socioeconomic determinants of the nutritional … · 2018-05-02 · relationships between child...

SOCIOECONOMIC DETERMINANTS OF THE NUTRITIONAL STATUS OF CHILDREN:

AN ORDERED PROBIT ANALYSIS*

Michael David A. Son Ralph M. Menchavez

ABSTRACT This study employs a multidimensional approach in the analysis of the socioeconomic determinants of the nutritional status of children aged five years and below. The paper uses data from the Community-Based Monitoring System Surveys conducted in the town of Sta. Elena, Camarines Norte in 2003 and in Barangay Salvacion, Puerto Princesa City, Palawan in 2001, analyzed within a theoretical framework based on a health production function. An ordered probit regression yielding maximum likelihood estimates for the specified reduced form model of health shows that results differ across communities, as the significant socioeconomic determinants and the direction of their effects vary from one barangay to another. In general, sex of the child; economic constraints like total household income per capita and the experience of food shortage; household characteristics like proportion of household excluding mother employed, access to safe water, and access to sanitary toilet facilities; and characteristics like age, years of schooling, employment status and membership in community organization of the household member who makes and implements critical health-related decisions were found to be significant in determining the nutritional status of children subject to the study. Examining two-way distribution tables indicating a child’s nutritional status and participation in various supplemental feeding programs reveals that while a considerable proportion of overnourished children still participate in these programs, a substantial majority of the undernourished children have yet to be included as participants. Finally, based on the results of the study, the paper proposes several policies aimed at improving the delivery of nutrition to more effectively address the problem of malnutrition.

* An undergraduate thesis submitted to the faculty of the Economics Department, Ateneo de Manila University

1

I. INTRODUCTION Malnutrition comes in four forms: undernutrition or protein-calorie malnutrition, secondary

undernutrition or chronic energy deficiency, micronutrient deficiency and overnutrition. While the widespread problem in the First World is overnutrition, with its related diseases like hypertension, heart disease and diabetes, it is undernutrition, with its two other variations, which is identified as the leading problem in the developing world (Felps 1992). In fact, the World Health Organization (WHO) estimated in 1998 that between half and two-thirds of deaths among children less than five years of age in developing countries can be attributed to undernutrition. Put simply, “undernutrition occurs when an individual simply does not get enough food. He or she is short on the calories or protein necessary for normal growth, body maintenance, and the energy necessary for ordinary human activities” (Gopalan & Rao 1979; quoted by Foster 1992, 16).

The Philippines is clearly not exempt from this problem of malnutrition. According to the National Nutrition Council, chronic energy deficiency is common among young children and pregnant and lactating women from low income households, while protein-calorie malnutrition, as manifested by growth deficits, is widespread among preschool and school children. Micronutrient deficiencies particularly in Vitamin A, iron and iodine, on the other hand, can be observed among a large group of the population of all ages (1991).

Data from national nutrition surveys conducted by the Food and Nutrition Research Institute (FNRI) for 1987 show that the dietary intake of children six months to six years old (excluding those who were fully or partially breastfed) was only about 65 percent of daily requirements. In 1990, the National Anthropometric Survey conducted by the FNRI revealed that among preschool children, 14 percent were underweight, 11.6 percent had less than the average height for their age and 9.0 percent recorded less than the average weight for their height (Herrin 1992). According to the National Statistics Office, the prevalence of malnutrition among children less than five years of age continued to decline to a level of 8.4 percent in 1996, but rose to 9.2 percent in 1998 due to the onset of the Asian financial crisis and the El Nino phenomenon (Reyes 2001). The most recent National Nutrition Survey, conducted by the FNRI in 2003, reported that 27.6 percent of children aged five years and below are underweight, while 30.4 percent are short, and 5.5 percent are thin. Among children six to ten years of age, 26.7 percent are underweight, and 36.5 percent are short. Finally, for the eleven to nineteen year-old group comprising preadolescents and adolescents, 15.5 percent are underweight.

The condition of nutrition in the Philippines thus remains to be a grave issue for policy-makers to consider and a compelling area of study for research aimed at its alleviation. Objectives of the Study

This study aims to answer the following questions: 1. How do socioeconomic variables affect the nutritional status of children in certain communities in

the Philippines? This research takes after a study made by Musgrove et al. (2003), which suggested

considering influences other than dietary intake, like water supplies, education levels, health services, and general sanitation levels, among others, in explaining undernutrition. The writers shall employ the static approach that uses cross-sectional data to establish and quantify relationships between child nutrition indicators measured in the current period and possible social and economic determinants measured in the same period.

And while there is sufficient data regarding the prevalence of undernutrition in the Philippines, as supplied by FNRI, Micro Impacts of Macroeconomic and Adjustment Policies (MIMAP)-Philippines and International Food Policy Research Institute (IFPRI), these statistics are

2

purely descriptive. Few studies have focused on econometrically understanding the mechanisms underlying nutritional status on the one hand, and the characteristics of the individual, his or her household, and his or her community on the other. By locating such a study in the Philippines, the researchers thus attempt to contribute to the body of previous studies on the socioeconomic determinants of child health which have been conducted in other countries.

2. How effective is the targeting of various health intervention programs with respect to children’s nutritional status?

Previous studies found that interventions which seek to alter single factors in the environment, such as food intake alone or the health environment alone, bring about insignificant improvement in children’s development or chances of survival (Payne 1992). With this in mind, this study aims to look into the nutritional status of children as associated with direct interventions in their community, assess whether these programs are adequately targeted to undernourished children, and find ways to improve these programs.

3. What are policy implications suggested by the results of the study? After having analyzed the effects of various socioeconomic determinants on the nutritional

status of children and having assessed the targeting of intervention programs, this paper aims to make the necessary policy recommendations to make for a more effective and economically efficient delivery of nutrition.

Significance of the Study A number of reasons urge these writers to conduct a study on the socioeconomic determinants of the nutritional status of children, and in so doing redress the country’s problems of malnutrition. The writers thus explain the rationale for conducting such a study by justifying every aspect of the research problem. Undernutrition

Undernutrition is a pressing concern being addressed by the government. The Medium-Term Philippine Development Plan (MTPDP) of the current administration, under the section Responding to Basic Needs of the Poor, stresses the need for protection of the vulnerable, specifically of children. Some of the measures outlined by the MTPDP include supporting the Bright Child Program as a holistic approach in providing interventions on food, nutrition, and health, among others; and putting greater emphasis on child health and nutrition in general. These measures echo Target 2 of the UN’s Millennium Development Goals: Halve the proportion of population below the minimum level of dietary energy consumption and halve the proportion of underweight children under five years of age. It is interesting to note that according to the Summary of Progress Towards Meeting the Millennium Development Goals, this target only has a low probability of being met, further asserting the gravity of the situation.

The effects of undernutrition further assign importance to its addressing. First, inadequate nutrition diminishes the body’s immune response, which may, in turn, lead to infection. As such, undernutrition causes intestinal disorders like diarrhea, and other diseases like pneumonia, influenza, and bronchitis—largely preventable diseases that account for over 40 percent of childhood deaths in Third World countries (Foster 1992; Dever 1983, cited by Foster 1992).

Also, undernutrition has been linked to serious physiological defects. Male adults subjected to undernutrition have been found to experience loss of memory, difficulty concentrating, and problems with physical dexterity (Keys et al. 1950, cited by Foster, 1992). For children, on the other hand, undernutrition and its associated diseases may result in growth deficits in height and brain size that are never made up. On the contrary, children who are better nourished will be more likely to be alert, active, and demanding of his or her environment and will therefore be more likely to advance intellectually to his or her fullest capacity (Myers 1988, cited by Foster 1992; Foster 1992, 26).

3

Furthermore, child undernutrition necessarily leads to poor school performance. Undernourished children have shorter attention spans, which make him or her apathetic to learning. They also miss more school days due to illness. Better health and nutrition leads to earlier and longer school enrollment, better school attendance, and more effective learning. Improving the health especially of children, then, serves to improve not only the effectiveness of schooling but also the individual child’s educational development (Foster 1992; Alderman et al 2001; Todaro and Smith 2003). Lastly, undernutrition places considerable burden on the macroeconomy. A level of dietary intake which can permit only substandard growth leads to loss of function relating to cognitive ability and a substantial decline in work time due to sickness—an erosion of the quality of human resource in a country. Causing a reduction in peak work capacity and diminished labor productivity, undernutrition consequently exerts downward pressure on national income (Arcand 2001; Foster 1992; Gopalan 1992; Payne 1992). Children The vulnerability of children to undernutrition creates moral impetus for making them subject of the study. The 2001 Human Development Report of the United Nations Development Program reported that in 1998 alone, 11 million children died before reaching the age of five from preventable causes, while as much as 163 million children under the age of four are underweight. While adults do die of hunger during a famine, the majority of deaths, whether from famine or chronic undernutrition, occur among preschoolers. Thus children, by far the most defenseless to undernutrition, are the top priority for any aims at intervention (Foster 1992; WHO 1998).

The irreversibility of undernutrition’s effects further necessitates addressing malnutrition during early childhood. With respect to mental development, studies agree that there is a level of deprivation in very early life at which some irreversible damage is done to brain function. As has been mentioned, undernutrition also causes growth deficits that cannot be made up for. In addition, many studies have shown that children’s peak work capacities are proportional to body weight, and more importantly, this holds when they become adults (Payne 1992).

Finally, from a statistical standpoint, the nutritional status of children is a rich and valid variable. Human physical growth and development are highly responsive to changes in dietary intake and is especially true for children under five years of age, when growth is so rapid. As such, assuming similar intra-household allocation of resources among households in a community, growth performance of children becomes a suitable and reliable indicator of community nutritional status. The findings of a study that looks into the nutritional status of children can thus be relevant not only for the age group of children but also for the entire population of a given community (Foster 1992; Gopalan 1992).

Socioeconomic determinants

While most studies on health and nutrition of both children and adults look into the effects of nutrient consumption and food availability, few studies focus on the relationship between nutritional status and non-nutritional factors, like educational attainment, availability of water and sanitation, etc. And among these already few studies that look into socioeconomic aspects, fewer still give emphasis to children’s nutritional status: most either focus on adult health on the one hand, or infant mortality on the other. In response, this study aims to address the gap in knowledge regarding the relationship between the nutritional status of children and their socioeconomic environment. In so doing, the study responds to the difficulty of measuring malnutrition. By examining the link between nutritional status and socioeconomic determinants in a community, one can find a criterion or set of criteria that can sufficiently be used to predict the nutritional status of children in other areas. This circumvents the difficulties in estimating nutrition and will especially be useful in assessing the nutritional status of children in areas where there is a lack of necessary data (Anand and Harris 1992).

4

Moreover, Phillips Foster, in his book The world food problem: tackling the causes of undernutrition in the third world (1992), insists that nutrition intervention programs treat only the symptoms and do not treat the causes of hunger. Anand and Harris (1992) add that in order to design policies that attempt to alleviate undernutrition, it is important to first understand the relationship between economic and social characteristics and undernutrition. Knowing which variables significantly affect nutrition status would provide valuable practical leads for combating the causes of undernutrition in the community (Gopalan 1992). As S. R. Osmani eloquently puts it:

Nutrition and poverty are two very closely related themes. Many elemental aspects of being poor, such as hunger, inadequate health-care, unhygienic living conditions, and the stress and strain of precarious living, tend to impair a person’s nutritional status. In consequence, being poor almost always means being deprived of full nutritional capabilities. . . An understanding of the processes through which chronic malnutrition comes to afflict a household or community can reveal a good deal about the process leading to endemic poverty. Such analyses may also yield useful guides for policy. For example, by delineating the relative roles of food, health care and environment in the genesis of malnutrition, such analyses may help policy-makers to rationalize priorities among different components of anti-poverty programs. (1992, 1)

Scope and Limitations There are inherent limitations in measuring nutritional status. While one approach is to compare caloric and nutrient intake with some standard of requirement, the other is to compare anthropometric measurements of the body with some reference standard. As such, while the former presupposes that the energy requirement of a given type of individual is fixed and the latter presupposes that there is a normal weight and height for a certain type of individual, both approaches assume that any shortfall from this norm must entail impairment of at least some nutritional functions. And by disregarding interpersonal and intrapersonal variation, anthropometry may even supply inaccurate measurements of a person’s true level of nutrition (Osmani 1992).

Furthermore, in choosing the anthropometric approach, the researchers arrest the contention that “the anthropometric achievement of a person may vary within a certain (fairly wide) limit without causing any damage to his functional competence” (Osmani 1992, 12), and instead work under the assumption of the genetic potential theory, which “contends that, if free from all nutritional constraints, most population groups in the world are capable of achieving the same physical dimensions” (Osmani 1992, 11). At this point therefore, the writers make all the necessary assumptions.

In addition, while there are several anthropometric proxies for nutritional status—birth weight, height, height-for-age, weight-for-age, and weight-for-height or body mass index—the study will employ only one: weight-for-age. The choice is thus made not only because weight-for-age is the standard used in the CBMS-Philippines data set to be used, but also because weight-for-age is an apt indicator of the nutritional status of a child, as low levels of such is a symptom of present undernutrition. Birth weight is more closely linked to undernutrition of mothers, while height and height-for-age are symptoms of past undernutrition (during growth years) and as such are more apt as indicator of nutritional status of adults and older children (Foster 1992).

Finally, the nature of the data necessarily imposes limitations on the type of methodology used and the extent of the conclusions drawn. There was an insurmountable difficulty in obtaining panel data involving the needed variables, and the data set acquired by the writers is only cross-sectional. Thus a more ideal dynamic approach to analysis may not be utilized and the writers are forced to employ the static approach to analyzing the socioeconomic determinants of health. As such, no conclusions can be made about the impact of intervention programs. The data set is also confined to a certain barangays or

5

municipalities and are by no means said to be representative of the entire country. Therefore, there can be no claims for generalizability beyond the locality in question, and the results will be interpreted only insofar as the subjects of the study are concerned.

II. REVIEW OF RELATED LITERATURE

Most available studies concerned with the socioeconomic determinants of child health emphasize

the effects on health of several constraints such as parental knowledge, physical resources, and government programs, along with nutritional intake. These studies were conducted in a variety of methods that range from simple analysis of descriptive statistics to more mathematical methods like reduced form equation estimation. Throughout the years, more sophisticated models and larger data sets were employed in order to better probe the causes of child malnutrition. Worth mentioning are the studies which analyzed data from developing countries, since determinants such as income and government programs bear greater importance in poorer countries than in developed ones (Behrman and Deolalikar 1987; Duflo 2000).

Early studies conducted in the Philippines about child nutrition aimed to find out whether family data sets could be used to predict the incidence of child malnutrition. Using data obtained in 1979 by the National Nutrition Council of the Philippines which covered 722 families from seven provinces, Arnold and colleagues (1981; cited by Foster 1992) simply estimated the coefficient of correlation between the child’s nutritional status, measured as a percentage of the standard weight-for-age, and several socioeconomic variables. Results showed that the educational attainment of the parents and family income both significantly and positively affect child health. A closer scrutiny of the method, however, will reveal sample selection bias, since only the preschool child with the lowest level of nutrition in the household was used as subject. The study neither considered siblings of the undernourished children, nor did the survey include data from households with children exceeding or of standard weight-for-age. The study, then, may draw conclusions concerning undernutrition only and not nutritional status in general.

Years later, studies shifted their focus to the obvious question of relating income to child health. Behrman and Deolalikar (1987) looked into this relationship by means of estimating the income elasticity of nutrient intake of a sample from rural south India. The study contradicts the World Bank argument that malnutrition is a manifestation of poverty, i.e. that an increase in income will increase expenditure on food, and subsequently the child’s intake of nutrients. After estimating the reduced form food and nutrient relations, they did not reject the null hypothesis that elasticity was equal to zero, i.e. there was no significant change in food intake after an increase in income: the subject households tend to devote the increase in income to concerns other than food quantity. The paper therefore, highlights the importance of factors other than income that mediate the production of health, and stresses the need for education on health production and nutrient intake to accompany any such direct interventions on income.

Several other studies that looked into the elasticities of income with respect to determining calorie or nutrient intake employed different methods to calculate the estimates. While Behrman and Deolalikar used the 2-Stage Least Squares (2SLS) method in the previously mentioned study, Pitt (1983) used the Tobit technique for a sample from Bangladesh, and Sahn (1988) used the Heckit system for his study on Sri Lanka. Later, Deaton and Subramanian (1996) used a nonparametric technique for another sample from India.

Towards the 1990s, emphasis increasingly shifted from the primacy of income to the impact of other socioeconomic determinants on health. Bhargava (1992) conducted a study on malnutrition using data from 368 individuals from six villages in South India and 2,047 individuals living within a 20-kilometer radius in Southern Bukidnon in the Philippines. Using the method of maximum likelihood, the study noted that proportions of variation in the intakes of dietary energy and protein due to interindividual and intraindividual differences are affected by socioeconomic factors like age, sex, and income, and that the

6

variation is likely to decline with rises in income. While it is nutrient inequality among households to which his study gives emphasis, findings underscore the significance of socioeconomic factors, income included, in explaining differences in nutritional status.

Many studies were also intended to examine the relation between child health and parental characteristics. A study conducted by Barrera (1990) in the Philippines used reduced form estimates and observed that although better-educated mothers tend to shorten breastfeeding time, they compensate this with better care. Analysis of height-for-age z-scores of children under the study confirm the positive effects of early albeit shorter period of weaning to child health. Likewise, the Cebu Longitudinal Health and Nutrition Survey conducted by the Cebu Study Team (1992) showed that mother's education correlates with knowledge on waste disposal and higher non-breast milk calorie intake of their infants, which both reduced the incidence of diarrhea among children.

Thomas (1994) also found out that parental education greatly affects child's height and moreover, the impact was gender dependent. Ordinary least squares (OLS) regression was applied to household data from the United States, Brazil and Ghana in order to estimate a health production model. Findings show that in the developing country of Brazil, women's non-labor income has a positive impact on the health of her daughter but not on her son's health, while in Ghana, the education of a woman who is better educated than her husband has a bigger impact on the height of her daughter and her son. The study therefore concludes that gender differences affect both technological differences in raising children and discrepancies in the preferences of parents.

An apparent trend in the analysis of health, as seen in the studies of Barrera and Thomas, is that of using (1) anthropometric measures, being a more observable proxy of the quality and quantity of nutrition that the child receives, instead of nutritional or caloric intake; and (2) the framework of a health production function to analyze the impact of several variables on the health of both adults and children. Following this trend is a study done by Mackinnon (1995). The study used data from the 1992 Integrated Household Survey in Uganda, which probed the determinants of child mortality and malnutrition over the last twenty years, and covered a sample of about 10,000 households. Maximization of a proposed household utility function constrained by a health demand function and subsequent demand functions for goods and services related to health practices yields a reduced form model of health, with inputs such as demographic characteristics, consumption, household income, labor characteristics, parental education, and various environmental characteristics such as sanitation facilities, water supply and garbage disposal practices. Child survival ratio (of children ever born), and z-scores for height-for-age and weight-for-height were used as dependent variables, leading to several versions of the reduced form model estimated through OLS. It was observed that the beliefs about health (e.g. knowledge of the causes of malaria and diarrhea) greatly influence child mortality and nutrition, since the pertinent variables were seen to be positive and significant for both mortality and malnutrition. And while parental education exhibited significant positive effects on child mortality, it is the environment endowments which exerted significant positive effects on child nutrition. In addition, similar to the results of the study by Thomas, the gender of the child was also shown to significantly affect health. Interestingly, girls do better than boys in the sample from Uganda. Another study that used a health production function was that of Glewwe (1999) which used evidence from Morocco and starts with the notion that mother's education is often found to be positively correlated with child health and nutrition in developing countries. He posits that the mechanisms which enable this notion are: (a) formal education directly taught health knowledge to future mothers; (b) literacy and numeracy skills in school assist future mothers in diagnosing and treating child problems; and (c) exposure to modern society from formal schooling makes women more receptive to modern medical treatments. As such, Glewwe proposed a health production function relating z-scores for child height-for-age to inputs such as father’s schooling, mother’s skills in literacy and numeracy, mother’s health knowledge (measured by a test on awareness regarding vaccination, boiling water before ingestion and

7

immediate illness treatment), and household income. Using OLS, 2SLS and Community Fixed Effects (FE), the author produced reduced form estimates of the determinants of child height and observed that mother’s health knowledge is the most significant pathway through which mother’s education raises child health. Similarly, Chaudhuri (2003) calculated reduced form health production function estimates using data obtained from a 1996 survey in rural Bangladesh to evaluate the impact of a maternal and child health (MCH) program. Dependent variables used were children’s z-scores of height-for-age, weight-for-age, and weight-for-height. The impact of the MCH program is analyzed by using a dummy variable indicating exposure to the program. Results show that the coefficient of this dummy variable, and therefore the MCH program, is statistically significant for long-term health (measured in height-for-age) as well as short-term health (measured in weight-for-age), but not on acute growth disturbances (measured in weight-for-height). Furthermore, a probit model estimation was performed by using a discrete dependent variable indicating stunting (low height-for-age), wasting (low weight-for-height) and being underweight (low weight-for-age) among children to capture the program effect on extreme health outcomes. The marginal effect of exposure to the MCH program is a decrease in the probability of stunting, wasting, and being underweight, although the treatment effect is statistically significant only for wasting and being underweight. An alternative technique of analyzing the impact of government policy on child health is proposed by Duflo (2000). Her study, aimed at analyzing the effects of the Old Age Pension program on child health used data gathered in 1993 from 9,000 randomly selected households in South Africa. What she did was to compare differences in height between children from households eligible for the program (where there is a member over the age of 60 or 65) and non-eligible ones. Nonparametric techniques using locally weighted regressions of height (and height-for-age z-scores) as a function of date of birth, as this tells how long the child has been affected by the program, suggest that the extension of Old Age Pension program in South Africa has led to an improvement in health and nutrition of children coming from households enjoying the pension program. To further evaluate the effectiveness of the program, cubic-spline regressions were run, yielding average derivatives. These derivatives suggest that there is a significant difference between height (and height-for-age) of girls relative to date of birth from eligible households before and after the breakpoint of December 1990, when the Old Age Pension program expanded rapidly. The health production function is expanded by Fedorov and Sahn (2003), in their analysis of the socioeconomic determinants of child health in Russia, by expressing the function through a dynamic approach. The dynamic health production function assumes that there are intertemporal connections between child health and its determinants, and that current health status is a function of not just current prices, resources, and exogenous characteristics, but also those of previous periods. As such, they specified a health production function which relates current health with health lagged by one period, current period inputs, and exogenous characteristics of the household. The lag variable would thus account for all other health inputs in the past. As with other studies, child height is chosen as the indicator of child health status. The reduced form of the dynamic health production function was estimated in the five-year panel setting by Generalized Least Squares. Confirming the conjecture, height of the child lagged by one period exhibits a positive effect at the 1% level. The estimated parameters also suggest that education of the mother, household income, location (rural over urban), and availability of hospitals exert significant and positive effects on the height of children under the study, while food prices have significant but negative effects. With the existing complexity of methods being employed over the years in order to determine the causes of child malnutrition in developing countries, this paper sees it fit to follow after the recent studies on the determinants of child health by using anthropometric data, involving a range of socioeconomic variables like gender, household income, parental education, environment endowments, presence of government programs, and finally, situating the analysis within the framework of a health production function, as will be described in the next section.

8

III. THEORETICAL FRAMEWORK

While it is clear that malnutrition is primarily caused by an inadequate diet, the literature suggests that malnutrition can also be caused by claims on that diet so great as to cause such despite a nutrient intake that in other circumstances might be deemed adequate (Fogel 1992). Various studies treat the nutrition problem as a multi-factorial one, necessitating action based on a multisectoral approach alongside policies aimed directly at food subsidization and nutrient fortification (Payne 1992; WHO 1998). Anand and Harris (1992), Gopalan (1992) and Tomkins and Watson (1989; cited by Payne 1992) agree that an individual’s command over resources greatly influences his or her health, and thus suggest approaching the problem of undernutrition through income subsidies, employment provision, strategies to improve housing and ensuring access to health care and medical care. In addition, Srinivasan (1992) and the WHO (1998) highlight the importance of clean water in the prevention of disease and maintaining health, while Todaro and Smith (2003) underscore the importance of parental literacy in the production of child health. Finally, Thomas (1994) and Gertler and Zeitlin (2002) add that the productivity of an individual’s production of health depends on individual characteristics such as education, age, gender, and family background, and household and community characteristics.

In response to the above propositions, the writers at this point propose a multidimensional approach to the analysis of nutritional status. The following theoretical framework is synthesized from J. R. Behrman and A. B. Deolalikar’s “Health and Nutrition” (1988), Paul Gertler and Jennifer Zeitlin’s “Do Investments in Child Education and Nutrition Improve Adult Health?” (2002), and Leonid Federov and David E. Sahn’s “Socio-Economic Determinants of Children’s Health in Russia” (2003), which in turn are based on Michael Grossman’s seminal work, “On the Concept of Health Capital and the Demand for Health” (1972). In the work cited, Grossman suggested that health can be thought of as a form of human capital: an individual’s health stock at any point in time is determined by an initial genetic endowment, subsequent behavioural choices (for example, nutrition, medical care, smoking, exercise), and exogenous shocks from the public health environment (for example, contracting cancer from toxic waste). Household preference function

The determinants of an individual’s health and nutrition usually are decisions made by the individual or the household in which he or she lives—given assets, prices, and community endowments. Therefore a natural starting point is the determination of individual health and nutrition at the household level.

The model is structured assuming that the household maximizes a single preference function subject to constraints (to be enumerated below). For simplicity, a static or one-period model is considered. The researchers now turn to an algebraic statement of the one-period, household model with constrained maximization of a joint utility function.

Assume that the household has a preference function:

U = U (Hi, C i, Cp, TiL, Ei|c, S ; ξ), i = 1, . . . , I (1) where Hi is the health of household member i, C i is the consumption of household member i of private goods, C p is the consumption of household member i of pure public goods, TiL is the leisure time of household member i, Ei|c is the education of household child i,

9

S is the number of surviving children, ξ are taste norms, and I is the number of individuals in the household. (All of these variables and others defined below may be vectors with multiple dimensions.) Utility is presumed to depend on the health of each of the household individuals, with a negative impact of poor health and mortality. Private goods consumption, pure public goods consumption and leisure of each household member have positive impact. The education of each child is included because of possible altruistic interests of the parents and concern about the child’s expected prospects as an adult which may affect the parents’ material well-being in their old age. The number of surviving children is presumed to improve parental welfare whether for altruistic, insurance, or other reasons. The utility function finally is conditional on norms, here assumed to be exogenous. What follows are the functions on which the maximization of the household preference function is constrained. Production function determining health

The health of the ith individual is produced by a number of choices relating to the consumption and time use of that individual, the education of that individual and of the key person(s) in the household making and implementing health-related decisions, and the individual, household, and barangay endowments:

Hi = H (Ni, C i, Cp, I, Ei, Em, TiL, TiH, TmH, ηi, ψ, Ω), (2)

where Ni is the nutrient intake of the ith individual, Ei is the education of the ith individual, Em is the education of the person—often the mother—who makes critical health-related

decisions and implements them within the household, hereafter referred to as “mother”), TiH is the time of the ith individual devoted to health-related procedures, TmH is the mother’s time devoted to health-related procedures, ηi is the endowment of the ith individual ψ is the endowment of the household Ω is the endowment of the barangay, and the other variables are defined above.

Nutrient intakes (Ni) are emphasized because of their presumed importance in health determination, and its impact on health is assumed to be positive. Other consumption items (C i, Cp) include goods and services with a range of direct effects on health (e.g. medicine, cigarettes, driving vehicles). The household size (I) is included to represent possible scale and congestion effects. The individual’s time use is included because the nature of his or her occupation (not explicitly included above), the extent of leisure time (TiL) and the time devoted to health-related activities (TiH) may have strong health effects. The individual’s education (Ei) and that of the mother may affect health through affecting the choice of health practices, through better information and through affecting the effectiveness of the use of given health-related inputs (e.g. food preparation, disease treatment, etc.).

The last three variables, the individual’s endowments (ηi), the household endowment (ψ), and the barangay endowments (Ω), differ from the other variables in that they are not presumed to be choice

10

variables of the household during the period being modeled. Examples, respectively, would be: individual’s age, initial health and genetic make-up; the household’s tenure status and type of housing; and access to water and basic utilities, and the presence of medical, health care and waste disposal facilities in the barangay. While the individual’s endowments may posit direct effects to his or her health through biological mechanisms, the endowments of the household and the barangay affect the individual’s health by way of its claims to the individual’s disease environment and subsequently, its impact to the individual’s nutrient-utilization capabilities.

Income constraint In the modeled period, the household faces a budget constraint: PCC + ΣPEEi + X = ΣP iL(T i – T iE – T iL – T iH) + ΣYi + F , (3) where PC is the price (or vector of prices) of consumption good(s) C, PEi is the cost of education specific to child i, X is all other household expenses for the period (e.g. rent, utilities expense, etc.), P iL is the wage rate of the ith individual, T i is total time of the ith individual, T iE is the school time of the ith child, Yi is the non-labor income of the ith individual, and F is transfers less taxes (assumed to be lump-sum for simplicity). The reduced form

Under the assumption that the underlying functions have desirable properties so that the maximization of (1) subject to (2) and (3) is obtained, the constrained maximization of preferences leads to the reduced form demand function for health:

Hi = H (N i, C i, Cp, I, Ei, Em, T i, T iL, T iH, T iE, T mH, PC, PE, R, P iL, Yi, F, ηi, ψ, Ω), (4) which provides a consistent framework within which to examine the impact of changes in, among others, household and barangay endowments, on the health-related consumption of different types of individuals. Equation (4) thus gives the relationship of primary interest in this study.

IV. RESEARCH METHODOLOGY Since the focus of the study is not merely the incidence of malnutrition in a community but the nutritional status of the individual child, health (Hi) would be represented by the anthropometric proxy of nutrional status. The weight and height of the children, however, were not recorded in the data set: the nutritional status of children aged five years and below was simply assigned into categories. The writers thus employ an econometric analysis involving discrete choice variables, the possible outcomes being that the child is (1) undernourished (either mildly, moderately or severely), (2) of normal nourishment, or (3) overnourished. Owing to the inherent relatedness of the possible outcomes—the categories of nutritional status being a continuous spectrum exhaustive of possible outcomes—an ordered choice technique of estimation is made applicable. As such, the Hi at the left-hand side of equation (4) is operationalized as an ordered dependent variable taking on values representing the possible qualitative outcomes for the nutritional status of the individual child outlined above.

11

Consequently, we assume that the endogenous disturbance to the function is normally distributed, thereby allowing for the use of a parametric ordered probit technique of estimation. While use of an ordered logit technique can also be considered, the probit method is chosen because of the ease in interpreting its results—as probability is easier to interpret than odds—thereby benefiting possible policymakers. The necessary statistical software which will allow for the evaluation of multiple integrals—a computational obstacle to the probit method—is also available. Specifying the reduced form equation (4) as an ordered probit function and estimating the function using data gathered would allow for the estimation of the changes in probability associated with differences in the right-hand variables (i.e. socioeconomic determinants present in the individual, household and barangay) on the severity and direction of malnutrition. This would then fulfill the first objective of the study, which is to see how socioeconomic variables affect the nutritional status of the individual child. The right-hand variables shall be operationalized as afforded by the data set. The writers note, though, that since the data was obtained from an external source, the survey employed was designed without the proposed theoretical framework in mind. As such, certain variables may be modified or instrumentalized, if not omitted altogether. Other information available from the data set would have to stand as proxies to measuring certain variables. For instance, since nutrient intake (Ni) was not measured, the number of meals eaten in a day or the household’s experience of food shortage, which is included in the data set, may be considered instead. Some variables, like time of the individual devoted to health-related procedures (TiH) may be dropped because children aged five years and below, the subjects of the study, can be assumed to exert almost no control in their own health production. In order to look into the targeting and association of various health intervention programs with children’s nutritional status, the third objective of the study, the household’s participation in feeding programs will be included as a dummy variable in the vector of household endowments (ψ). With the model properly estimated, an understanding of the dynamics between health and nutrition production and its significant inputs would be established. This understanding would greatly inform policy recommendations to be suggested and in so doing will the writers fulfill the last objective of this study. These policy recommendations would, in turn, propose ways in which the delivery of nutrition may be made more efficient for the ultimate purpose of improving the nutritional status of children in the Philippines. The Data The data analyzed in this paper come from the Community-Based Monitoring System (CBMS) Survey 2003 of the Municipal Government of Sta. Elena, Camarines Norte and the CBMS Survey 2001, Barangay Salvacion, Puerto Princesa City, Palawan. Authorization to use the data was provided by the Municipal Planning and Development Office of the Municipal Government of Sta. Elena and the City Planning and Development Office of the City Government of Puerto Princesa, respectively, obtained through the CBMS Network Coordinating Team of the Angelo King Institute for Economic and Business Studies. The CBMS Survey of Sta. Elena was conducted in August, 2003 and covers the entire population of 7,521 households from all nineteen barangays in Sta. Elena, Camarines Norte. Of the 38,091 population, 6,244 are children aged five years and below and are thus the subjects of the study. On the other hand, the CBMS Survey of Brgy. Salvacion was completed in November, 2001 and covers Brgy. Salvacion’s population of 191 households. Of the 953 residents, 176 are children aged five years and below and are likewise included in this study.

Appendix A lists down the variables found in the CBMS Survey. Nutritional status of the child, represented by the variable MNUTIND, is operationalized as a discrete choice variable, and is based on a table of standard weight-for-age (see Appendix B). The table supplies weight intervals or boundaries (in

12

kilograms) according to age (in months) that correspond to the categories of nutritional status. After data is gathered at the household level, barangay health workers supply information as provided by records of the barangay health center regarding the nutritional status of the children in the household.

Most of the other variables are dummies taking on the value of either 1 or 0, depending on the information supplied. Included also are continuous variables like age and income, which will be used as regressants, together with the dummy variables mentioned, in the reduced form equation being estimated.

The variables with the -MOM suffix pertain to the mother, or the person assumed to make critical health-related decisions and implements them within the household. The variable is assigned to the actual mother for households with distinct nuclear families. For grandchildren living with grandparents only, the grandmother is assumed to be their caretaker. For households without a single matriarchal figure, the household head is assumed to be the “mother.” For households where the identity of the “mother” is ambiguous, no assumptions are made. Those particular observations are then omitted by the regression software.

Empirical Model Based on the data provided by the CBMS Surveys, the writers specify the reduced form function (4) into an empirical model, where nutritional status of the individual child, represented by the variable MNUTIND, is expressed as an ordered probit function of socioeconomic determinants: MNUTIND = H (SEX, MWJOBPROP2, INCOMETOTALPC, WATERSAFE, TOILSAN, ELECHOURS, FSHORT, M05FEED, AGEMOM, SCHOOLNGMOM, JOBINDMOM, COOPASSOCMOM), (5) where the variables are as defined in Appendix A. The model includes a dummy variable for sex, a characteristic representative of the individual child’s endowments, as well as dummies for access to safe water and access to sanitary toilet facilities, and number of hours of access to electricity per day to capture the effects on nutrition of endowments to the household and to the community. A variable measuring total household income per capita is also included to measure effects of the budget constraint relative to household size, as are dummies for the experience of food shortage and participation in feeding programs, taken as proxies for nutrient intake. The dummy for participation in feeding programs is likewise representative of the consumption of a public good. Moreover, variables representative of characteristics of the mother are included like employment status to control for the effects on nutrition of time devoted to health-related procedures, age and schooling to examine effects of education and accumulation of experience, and membership in community organization to look into possible effects of social capital. A variable measuring the proportion of household members excluding mother employed is also included to capture the effects of all other time-related variables included in the reduced form model. V. PRESENTATION AND ANALYSIS OF RESULTS Reported first are general descriptive statistics showing the distribution of children according to categories of nutritional status for all barangays represented by the CBMS data sets. Accompanying these are the summary of each variable’s mean and standard deviation for each of the ten barangays whose regression results meet acceptable statistical criteria and as such are presented below. Following this, the writers present and analyze the results of the ordered probit estimation of equation (5), the specified reduced form function. Finally, the writers assess the feeding programs implemented in Sta. Elena by way of presenting

13

two-way distribution tables indicating the individual child’s nutritional status and his or her participation in a feeding program. Descriptive Statistics Table 1 shows the magnitude and proportion of the population of children aged five years and below in all nineteen barangays of Sta. Elena, Camarines Norte distributed in each of the four categories of nutritional status defined in the survey. Note that the children of mild undernourishment and normal nourishment are lumped together in one category, as this was the manner of recording employed by the survey. Of these barangays, Poblacion, or the town proper of Sta. Elena, registers the highest proportion of children reporting mild undernourishment or normal nourishment, with 99.12% of the population of children falling under this category, while the proportion for the same category is smallest for Bulala, at 76.33%. Also, Bulala and Kabuluan have the highest proportion of moderately undernourished children, reporting 12.33% and 12.08% respectively. As observed, the proportion of children with severe undernutrition does not vary largely across barangays, with the proportion ranging from 0% (in ten of the nineteen barangays) to 2.61% in Kagtalaba. One can also see that overnourishment is most prounounced in Basiad, with 14.06% of its children falling under the category, followed by Bulala with 11.33%.

14

Table 1 Distribution of Children Across Categories of Nutritional Status

(Sta. Elena, Camarines Norte)

Basiad Bulala Don Tomas Guitol Kabuluan Camarines Norte Camarines Norte Camarines Norte Camarines Norte Camarines Norte

Magnitude Proportion Magnitude Proportion Magnitude Proportion Magnitude Proportion Magnitude Proportion

Severely Undernourished 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00

Moderately Undernourished 7 1.58 37 12.33 0 0.00 14 12.96 29 12.08 Mildly Undernourished and Normal Nourishment

372 84.35 229 76.33 326 97.60 94 87.04 209 87.08

Overnourished 62 14.06 34 11.33 8 2.40 0 0.00 2 0.83

Total 441 100.00 300 100.00 334 100.00 108 100.00 240 100.00

Kagtalaba Maulawin Patag Ibaba Patag Ilaya Plaridel Camarines Norte Camarines Norte Camarines Norte Camarines Norte Camarines Norte




203 92.69 390 90.49 100 99.01 79 88.76 194 88.18

Overnourished 2 0.91 0 0.00 1 0.99 0 0.00 1 0.45

Total 219 100.00 431 100.00 101 100.00 89 100.00 220 100.00

Poblacion Pulong Guit-Guit Rizal Salvacion San Lorenzo Camarines Norte Camarines Norte Camarines Norte Camarines Norte Camarines Norte




1235 99.12 261 92.23 284 91.32 99 86.08 1036 95.40

Overnourished 2 0.16 0 0.00 0 0.00 5 4.35 19 1.75

Total 1246 100.00 283 100.00 311 100.00 115 100.00 1086 100.00

15

San Pedro San Vicente Tabugon Villa San Isidro Total for Sta. Elena Camarines Norte Camarines Norte Camarines Norte Camarines Norte Camarines Norte


Severely Undernourished 0 0.00 0 0.00 0 0 1 2.08 23 0.37


248 100.00 169 98.83 239 94.46 47 97.92 5814 93.11

Overnourished 0 0.00 0 0.00 7 2.77 0 0.00 143 2.29

Total 248 100.00 171 100.00 253 100.00 48 100.00 6244 100.00

Source: Community-Based Monitoring System (CBMS) Survey 2003 of the Municipal Government of Sta. Elena, Camarines Norte

16

Table 2, on the other hand, presents data from the single barangay of Salvacion, Puerto Princesa in Palawan. In this data set, the children of mild undernourishment and of normal nourishment are assigned to distinct categories. One can see that as much as 26.71% of the children in Salvacion are malnourished, with 22.16% reporting mild undernourishment.

Table 2 Distribution of Children Across Categories of Nutrition

(Brgy. Salvacion, Puerto Princesa, Palawan)

Salvacion Palawan

Magnitude Proportion

Severely Undernourished 0 0 Moderately Undernourished 6 3.41 Mildly Undernourished 39 22.16 Normal Nourishment 129 73.30 Overnourished 2 1.14

Total 176 100.00

Source: CBMS Survey 2001, Brgy. Salvacion, Puerto Princesa City, Palawan.

Table 3 shows the statistical summary in terms of means and standard deviations of variables used in the study for the ten barangays whose regression results are reported in the following section. From the given means, one could see that most barangays have almost equal number of boys and girls; only in San Lorenzo and Tabugon do girls outnumber boys. The inadequate access to safe water in Basiad is quite noticeable as more than half of its children does not have access to safe water, defined as water coming from the community water system, deep wells or artesian wells. This is compounded by poor access to sanitary toilet facilities in Basiad, as well as in other barangays. For eight out of the ten barangays, even if most households have toilets, less than half of the population of children has access to sanitary toilets, defined as shared or private toilets with sealed flush. The high occurrence of food shortage experienced within the six months prior to administering the survey in Brgy. Salvacion may have yielded a relatively high case of undernourishment, with as much as 25.57% of the children in Salvacion being undernourished. The fact that Brgys. Bulala and Kabuluan enjoy the highest proportion of participants in the feeding program, with proportions of 63.33% and 89.58% respectively, while at the same time having high proportions of moderately malnourished children, may point to adequate site targeting, i.e. programs are implemented where they are needed. Most barangays have approximately one-fifth of their household members excluding mothers employed. So on the average, in a family of five, at least one person is working and providing for the rest. Also, most mothers are in their early 30s, and so can still be considered young and able to learn and apply new methods of caring for children. It should also be noted that most mothers in these areas had not been able to finish secondary schooling, as the mean for years of schooling never exceeded 11. This may have influenced the low incidence of mothers employed, with these proportions ranging from 8.68% in Don Tomas to only 30.11% in Salvacion.

17

Table 3 Summary of Means and Standard Deviations of Variables

(Sta. Elena, Camarines Norte) Basiad Bulala Don Tomas Kabuluan Maulawin

Camarines Norte Camarines Norte Camarines Norte Camarines Norte Camarines Norte

Mean Standard Dev. Mean Standard Dev. Mean Standard Dev. Mean Standard Dev. Mean Standard Dev.

Sex of child 0.5397 0.4990 0.5067 0.5008 0.5868 0.4931 0.5167 0.5008 0.5360 0.4993 Proportion of

household members excluding mother employed

0.1730 0.0910 0.2010 0.0900 0.2130 0.0900 0.2100 0.0920 0.2040 0.0970

Access to safe water 0.3923 0.4888 0.7700 0.4215 0.5778 0.4946 0.8917 0.3115 0.5522 0.4978 Access to sanitary toilet

facility 0.4014 .4907 0.4667 0.4970 0.4042 0.4915 0.5958 0.4918 0.3039 0.4605

Access to toilet facility1 0.5011 0.5006 0.5167 0.5006 0.9940 0.0773 0.9500 0.2184 0.6543 0.4762 Total household

income per capita 4590.4039 5667.2514 5236.4529 4616.6815 4489.5958 3476.7850 9207.7489 7226.7250 9048.8496 7916.9655

Food shortage within last six months

0.0068 0.0823 0.0800 0.2717 0.0000 0.0000 0.0333 0.1799 0.1949 0.3966

Feeding program participation

0.0091 0.0949 0.6333 0.4827 0.0240 0.1531 0.8958 0.3061 0.1044 0.3061

Age of mother 32.6848 7.8209 33.0033 9.2282 30.8413 6.6207 32.1458 9.5466 32.3039 8.4939 Years of schooling of

mother 9.2808 2.9583 9.1200 2.5272 9.4820 1.9920 10.4583 2.7577 8.7935 2.4469

Employment status of mother

0.1451 0.3526 0.2333 0.4237 0.0868 0.2820 0.1292 0.3361 0.2390 0.4270

1 Variable not included in empirical model; included in table only for illustration purposes.

18

Table 3 (cont.) Summary of Means and Standard Deviations of Variables

(Sta. Elena, Camarines Norte and Salvacion, Puerto Princesa, Palawan) Plaridel Poblacion San Lorenzo Tabugon Salvacion

Camarines Norte Camarines Norte Camarines Norte Camarines Norte Palawan

Mean Standard Dev. Mean Standard Dev. Mean Standard Dev. Mean Standard Dev. Mean Standard Dev.

Sex of child 0.5364 0.4998 0.5177 0.4999 0.4991 0.5002 0.4901 0.5009 0.5170 0.5011 Proportion of household

members excluding mother employed1

0.2100 0.0870 0.1930 0.1000 0.2060 0.2050 0.1764 0.1195 -- --

Access to safe water 0.2500 0.4340 0.7689 0.4217 0.7772 0.4163 0.5968 0.4915 0.5568 0.4982 Access to sanitary toilet

facility 0.4455 0.4981 0.8828 0.3218 0.4567 0.4984 0.3992 0.4907 0.4318 0.4967

Access to toilet facility2 0.7182 0.4509 0.9222 0.2680 0.5092 0.5001 0.8182 0.3865 0.4489 0.4988 Total household income per

capita 8312.9790 8650.3623 15629.5928 23787.4486 13214.3463 45985.6690 7727.7930 9567.8467 8463.7995 5443.5747

Food shortage within last six months

0.0000 0.0000 0.0016 0.0400 0.0267 0.1613 0.0395 0.1952 0.6023 0.5023

Feeding program participation1

0.0364 0.1876 0.1132 0.3169 0.2127 0.4094 0.0000 0.0000 -- --

Age of mother 32.9635 8.9231 32.5883 8.2922 32.8783 9.7016 33.4071 8.6875 31.0682 7.5355

Years of schooling of mother 9.9452 2.3061 10.6571 2.9784 9.5171 2.7576 9.5257 2.4410 10.1591 3.1583

Employment status of mother 0.2283 0.4207 0.2245 0.4174 0.1889 0.3916 0.2925 0.4558 0.3011 0.4723

Membership of mother in community organization3

-- -- -- -- -- -- -- -- 0.46022 0.51114

Number of hours of access to electricity per day3

-- -- -- -- -- -- -- -- 1.03409 1.76278

1 Variable present in data from Sta. Elena, Camarines Norte only. 2 Variable not included in empirical model; included in table only for illustration purposes. 3 Variable present in data from Brgy. Salvacion, Puerto Princesa, Palawan only.

19

Ordered Probit Results The results of the ordered probit regression are reported in Table 4. Don Tomas, having only children who were either overnourished or of normal nourishment (see Table 1), was estimated using binary probit instead. Results for Don Tomas are reported in Table 5. Residuals are normally distributed at a level of significance of less than 1% for all barangays, thereby validating the assumption of normal distribution and allowing for the use of parametric estimation. Reporting separate results for each barangay fulfils the very purpose of the survey: supplying community-based information in order to monitor trends specific to the community and guide the formation of effective policy. This also allows for comparability between the data for the barangays of Sta. Elena and that of Brgy. Salvacion. Different sets of variables appear to be correctly specified for certain barangays, and as such, presence of variables is not uniform across all regression equations. The LR statistic and LR index for each equation are also indicated in the table. In the interest of space, only regression results in barangays where the LR statistic is accepted at the 10% level of significance are presented here, as the data from the remaining barangays do not meet acceptable statistical criteria. Accordingly, reported only are coefficients (and standard errors) of the estimated equation for each of nine barangays in Sta. Elena, Camarines Norte and the single barangay of Salvacion in Puerto Princesa, Palawan. (See Appendix C for actual regression results.) Sex of the child appears to be significant in explaining the nutritional status of children in four of the ten barangays. Male children have a higher probability of being of normal nourishment in Bulala, Kabuluan and Salvacion, most especially, the coefficient in Salvacion being accepted at the 1% level of significance. Anecdotal evidence from Salvacion attribute this to the larger appetite observed of male children relative to female children. This may also be explained by the theory mentioned earlier: that there are physiological and genetic differences between people—here put forth in terms of gender difference—in the way their bodies harness food nutrients.

20

Table 4 Maximum Likelihood Estimates of Coefficients in Ordered Probit Regression

Dependent Variable: Nutritional Status

Basiad Bulala Kabuluan Maulawin Plaridel Camarines Norte Camarines Norte Camarines Norte Camarines Norte Camarines Norte

Sex of child 0.034906 0.328798** 0.358948* 0.170524 0.251581 (0.149229) (0.147904) (0.217284) (0.172901) (0.225693)

Proportion of household members -3.594549*** -1.198696 1.638058 -0.121054 2.846534** excluding mother employed (0.955703) (0.838265) (1.219133) (0.885802) (1.398202)

Total household income 0.0000659*** 0.0000222 0.00000465 0.000031 -0.0000331** per capita (0.0000137) (0.0000174) (0.0000175) (0.0000191) (0.0000148)

Access to safe water 0.552499*** -0.118798 -0.339862 -0.203851 0.673733** (0.155949) (0.183391) (0.384475) (0.184351) (0.33452)

Access to sanitary 0.139998 -0.142953 0.157656 0.447868** 0.029731 toilet facility (0.162412) (0.159157) (0.236476) (0.217209) (0.235061)

Food shortage within -0.011228 -0.465867* -1.424359*** 0.123727 -- last six months (1.03999) (0.273027) (0.494341) (0.233479) --

Age of mother -0.010937 -0.009511 0.005885 -0.015817 0.006153 (0.010152) (0.008672) (0.011421) (0.010906) (0.0134659)

Years of schooling -0.008457 -0.018189 0.026374 -0.024348 0.047765 of mother (0.028061) (0.029578) (0.042747) (0.04259) (0.053193)

Employment status -1.147135*** 0.092874 -0.051905 -0.272873 0.513463 of mother (0.265872) (0.177972) (0.327616) (0.198327) (0.317871)

LR statistic 60.02111 16.12669 19.96477 15.70612 14.4825 Probability (LR Stat) 0.0000000013 0.064283 0.018131 0.073278 0.070024 LR index (Pseudo-R2) 0.140615 0.037801 0.100056 0.054752 0.082069 No. of included observations 438 300 240 431 219 *** significant at the 1% level ** significant at the 5% level * significant at the 10% level (standard errors reported in parentheses)

21

Table 4 (cont.)

Maximum Likelihood Estimates of Coefficients in Ordered Probit Regression Dependent Variable: Nutritional Status

Poblacion San

Lorenzo Tabugon Salvacion Camarines Norte Camarines Norte Camarines Norte Palawan

Sex of child 0.264113 -0.033535 0.23001 0.676487*** (0.250041) (0.122739) (0.268377) (0.211170)

1.659193 0.452433 -0.01567 -- Proportion of household members

excluding mother employed (1.266097) (0.599652) (1.274409) --

Total household income 0.000000586 -0.00000171** -0.00000383 0.0000569** per capita (0.00000467) (0.000000748) (0.0000159) (0.0000245)

Access to safe water 0.28432 0.348284** -0.328475 -0.113685 (0.279636) (0.145028) (0.278849) (0.233756)

Access to sanitary -0.561148 -- -0.046985 0.218153 toilet facility (0.454414) -- (0.309785) (0.237684)

Food shortage within -2.914626*** -0.775896** -0.002872 -0.175694 last six months (0.906098) (0.312381) (0.738949) (0.237417)

Age of mother 0.000234 -0.005183 -0.062886*** -0.008557 (0.014503) (0.006573) (0.016732) (0.015111)

Years of schooling 0.123012*** 0.009974 -0.024598 0.101573*** of mother (0.041501) (0.02296) (0.062877) (0.037456)

Employment status 0.058622 -- -0.234291 -- of mother (0.332769) -- (0.333495) --

Membership of mother 0.426438* in community organization (0.232504)

Number of hours of access -0.165534** to electricity per day (0.071239)

LR statistic 25.81175 19.74876 31.67569 29.96476 Probability (LR Stat) 0.006922 0.031721 0.00086 0.000445 LR index (Pseudo-R2) 0.177562 0.108348 0.065904 0.116983 No. of included observations 1215 304 1085 176 *** significant at the 1% level ** significant at the 5% level * significant at the 10% level (standard errors reported in parentheses)

22

Table 5 Maximum Likelihood Estimates of Coefficients in Binary Probit Regression

Dependent Variable: Nutritional Status

Don Tomas Camarines Norte

Sex of child -0.799886**

(0.406708)

Proportion of household members -1.919054

excluding mother employed (2.257867)

Total household income 0.0000535

per capita (0.0000428)

Access to safe water -0.065774

(0.447503)

Access to sanitary 0.898117**

toilet facility (0.465518)

Age of mother -0.047855**

(0.022548)

Years of schooling -0.074506

of mother (0.078890)

Employment status 0.670754

of mother (0.605000) S.E. of regression 0.145763 Sum squared residuals 6.926503 No. of included observations 334 *** significant at the 1% level ** significant at the 5% level * significant at the 10% level (standard errors reported in parentheses)

This intrapersonal variation clearly cannot be captured by having a single standard of nutritional status for both male and female children. For Don Tomas, though the coefficient is negative at 5%, the same is true: male children have a higher probability of being of normal nourishment, since Don Tomas only has normally nourished and overnourished children. This is further evidence of the social and cultural reality that there are inherent biases in child-rearing within the household, usually favoring one gender over another. The proportion of household members, excluding the mother, economically active in the last twelve months evidently exerts a negative effect on the nutritional outcome of the population of children in Basiad, the coefficient being significant at 1%. This means that as the proportion of working household members excluding the mother increases, the probability of being undernourished for the children of that household also increases. This can be explained by considering this variable, as

23

mentioned, as a proxy for time devoted to health care of children. Working requires time outside of the house and necessarily, takes away time that could have otherwise been used to supervise the eating habits and respond to the health needs of very young children. Having only one or few adult members staying home may exert downward pressure on the nutritional status of children, especially as the number of children increases relative to the number of members who devote full attention to the care of children. The opposite, however, is true for Plaridel, where the coefficient is positive at 5%. In this particular barangay, absence from the home does not have an adverse relation to nutritional status of child and may be compensated for by better household management, though the contrary is observed in most of the other barangays. The important role that household income plays in determining the nutritional status of children is statistically verified in Basiad and Salvacion. As expected, income exerts a positive effect on the probability of better nourishment, as food security and the capacity for the medication of illnesses accompany higher levels of income. More affluent households can also purchase assets that support health, like refrigerators for keeping food safe and readily available. However, the same cannot be said for Plaridel and San Lorenzo, where the effect of income per capita is significant but negative. This may be because higher levels of income can also mean more time invested in economic activities, hectic work schedules of parents, and consequently, less time with children at home. Income does not seem to be significant in six out of ten barangays, thereby affirming for these locales the conjecture of Behrman and Deolalikar (1987) that income does not exhibit as great an effect on nutrition as do other socioeconomic factors. Access to safe drinking water, defined as water coming from the community water system, deep well, or artesian well, whether private or shared, shows a significant effect in explaining nutritional status of children in Plaridel, San Lorenzo, and more so in Basiad. The positive sign of the coefficients for these three barangays agrees with intuition: safe drinking water would more certainly protect children from infection and diseases and would subsequently increase their weight-for-age, than would water coming from dug or shallow wells, rivers, or streams. A similar trend can be observed with sanitary toilet facilities. In Don Tomas and Maulawin, having access to a sanitary toilet, whether a private or shared closed-flush facility, significantly increases the probability that a child is not undernourished. The positive effect observed in both barangays is as expected: adequate sanitation facilities control the disease environment and thus serve to assist the preservation of health and the maintenance of weight among children.

Ironically, number of hours of household access to electricity does not accompany greater chances of nutrition. It is surprising that although this is a significant variable for Salvacion, it implies an inverse relationship. Counter-intuitive as it may be, this only shows that electricity does not directly influence nutritional status. With an average of 1.034 hours of electricity a day (and a standard deviation of 1.763 hours), this kind of access to electricity may only be used for lighting at night, and as such does not immediately translate to the delivery of nutrition for children and may simply exhibit coincidence: children from households with access to more electricity tend to be undernourished. It is to be noted here that there is no data on access to electricity for Sta. Elena. Experiencing food shortage in the household sometime within the last six months before the survey was conducted considerably decreased the probability that a child is of better nutrition. This is observed in Bulala, Kabuluan, Poblacion and San Lorenzo, all four barangays showing negative coefficients, with the coefficients for Kabuluan and Poblacion significant at 1%. In fact, estimates are negative for all ten barangays. These findings simply confirm the direct link between food intake and nutritional status even for a short period of six months. The succeeding variables measure attributes of the mother, or the household member assumed to make key household decisions and take on the task childcare. The coefficient of the mother’s age is seen to be consistently negative where it is significant, namely in Don Tomas, and Tabugon, in decreasing

24

levels of significance. This says that the older the mother, the more chances there are that the child or children under her (or his) care has low weight-for-age. Older people, like grandparents, may be less able to take care of children properly because of deteriorating mental and physical health. It would also be more difficult for older people to keep up with new health practices and modern methods of healthcare.

Number of years of schooling of the mother also shows a positive effect in Poblacion and Salvacion, the variable being significant at 1% in both barangays. This observation confirms the notions postulated by Glewwe (1999) that (a) formal education directly taught health knowledge to future mothers; (b) literacy and numeracy skills in school assist future mothers in diagnosing and treating child problems; and (c) exposure to modern society from formal schooling makes women more receptive to modern medical treatments. It is interesting to note that both Poblacion and Salvacion are urban areas, showing that schooling gives comparative advantage to mothers and improves the nutritional status of their children especially in urban communities. Most of the barangays reported no statistical difference in probability of better nourishment between children with a working mother, and children whose mother is economically inactive. The variable is significant, however, in Basiad, where it exhibits a negative effect at 1%. This means that for this community, higher chances of being of normal nourishment accompany children whose mothers do not work, for these mothers probably have more time to spend at home giving full attention to their children, and responding as necessary to any of their children’s healthcare needs.

The Salvacion data set also shows that the mother’s membership in a community organization or association also significantly and positively influences nutritional status of children. This can be explained by thinking of community organizations as networking mechanisms between mothers where they share health knowledge among themselves and assist one another in times of need, whether through physical or financial support. Assessment of Feeding Programs

To assess the targeting of several feeding programs administered in Sta. Elena, Camarines Norte, a dummy variable for household participation in various supplemental feeding programs was included in equation (5), the empirical model to be estimated. These feeding programs, like the Department of Health’s Garantisadong Pambata, Oplan Busog, Sentrong Sigla and Tatak Sangkap Pinoy, are long-standing interventions and regular in their implementation. Including the dummy variable in the actual regression, however, led to estimation problems caused by near-singularity, which does not allow for estimation of coefficients. As such, the researchers omitted this variable and instead made use of two-way distribution tables to fulfill the second objective of this study. These two-way distribution tables for the twelve barangays of Sta. Elena where the feeding programs were administered, reported as Table 6, indicate the individual child’s nutritional status and whether he or she participated in a feeding program. As can be observed in Table 6.2, 85.3% of the overnourished children in Basiad participated in the feeding programs. Similarly, Table 6.11 shows that 26.3% of overnourished children in San Lorenzo also participated in these programs, as well as 1 child in Kabuluan. Whether they achieved this nutritional status by way of the feeding program, however, cannot be determined by this cross- sectional analysis. Only a dynamic approach would allow for examining changes to nutritional status over a period of time, while at the same time controlling for participation or non-participation in the programs. All other barangays report that their overnourished children do not participate in the feeding program. What is more alarming is the presence of undernourished children who do not participate in the feeding programs. As much as 17 out of 18 (94.44%) of the severely undernourished children from the twelve barangays where the programs are implemented are not recipients of the intervention. Likewise, 140 out of 225 (71.1%) of the moderately undernourished children also do not participate in the programs. The proportion of undernourished children who are still not exposed to the programs is especially pronounced in Bulala, Kagtalaba, Maulawin, Plaridel, Poblacion, Pulong Guit-guit and San

25

Lorenzo. As such, one can see that the feeding programs require better targeting, since substantially more of the undernourished children have yet to participate in the various supplemental feeding programs being administered.

Table 6

Distribution of Children According to Nutritional Status and Feeding Program Participation (Sta. Elena, Camarines Norte)

6.1 Basiad

Moderately Undernourished

Mildly Undernourished/ Normal Nourishment Overnourished Feeding Program

Participation Magnitude Proportion Magnitude Proportion Magnitude Proportion Total No 7 100 368 98.9 62 100 437

Yes 0 0 4 1.1 0 0 4

Total 7 100 372 100 62 100 441

6.2 Bulala

Moderately Undernourished

Mildly Undernourished/ Normal Nourishment Overnourished Feeding Program

Participation Magnitude Proportion Magnitude Proportion Magnitude Proportion Total

No 17 45.9 88 38.4 5 14.7 110

Yes 20 55.1 141 61.6 29 85.3 190

Total 37 100 229 100 34 100 300

6.3 Don Tomas Mildly Undernourished/

Normal Nourishment Overnourished Feeding Program Participation Magnitude Proportion Magnitude Proportion Total

No 318 97.5 8 100 326

Yes 8 2.5 0 0 8

Total 326 100 8 100 334

6.4 Kabuluan Moderately

Undernourished Mildly Undernourished/

Normal Nourishment Overnourished Feeding Program Participation Magnitude Proportion Magnitude Proportion Magnitude Proportion Total

No 6 20.7 18 8.6 1 50 25

Yes 23 79.3 191 91.4 1 50 215

Total 29 100 209 100 2 100 240

26

6.6 Maulawin

Severely Undernourished Moderately


Normal Nourishment Feeding Program Participation Magnitude Proportion Magnitude Proportion Magnitude Proportion Total

No 2 100 30 76.9 354 90.8 386

Yes 0 0 9 23.1 36 9.2 45

Total 2 100 39 100 390 100 431

6.9 Pulong Guit-guit



Normal Nourishment Feeding Program Participation Magnitude Proportion Magnitude Proportion Magnitude Proportion Total

No 6 100 16 100 258 98.9 280

Yes 0 0 0 0 3 1.1 3

Total 6 100 16 100 261 100 283

6.5 Kagtalaba Severely

Undernourished Moderately

Undernourished Mildly Undernourished/ Normal Nourishment Overnourished

Feeding Program Participation Magnitude

Proportion Magnitude



Proportion Total

No 5 100 9 100 195 96.1 2 100 211

Yes 0 0 0 0 8 3.9 0 0 8

Total 5 100 9 100 203 100 2 100 219

6.7 Plaridel



Normal Nourishment Overnourished Feeding Program Participation Magnitude Proportion Magnitude Proportion Magnitude Proportion Magnitude Proportion Total

No 1 100 23 95.8 187 96.4 1 100 212

Yes 0 0 1 4.2 7 3.6 0 0 8

Total 1 100 24 100 194 100 1 100 220

6.8 Poblacion Severely



Normal Nourishment Overnourished Feeding Program Participation Magnitude


Proportion Magnitude Proportion Magnitude Proportion Total

No 2 100 6 85.7 1095 88.7 2 100 1105 Yes 0 0 1 14.3 140 11.3 0 0 141

Total 2 100 7 100 1235 100 2 100 1246

27

6.10 Rizal Moderately


Normal Nourishment Feeding Program Participation Magnitude Proportion Magnitude Proportion Total

No 3 11.1 63 22.2 66

Yes 24 88.9 221 77.8 245

Total 27 100 284 100 311

6.12 Villa San Isidro

Severely Undernourished Mildly Undernourished/

Normal Nourishment Feeding Program Participation Magnitude Proportion Magnitude Proportion Total

No 1 100 46 97.9 47

Yes 0 0 1 2.1 1

Total 1 100 47 100 48

*Covers only the twelve barangays where the feeding programs are administered.

6.11 San Lorenzo Severely







Proportion Total

No 0 0 23 76.7 818 79.0 14 73.7 855

Yes 1 100 7 23.3 218 11.0 5 26.3 231

Total 1 100 30 100 1036 100 19 100 1246

6.13 Santa Elena* Severely







Proportion Total

No 17 94.4 140 71.1 3848 79.3 94 72.9 4099

Yes 1 5.6 85 28.9 978 20.7 35 27.1 1099

Total 18 100 225 100 4826 100 129 100 5198

28

VI. CONCLUSIONS AND RECOMMENDATIONS In summary, running an ordered probit analysis on nutritional status of children revealed that several socioeconomic determinants contributed significantly to the child’s production of health. Male children appear to be associated with normal nourishment moreso than female children. Also, household income per capita exhibits contrasting effects on nutritional status of children, as does the proportion of household members other than the mother employed. Household endowments also generally affect child nutrition positively: children from households with access to a sanitary toilet facility and a source of safe drinking water have greater probability of adequate nourishment. Attributes of the person taking care of the child also significantly affected the nourishment of children under his or her care. Negative effects accompany increases in the caretaker’s age, while positive effects come with longer periods of formal schooling. Having an employed caretaker was also observed to have a negative effect on their ward’s nutrional status in one barangay. Social capital, observed through a parent’s membership in a community organization or association, is also shown to be associated with better child nutrition. The significant socioeconomic determinants vary from one barangay to another as results vary across different communities. The mentioned variables, however, were significant determinants in at least one of the ten barangays where estimation met sufficient statistical criteria. As such, these socioeconomic factors can be said to be of importance in explaining nutritional disparities between Filipino children. Corollary to this, it could be said that the determinants of nutritional status are closely identified with peculiarities of the specific community. While food shortage was experienced throughout the barangays and is in fact associated with undernourishment for all ten barangays included in the study, this study can make no conclusions as regards the impact on nutrition of supplemental feeding programs implemented in Sta. Elena, Camarines Norte. All that can be assessed is the programs’ targeting, which can be said to require much improvement since these programs are not enjoyed by most of the undernourished children where these programs are implemented. Policy Implications Socioeconomic determinants influence barangays differently, as has been showed by this study. The method employed—that of using an ordered probit technique of estimation—appears to be highly appropriate for the analysis of data coming from the CBMS Survey. Since the CBMS Survey is targeted to be conducted nationwide, the employed method of analysis can be used alongside every CBMS site in the Philippines, and even in other CBMS sites throughout the developing world. This would greatly enhance the quality of information gathered from the particular community. And since the government intends to combat the prevalence and severity of malnutrition, it is of the essence that its causes are made known and its processes are understood. Those are in fact what this ordered probit analysis aims to fulfill.

The employed method of analysis also allows for significant socioeconomic variables to be identified. While some barangays report several significant variables, only one variable is reported to be significant in other communities. Also, certain variables are seen to be more significant than others. As such, this information would greatly inform policymakers and local government officials as to what are the gravest needs of a certain community, in order for priorities in any such efforts at intervention to be rationalized. Funds would then be allocated more efficiently and nutrition intervention programs would be structured more effectively. And since food shortage has been shown to be closely associated to undernourishment even in the short run, feeding programs aimed at ensuring food sufficiency during lean months and instances of calamity may improve child nutrition significantly.

29

As regards the implementation of supplemental feeding programs in Sta. Elena, much can be improved as regards their targeting. Key in addressing undernourishment is first identifying who the undernourished children are and where they are located. Using the estimated equation for each barangay may help predict the probability of a certain child’s nutritional status and possibly aid in correctly identifying priority cases. Subsequently, efforts should be devoted in making sure that the feeding programs reach these undernourished children first and foremost. Effects of the feeding programs should also be strictly monitored to allow for efficient reallocation of resources according to the needs of the child. This would ensure minimal leakages and maximum benefits for all children.

Moreover, the direction of the effects of socioeconomic variables, supplied by regression analysis, is crucial in determining the apt strategies to address malnutrition. Public investments in expanding household access to sources of safe drinking water and sanitation facilities would improve nutritional status of children, as its effects are generally positive where significant. Also, income transfers and subsidies may be fitting for barangays where the effect of income is positive; but for communities where the effect of income is negative, that may not be the best strategy. A daycare center for children in Basiad, for example, may arrest the negative effect of increasing number of working household members. In addition, promotion of health knowledge, being a good in itself, may be most beneficial if targeted also towards older parents, as the age of the child caretaker is shown to have an impact on nutritional status of children. Furthermore, health knowledge would also be best shared among relatively less schooled caretakers of children to compensate for their assumed lack of literacy and numeracy skills essential for health practices, as schooling also exhibits significant positive effects in certain areas. Finally, encouraging the formation of community organizations and providing other avenues for accumulating social capital may also positively affect the nutritional status of children. At bottom, a multidimensional approach is seen as necessary in effectively combatting malnutrition. The aforementioned actions are thus proposed for policymakers to consider. Recommendations for Further Studies The main area for consideration for further studies is using a dynamic approach in specifying a reduced form relation between health and the inputs to the health production function. A dynamic analysis would not only examine changes to nutritional status across time but also better capture the impact of any such intervention program implemented in the different communities. Since the CBMS Survey is currently being conducted on its second round for pilot areas in the country, consistent pooled data may be available for analysis in the near future. Secondly, more and various other socioeconomic determinants may be sought for and used in studies to be undertaken on the subject of nutritional status. Examples of these other variables are distance from community health center, dummy variables to mark lean months or incidences of calamity, price levels of food, medicine and other means of healthcare, and precise measures for nutrient intake, such as food intake recall or amount of food purchased. Finally, improvements in nutritional status can be better captured if there was access to anthropometric data of the children, instead of nutritional categories, as measurement of nutritional status would be more precise and more sensitive to changes. As such, the writers suggest using anthropometric data of children’s height and weight, coupled with data on their age measured in months, rather than merely in years.

30

APPENDIX A

List of Variables from CBMS Survey

MNUTIND nutritional status of child =1 if overnourished =0 if of normal nourishment = -1 if 1st degree (mildly) undernourished = -2 if 2nd degree (moderately) undernourished = -3 if 3rd degree (severely) undernourished SEX =1 if child is male, 0 otherwise AGE age of child in years HSIZE household size MWJOB number of household members employed MWJOBPROP proportion of household members employed MWJOBPROP2 proportion of household members excluding mother employed WATER household's main source of drinking water WATERCOM =1 if household's main source of drinking water is the community water system,

0 otherwise WATERSAFE =1 if household's main source of drinking water is the community water system,

deep well or artesian well, 0 otherwise TOIL kind of toilet facility household uses TOILHAV =1 if household has access to toilet facility, 0 otherwise TOILSAN =1 if household has access to sanitary (with water-sealed flush to septic tank/sewerage

system) toilet facility, 0 otherwise TENUR tenure status of lot occupied by household OWNHOUSE =1 if household owns house, 0 otherwise OWNLOT =1 if household owns lot, 0 otherwise WALL construction materials used for the walls of the house ROOF construction materials used for the walls of the house INCOMEENTREP household entrepreneurial income INCOMEWAGE household labor income INCOMEOTHER household income from other sources INCOMETOTAL total household income INCOMEENTREPPC household entrepreneurial income per capita INCOMEWAGEPC household labor income per capita INCOMEOTHERPC household income from other sources per capita INCOMETOTALPC total household income per capita FSHORT =1 if household experienced food shortage within the last 6 months, 0 otherwise CALMIND =1 if household had been displaced by natural calamity within the last 12 months, ]

0 otherwise M05FEED =1 if household participated in supplemental feeding program for 0-5 year old children,

0 otherwise AGEDAD age of household head in years SCHOOLNGDAD no. of years of schooling of household head ELEMGRADDAD =1 if household head is at least an elementary graduate, 0 otherwise HISCHGRADDAD =1 if household head is at least a high school graduate, 0 otherwise

31

COLLGRADDAD =1 if household head is at least a college graduate, 0 otherwise JOBINDDAD =1 if household head has been economically active during the past 12 months, 0

otherwise AGEMOM age of mother in years SCHOOLNGMOM no. of years of schooling of mother ELEMGRADMOM =1 if mother is at least an elementary graduate, 0 otherwise HISCHGRADMOM =1 if mother is at least a high school graduate, 0 otherwise COLLGRADMOM =1 if mother is at least a college graduate, 0 otherwise JOBINDMOM =1 if mother has been economically active during the past 12 months, 0 otherwise SEXMOM =1 if mother is female, 0 otherwise Additional variables from the Salvacion, Puerto Princesa data set: ELEC =1 if household has access to electricity, 0 otherwise ELECHOURS no. of hours electricity is available to household CREDIT =1 if at least 1 household availed of credit or loan program in the last 6 months,

0 otherwise DISPOSAL =1 if household burns their garbage, 0 otherwise SEGREG =1 if household segregates their garbage, 0 otherwise LANDUSEOWN =1 if household tills land and owns it, 0 otherwise LANDUSERENT =1 if household tills land and rents it, 0 otherwise BOATUSEOWN =1 if household engages in fishing and owns boat, 0 otherwise BOATUSERENT =1 if household engages in fishing and rents boat, 0 otherswise COOPASSOCDAD =1 if household head is member of at least 1 community organization or association,

0 otherwise COOPASSOCMOM =1 if mother is member of at least 1 community organization or association,

0 otherwise

32

APPENDIX B

Standard Weight-for-Age (as according to Barangay Health Center)

WEIGHT (kilograms) UNDERWEIGHT OVERWEIGHT

AGE SEVERE MODERATE MILD

NORMAL/ AVERAGE

MILD MOD. - SEVERE

(months) =< From To From To From To From To > =

0 2.1 2.2 2.6 2.7 3.0 3.1 4.6 4.7 5.4 5.5 1 2.6 2.7 3.3 3.4 3.8 3.9 5.6 5.7 6.5 6.6 2 3.2 3.3 3.9 4.0 4.5 4.6 6.5 6.6 7.5 7.6 3 3.6 3.7 4.4 4.5 5.1 5.2 7.2 7.3 8.2 8.3 4 4.0 4.1 4.9 5.0 5.6 5.7 7.9 8.0 8.9 9.0 5 4.3 4.4 5.2 5.3 6.0 6.1 8.3 8.4 9.4 9.5 6 4.5 4.6 5.5 5.6 6.3 6.4 8.8 8.9 9.9 10.0 7 4.7 4.8 5.8 5.9 6.7 6.8 9.3 9.4 10.3 10.4 8 4.9 5.0 6.0 6.1 7.0 7.1 9.7 9.8 10.8 10.9 9 5.1 5.2 6.3 6.4 7.3 7.4 10.0 10.1 11.2 11.3 10 5.3 5.4 6.0 6.1 7.5 7.6 10.4 10.5 11.5 11.6 11 5.5 5.6 6.7 6.8 7.8 7.9 10.8 10.9 11.8 11.9 12 5.7 5.8 7.0 7.1 8.0 8.1 11.1 11.2 12.1 12.2 13 5.9 6.0 7.1 7.2 8.2 8.3 11.3 11.4 12.4 12.5 14 6.0 6.1 7.3 7.4 8.4 8.5 11.6 11.7 12.7 12.8 15 6.1 6.2 7.4 7.5 8.5 8.6 11.8 11.9 13.0 13.1 16 6.2 6.3 7.6 7.7 8.7 8.8 12.0 12.1 13.3 13.4 17 6.3 6.4 7.7 7.8 8.8 8.9 12.2 12.3 13.6 13.7 18 6.4 6.5 7.8 7.9 9.0 9.1 12.5 12.6 13.9 14.0 19 6.5 6.6 7.9 8.0 9.2 9.3 12.7 12.8 14.2 14.3 20 6.6 6.7 8.1 8.2 9.3 9.4 12.9 13.0 14.4 14.5 21 6.7 6.8 8.2 8.3 9.4 9.5 13.1 13.2 14.7 14.8 22 6.8 6.9 8.3 8.4 9.6 9.7 13.2 13.3 14.9 15.0 23 6.9 7.0 8.4 8.5 9.7 9.8 13.4 13.5 15.2 15.3 24 7.0 7.1 8.5 8.6 9.8 9.9 13.6 13.7 15.4 15.5 25 7.1 7.2 8.7 8.8 10.0 10.1 13.7 13.8 15.7 15.8 26 7.2 7.3 8.8 8.9 10.1 10.2 13.9 14.0 15.9 16.0 27 7.3 7.4 8.9 9.0 10.2 10.3 14.1 14.2 16.1 16.2 28 7.4 7.5 9.0 9.1 10.3 10.4 14.2 14.3 16.3 16.4 29 7.5 7.6 9.1 9.2 10.4 10.5 14.4 14.5 16.6 16.7 30 7.6 7.7 9.2 9.3 10.6 10.7 14.6 14.7 16.8 16.9 31 7.7 7.8 9.3 9.4 10.7 10.8 14.7 14.8 17.0 17.1 32 7.8 7.9 9.4 9.5 10.8 10.9 14.9 15.0 17.2 17.3 33 7.9 8.0 9.5 9.6 10.9 11.0 15.0 15.1 17.4 17.5 34 8.0 8.1 9.6 9.7 11.1 11.2 15.2 15.3 17.6 17.7 35 8.1 8.2 9.7 9.8 11.2 11.3 15.4 15.5 17.8 17.9 36 8.2 8.3 9.8 9.9 11.3 11.4 15.5 15.6 17.9 18.0 37 8.3 8.4 10.0 10.1 11.4 11.5 15.7 15.8 18.1 18.2 38 8.4 8.5 10.1 10.2 11.5 11.6 15.9 16.0 18.3 18.4 39 8.5 8.6 10.2 10.3 11.7 11.8 16.0 16.1 18.5 18.6

33

40 8.6 8.7 10.3 10.4 11.8 11.9 16.2 16.3 18.6 18.7 41 8.7 8.8 10.5 10.6 11.9 12.0 16.3 16.4 18.8 18.9 42 8.7 8.8 10.6 10.7 12.0 12.1 16.0 16.1 19.0 19.1 43 8.9 9.0 10.7 10.8 12.1 12.2 16.6 16.7 19.1 19.2 44 9.0 9.1 10.8 10.9 12.2 12.3 16.7 16.8 19.3 19.4 45 9.1 9.2 10.9 11.0 12.3 12.4 16.9 17.0 19.4 19.5 46 9.2 9.3 11.0 11.1 12.4 12.5 17.1 17.2 19.6 19.7 47 9.3 9.4 11.1 11.2 12.5 12.6 17.2 17.3 19.8 19.9 48 9.4 9.5 11.3 11.4 12.7 12.8 17.3 17.4 19.9 20.0 49 9.5 9.6 11.4 11.5 12.8 12.9 17.5 17.6 20.1 20.2 50 9.6 9.7 11.5 11.6 12.9 13.0 17.6 17.7 20.2 20.3 51 9.6 9.7 11.6 11.7 13.0 13.1 17.7 17.8 20.3 20.4 52 9.7 9.8 11.7 11.8 13.1 13.2 17.9 18.0 20.5 20.6 53 9.8 9.9 11.8 11.9 13.2 13.3 18.0 18.1 20.6 20.7 54 9.9 10.0 11.9 12.0 13.3 13.4 18.1 18.2 20.8 20.9 55 10.0 10.1 12.0 12.1 13.4 13.5 18.2 18.3 20.9 21.0 56 10.1 10.2 12.1 12.2 13.5 13.6 18.4 18.5 21.0 21.1 57 10.3 10.4 12.2 12.3 13.6 13.7 18.5 18.6 21.2 21.3 58 10.4 10.5 12.3 12.4 13.7 13.8 18.6 18.7 21.3 21.4 59 10.5 10.6 12.4 12.5 13.9 14.0 18.7 18.8 21.4 21.5 60 10.6 10.7 12.5 12.6 14.0 14.1 18.9 19.0 21.6 21.7

34

APPENDIX C Econometrics Views Regression Results

Basiad, Sta. Elena, Camarines Norte

Dependent Variable: MNUTIND Method: ML - Ordered Probit Date: 09/21/05 Time: 08:52 Sample: 1 441 Included observations: 438 Excluded observations: 3 Number of ordered indicator values: 3 Convergence achieved after 10 iterations Covariance matrix computed using second derivatives

Coefficient Std. Error z-Statistic Prob. SEX 0.034906 0.149229 0.233912 0.8151 MWJOBPROP2 -3.594549 0.955703 -3.761157 0.0002 INCOMETOTALPC 0.0000659 0.0000137 4.822326 0 WATERSAFE 0.552499 0.155949 3.542819 0.0004 TOILSAN 0.139998 0.162412 0.861994 0.3887 FSHORT -0.011228 1.03999 -0.010796 0.9914 AGEMOM -0.010937 0.010152 -1.077298 0.2813 SCHOOLNGMOM -0.008457 0.028061 -0.301363 0.7631 JOBINDMOM -1.147135 0.265872 -4.314618 0

Limit Points

LIMIT_0:C(10) -3.089637 0.529503 -5.834979 0 LIMIT_1:C(11) 0.615728 0.473183 1.301246 0.1932

Akaike info criterion 0.887734 Schwarz criterion 0.990255 Log likelihood -183.4137 Hannan-Quinn criter. 0.928186 Restr. log likelihood -213.4242 Avg. log likelihood -0.418753 LR statistic (9 df) 60.02111 LR index (Pseudo-R2) 0.140615 Probability (LR stat) 1.33E-09

35

Brgy. Bulala, Sta. Elena, Camarines Norte Dependent Variable: MNUTIND Method: ML - Ordered Probit Date: 09/21/05 Time: 08:53 Sample: 442 741 Included observations: 300 Number of ordered indicator values: 3 Convergence achieved after 8 iterations Covariance matrix computed using second derivatives

Coefficient Std. Error z-Statistic Prob. SEX 0.328798 0.147904 2.223041 0.0262 MWJOBPROP2 -1.198696 0.838265 -1.429972 0.1527 INCOMETOTALPC 0.0000222 0.0000174 1.276696 0.2017 WATERSAFE -0.118798 0.183391 -0.647784 0.5171 TOILSAN -0.142953 0.159157 -0.898187 0.3691 FSHORT -0.465867 0.273027 -1.706302 0.088 AGEMOM -0.009511 0.008672 -1.096843 0.2727 SCHOOLNGMOM -0.018189 0.029578 -0.614956 0.5386 JOBINDMOM 0.092874 0.177972 0.521843 0.6018

Limit Points

LIMIT_0:C(10) -1.824798 0.505804 -3.607718 0.0003 LIMIT_1:C(11) 0.641241 0.493223 1.300104 0.1936

Akaike info criterion 1.441659 Schwarz criterion 1.577464 Log likelihood -205.2488 Hannan-Quinn criter. 1.496008 Restr. log likelihood -213.3122 Avg. log likelihood -0.684163 LR statistic (9 df) 16.12669 LR index (Pseudo-R2) 0.037801 Probability (LR stat) 0.064283

36

Brgy. Kabuluan, Sta. Elena, Camarines Norte

Dependent Variable: MNUTIND Method: ML - Ordered Probit Date: 09/21/05 Time: 08:53 Sample: 1184 1423 Included observations: 240 Number of ordered indicator values: 3 Convergence achieved after 9 iterations Covariance matrix computed using second derivatives

Coefficient Std. Error z-Statistic Prob. SEX 0.358948 0.217284 1.651976 0.0985 MWJOBPROP2 1.638058 1.219133 1.343626 0.1791 INCOMETOTALPC 0.00000465 0.0000175 0.26646 0.7899 WATERSAFE -0.339862 0.384475 -0.883963 0.3767 TOILSAN 0.157656 0.236476 0.666691 0.505 FSHORT -1.424359 0.494341 -2.88133 0.004 AGEMOM 0.005885 0.011421 0.515262 0.6064 SCHOOLNGMOM 0.026374 0.042747 0.616976 0.5373 JOBINDMOM -0.051905 0.327616 -0.158432 0.8741

Limit Points

LIMIT_0:C(10) -0.481195 0.770538 -0.624491 0.5323 LIMIT_1:C(11) 3.358303 0.860403 3.903176 0.0001

Akaike info criterion 0.839877 Schwarz criterion 0.999406

Log likelihood -89.78522 Hannan-Quinn criter. 0.904156

Restr. log likelihood -99.76761 Avg. log likelihood -0.374105 LR statistic (9 df) 19.96477 LR index (Pseudo-R2) 0.100056 Probability (LR stat) 0.018131

37

Brgy. Maulawin, Sta. Elena, Camarines Norte


Coefficient Std. Error z-Statistic Prob. SEX 0.170524 0.172901 0.98625 0.324 MWJOBPROP2 -0.121054 0.885802 -0.13666 0.8913 INCOMETOTALPC 0.000031 0.0000191 1.623282 0.1045 WATERSAFE -0.203851 0.184351 -1.10578 0.2688 TOILSAN 0.447868** 0.217209 2.061923 0.0392 FSHORT 0.123727 0.233479 0.529929 0.5962 AGEMOM -0.015817 0.010906 -1.450335 0.147 SCHOOLNGMOM -0.024348 0.04259 -0.571683 0.5675 JOBINDMOM -0.272873 0.198327 -1.375876 0.1689

Limit Points

LIMIT_-2:C(10) -3.166152 0.674855 -4.691605 0 LIMIT_0:C(11) -1.790022 0.61641 -2.903948 0.0037

Akaike info criterion 0.680172 Schwarz criterion 0.783948 Log likelihood -135.5771 Hannan-Quinn criter. 0.721146 Restr. log likelihood -143.4302 Avg. log likelihood -0.314564 LR statistic (9 df) 15.70612 LR index (Pseudo-R2) 0.054752 Probability(LR stat) 0.073278

38

Brgy. Plaridel, Sta. Elena, Camarines Norte


Coefficient Std. Error z-Statistic Prob. SEX 0.251581 0.225693 1.114704 0.265 MWJOBPROP2 2.846534 1.398202 2.035854 0.0418 INCOMETOTALPC -0.0000331 0.0000148 -2.231707 0.0256 WATERSAFE 0.673733 0.33452 2.014028 0.044 TOILSAN 0.029731 0.235061 0.126482 0.8994 AGEMOM 0.006153 0.013465 0.456949 0.6477 SCHOOLNGMOM 0.047765 0.053193 0.897964 0.3692 JOBINDMOM 0.513463 0.317871 1.615322 0.1062

Limit Points

LIMIT_-2:C(9) -1.370532 0.876016 -1.564506 0.1177 LIMIT_0:C(10) 0.103858 0.825198 0.125858 0.8998 LIMIT_1:C(11) 4.289156 0.968901 4.426827 0


39

Brgy. Poblacion, Sta. Elena, Camarines Norte


Coefficient Std. Error z-Statistic Prob. SEX 0.264113 0.250041 1.056281 0.2908 MWJOBPROP2 1.659193 1.266097 1.310478 0.19 INCOMETOTALPC 0.000000586 0.00000467 0.12571 0.9 WATERSAFE 0.28432 0.279636 1.016749 0.3093 TOILSAN -0.561148 0.454414 -1.234882 0.2169 FSHORT -2.914626 0.906098 -3.21668 0.0013 AGEMOM 0.000234 0.014503 0.016156 0.9871 SCHOOLNGMOM 0.123012 0.041501 2.964082 0.003 JOBINDMOM 0.058622 0.332769 0.176163 0.8602

Limit Points

LIMIT_-2:C(10) -1.966281 0.847285 -2.320685 0.0203 LIMIT_0:C(11) -1.271039 0.810014 -1.569156 0.1166 LIMIT_1:C(12) 4.740005 0.961516 4.929721 0


40

Brgy. San Lorenzo, Sta. Elena, Camarines Norte


Coefficient Std. Error z-Statistic Prob. SEX -0.033535 0.122739 -0.273224 0.7847 MWJOBPROP2 0.452433 0.599652 0.754492 0.4506 INCOMETOTALPC -0.00000171 0.000000748 -2.291012 0.022 WATERSAFE 0.348284 0.145028 2.401487 0.0163 FSHORT -0.775896 0.312381 -2.483809 0.013 AGEMOM -0.005183 0.006573 -0.788475 0.4304 SCHOOLNGMOM 0.009974 0.02296 0.434393 0.664

Limit Points

LIMIT_-2:C(8) -3.008885 0.480231 -6.265491 0 LIMIT_0:C(9) -1.730794 0.379911 -4.555789 0 LIMIT_1:C(10) 2.376449 0.390841 6.080348 0


41

Brgy. Tabugon, Sta. Elena, Camarines Norte


Coefficient Std. Error z-Statistic Prob. SEX 0.23001 0.268377 0.857041 0.3914 MWJOBPROP2 -0.01567 1.274409 -0.012296 0.9902 INCOMETOTALPC -0.00000383 0.0000159 -0.240541 0.8099 WATERSAFE -0.328475 0.278849 -1.177967 0.2388 TOILSAN -0.046985 0.309785 -0.151668 0.8794 FSHORT -0.002872 0.738949 -0.003886 0.9969 AGEMOM -0.062886 0.016732 -3.758337 0.0002 SCHOOLNGMOM -0.024598 0.062877 -0.3912 0.6956 JOBINDMOM -0.234291 0.333495 -0.702531 0.4823

Limit Points

LIMIT_0:C(10) -4.82153 1.07727 -4.475691 0 LIMIT_1:C(11) -0.34917 0.902068 -0.387078 0.6987


42

Brgy. Salvacion, Puerto Princesa, Palawan


Coefficient Std. Error z-Statistic Prob. SEX 0.676487 0.211170 3.203525 0.0014 INCOMETOTALPC 5.69E-05 2.45E-05 2.320975 0.0203 WATERSAFE -0.113685 0.233756 -0.486339 0.6267 TOILSAN 0.218153 0.237684 0.917830 0.3587 FSHORT -0.175694 0.237417 -0.740022 0.4593 AGEMOM -0.008557 0.015111 -0.566259 0.5712 SCHOOLNGMOM 0.101573 0.037456 2.711806 0.0067 COOPASSOCMOM 0.42643 0.232504 1.834114 0.0666 ELECHOURS -0.165534 0.071239 -2.323622 0.0201

Limit Points

LIMIT_-1:C(10) -0.526825 0.606723 -0.868313 0.3852 LIMIT_0:C(11) 0.774235 0.594958 1.301326 0.1931 LIMIT_1:C(12) 4.216324 0.748707 5.631475 0.0000


43

Brgy. Don Tomas, Sta. Elena, Camarines Norte

Dependent Variable: MNUTIND Method: ML - Binary Probit Date: 10/03/05 Time: 20:05 Sample: 742 1075 Included observations: 334 Convergence achieved after 9 iterations Covariance matrix computed using second derivatives

Variable Coefficient Std. Error z-Statistic Prob. SEX -0.799886 0.406708 -1.966734 0.0492 MWJOBPROP2 -1.919054 2.257867 -0.849941 0.3954 INCOMETOTALPC 5.35E-05 4.28E-05 1.250863 0.211 WATERSAFE -0.065774 0.447503 -0.14698 0.8831 TOILSAN 0.898117 0.465518 1.929286 0.0537 AGEMOM -0.047855 0.022548 -2.12238 0.0338 SCHOOLNGMOM -0.074506 0.07889 -0.944433 0.3449 JOBINDMOM 0.670754 0.605 1.108683 0.2676

Mean dependent var 0.023952 S.D. dependent var 0.153129 S.E. of regression 0.145763 Akaike info criterion 0.21925 Sum squared resid 6.926503 Schwarz criterion 0.310535 Log likelihood -28.61471 Hannan-Quinn criter. 0.255646 Avg. log likelihood -0.085673

Obs with Dep=0 326 Total obs 334 Obs with Dep=1 8

44

SELECTED REFERENCES

Alderman, Harold, Jere R. Behrman, Victor Lavy, and Rekha Menon. 2001. “Child Health and School

Enrollment: A Longitudinal Analysis” The Journal of Human Resources 36(1): 185-205. Anand, S. and C. J. Harris. 1992. “Issues in the Measurement of Undernutrition.” In Nutrition and

poverty. New York: Oxford University Press Inc. Arcand, Jean-Louis. 2001. “Undernourishment and Economic Growth: The Efficiency Cost of Hunger.”

Speech. European Media Seminar on Global Food Security. <www.fao.org/News/2001/stockholm/arcandfull.pdf>. Accessed 4 July 2005.

Arnold, Jesse C., R. W. Engel, D. B. Aguillon, and M. Caedo. 1981. “Utilization of Family Characteristics

in nutritional classification of preschool children.” American Journal of Clinical Nutrition 34: 2546-2550.

Barrera, Albino. 1990. “The Interactive Effects of Mother's Schooling and Unsupplemented Breastfeeding on Child Health.” Journal of Development Economics 34(1): 81-98. Behrman, Jere R and Anil B. Deolalikar. 1987. “Will Developing Country Nutrition Improve with

Income? A Case Study for Rural South India.” The Journal of Political Economy 95(3): 492-507. __________. 1988. “Health and Nutrition.” In Behrman, J. and Srinivasan, T. N. (eds.), Handbook of

Development Economics 1: 631-711. Amsterdam: Elsevier. Bhargava, Alok. 1992. “Malnutrition and the Role of Individual Variation with Evidence from the India

and the Philippines. Journal of the Royal Statistical Society, Series A (Statistic in Society) 155 (2): 221-231.

__________. 1994. “Modelling the Health of Filipino Children.” Journal of the Royal Statistical Society.

Series A (Statistic in Society) 157(3): 417-432. Cebu Study Team. 1992. “A Child Health Production Function Estimated from Longitudinal Data.” Journal of Development Economics 38(2): 323-51. Chaudhuri, Anoshua. 2003. “Program Impact on the Health and Nutritional Status of Children: Evidence

from rural Bangladesh.” Ph. D diss., University of Washington. <depts.washington.edu/crfam/WorkingPapers/CRF%202002-08_Chaudhuri.pdf>. Accessed 6 July 2005.

Cranfield, J. A. L. and Erik Magnusson. 2003. “Canadian Consumer’s Willingness-To-Pay for Pesticide

Free Food Products: An Ordered Probit Analysist.” International Food and Agribusiness Management Review 6(4): 13-30.

Duflo, Esther. 2000. “Child Health and Household Resources in South Africa: Evidence from the Old

Age Pension Program.” The American Economic Review 90(2): 393-398.

45

Federov, Leonid and David E. Sahn. 2003. “Socio-Economic Determinants of Children’s Health in Russia: Estimating a Dynamic Health Production Function.” Cornell University working paper. <www.hgu.gu.se/files/nationalekonomi/sem/050920%20anton%20ppr3.doc>. Accessed 5 July 2005.

Fogel, R. W. 1992. “The European Escape from Hunger.” In Nutrition and poverty. New York: Oxford

University Press Inc. Foster, Phillips. 1992. The world food problem: tackling the causes of undernutrition in the third world.

Boulder: Lynne Rienner Publishers, Inc. Gertler, Paul and Jennifer Zeitlin. 2002. “Do Investments in Child Education and Nutrition Improve

Adult Health?” In The economics of health care in Asia-Pacific countries. Cheltenham: Edward Elgar Publishing Limited.

Grossman, Michael. 1972. “On the Concept of Health Capital and the Demand for Health.” The Journal

of Political Economy 80(2): 223-255. Gopalan, G. 1992. “Undernutrition: Measurement and Implications.” In Nutrition and poverty. New

York: Oxford University Press Inc. Glewwe, Paul. 1999. “Why does Mother's Schooling Raise Child Health in Developing Countries?

Evidence from Morocco” The Journal of Human Resources 34(1): 124-159. Herrin, Alejandro N. 1992. “MIMAP Research Paper No. 7: MIMAP on Health, Nutrition and Education.” Paper Presented During the Technical Workshop on the Results of the Second Phase of the MIMAP Project. Published as PIDS Working Paper Series No. 92-18. Mackinnon, John. 1995. “Health as an Informational Good: The Determinants of Child Nutrition

and Mortality During the Political and Economic Recovery in Uganda.” Working Paper Series No. 95-9. Centre for the Study of African Economies, University of Oxford.

Musgrove, P. with Dean T. Jamison and Joanna Leslie. 2004. “Malnutrition and Dietary Protein.” In

Health economics in development. Washington D.C.: The World Bank. Original, United Nations University, 2003.

Osmani, S. R. 1992. “Introduction.” In Nutrition and poverty. New York: Oxford University Press Inc. Payne, P. 1992. “Assessing Undernutrition.” In Nutrition and poverty. New York: Oxford University

Press Inc. Pitt, M. 1983. “Food Preferences and Nutrition in Rural Bangladesh.” Review of Economics and Statistics 65(1): 105-114. Reyes, Celia M. 2001. “MIMAP Research Paper No. 50: Poverty Profile of the Philippines.”

Paper presented during the Fourth Meeting of the MIMAP Research Network. (Unpublished) Sahn, D. 1988. “The Effect of Price and Income Changes in Food-Energy Intake in Sri Lanka.”

46

Economic Development and Cultural Change 36(2): 315-340. Smith, Lisa C., Marie T. Ruel, and Aida Ndiaye. 2004. “Why is Child Malnutrition Lower in Urban than

Rural Areas: Evidence from 36 Developing Countries.” Food Consumption and Nutrition Division of the International Food Policy Research Institute Discussion Paper 176.

Srinivasan, T. N. 1992. “Undernutrition: Measurement and Policy.” In Nutrition and poverty. New York:

Oxford University Press Inc. Subramanian, Shankar & Deaton, Angus. 1996. “The Demand for Food and Calories: Further Evidence

from India.” Journal of Political Economy 104(1): 133-62. Thomas, Duncan. 1994. “Like Father, like Son; Like Mother, like Daughter: Parental Resources and

Child Height.” The Journal of Human Resources (Special Issue: The Family and Intergenerational Relations) 29(4): 950-988.

Todaro, Michael P. and Stephen C. Smith. 2003. Economic development. 8th ed. Singapore: Pearson

Education. World Health Organization. 1998. Fifty years of the World Wealth Organization in the western Pacific

region: report of the regional director to the regional committee for the western Pacific, part 2. Manila: World Health Organization.

socioeconomic determinants of the nutritional … · 2018-05-02 · relationships between child...

Documents