With the help of one's neighbors: externalities in the production of nutrition in Peru

Abstract

Both public and private resources contribute to the nutritional status of children. In addition, the investments made by one household may contribute to the health of other households in the neighborhood through improvements in the sanitation environment and through increases in shared knowledge.

This paper measures the externalities of investments in nutrition by indicating the impact of the education of women in Peruvian neighborhoods on the nutrition of children in other households, after controlling for the education and income of those households. We find that in rural areas this shared knowledge has a significant impact on nutrition, with the coefficient of an increase in the average education of women in the neighborhood being appreciable larger than the coefficient of education in isolation.

In addition, we indicate the impact of the water and sanitation environment in the neighborhood, again controlling for the household's own access to sanitation and water. In both urban and rural areas, we observe externalities from investments in such household level infrastructure with the evidence particularly strong for sanitation made by neighboring households.

Introduction

The nutritional status of children reflects the interplay of household level decisions regarding food, health, and childcare. These decisions are conditioned, in part, by a household's access to health and sanitation services as well as its ability to make optimal use of its own and community resources. A number of empirical studies have explored the contribution of such community resources to nutritional status as well as the role education plays in reducing malnutrition.¹ Most often these investigations are framed, implicitly, if not explicitly, in terms of individual resources and public goods.

This current study amplifies such approaches by recognizing an additional contribution to child health through externalities of the investments of other households in the neighborhood, hence the importance of individual resources of other households on the nutritional status of children. In particular, we look at the contribution to nutrition outcomes attributed to the education, water and sanitation investments of neighborhood members that is additional to the impact of a household's own assets.²

Consider, first, the different possible contributions of education on the nutritional status of children. Both male and female education has an indirect impact on nutrition through the relationship of household income and education, with both directions of causality possible. However, while controlling for income may attenuate the measured impact of female education on nutrition or on child mortality, most studies make it clear that impact of female education goes beyond the influence of education on household income (Cleland and van Ginneken, 1988). At least five pathways establishing a direct link between education of the caregiver and their children may account for the additional role of schooling in improving child nutrition: (i) schooling may transmit information about health and nutrition directly; (ii) it teaches numeracy and literacy, thereby assisting care givers in acquiring information, for example, via access to newspapers; (iii) by exposing individuals to new environments schooling makes them receptive to modern medical treatment; (iv) it imparts self confidence which enhances women's role in intra-household decision making and all individuals in their interaction with health care professionals; (v) it provides women with the opportunity to form social networks which can be of particular importance in isolated rural areas. Such social networks can support individual women and families in times of need.³

Some researchers have attempted to determine the relative importance of these (and similar) pathways for either a reduction in malnutrition or child mortality. For example, Thomas, Strauss, and Henriques (1991) determined that education remained important for nutrition in Brazil even when controlling for assets or income and that the principal pathway in their study was via the interaction with newspapers. Glewwe (1999) tested the relative role of the first three of the avenues mentioned above and found that the health knowledge itself contributes the most to child health in Morocco. However, he also observed that this is not part of the school curriculum.

Other studies in the literature have explored the role of information by examining the influence of education on knowledge and specific health related behaviors.⁴ Investigations of the means by which education affect health outcomes also have looked at the type of behaviors that are affected—for example, service utilization or child care practices—and whether education substitutes for service access and quality or whether it complements such services.

While these studies have assisted in identifying actions that may enhance investments in school or in reaching women who have missed out on educational opportunities in their childhood, they differ from our goal here which is to ascertain whether externalities of one woman's schooling diffuses to neighbors. An uneducated woman may obtain specific information important to health care from interaction with one's neighbors. For example, an illiterate caregiver may obtain information by prevailing upon her neighbor to explain the instructions that come on the label of a medicine. Similarly, her cultural attitudes may be modified in the course of such interactions; while schooling enhances receptivity to modern medical approaches this openness may also occur via diffusion across neighbors. Even enhancement of status, which often is a relative position, may benefit uneducated women if female education collectively influences the role of women in community activities. Thus, the schooling of some individuals may contribute to health over a wider net than only over the individual's immediate household.

Basu and Foster (1998) have formalized a definition of proximate literacy that is in keeping with this view that knowledge commonly associated with formal schooling can be transmitted by interaction with others who have obtained this schooling. Recently, Gibson (2001) has shown empirically that externalities from literacy may be appreciable within a family. The current study takes the measurement of externalities one step further by asking if these externalities are inter-household as well.

In a similar vein, it is possible that investments in water and sanitation by some individuals will have an impact across the wider neighborhood. Clearly, investments in household sanitation have the potential to reduce the bacterial count in the air and soil of the neighborhood and, thus, indirectly benefit their neighbors. It is less clear that individual investments in water supply assist other households since the link to the community environment is weaker. While access to uncontaminated water at home may improve nutrition due to the reduction in its price and the attendant increase of the quantity used for cleaning as much as due to any increase in the quality of water that is drunk (Burger & Esrey, 1994), the impact on the bacterial count in the neighborhood should be relatively small. A possible externality could arise, however, if families with access to quality water share this source with neighbors that do not have such access. Even in such circumstances, households may permit access to their own water source for small amounts of drinking water but be less willing to provide larger quantities (e.g., for bathing).

Knowledge of such externalities can potentially influence public policy with respect to investment decisions, or pricing policies and targeting decisions. If externalities in the production of nutrition are significant for public services such as education, this would justify additional financing for education investments or education current expenditures that would raise enrolment and literacy rates. Regarding water and sanitation, even if these services were not directly provided by public utilities, the existence of externalities could justify price subsidies, thereby increasing the number of households that can afford such services. Lastly, community externalities for individual nutrition outcomes would also have an influence for targeting social policies. The appropriate geographic targeting unit would then be the community or neighborhood rather than individual households or members thereof. The social benefits of some forms of mistargeting—often referred to as leakage—would not be as heavily discounted as is often the case.

Despite the importance of knowing more about these externalities and the existence of studies in the literature on technological diffusion (Foster & Rosenzweig, 1995), there is little empirical work on the externalities across households from individual access to basic services or education at the neighborhood level. That is, there have been few attempts at ascertaining potential positive (or negative) externalities in the production of child health. One exception is Hughes and Dunleavy (2000) who examine the influence of community provision of basic services such as water and sanitation as well as community female education in a reduced form mortality hazard model for children under the age of five in India. They find that community access to toilets alone has no significant impact on child mortality rates and that the same is true for community access to water (alone) in rural areas. However, when sanitation and water access at the community level are combined, a joint externality effect emerges which is statistically significant.

Another study that is closely related to the analysis presented here is Gragnolati (1999). He examines the determinants of children's growth in rural Guatemala, specifically considering the importance of a number of community variables that might be associated with child anthropometry. Gragnolati finds that the proportion of households with piped water connections is positively associated with children's nutritional status. Surprisingly, however, the height of rural Guatemalan children is inversely associated with the proportion of households (per community) with flush toilets.

Our approach, however, differs from Gragnolati's in an important respect. Gragnolati treats own access to basic water and sanitation services as endogenous to child health, i.e. parents decide on both child health and access to basic services jointly. To accommodate this, he does not include individual access to basic services in his model. In such formulation, it is not possible to distinguish the additional impact of the investments of the other households in the community, separate from own access to such services. As this is the primary focus of the current paper, it is necessary that we include both household and neighborhood infrastructure in the regressions reported.