Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

Food policy

Micro-economic theory of the household and nutrition programmes

Dov Chernichovsky and Linda Zangwill


It is now generally accepted that malnutrition and hunger are problems of distribution rather than of production, and of households rather than of economies [1]. The supply of food is still an issue in many developing areas, those of sub-Saharan Africa in particular. The overall growth of this supply still does not match the rate of population growth [2] and may not match it for some time [3]. Yet lack of global food supply has ceased to be the major cause of malnutrition. Three important developing countries, China, India, and Indonesia, whose combined populations constitute the bulk of the human race, appear to have enough stocks of food energy to feed their people.

Indeed, economic growth and development, with related macro-economic policy, that are not offset by population growth can secure availability of food in the long run. Even then, however, the risk of malnutrition may persist at least in the short run, requiring appropriate policies and interventions. Many households and individuals remain malnourished even where there is an overall adequate supply of food.

Various household factors are associated with the risk of malnutrition: size and composition, command over human and non-human resources, environmental conditions, and a host of cultural and social attributes. These affect households' access to food, the way they use it, and how well food is absorbed biologically. The distribution of these factors in the population usually determines which and how many households are at risk of malnutrition, the magnitude of the problem, and the resources that may be required for its solution.

Any increase in household resources, whether through growth and development or policy programmes, stops at the household; the family can allocate these added resources in any manner it sees fit, and often in ways that are incompatible with improved nutrition and related policy goals. In fact, families do not necessarily buy efficient diets from a nutritional perspective. Malnutrition is therefore not just an income problem, as many if not all households can afford a technically defined minimal diet when food is available [4]. Economic development and social policy can affect households by changing tastes and attitudes, incomes, prices, and even family size. How these in turn affect health and nutrition remains in many ways unclear.

If policies and programmes are to succeed, they must consider household behaviour. This, combined with programme design and operations, is a major determinant of programmes' impact and hence internal efficiency - how much is gained per unit of resources. Moreover, nutrition policy and programmes have potential benefits in other sectors, such as education; meals in school may increase school attendance. Knowledge of the household's response to nutrition programmes can thus help in evaluating the social returns. Many policies and programmes to date are designed and implemented on the basis of limited knowledge of this response, which is a major determinant of internal and social efficiency.

The economic theory of the household and econometrics portray and measure household behaviour in response to external stimuli such as those generated by market forces and policy interventions. The objective of this paper is to outline the potential contribution of this theory, through a series of hypotheses and their empirical testing, to better policy making and programming. This will serve the following functions: (1) establishing which determinants of risk should be monitored to anticipate malnutrition problems that are not related to overall food supply; (2) targeting interventions according to the hypothesized or observed determinants of the risks rather than on the basis of costly screening; (3) deciding whether to follow a health, nutrition, or combined policy to improve nutritional status; (4) designing appropriate intervention; (5) evaluating the household's response to the intervention; and (6) evaluating programme impact.

An economic view of the household

The economic theory of household behaviour is a theory of choices. It focuses on the household's responses to changes in mostly external factors as a way of increasing or protecting the welfare of its members.

The (new) theory of household economics views the household as a harmonious microcosm or entity that shares the same resources and aims to increase its utility or welfare through the production and consumption of commodities such as good health, and aesthetic and gastronomic utility from food. Home-produced commodities are distinguished from market-purchased goods. By viewing the household as a production unit rather than just a consumption unit, this theory (in contrast with the traditional theory) also permits us to deal with behaviour concerning the production and consumption of non-market commodities, such as health. In addition, it enables us to deal with farm households, which are common in developing economies and which often combine food-consumption decisions with food-production decisions [5]. The household produces these commodities by combining goods and services purchased in the market with time inputs and skills of its members [6; 7].

Real income and available time limit the household's ability to increase its welfare level. The first constrains ability to buy goods and services in the market. The second limits ability to produce income through labour (when work is available) and household commodities through home production.

For households that depend on wages and income from capital assets for their livelihood, variations in wage rates and in interest and rental rates alter their nominal income, or money value. Similarly, variations in market prices of goods and services change the households' real income, or how much they can buy in the market place with a given nominal income. Relatively higher prices, for example, reduce real income.

In contrast, households that sell or consume their own produce benefit from higher prices of their products and lower prices of inputs. Consequently, at any given moment their command over market goods and services is determined by their own production' market prices and money wages, and interest and rental rates.

Behaviour is interpreted as the allocation and re-allocation of scarce resources among competing utilitarian objectives or commodities whose consumption the household strives to maximize. The allocation of any given level of resources, available through time and income, toward meeting competing ends depends on the opportunity cost, or shadow price, of attaining any such end. The shadow price of diet encompasses the market value of the foods for which other goods and services could have been bought, and the value of time invested in food preparation that could have been used elsewhere. This implies that the shadow price of a commodity, such as a particular diet, may increase with a rise in the market price of goods used for it, and the wage rate or any other variable that would increase the value of time employed in its production. For example, as a household's income rises, especially through wages, the value of its members' time also increases. It is, therefore, inclined to spend less time on food preparation, for example, by eating more processed foods, employing others to cook, and eating in restaurants. Higher market prices are likely to raise the relative shadow price of those commodities that are relatively intensive in market goods, while higher incomes and wages change the relative shadow price of those commodities that are relatively time-intensive [6].

Behaviour is viewed as a result of two effects: income and substitution. The income effect leads to more consumption of all so-called normal commodities when real income rises. The substitution effect induces more consumption of those commodities whose relative price has declined. At times, the two effects induce conflicting behaviour. For example, in the case of people who grow their own food and are net sellers, an increase in prices of produce induces an income effect in favour of more food purchasing. The substitution effect induces the opposite, because selling the produce rather than consuming it is more rewarding financially when prices are higher. Only an empirical analysis can determine which effect dominates, or how those households would actually behave when prices change.

Optimal behaviour suggests the allocation of each additional unit of resources to the activity or commodity that renders the highest marginal utility or gain in satisfaction. Maximum possible welfare from given resources is attained when the allocation of resources from one activity to another does not bring about any net gain in welfare or utility.

Some of the limitations of the economic theory of the household must be highlighted. The theory strives to explain all behaviour: family formation through marriage and procreation, income generation, human and non-human capital formation, and so on. Practically, however, it cannot adequately deal with all behaviour because it is an analysis of the effects of external or predisposing factors on behaviour. The more behaviour it attempts to explain, the fewer remaining predisposing factors it can be based on. Therefore, it deals with so-called partial equilibria. It identifies a subset of behaviour which is the subject of analysis, and assumes other behaviour external to this subset or ignores it altogether. The focus on external factors, largely income and prices, ignores internal issues that may be crucial to resource mobilization and allocation, such as motivation, cognition, and a host of psychological and cultural factors.

The view of the household as a harmonious microcosm is clearly limiting. Economic theory needs to move to understand intra-household resource allocation. While for institutional and cultural reasons some role allocation is known, such as who goes to school or who cooks, discriminating behaviour, such as who may receive more food within the household and why, is still beyond the grasp of economic theory.

Central to the theory is the assumption that the consumer or the household has full knowledge about the values and attributes of its resources and the consequences of their allocation. This is a dubious assumption, especially with regard to health and nutrition. Most households cannot be expected to know the nutritional value of the food they consume and the health consequences of their behaviour.

These theoretical limitations are well recognized by students of household economics, and are dealt with to a substantial extent by econometrics, the empirical theory complementing economics. Central to econometrics is the notion that some factors explain behaviour across households, and over time are unrecognized, at least by economic theory, unobserved, or simply incorrectly measured. That is, residual behaviour cannot be explained by theory but can be handled in its empirical testing. Therefore, non-economic variables such as religion, ethnicity, and location that affect behaviour are incorporated in the empirical study as control variables that qualify the effects of the economic variables, but in ways economic theory cannot always predict.

Econometrics also deals extensively with interdependent circular behaviour or simultaneous relationships. It can establish the direction of the simultaneity bias associated with disregarding such relationships. For example, in low-income settings, the income determines the level of food consumption, but food consumption may determine levels of energy and income. Disregarding this simultaneous relationship in the estimation of, say, the effect of income on food consumption would produce upward-biased estimates of that effect. Econometrics suggests mathematical and statistical solutions to such interdependence that can substantially improve microeconomic research in nutrition.

The economics of household nutrition

Basic relationships

The nutritional and health status of an individual is based on the complex interaction of genetic, behavioural, and environmental factors on the intake and absorption of nutrients. In addition, since the intake and absorption of nutrients are affected primarily by the presence or absence of disease, nutritional status is largely affected by health. Thus a strong synergistic relationship exists between infection and food absorption and vice versa. A general diagram of these relationships is presented in figure 1 (see FIG. 1. A schematic outline for organizing research in nutrition), where they are separated into possible topical study areas (left) and their relevance to policy making and programming (right).

Health and nutritional status are determined by food, health care, housing, and hygienic practices (topical area 1). which are in turn affected by market prices, incomes, family size and composition, education, and other taste variables (topical area 2). These are all affected by economic development and growth. as well as policy. Policy and programming would naturally follow issues I through III listed at the right of the figure.

Although it may be difficult to single out the effects of each factor on nutritional status and health, it is important to try to do so. Such an identification is necessary in order to anticipate, design, manage, and monitor inventions appropriately through the determinants of risk to identify nutritionally at-risk households.

Although no single definition of nutritional risk exists, it can be considered "the chance of death, ill health, malfunction, poor achievement in body size or hunger due to insufficient food'' [8]. In that light, we specify a set of structural relationships that are assumed to portray the nutritional and health aspects of household behaviour. In the paradigm of partial equilibria, the micro-economic study of nutrition has focused on several critical relationships. These relationships, as portrayed in figure 1, depict an economic view of common factors affecting diet and nutritional status:

- income and prices - purchasing power and food availability in the household (famine is not considered because it is beyond household control);
- tastes - e.g. food preferences - education, etc.;
- family size and composition - per capita purchasing power and food availability;
- food consumption - quantity and quality;
- health care and practices;
- environment;
- development and policy.

The type of intervention that will be most efficient in alleviating malnutrition depends on whether, and to what extent, these causal factors contribute to the problem at the household level. This would help identify the means and the social cost of the intervention. Clearly, many interventions aim at particular members or groups within the household, such as children and pregnant women. In this regard we lack a clear theory that can predict behaviour that would affect programme efficiency.

We start with a household utility function (u) that outlines the behavioural aspects that the household wishes to maximize and that are relevant to the discussion. They are health (H), nutritional status (NS), diet (D), and all other utilitarian commodities (Z) as well as leisure time (Tl). The last two are not of direct concern to this discussion. That is,

U=u(H, NS, D, Z, Tl). (1)

This function, which is not directly observable but inferred from behaviour, determines how much the household values different commodities at different levels of consumption. It usually assumes that the additional or marginal gain in utility falls with increased consumption. This is the economic formulation of the sense of approaching saturation.

The second relationship concerns the production of the diet:

D = d(Xd, Td, NS; E ). (2)

Household diet (D) is produced through a vector of market goods and services (Xd), which include foods, appliances, and so on, and the time (Td) needed to prepare it. This function can be spelt out in terms of the probability of being malnourished or at risk of malnutrition. In that case, D would be a qualitative (dummy or categorical) variable standing for being below a particular level of nutritional requirement. In addition, the level of the diet is assumed to be conditioned by the nutritional status (NS) of household members, as can be estimated by their heights and weights; for example, heavier and taller persons may require more calories than lighter, shorter ones. The production of the specific diet is also determined by environmental variables (E) such as ethnicity, tradition, and homemaker's education, which may determine food preparation patterns; educated homemakers may avoid overcooking to prevent loss of

food nutrients. This relationship refers to the lines marked A in figure 1.

The third relationship deals with the determination of nutritional status:

NS = n(D, G, H). (3)

where NS is assumed to be determined by diet (D), pertinent genetic factors (G), and health (H), as indicated by lines B in the figure. Health is believed to determine the efficiency of the diet in the production of NS. For example, disease may limit the absorption of nutrients.

The fourth relationship concerns health:

H = h(NS, Xh, The; E ). (4)

Good health is assumed to be produced by nutritional status (NS), goods and services (Xh), such as medical care, and time (The). Here again, production can be conditioned by environmental variables (E): education of household members as well as community-level variables such as safe water and sanitation. These are outlined by the family of lines marked C in figure 1. As NS and H are stocks, compared with the flow of the diet, it is often common to use recursive models where the stock of period t is determined by, among other things, the stock in the previous period, t- 1, e.g. Ht = h(Ht - 1 . . .). This approach would lead to inclusion of initial endowments, e.g. birth weight in a nutrition status equation, especially of children [912]. This approach is not taken here? as we wish to keep the discussion simple without too much loss of generality. Related statistical issues are beyond the scope of this paper. While most analyses use cross-sectional data which are more readily available, panel data, preferable generated under experimental conditions, would be more appropriate for measurement of the relationships discussed here.

Equations 2 through 4 outline periodic flows of food consumption and diets, and accumulated stocks of health and nutritional status produced over time. The synergistic relationship between health and nutritional status is depicted in equations 3 and 4.

Apart from the diet, the household enjoys other commodities (Z). The production of these is depicted by

Z = z(Xz, Tz; E). (5)

That is, Z is produced by market goods and services (Xz), household members' time (Tz), and pertinent environmental variables (E).

The next three relationships deal with income and productivity of household members. A farm household can be characterized by a farm-production function:

Q = q(Tif, A, S, NS, D ), (6)

which links household resources with the product (Q) it produces through a particular technology. Q is stated here in general terms to include food cultivation and may stand for more than one product. It may be a composite product made up of several goods with adjustment for their relative prices. This product, which can be sold for the price P, is produced by the labour, the time (Tif) household members devote to work on the farm, physical assets (A) (e.g. land and equipment when they apply), skill levels (S), nutritional status (NS) when physical strength may be required, and the diet (D) largely as a determinant of energy levels which may determine productivity. In addition, family members can work part of their time (Tiw) as employees for wage rate (W) and earn (WTiw) in wage income.

Household income may vary not just because of changes in household resources but also because of changes in farm technology and market conditions: improved marketing systems, farm prices, and higher wages. All can increase family incomes with identical resources.

To the income produced by the household, transfers or resources given to it by social programmes (V) are added. These are obtained by

V = v(Xv, Tv; E). (7)

indicating that the household can obtain such transfers through investment of some of its own resources (Xv) (e.g. school uniforms, transportation, etc.) and time (Tv) and environmental variables (E). If the cost of these exceeds the perceived gain from the transfer (V), the household will not participate in the programme.

To close this system of relationships, two resource constraints that limit household production and consumption possibilities, are identified. The first is the income constraint:

I = PQ + WTiw + V = Pxd Xd + Pxh Xh + Pxz Xz + Pxv Xv. (8)

This relationship indicates that the household's income from all sources, own production, wages, and transfers, is exhausted on all goods and services purchased in the market: foods and related goods and services (Xd), investment in health (Xh), and goods and services for use in all other commodities (Xz), as well as for use of public programmes (Xv). The second constraint is time:

T = Td + The + Tz + Tv + Tl + Tif + Tiw, (9)

which indicates that the household's time endowment is allocated between labour (Tif, Tiw), on the one hand, and household production of D, H, Z, and V, and leisure, on the other.

While equation 1 determines how much the household values the different commodities, equations 2-5 and 7 determine how much it would cost to produce them, subject to resource availability determined by income, time, and market wages.

Of the above, D, H, Z, and I are choice variables, and the relationships whereby they are determined are behavioural. In other words, the household must decide what levels of scarce resources it allocates to the production of any of these. Given the contribution of each commodity to its welfare and the cost of achieving it, the household decides how much it will produce of each. Thus diet (D), health levels (H), and nutritional status (NS ) are co-determined by choice.

There are numerous ways by which even this relatively simple set of relationships could become complicated, making it a more realistic portrayal of reality, but probably less manageable analytically. For instance, days worked (Tif) or working time (Tlw) could be related to health and nutritional status. Work could be assumed to be a determinant of NS, for example, inasmuch as deficient energy for physical activity may reduce body weight. But we may not be able to solve or establish how particular variables are determined even in this relatively simple model, because of the limited number of predisposing variables it assumes at any particular time - G, S, A, T, and E - compared with the number of endogenously co-determined variables, H, M, S, D, Z, and T. Eliminating NS and D from equation 6 can facilitate a solution at the expense of assuming that NS and D do not affect productivity and income. While such a trade-off is probably of no consequence in well-nourished populations, it might be significant in malnourished populations. This exemplifies the importance of taking into account the nature and environment of the population under study and the specific objectives of the study.

Any of these structural relationships can be estimated separately. All should and, under particular conditions, can be estimated together because of their interdependence. An example of related estimates is given in table 1 [9]. Various measures of NS are codetermined with health (colds) by C and E, which stand for a host of socioeconomic and environmental variables. The estimated parameters, even when biased, are crucial to programme and policy formulation for diverse populations.

Specific relationships are discussed below in more detail for their policy and programme relevance.


Analyses of household consumption, including food, is one of the oldest and most established of economic analyses. Equation 2 is an association between the diet and how it is prepared. It includes the level and composition of the diet, given particular inputs, and implicitly its shadow price, reflecting how it is produced.

The approach taken here views the demand for food as derived from the demand for a particular diet and the gastronomic and aesthetic utility of food. This formulation [7], which is central to the theory of the (new) household economics, may be quite debatable. It assumes that consumers have full knowledge of the nutritional value of foods. This is a strong assumption that is not borne out by some research [13] and would clearly be questionable in developing economies. As argued below, the approach is nonetheless useful for understanding and predicting how the household produces its diet in view of changing market conditions.

For practical purposes, however, this assumption is not central to explaining the composition of food consumption and the diet. From policy and programme perspectives, it is important to know what people consume and explain this behaviour by variables that can possibly be manipulated, such as incomes through wages (W), transfers (V), and prices (P). As there is a strict linear relationship between foods purchased by the household and their dietary value, we can look at the chosen items, derive the diet, and explain the correlates of this choice, either of foods or of nutrients.

That is, we can deal with food or diet consumption as identical choices and work around the traditional consumption analysis, where

F = f(I, Pf, Po; E), (10)

where l is household income, Pf is a vector of food prices, and Po is a vector of prices of other related goods and services (e.g. appliances). E would include the homemaker's education as a proxy for the value of her time and hence as a determinant of the diet's shadow price. As a particular vector of D can be produced by many food combinations and in many ways, the household, given its taste for food and other items, will choose the least costly diet with its endowments and production technology.

The major objective of this analysis is to determine the effect of household income or expenditures, food prices, and other relevant variables on food consumption and the diet. This effect is customarily measured in terms of so-called sensitivity or responsiveness measures - income and price elasticities.

Income elasticity (the percentage change in quantity of food consumed as a result of a given percentage change in income) is made up of two parameters: (a) the share of expenditures on foods in income (PfF/I); and (b) the marginal propensity to consume [MPC = ((D PfF)/(D I ], or the change in expenditures on foods ((D PfF) that follows a particular change in income ((D I). The higher the MPC for food, the more the added spending on food with a change in household income. For example, an MPC of $0.60 would imply that from each additional dollar in income the household would increase food consumption by $0.60. Or, to induce the household to raise its expenditures on food by $1.00, its income must be raised by about $1.67. The relative effect of a change in income is higher, the higher its income elasticity.

One of the basic laws established for food consumption is Engel's Law, which states that, while food consumption rises with income, the share of expenditures on food falls, because the MPC for food declines as income rises and there is a saturation process with regard to food.

From a programmatic viewpoint, the higher the MPC for food, the higher the impact of an income transfer. This value may depend on the permanence of the change in income, its source, and who in the household receives it. A clear distinction is made in economics between the MPC from a transitory change in income and that from a permanent change. The MPC from the former is lower because the household does not adjust long-term consumption patterns to a transient change in income. It may adjust consumption only to a fraction of that change. Consequently, a change in income from a source of permanent nature will bring about a higher MPC. In addition, income received in kind - in food, for example - will result in a higher MPC for the food, because the household cannot exchange this food for other commodities as easily (and for the same value) as it could with cash. It is also argued that income received by women induces higher expenditures on food than that received by men [14].

The (own) price elasticity (the percentage change in the quantity consumed of any food as a result of a percentage change in its price) is determined by two effects related to income and substitution effects. When prices of particular goods rise, consumption will fall, because higher prices mean lower real incomes, hence the income effect, and a shift away from these foods for substitutes whose relative prices are lower, hence the substitution effect. The effect of a rise in the price of one good on the consumption of others is measured by cross price elasticities. It can be shown that when the price elasticity of a commodity is low, as may be the case for basic foods, an increase in price will result in a decrease in other consumption as well.

The appropriate income and price elasticities for specific nutrients with respect to income and food prices can be established. The change in consumption of a particular nutrient with regard to a change in income or prices depends on the income or price elasticities of the foods and the contribution of any particular food item to the total consumption of that nutrient. Further information on technical relationships and variations can be obtained from the authors.

Much can be said about the relative magnitudes of income elasticities from general knowledge. For instance, relatively low-income groups are likely to have high shares of expenditures on foods and high MPCs, usually leading to high income elasticities and low price elasticities for basic foods that have no substitutes. Staples such as rice and wheat are likely to be major contributors to calories and protein in low-income populations and thus their consumption is sensitive to changes in prices. At the same time, the actual values of the elasticities are a matter of empirical evaluation.

Table 2 (columns 2-4) and table 3 (last three columns) exemplify estimates of income elasticities for foods and nutrients based on Indonesian data [15]. It is noteworthy that, while the estimated income elasticity for rice falls with income, it rises for dairy products. The elasticities fall, however, for most nutrients, but their levels of consumption rise with income (see table 3, columns 2-4).

In addition to the quantitative composition of the food basket, there is likely to be a qualitative change in food consumption as income changes. This change may take several dimensions. Food items may be of different nutritional quality. They may also require different levels of preparation. The last three columns of table 2 indicate' for example, that Indonesian households with higher income pay higher prices for their foods than their lower-income counterparts.

The new economic theory of the household emphasizes one qualitative dimension of food preparation through its preoccupation with the shadow price of a diet that includes also the price or value of time. Whatever causes a rise in the value of time, such as an increase in household income, through employment opportunities and wages of women in particular, will induce time-saving production of diets. This can imply a host of behavioural changes, from the replacement of breast-feeding by bottle-feeding to the replacement of labour-intensive home cooking by appliance-intensive cooking, ready-made foods, and foods eaten away from home. Data from the Philippines indicate that women working outside the home are more likely to initiate mixed feedings by adding breast-milk substitutes after the third month [16]. Additional data. presented in table 4 , based on the Indonesian experience, show that better-educated homemakers, presumably with higher incomes and value of time, tend to have a lower consumption of nutrients, all other things being equal.

Nutritional status

The NS relationship measures or accounts for the way the household produces nutritional status, subject to its genetic endowments, its knowledge, and the private cost of producing nutritional status through health and diet. It focuses on the intervening variables health and diet, through which socioeconomic status affects nutritional status. This relationship highlights the way the efficiency of a diet relates to health status: substantial waste of a diet may result from the presence of disease. In a world of perfect information, knowledge of this relationship would help decide on the optimum combination of diet and health the household or society should choose to produce a particular level of nutritional status. That is, given the shadow prices of diet and health, an optimum behaviour and policy would be to spend any given amount of additional resources on the diet or health that would yield the highest gain in nutritional status (at the margin). At the optimum, the gain in nutritional status from spending a unit of resources on either diet or health should be the same.

Knowledge of this interaction would help to determine whether to follow a health policy or a nutrition policy or some combination thereof. The Narangwal experience in India, for example, shows that the presence of diarrhoea has a negative effect on nutritional status when measured in height . Indeed , a combined nutrition - medical-care programme proved more efficient than free-standing interventions [17].

TABLE 1. Structural equation estimates for children's growth and health three-stage least squares (N= 2.515)

Independent variables

Dependent variables

Genetic factors (G)

Health factor (H), colds
Height Weight Head circumference  
Dietary factors (D)
protein 0.087 (4.39)     0.22 (11.76) 0.01 (1.49)
calories     0.002 (5.02)        
vitamin C             0.011 (0.10)
Genetic factors (G)
age 0.84 (31.52) 0.09 (5.71) 0.24 (8.57) -0.17 (-8.60)
age squared -0.002 (- 9.45) 0.001 (5.10) -0.002 (-7.10) 0.001 (6.91)
sex 0.48 (2.63) 0.13 (1.28) 0.36 (1.84) -0 82 (-6 11)
birth weight 0.003 (8.53) 0.02 (11.06) 0.002 (0.72)    
birth order -0.19 (-3.82) -0.07 (-4 21) 0.10 (3.36)    
mother's height 0.51 (20.80) -0.11 (-0 92) 0.42 (23.96)    
father's height 0.40 (18.11) -0.06 (6.13)        
mother's weight 0.002 (1.03) -0.008 (5.61) -0.01 (-7.30)    
race 1.24 (5.19) 3.10 (2.53) -0.71 (-3.24)    
Income and environmental factors (I, E)
income             -0.001 (-6.44)
household size             0.06 (1.86)
schooling 2             0.72 (-4.90)
schooling 3             -1.40 (-6.83)
schooling 4             -0.79 (22.84)

Source: Ref 9, p 120

TABLE 2. Income-related parameters for food consumption—Java. 1978

  Total expenditure elasticities of demand Proportion of food budget allocated (%) Daily consumption per capita (g)a,b Prices (rupiahs per kg)a ,c
Rice 3.022 0.914 0.034 36.29 36.86 28.01 310.7 346.3 369.3 139 141 149
Corn -0.622 -0.425 0.203 6.41 2.68 1.15 261.6 194.6 155.2 63 68 74
Wheat 0.061 -0.027 0.943 0.82 0.48 0.47 110.1 73.1 49.8 115 121 131
Cassava 0.238 0.790 -0.074 2.80 1.65 0.96 215.2 170.2 139.5 27 27 31
Potatoes 0.539 1.238 1.673 0.77 0.69 0.74 157.7 117.4 76.6 59 67 86
Fish 1.317 1.825 0.979 5.88 6.75 7.24 34.2 41.4 54.8 324 349 419
Meat and poultry 3.948 2.162 2.534 0.86 2. 04 4.94 32.2 30.2 33.1 902 936 994
Eggs 1.143 2.871 2.544 0.59 0.96 1.77 0.16 0.15 0.20 34 38 42
Dairy products 0.076 0.783 2.203 0.14 0.49 1.66 24.6 18.2 20.2 738 733 737
Vegetables 0.953 0.990 0.559 8.04 7.05 6.62 143.4 142.0 158.8 91 105 125
Legumes 2.613 1.991 0.653 2.75 3.14 3.68 41.2 43.3 62.1 166 176 194
Fruit 1.901 3.708 2.617 1.67 2.42 3.71 88.1 92.8 113.0 96 107 129
Other 1.243 0.911 0.696 32.99 34.78 39.04 181.0 211.5 295.5 335 375 651

L = lower 40% expenditure group.
M = middle 30% expenditure group
U = upper 30% expenditure group.
a. For households reporting consumption.
b. Consumption in grams, except for eggs. which arc in units.
c. Prices in rupiahs per kilogram. except for eggs, which are in rupiahs per unit.

TABLE 3. Income-related parameters for nutrient consumption—Java, 1978

  Daily consumption per capita Regression coefficients
Calories 1,747 1,988 2,279 0.789 0.543 0.298
Protein (g) 41.97 49.95 62.90 0.914 0.682 0.424
Fat (g) 23.20 28.63 41.54 1.224 0.952 0.604
Carbohydrates (g) 346 385 418 0.702 0.479 0.218
Calcium (mg) 254 274 349 0.805 0.900 0.611
Iron (mg) 8.56 9.58 11.30 0.759 0.660 0.438
Vitamin A (IU) 5,367 5,337 6,423 0.992 1.535 0.836
Thiamine (mg) 0.76 0.88 1.05 0.933 0.652 0.366
Riboflavin (mg) 0.62 0.67 0.81 0.753 0.642 0.507
Niacin (mg) 12.6 14.6 17.6 0.790 0.559 0.362
Vitamin C (mg) 146 146 165 0.876 1.450 0.820

L - lower 40% expenditure group.
M = middle 30% expenditure group.
U = upper 30% expenditure

TABLE 4. Regression coefficients on the education of the spouse of the head of household, with the consumption of nutrients as dependent variables—Indonesia

  Java Outer islands
Elementary school Junior high school Senior high school Higher education Elementary school Junior high school Senior high school Higher Education
Calories   -0.1397 -0.1939     - 0.1011 -0.1445 -0.2358
Protein     -0.1654       -0.1022  
Fat 0.1245       0.0699      
Carbohydrates   -0.1409 -0.1755     -0.0924 -0.1524 -0.2893
Iron     -0.1800 -0.3482   -0.0727 -0.1028 -0.2113
Vitamin A                
Thiamine   -0.1264 -0.2570 - 0.3920   -0.0982 -0.1078 -0.2610
Niacin   -0.1303 - 0.1802     - 0.0956 0.1199 -0.2022
Vitamin C                


Contents - Previous - Next