Contents - Previous - Next

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

I. Conceptual frameworks

1.Conceptual frameworks
2. Programme interventions, intra-household allocation, and the welfare of individuals: economic models of the household
Person-specific prices and goods-specific prices
3. Peeking into the black box of economic models of the household
A theoretical peek into the black box of economic household models
4. Intra-household allocation of resources: perspectives from anthropology
Social distribution of rights and responsibilities
5. Intra-household allocation of resources: perspectives from psychology
Distribution Rules : Rules of Social Exchange

Conceptual frameworks

Tufts University School of Nutrition, Medford, Massachusetts, USA

Office of Nutrition, US Agency for International Development, Washington, D.C., USA

The papers in part I present three frameworks for studying the household, derived from the disciplines of economics, anthropology, and psychology. Rosenzweig, in his paper, stresses the importance of constructing a model of the household which contains measurable variables, and which leads to the development of testable hypotheses about the determinants of different patterns of intra-household allocation. His economic model of the household rests on the concept of a "unified household preference function," a model of household behaviour that suggests that the household, at least from the outside observer's point of view, allocates resources among its members according to some jointly held set of allocative rules.

Both Messer and Engle criticize the unified household model as a misrepresentation of reality: they argue that allocation rules are the result of conflict and conflict resolution based on different members' power and influence within the household. In fact, the unified household model does acknowledge the importance of the varying characteristics of each member. Moreover, it does not suggest that the behavioural outcome, a set of implicit rules for the allocation of resources, is reached without conflict. The model simply does not address the interpersonal dynamics by which the household preference function emerges. It does, however, include individual-level variables which have been identified as important determinants of household preferences, provided they are measurable characteristics (such as earning power). Rosenzweig acknowledges that the idea of a joint household welfare function "does violence to reality," but suggests that the conflict resolution model must be shown to predict distinct outcomes before it can prove its value empirically.

Behrman's comment on Rosenzweig's economic model introduces the idea, developed further by Engle, that it would be useful to identify the allocative rules for different types of households and individuals, and what determines the choice of a rule. Behrman distinguishes allocation based on equity (each member is entitled to a fair share of any resource), equal outcome (resources are distributed in such a way as to equalize the welfare of members, so that the least endowed member might receive the most), fair return (each member is entitled to resources in proportion to his/her material contribution to the household), and maximization (members most likely to benefit from a resource are given the largest share). This latter rule may produce the greatest inequality among members, while maximizing returns to the household as a whole. Moreover, different rules may be followed for different resources (e.g. health care may be allocated equally among children while only the brightest child might be sent to school), and households facing different kinds of constraints may follow different rules. For example, households under severe resource constraints may try to maximize the household's resources at a cost to the welfare of some members, while households with relatively abundant resources may follow an equal-share or equal-outcome strategy.

Messer, writing from the perspective of anthropology, stresses the importance of understanding household behaviour from the point of view of the people being studied (the emic perspective), as well as in the outsider's, objectively measurable terms (the etic perspective). Messer and then Engle argue that the subjective meanings inherent in behaviour must be taken into account in project planning because they affect the ways in which households and individuals respond to a changed environment. They hold that knowledge of how the local population perceives a given activity or resource will be invaluable in predicting whether people will accept the changes effected by a project, and whether new resources will be used in the manner anticipated by its planners.

Messer further cautions that cultural rules and practices, and culturally determined systems of rights and obligations, condition the acceptance (or rejection) of new forms of behaviour and new relationships resulting from project-induced changes. The culture is the context into which programmes are introduced. Programmes which distort existing patterns (e.g. introducing a new technology which results in devaluation of the older generation's knowledge and skills) may engender resistance or, worse, may disrupt the culture in profound, unanticipated ways.

The economic approach to analysis typically depends on a model of behaviour that has already been formulated. The ethnographic method is less structured, more openended. This approach, commonly used by anthropologists, is therefore particularly well-suited to identifying variables whose importance might not have been revealed without a sensitivity to the culture. Cultural constraints on behaviour are, in fact, concrete and measurable variables. Once their existence has been documented, it is possible to incorporate them into a more formal model such as Rosenzweig's. For example, certain categories of people (the elderly, women, children) may be prohibited from engaging in specific types of labour. Returns to human capital such as education may also differ among groups. These cultural variables can be modelled by introducing, for instance, separate imputed wage rates for males and females of equal skill level, or by distinguishing several categories of market labour time. The anthropological approach clearly does not conflict with the economic, but complements it by focusing attention on important variables which may have been missed in a purely economic framework. Including these variables ensures a more accurate and complete model of household behaviour.

Engle emphasizes that the important contribution of psychology is the recognition of individual psychological characteristics as determinants of behaviour. She posits that personality may be a significant factor in both control over resource allocation and access to consumption goods. One empirical application of this perception is that personality characteristics themselves may be affected by externally introduced changes. If, for instance, women are given income-earning opportunities outside the home, they may develop greater self-esteem, which in turn may increase their ability and willingness to influence allocation decisions.

Engle stresses the conflict-resolution model rather than the unified household model of intra-household allocation, noting that household priorities may be influenced by individual members in a variety of ways. Members with the power to enforce their preferences may coerce, but household members without authority can still exercise influence through their own special knowledge and expertise, by persuasion, or by quiet resistance. Engle also distinguishes between the power to influence allocation decisions and the access to consumption goods and productive resources.

Both Messer and Engle discuss the widely held notion that resources under the control of women are more likely to be used to enhance child welfare (through the purchase of food, health care, and other basic needs) than are resources controlled by men. Engle offers a psychological explanation for this phenomenon, based on the closer relationship and greater degree of bonding between mothers and their children than between fathers and children. She cautions, however, that this theory has not yet been empirically tested in a rigorous manner.

These papers taken together underscore the need for integrating a wide variety of types of information in order to achieve a complete understanding of patterns of intrahousehold resource allocation. Rosenzweig emphasizes that models of the household are only useful if they help to predict behaviour. If two alternative explanations of an observed phenomenon are equally plausible and cannot be distinguished in practice, then, he suggests, it does not matter for planning purposes which one is in operation.

Both Engle and Messer suggest, to the contrary, that comprehending the subjective perceptions of a target population is essential precisely because these perceptions do affect behaviour. Moreover, cultural perceptions of mutual obligations may result in dispersion of some programme benefits beyond the target household or individual for which or whom it was intended. The psychological consequences of changing accepted lines of authority in the household may lead to resistance, conflict, or even violence. The outsider's (etic) perception of efficiency in the use of a resource may conflict with the native's (emic) perception of individual rights to its use. In such cases, it is essential to know both perspectives in order to assess the desirability and the likely adoption of a change.

The economic and material, the cultural, and the psychodynamic perspectives each furnish elements instrumental to the analysis of intra-household allocation processes. Operational definitions of psychological and cultural constructs may be difficult to derive, but recognizing their importance in shaping intra-household behaviour is the first step in solving the measurement problem. The very different approaches implied by these papers are not irreconcilable. On the contrary, they complement each other in developing a complete understanding of current intra-household allocation practices, and of the likely directions for change.

2 Programme interventions, intra-household allocation, and the welfare of individuals: economic models of the household

Department of Economics, University of Minnesota, St Paul, Minnesota, USA

Many policy initiatives and programmes have as their objective the improvement of the welfare of individuals. Such policies and programmes alter the environment in which agents make decisions, thereby affecting resource allocations, and/or directly transfer resources across groups or individuals. Almost all individuals, however, reside within multimember entities, families, or households;1 moreover, many of the most important components of well-being, such as health and education, are principally determined within households. To the extent that the household or the family is the ultimate decision-making unit, it is necessary to understand how households allocate resources among their individual members and across household activities in order to predict how policies will affect individuals. The household is thus an important intermediary between aggregate policies, local programmes, and the individual.

In this essay, I will review what economic models of the household have to say about the effects of various policy-relevant interventions on the intra-household distribution of resources, and thus on the welfare of individuals. I also review empirical evidence, informed by such models, on these issues. To focus the discussion, I will pay particular attention to the within-family allocation of two household resources - time and food and to three individual-specific outcomes- schooling, earnings (labour supply), and health status. Policies and/or programmes considered are: direct food or nutritional supplements provided to individuals, female employment-creation schemes, interventions that have as direct or indirect consequences the alteration of relative food prices, and subventions of health services or schooling.

Most of the discussion will be devoted to household models that (1) assume that the size and composition of family members (with the exception of the number of children) are exogenously determined and (2) treat the household as if it were maximizing a single welfare function. Clearly these assumptions do violence to reality. However, these models are the most developed, have seen the most rigorous empirical applications, and, most importantly, provide a clear foundation with which to begin to develop a more comprehensive model of family behaviour. Moreover, most policies and programmes effect marginal changes in the environment that are not likely to alter significantly the structure of households. For predicting the consequences of such policies, as listed above, the models appear useful. For predicting the long-term consequences of economic development, more attention would need to be paid to understanding the determinants of the size and composition of families and/or households and the processes of household decision-making (Pollak, 1985; Rosenzweig, 1988; Safilios-Rothschild, this volume).

Whatever the limitations of existing economic models of household behaviour, they do represent a comprehensive framework for describing and predicting important aspects of intra-household resource allocations. Policy evaluation, data collection, and data analysis cannot be executed well in the absence of an organizing framework. Collections of "precautions," "insights," and/or examples are surely not sufficient for discerning what data to collect, what statistical techniques to use in order to learn something from the data that inform policy, or what consequences to expect from policy interventions.

The first part presents a prototypical economic model of the household, incorporating multiple family members with differing characteristics, household production, labourmarket relations, and time and income constraints. The equilibrium of the household is described in terms of the optimal intra-household allocation of goods across individuals and household activities. The second part assesses the effects of in-kind or pecuniary income transfers to the household on the intra-household distribution of resources as implied by the model. The third part investigates the consequences of employmentcreation programmes and of alterations in relative food prices on household time allocation and on the distribution of household human capital investments (schooling or health). Datasets describing household behaviour in India, the Philippines, and Indonesia provide empirical examples. The fourth part discusses the implications of intra-household allocative behaviour for discerning how household resources affect such individual outcomes as health and earnings potential when certain traits of individual household members are not measured. The final section briefly recommends future research directions and identifies priorities in data collection for policy-relevant research, as implied by the economic framework.


The economic model of the household has three central features: (1) the household welfare function, describing how the household decision-makers value the outcomes of the allocation of household resources; (2) the household technology, describing how income allocations - goods and time - affect outcomes valued by the household, such as health, earnings potential, and a clean house; and (3) the household budget constraint, describing the total resources of time and goods the household can command. No allocation of resources can violate this constraint.

Consider a household with n members. Each member i consumes food xi, spends some time in leisure Ti, has a stock of human capital Hi (e.g. health, earnings potential), and receives a wage rate wi in the labour market. The amount of human capital for individual i is determined by the food he or she consumes, other goods Zi (e.g. schooling or medical services), and the time he/she and other family members allocate to given activities tj, j = I, . . ., n. Each individual i also has an exogenously given set of characteristics i (personal ability, gender, etc.) that possibly affect the productivity of inputs. The relationship between household or individual inputs, person-specific characteristics, and the human capital for each family member is given by the production function:


The wage paid to each family member i also may depend on his/her human capital Hi and on personal characteristics (i), such that:

Wi = riHi (2)

where ri = r(i).2

Household resources may also be used to produce a set of commodities G. consumed jointly by all household members, including household sanitation, cleanliness, or meals. Such household "public" goods employ the time of individual family members as well as purchased goods Y; i.e.

G = G(Y;tGl, . . . ,tGl), (3)

but all household members benefit equally from the production of those goods.

The household decision-makers care about individual food consumption, the level of human capital, the amount of leisure time, and the earnings potential of every household member and the level of the public good G(3) The household welfare function is thus:

U = U(Xl, . . ., Xn;Hl, . . ., Hn;Tl, . . ., Tn;Wl, . . ., Wn;G) (4)

If each family member i has W amount of time to allocate among all household production activities Ti H. work Tiw, and leisure Ti, then the household budget constraint is:


where px is the price of food, pz and py are the prices of other goods, and Vi is the wealth income of each member i.

This model ignores financial bequests from parents to children that are the focus of attention in Becker and Tomes (1979) and in Behrman et al. (1982), but incorporates the pecuniary contribution of children to family income, as in Rosenzweig and Schultz (1982). However, the implications of the model are retained even if parents make bequests or financial transfers and care about the total income of each family member, as long as they also care, as here, about the health, earnings potential, and/or food consumption of each member.

The household is assumed to allocate all resources across activities and individuals in order to maximize the welfare function (4), subject to the technology (1) and (3), earnings function (2), and budget constraints (5). For given prices and the set of endowments of all family members, the optimal allocation of food, for example, across any two family members i and j is given by:


where A is the marginal utility of wealth.

Expression (6), while formidable in appearance, merely says that the marginal returns to food allocated to each individual in terms of family welfare dU and earnings TiWdWi be equal for each individual and equal to the (marginal) cost per unit of food (the food price px). Food will thus be allocated equally among family members only if exogenous endowments or characteristics i are the same for all members. Differences in market returns r(i) to human capital, in endowment levels , in returns to human capital inputs , or in employment Tiw among family members will lead to an unequal intrafamily distribution of resources. Rosenzweig and Schultz (1982) used a simplified version of this model and showed that where opportunities for female employment and earnings were exogenously higher, female infants in India had higher survival probabilities and, by inference, received a larger share of household resources.


Any change in the constraints, technology, or prices facing the household will induce the household to reallocate resources in order to conform to the optimizing allocation, of which expression (6) is one condition. More direct interventions will also lead to reallocations. Consider a programme that provides food at no cost to an individual i, such as in a school-lunch programme. Given the usual neo-classical properties of the welfare function (diminishing marginal utility), such an intervention will lower the left-hand side of (6) relative to the two right-hand side expressions (individual i will be more well-off than any other individual j compared to his/her condition before the intervention); the family will thus attempt to correct this imbalance and increase family welfare by reducing familycontrolled resources given to individual i and redistributing those resources to other members j, in order to maintain the equalities in (6). If family resources are perfect substitutes for the resource provided in the programme, the person-specific subsidy will thus be no more successful in raising individual i's food consumption, health, or welfare compared to other family members than would a programme providing an equivalent amount of money to the family as a whole.

In the neo-classical model of the household, in which all resources are essentially pooled and reallocated to individuals, it thus does not make any difference to whom income subsidies are given or, for the most part, whether such subsidies are in cash or in kind. The effect on the intra-family consumption of resources will be the same and will correspond to that of a general income effect. This strong prediction - the irrelevance of the source of income transfers (as opposed to earnings) - is in sharp contrast to models that emphasize conflict among family members and depict the household allocation of resources as the outcome of a bargaining process (cf. Engle, Safilios-Rothschild, this volume). In such models, a rise in assets owned or resources received by an individual i from extra-family resources increases that person's relative bargaining position or power (McElroy and Homey, 1981; Folbre, 1984). Such models, while more realistic, have not provided clear directions, however, as to how intra-household allocations will differ when some individual attains more bargaining strength (Rosenzweig and Schultz, 1984); nor has there been any clear empirical rejection of the implications of the unified household model. Whichever model is adopted, unless each individual in the family has full control over all resources (not true even in household bargaining models), person-specific transfer programmes which provide resources to an individual may not succeed in effecting a significant net redistribution of resources to that individual because of family redistribution rules over which policy-makers have no direct control.

Contents - Previous - Next