considered as conditions for other needs to satisfy (Deaton, 1997: 211)

Recent achievements in economic growth in Vietnam have increased substantially income of people However, the increase in quantity alone is not enough It is the quality of· such increase to be important, too Whether additional income contributes to welfare of people is really a matter This thesis tries to examine the impact of income on nutrient intake of people, so it may help policymakers and food suppliers to have a real picture of nutrient consumption of people Based on this, policymakers and food suppliers may have different strategies to influence the

nu~rient consumption of people in the way that improve their health and productivity In detail, the thesis plans to address the following question:

(1) Is there a positive effect of income level of households on daily calorie

intake per capita of households in the Southeast region of Vietnam in 2000?

Besides, in the context that Vietnamese females play an important role on food preparation and consumption in the household, the thesis tries to answer two more questions:

(2) Is there a positive effect of female head on daily calorie intake per capita of

households?

(3) Is there a positive effect of highest level of formal education of female

adults on daily calorie intake per capita of households?

Because factors such as income and female characteristics seem to be complementary for nutrition status of a person based on many empirical studies, the following relationships are hypothesized for testing:

(1) There is a positive effect of income on daily calorie intake per capita of

households in the Southeast region of Vietnam in 2000

(2) There is a positive effect of female head on daily calorie intake per capita of

households

(3) There is a positive effect of the highest level of formal education of female

adult on daily calorie intake per capita of households

This thesis studies the impacts of income, female headship and education on calorie intake of households in the Southeast region of Vietnam in 2000 by using the household-level data set done by the NIN The tertiary and secondary data (data

and reports from specialized organizations in Vietnam and abroad) will be used to analyze the current food pattern of Vietnamese households and to compare it with others in some developing countries Besides, some specialized documents will be used to correctly evaluate impacts of the above factors on calorie intake of households

The research questions will be handled by using descriptive statistics, quantitative analyses and OLS regression to analyze factors affecting calor1e intake The first approach analyzes the distribution of dependent and independent variables, and then studies the relationships of calorie per capita per day with expenditure, female headship and education via using numerical summaries and graphs The effects of all factors in determining calorie intake are estimated by a multivariate model of calorie demand Finally, based on the regression model, results will be interpreted for policy recommendations

The thesis consists of six chapters, annexes and a bibliography The outline is as follows:

Chapter 1 presents the introduction, including the problem statement, research questions and hypothesis, research methodology, and the organizati<;m of the thesis

Chapter 2 provides the theoretical framework and empirical studies related to impacts of income (expenditure), female headship and education on calorie intake The chapter includes the definitions of key concepts such as household and nutrient intake, traditional consumer theory and the theory of intra-household allocation

Chapter 3 presents general information about food patterns and calorie intake in Vietnam and in the Southeast region Government policy as well as remarkable results will also be outlined

Chapter 4 focuses on model specification and variables choice justification

The practical results are analyzed m chapter 5 vm descriptive statistics and econometrics analysis

Concluding remarks and policy implications to influence the nutrition problem in the Southeast region are shown in chapter 6

•

Chapter 2: LITERATURE REVIEW AND EMPIRICAL STUDIES

This chapter includes three sections First, some important concepts such as household and nutrient intake are defined Second, it reviews the traditional consumer theory relating to consumer choice with the budget constraint, the intra-household allocation theory to specify the preferences of male and female over nutrient-related expenditure Third, some empirical studies about the important determinants of household nutrient consumption are summarized

2.1.1 NUTRIENT INTAKE

The nutrient intakes are nutrients consumed that are necessary for maintaining the body and different physical activities of individuals (Ray, 1998) Traditionally, human beings consume five categories of nutrients, including proteins, carbohydrates, fats, vitamins, and minerals (Dasgupta, 1995) These nutrients are complementary to each other, so a deficiency of one nutrient cannot be added by a redundancy of others The main tasks of nutrients are to create the growth of human body, to maintain enough energy for activities such as working and learning and to reinforce the resistance and immunity system against infections and diseases (King and Burgess, 1996) Therefore, a person is malnutrition if he cannot obtain one or all of these above effects of nutrients

One way to avoid being undernutrition is increasing the amount of food that a person eat because food is a rich source of nutrient supply for human, especially create energy through a chemical form (Dasgupta, 1995; Foster and Leathers, 1999) Empirical evidence relating to food consumption shows that calorie deficiency is 'a more widespread problem' in developing countries than protein deficiency (Foster and Leathers, 1999: 28) Thus, this thesis will focus on examining daily calorie intake per capita of households

2.1.2 HOUSEHOLD

In general, a household is often defined as a group of people who have kinship and

do some activities together such as eating, sharing accommodation, taking care of the house as well as each other and pooling some money to form a budget for daily consumption (The Open University, 1998) Roberts (1991: 62) considers a household as 'a basic unit of the society in which activities of production, reproduction, consumption and the socialization of children taking place' Messer (1982: 52) adds that a household is also 'a group who have a common fund of material and human resources and rules for practices and exchange within it' In an attempt to define household, the WB (2001a: 150) stresses on the function of giving birth, bearing, rearing and training children to become useful citizens of the society, especially within the context of limited resources 'across activities, including consumption, production and investment'

According to the UN (quoted in Glewwe, Staphanie and Bui, 2002:135), wpen people do economic surveys in developing countries, they often define the household as 'a group of people who live together, pool their money, and eat at least one meal together per day' This thesis uses the data from the survey of the

NIN, so it will follow the definition given by them According to NIN (2001), a household is a group of three or more persons who eat in the same meal, cook in the same kitchen, have a common budget and include one or more generations

2.1.3 HOUSEHOLD INCOME

Household income includes wages, net revenue from agriculture and non-farm enterprises and from other sources (Haughton, et al., 2001 ) Theoretically, household income should be used to consider its impact on nutritional inputs of members of households However, this kind of data exposes some disadvantages for analytical results First, between income and expenditure, the first is the more

sensitive information and thus it is often supplied at low accuracy (Eatwell, et al., eds., 1991) Second, because income varies substantially from year to year, it may

not express fully the typical economic position of the household (Vu, 1999) Third, expenditure is easier to approach and closer to permanent income of household, so

it is 'a realistic measure of permanent income' (Haddad, et al., 1996: 40) Due to those reasons, total expenditure is considered a good proxy for income in explaining the pattern of expenditure In this thesis, total expenditure of households will be used to study its impact on calorie intake of household

2.1.4 HOUSEHOLD HEADSHIP

According to Fuwa (2000: 3), in defining household headship, two different household characteristics should be emphasized, including 'demographic composition and (relative) income/economic contribution to the household resources' Based on the objectives of each study, there exist many alternative optional definitions of household headship For example, aiming at examining living standards of Vietnamese in a certain year, the VLSS 1997-98 (GSO, 1997) defined a household head as a person who earns the highest income in the household as well as the main decision-maker for the household and knows about the economic and professional activities of every member in the household In this thesis, the definition of household head is given by the NIN (200 1) from their food and nutrition survey in 2000 and it is similar to the definition of the VLSS 1997-98 above

2.2.1 THE THEORY OF CONSUMER CHOICE

2.2.1.1 The Traditional Model ofConsumer Behaviour

Food is among many goods that people consume every day, so it is important to know what and why they purchase, and how they behave when prices of goods and their income change This part presents some assumptions and basic elements of the model Then some advantages and disadvantages of the model in reality are drawn

2.2.1.1.1 Assumptions ofthe model

As presented by Kooreman and Wunderink (1997), the traditional model of consumer behaviour has some following assumptions

First, the theory assumes that the consumers' satisfaction may be measured in utility The more utility a good offers, the more enjoyment a person has from consuming it to compare with other goods (Pearce, ed., 1992)

Second, consumers can decide which bundle of goods they like most to consume The order of their preferences is summarized as follows (Kooreman and Wunderink, 1997: 11):

Reflexive: A ~ B for all bundles of A;

Transitive: if A ~ B and B ~ C, then A ~ C;

Complete: either A ~ B, B ~ A or A = B

This also implies that given incomes and prices, consumers will choose the best bundle of goods that brings them maximum satisfaction

Third, consumers' taste faces a diminishing marginal rate of substitution (Beg et al., 1991) Consumers' interest in consuming one good reduces accordingly with its additional amount of consumption, given utility

Fourth, there is no asymmetric information on prices and qualities in the market Besides, the supply of goods and services is so large that consumers can buy at any amount they want, given their income and prices

2.2.1.1.2 The Budget Constraint and The Budget Line

According to Begg et al (1991: 76), a budget constraint refers to different combinations of goods that consumers can afford or 'the maximum affordable quantity of one good given the quantity of the other good being purchased' Consumers' choice must meet the budget restriction, based on prices of goods they face in the market and their income That is:

Where:

Pi: the price of good i qi: the quantity of good i consumed y: consumer's income

A budget line is a line that reflects a budget constraint in a graph It also indicates a maximum combination of goods that consumers can buy with a given income and prices (Pearce, 1992) Assuming that consumers spend all of their income Y on food (F) at the price pF and non-food commodities (NF) at the prices pNF, then Y

An indifference curve has three basic characteristics (Begg et al., 1991) First, it must slope downwards Second, it becomes flatter to the right to satisfy the assumption that consumers' taste exhibits a diminishing marginal rate of substitution Third, it never cuts any other indifference curve If this happened, the intersection point would yield two different levels of utility

In brief, the theory of consumer choice explains how consumers reconcile their desired bundles (indifference curves) and their available access (the budget line) with their rationality (Timmer et al., 1983) Figure 2.1 illustrates the above conclusion

Figure 2.1: Matching Consumer Desires with Available Consumer Choices

Food

c

Source: Timmer eta! (1983: 38)

2.2.1.1.41ncome effect on Consumption

When incomes of people change, they will shift their consumption as well However, such change may be positive, negative or idle related to changes in

I

incomes To examine these relationships, this sub-sector presents income effect on consumption through its representative, i.e income elasticity for demand First, income effect on demand or income-consumption curves are examined as presented

by Hicks (1995)

income-Another name of income-consumption curve is Engle curve 'An Engle curve shows the relationship between the quantity demanded of X and the income of the consumer with prices held constant' (Pashigian, 1995: 94) When income and quantity demanded of one good change in the same direction, that good is normal good and the Engle curve has a positive slope If they move in the opposite direction, that good is inferior good and the Engle curve has a negative slope However, it is up to people to consider which good is inferior or normal although they may have the same level of income or income change (Pashigian, 1995)

The magnitude of income effect on consumption is measured by income elasticity

of demand As defined by Hirshleifer (1988: 80), 'The income elasticity of demand

is the proportional change in the quantity purchased divided by the proportional change in income' The formula is as below:

X

£ix : the percentage change in the quantity of good x

it is called normal good Finally, when income elasticity for demand of one good is negative, people shift their spending to other goods as their incomes increase and it

is called inferior good In household expenditure structure, food is often considered

as inferior commodity, clothes as normal commodities and holidays as luxurious commodities (Pashigian, 1995)

2.2.1.2 Criticisms on The Theory ofConsumer Choice

The theory of consumer choice is often presented in microeconomics under the consumer behavior part because of its simplicity and plausibility Since the middle

of the 19th century, there has been some development of this theory based on shortages of the original one (Eatwell et al., eds., 1991) This part summarizes advantages and disadvantages of the theory of consumer choice

2.2.1.2.1 Advantages ofThe Theory ofConsumer Choice

The Theory of Consumer Choice is successful in modeling consumer behavior to satisfy their taste with their budget constraint and the prevailing prices in the market What contributed to that success? Here are the reasons

The Theory of Consumer Choice allows us to determine a bundle of goods that consumers choose to get the highest utility given their income and market prices There is no need to discover many complicated motivations to lead to such choice (Rice, 2001) It aims at results rather than the process ofmaking.decisions

Another advantage of the Theory of Consumer Choice is freedom of choice with respect to consumers' budget constraint (Timmer et al., 1983) Facing the same slope of the budget line given by the ratio of prices in the market, we can calculate the rate of substitution of one good for other good for people who have their highest indifference curve tangent to the budget line

2.2.1.2.2 Disadvantages of the Theory ofConsumer Choice

In the theory, consumers can maximize the utility from their choice with the assumptions that they are rational and get full information for their decision However, in reality, everything seems more complicated than those simple assumptions

In reality, when facing many choices, it is difficult for consumers to decide a choice that makes them best off (Davis, 2001 ) Between drug and fruits, the latter is better for their health and cheaper for their budget than the first, but some consumers still choose drug because of curiosity or just addiction

Another factor that challenges the assumption of the model is that there always exists uncertain information in reality (Kooreman and Wunderink, 1997) Income and prices often fluctuate largely through time Therefore, it is impossible for consumers to maximize their utility through their choice like in the model

The model focuses on explaining consumer choice in a static way, but in reality such choice is affected by the past or influences the future of the customers and their changing environment (Kooreman and Wunderink, 1997) An economy in recession may encourage people to spend less and save more although they still can afford for a bundle of goods that yield the highest utility

The model assumes that consumers can buy any amount of goods in the market within their budget but it is not true in reality, especially when distinguishing '

consumption and expenditure (Kooreman and Wunderink, 1997) With the high

speed of technology development, demands of people are increasingly diversified and satisfied as much as possible

One important disadvantage of the Theory of Consumer Choice is that it just aims

at individual behavior (Kooreman and Wunderink, 1997) In reality, some goods are consumed by a group of people such as a household Therefore it is difficult to determine who consumes what This situation often happens in food and utility consumption Besides, researchers often conduct surveys in households rather than

in individuals (Eatwell et al., eds., 1991) So, it is better to have a model to explain the behavior of the household and preferences of members in the household with the household's budget

2.2.2 MODELS OF INTRA-HOUSEHOLD ALLOCATION

Based on the Theory of Consumer Choice, we can examine a static behavior of individuals in their utility maximizing choice with respect to their budget constraint What happens if consumers are a group of people like a household? This part presents two model of intra-household allocation, including the unitary model and the collective model

2.2.2.1 The Unitary Model

Gary Becker first introduced the Unitary Model in the mid-1960s (Quisumbing and Maluccio, 2000) This model considers a household as one unit that has a common budget and allocates resources for household activities such as consumption, investment and production consistent with a single set of preferences to maximize some common welfare index (WB, 2001a)

The Unitary Model is applied largely in research related to household behavior because of 'the relative simplicity of comparative static generated and the diversity

of i~sues it can address' such as examining determinants of education, health and fertility in household (Quisumbing and Maluccio, 2000: 4) Quisumbing and

Maluccio (2000) quoted a study of Pitt, Rosenzweig and Hassan (1990), in which they add individual work effort as a choice variable in the household model developed by Singh, Squire and Strauss (1986) They find that unequal calorie allocations across gender and age groups may reveal different distributions of activities within those groups

However, according to Strauss and Thomas (1995), the model does not seem to be consistent with the Theory of Consumer Choice that a decision maker is a single person with his own preferences; and in reality, members within a household may have distinct preferences Some households may not pool their income and have different responsibility for resource allocations because of their customs such as households in West Africa (WB, 2001a)

2.2.2.2 The Collective Model

The Collective Model focuses on individuality of each member in a household and their preferences, which may be different from each other (Quisumbing and Maluccio, 2000) According to Chiappori (quoted in Quisumbing and Mallucio, 2000), all collective models have two common features First, they admit each member in the household is one decision maker who has different preferences Second, they present a household welfare index that is not unique but depends on prices, incomes and tastes Thus, like the Unitary Model, the Collective Model also allows rules in the intra-household resource allocation process to exist and such rules determine household responses to public policy However, the Collective Model differs from the Unitary Model in permitting public policy to affect those rules

In the Collective Model, households make allocation decisions through a bargaining process (Strauss and Thomas, 1995) In the bargaining process, each, metpber tries to allocate resources over their favorite goods belong to their control This results in two subclasses of collective models They rely on the cooperative and non-cooperative game theory (Quisumbing and Mallucio, 2000)

In the cooperative approach, individuals may choose to be single or to join a household This depends on advantages they get from which case are larger than the other If they find that forming a household yields more welfare, they will prefer the latter

In the non-cooperative approach, the 'fallback' or 'threat point' of individuals, is emphasized It means a certain level of utility a member can achieve If the welfare

of this member falls below the 'threat point', he or she will choose to be single rather than forming a household While all cooperative models are Pareto efficient, which means that resource allocation always happens in a way that makes one member better off together with other member worse off, some non-cooperative models do not satisfy the Pareto efficiency

An important note about the Collective Model is that there is nothing to take priority during the decision process (Quisumbing and Mallucio, 2000) It does not try to explain directly how individual preferences lead to a collective choice Therefore, although the rule to affect intra-household allocation is important, it is necessary to estimate from the data rather than to determine in advance

2.2.3 EQUIVALENT LAWS OF ECONOMICS

Although the theory of consumer behavior and the theory of intra-household allocation are important in determining how people will consume to maximize utility with their budget constraint, they do not show how income changes affect people's consumption Therefore, economists have tried to make out some equivalent laws that best describe changes of people's consumption that is consistent with their income variation Based on the objectives of this thesis, this part will present some equivalent laws of economics in food consumption, including Engel's law, Bennett's law and Slutsky's law

2.2.3.1 Engel's Law

Engel's Law is an empirical law of consumption introduced by Ernst Engle (Pearce, ed., 1992) It states that the share of food expenditure in total expenditure

of a family is a decreasing function of the level of richness of that family (Eatwell

et al., eds., 1991) This law is not only correct for a household but also for a country A survey in East Java, 1997 - 1998, shows that when income increases, the percentage of food expenditure declines from 70 for the poorest households to

65 for the richest ones (Foster and Leathers, 1999) Based on this law, the income elasticity of demand for food is often thought to be less than one (Timmer et · al., 1983)

According to Engel's Law, together with an increase in income, at first, the budget share for one good increases and then decreases (Eatwell et al., eds., 1991) Depending on income level of each household, a good may be a luxury one in a very poor household, a necessity in a normal household and an inferior one in a very rich household

2 2 3 2 Bennett's Law

Going into details of food budget share, Bennett states that as income rises, people will spend less on starchy staples such as cassava, rice, maize, and wheat flour and more on processed foods, higher-price calories and diversified foods such as animal products, vegetables, fruits and milk (Foster and Leathers, 1999) Therefore, Bennett's Law stresses on food patterns' changes from which people get energy as their incomes grow Households with better income may improve their diets in terms of increasing the proportion of higher priced foods that suit their tastes, diversification, and convenience As a result, starchy foods are also considered as inferior goods

Bennett's law describes correctly not only food preference shifts of household but also those of countries (Timmer et al., 1983) Based on a survey of 85 countries in

the world in 1962, it is said that the lowest-income countries get about 75 percent

of their dietary calories from starchy staple carbohydrates, while the income countries get only 30 percent of their dietary calories from those foods (Foster and Leathers, 1999)

In brief, Engel's law, Bennett's law and Slutsky's law helps to estimate household behaviors in food consumption and food expenditure as their income change More broadly, they help to compare food patterns of different countries in the world However, in explaining such relationships in individual of very poor households, they still mislead the results (Timmer et al., 1983) Besides, analyzing income effects on nutrient intake derived from food consumed also depends on how nutrient intake is measured The next part will deal with measurement errors in nutrient intake

2.2.4 MEASUREMENT ERRORS OF NUTRIENT INTAKE

Data measurement affects strongly results of a research When measurement errors exist, the interpretation may involve controversies that lead to bias in final results and policy recommendation (Behrman and Deolalikar, 1988) This part reviews those issues

First, it is difficult to measure exactly quantities of food people actually eat because there are wastage and leakages during processing and after meals (Strauss and Thomas, 1995) Often nutrient intake is overstated in high-income households who waste more foods or have more guests at meals Conversely, it is understated in low-income households that receive food aid or go away from many meals As a result, calorie consumption is calculated with bias

Second, there are likely measurement problems when people use fixed conversion factors for food with different qualities, way of processing and thus different nutrient contents (Behrman and Deolalikar, 1988) In one kind of food such as rice, there are varieties of rice that have different calorie quantities Also, energy people get from fruits that are fresh or dried is very different Therefore, income effect on nutrient intake may have bias in the use of fixed nutrient to food conversion factors

Third, food consumption survey is often collected in groups such as rice a.nd cereals rather than rice or cereals separately This may give wrong calorie contents because the aggregate level is emphasized, and which food in one group that has the highest calorie content may dominate the calorie content of the whole group (Behrman and Deolalikar, 1988)

Finally, data measurement errors happen because of unreliable information (Diskin, 1995) For example, respondent may recall insufficiently or misunderstand

'

questions of the survey or enumerators Not only food consumed but also other

factors may affect calorie contents that people get such as bad sanitary conditions

or careless food storage

2.3 EMPIRICAL STUDIES

This section reviews some empirical studies in developing countries about the determinants of nutrient intake such as income or expenditure and household characteristics to justify the literature

2.3.1 NUTRIENT INTAKE AND INCOME

Since good health is one basic need of people (Foster and Leathers, 1999), food consumption and nutrient intake become central issues of many studies in developing countries Through those studies, it seems that low calorie intake' in those countries is a consequence of low income However, the effect of income changes on calorie intake - one popular and important nutrient - varies in studies and countries (Strauss and Thomas, 1995)

2.3.1.1 Studies with low calorie income elasticities

When estimating income elasticity of demand for calories of households in some developing countries, studies of Wolfe and Behrman (1983), Behrman and Deolalikar (1987) and Timmer and Alderman (1979) find that this indicator is very small Due to this, the conclusion is that although belonged to the lowest-income group, such households still consume additional foods that are consistent with their tastes and have non-calorie nutrient Details of these studies are presented below

2.3.1.1.1 Study of Behrman and Wolfe (1983)

Behrman and Wolfe (1983) have conducted a study on nutrient demand for households in Managua - the central metropolis of Nicaragua, on an attempt to test the' idea of WB when considering income as a very important determinant of nutrient consumption The sample of the study consists of 1,294 households, in

which only 1,167 households with full information on all major variables such as household income and food consumption are used for estimating nutrient demand model The nutrient variables are calculated by summing nutrients from each food groups had per week by nutrient content table and international standards To estimate income effects on nutrient consumption, these authors apply OLS method

After estimation, tpe study finds that income elasticity for calorie intake at mean level is only 0.01 This means that an increase in income almost has no effect on calorie intake The authors give some reasons to explain for the low effect above First, it seems that there is a problem with income definition Household income used in the data set is short-run income that fluctuates often and thus cannot represent long run or permanent income Instead, expected income seems to satisfy the requirement Besides, incomes in the data are collected separately from men or women and effects from each source of income are different, especially when women are mainly responsible for food buying and cooking._ Second, there may have bias in income effect because there is big income variation with large standard deviation relative to the mean Third, measurement errors of income may exist in the data set and thus underestimate income effects on nutrient intake Fourth, there may have the simultaneity between income and nutrition and thus lead OLS estimates of the income effect to be close to zero than it would

In conclusion, the authors suggest that nutrient intake of households in Managua is not strongly positively related to income changes It may be because incomes of households in this region increase at such a low level that cannot give clear effects

on nutrient intake Therefore, policymakers and international institutions such as the WB should focus on the real and particular situation in each region rather than only income augmentation to improve nutrient intake and thus health of population

(1987) use panel data set from the International Crops Research Institute for rural south India to test this issue The data sample includes 240 households in six villages in semiarid tropical India These authors examine food and nutrient expenditure by running two stage-least squares and estimating directly expenditure-nutrient elasticities

In the study, Behrman and Deolalikar (1987) clarify why they choose direct nutrient elasticity estimation rather than indirect one In direct method, nutrient expenditures are estimated in a function of relevant nutrient price, total expenditure and other factors The indirect method tends to give higher nutrient expenditure elasticities than in the direct method because upward biases seem to exist in food aggregation Thus, these authors think that the direct estimates may give more realistic results

The study estimates food expenditure relations for six food groups (grains, sugar, pulses, vegetables, milk and meat), and nutrient expenditure relations for nine nutrients (calorie, protein, calcium, iron, carotene, thiamine, riboflavin, niacin and ascorbic acid) It finds that expenditure calorie elasticity at mean level is 0, 17, which means that calorie intake increases at low level with income improvement This result sets a question for policymakers in finding alternative solutions for nutrient improvement in developing countries rather than depending on income alone

2.3.1.1.3 The study ofTimmer and Alderman (1979)

Another study related to estimating price and income elasticities for food consumption and calorie intake conducted by Timmer and Alderman in 1979 The objective of the model is to find substitutive possibility of one bad quality to higher quality food due to income improvement and income effects on calorie intake of households at different levels of income

The study used data set from the Indonesian Socio-Economic survey V in 1976, including 18,000 households The authors estimate income elasticity of demand for

rice and cassava and calorie by Engle function with a double-log quadratic form This allows income elasticity of calorie demand to vary with different income classes The calorie income elasticity at mean level of urban households is 0.261 and that of rural households is 0.471 However, while the calorie income elasticities

of the lowest income class are 0.740 for urban households and 0.776 for rural households, those of the highest income class are 0.13 for urban households and 0.246 for rural households In conclusion, the authors find that the poor may have good chance to improve their calorie intake if policymakers direct income to them

in an efficient way

2 3.1 2 Studies with high calorie income elasticities

Studies with high calorie income elasticities often happen in the case of indirect method In these studies, only elasticities of demand for some food groups are calculated, then based on standard food composition table, they will be converted into elasticities of demand for calories Applying indirect approach, it is implicitly assumed that there is no shift from lower-priced calories to higher-priced calories within each food group (Strauss and Thomas, 1995) Among these studies are the study of Pitt (1983) and the study ofPinstrup-Andersen and Caicedo (1978)

2 3.1 2.1 The study of Pitt (1983)

In this study, the author wants to estimate own-price and cross-price elasticities and expenditure elasticities for main commodities that contribute a large part of calorie and protein for households in rural Bangladesh

The data used for the study are taken from the Household Expenditure Survey of Bangladesh in 1973-1974, which includes 5,750 households in about 800 villages

in Bangladesh Because there are some households who do not consume some foods observed, to avoid biased and inconsistent estimates due to non-norrp.al dist,ribution of disturbances, the author applies Tobit model for estimation

Households are divided into two groups The percentile 25 households are the ones who have per capita food expenditure larger than 75 percent of total household sample; the percentile 90 households are the ones who have per capita food expenditure greater than 10 percent of total household sample The first group represents households who do not suffer from nutritional deficiencies but the second group represents households who do suffer from such matter

After estimation, the author finds that calorie expenditure elasticities for percentile

25 households and percentile 90 households are 0.82 and 0 78 respectively These results suggest 'that even very poorly nourished households can improve nutrition

by altering their diet' (Pitt, 1983)

2.3.1.2.2 The study of Pinstrup-Andersen and Caicedo (1978)

Concerning malnutrition problem in developing countries and especially in Cali, Columbia, Pinstrup-Andersen and Caicedo (1978: 402) remarks that this problem is caused by 'unequal distribution of available nutrients rather than absolute scarcity' Malnutrition is more popular in low-income households who spend most of their income on food but still catmot meet nutritional requirements

To check these hypotheses, the authors use the model based on neoclassical demand theory They estimate calorie income elascitities for each income class by using data from a survey conducted in Cali, Colombia, which includes 230 families and quantities and prices of 22 foods or groups of foods consumed The study finds that calorie income elasticity at mean level is 0.51 Calorie income elasticities become smaller when the households belong to higher-income groups For example, while the calorie income elasticity of the poorest group is 0.69, that indicator of the richest group is 0.23 The study also finds that income transfer from the rich to the poor may have great effect on food demands and thus help to improve nutrient intake of the poor Therefore, the authors suggest that because any policy aiming at improving nutrition of people would have direct and indirect impact, policymakers should consider how to combine these policies so that to have good impacts in the long run

Finally, the authors state some limitations and needs for further research of the

study First, the estimation of elasticity may be biased because there are large

changes in prices, incomes and quantities consumed Second, there may have

intra-family allocation of resources, so although a intra-family have met the standard nutrient

requirement, some members in the family may not Third, the study only estimates

calorie and protein income elasticity although other nutrients are also important for

human being such as vitamins

Table 2.1 Income and expenditure elasticities of calorie demand

A INDIRECT ESTIMATES (Calculated from food demand equations)

& Caicedo

B DIRECT ESTIMATES (Calculated from calorie demand equations)

Source: extracted from Strauss and Thomas (1995: 1894-1895)

Availability: Calorie measures are household availability based on food purchases and changes in

stocks of own production; Intake 24-hr recall: recall for last 24 hours

System: Estimation method is demand system

2SLS: Estimation method is two stage least squares

OLS: Estimation method is ordinary least squares

2.3.2 NUTRIENT INTAKE AND HOUSEHOLD CHARACTERISTICS

Apart from income, household characteristics are also important determinants of

nutrient demand Among them are household size and demographic structure,

headship of the household and education of females in households This sub-sector

reviews empirical studies of each character above

2.3.2.1 Nutrient intake and household size and demographic structure

Because a household often consists of members with different ages, sexes, and preferences, there are some expectations about household calorie demand per capita For example, a household with more adult members may require higher level of calories or people who share one kitchen and have meals together may save foods from wastage (Deaton, 1997) Therefore, it is necessary to add household size and demographic structure into estimating model of calorie expenditure

Based on the study of Subramanian and Deaton (1996) about calorie consumption

in Maharashtra in India 1983, results from OLS method show that large households have lower calorie consumption per capita, partly because they have more children However, the household sizes do not affect the calorie expenditure elasticities strongly due to the fact that regression lines of different household sizes are nearly parallel to each other If the model includes only two variables that are daily calorie intake per capita and total · expenditure per capita, the calorie expenditure elasticities tend to be upwards because if total expenditure per capita increases but household size falls, the positive effect on calorie per capita may add to total expenditure per capita To test this situation, the authors add household size in the model, the calorie expenditure falls from 0.4 to 0.35

Ravallion (1990) has a study about income effects on undernutrition The author uses the data set from the 1978 National Socioeconomic Survey in Indonesia to estimate nutrient demand functions The model applies OLS method but includes some variables in quadratic forms such as total expenditure, price of rice, number

of adults, number of children, age of head and sex of head The study finds that calorie intake increases for number of adults, age of head and schooling of head at mean level

2 3 2 2 Calorie intake and household head

When examining the relationship between calorie intake and household head, many studies find that incomes that are controlled by female head often have better effects to food consumption and nutrient intake of households (Diskin, 1995) Conversely, other studies show that female-headed households have disadvantages

on households' welfare, including nutritional status of individual members in households

Because females tend to spend income directly on foods and services in favor of the nutrition, health, and the well-being of the household, an increase in income of females can have a greater effect on the above matters than that of males (Kurz and Johnson-Welch, 2000) A study of Thomas (1997), which analyzes data from a household budget survey in Brazil in 1974-75, supports this argument The study finds that while income managed by females created a three percent increase in food expenditure, income managed by males only contributed 0.6 percent in food expenditure Such difference is significant and reflected in child nutritional status According to the study, child weight-for-height increases eight times faster in female-headed households than in male-headed households Similarly, a study of Kennedy and Cogill (1987) in Kenya finds that there is a positive effect of income

of females but a negative effect of income of males on household calorie level, despite the fact that female heads earn less than male heads among households

Because of income constraint in female-headed households, some studies find that members in male-headed households are better off in many fields, including higher quality food consumption and nutrient intake As Panda (1997) reports that female-headed households spend relatively less on higher quality foods such as meat, vegetables, milk and other dairy products At the same time, children in female-headed households are worse off both in terms of access to social services and actual welfare outcomes Ayalew (2000) also concludes that the positive effect of female control of household resources on favoring basic needs seem to disappear when a panel of households is considered

In conclusion, the effect of female head on welfare of the household, especially in food consumption and nutrient intake is still unclear Therefore, whether such effect exists depends on the data and definition of household head in the survey

2 3 2 3 Calorie intake and female education

Female education should be focused on estimating nutrient demand of households because female are often responsible for food purchase and preparation It is expected that females with higher education may improve nutrient intake of members within a household However, this effect seems controversial because it is not significant in the study of Behrman and Deolalikar in 1988 for the rural South Indian sample, but it is positive in the study of Wolfe and Behrman in 1983 for the Nicaraguan sample (Behrman, Deolalikar and Wolfe, 1988) In general, effect of female educatio~ on nutrient intake is difficult to measure because female education depends on many subjective factors such as ability, motivation and expenence

2.4 CONCLUSION

Nutrient, like other consumer goods, is considered to follow the theory of consumer choice, in which consumers face the budget constraint and their indifference curves However, because food and nutrition surveys are often conducted in terms

of households, to examine effect of income level of households in nutrient intake of members in households, it is necessary to consider the theory of intra-household allocation Besides, some economical laws such as Engle's law should be reviewed

to determine spending trends of consumers in food and nutrient, so that to outline tastes of consumers in these fields As a result, the two theories and those economical laws will be combined in analyzing data in the thesis

There are many studies focused on the effect of income or expenditure on food

'

consumption and nutrient intake of households nowadays because nutrition problems seem to be increasingly important, especially in developing countries It