Appendix 6: Explaining Trends in Child Underweight in the Developing World
The causes of child undernutrition are complex, multidimensional, and interrelated, ranging from factors as fundamental as political instability and slow economic growth to those as specific in their manifestation as respiratory infection and diarrhoeal disease. This is well illustrated by the framework in Appendix 1. Determinants also differ considerably across geographical areas. Using cross-country regression analysis, it is nonetheless possible to gain a general sense of the relative importance and contribution of some broad causal factors for the developing world as a whole. A recent study undertook such an analysis using data on countries’ underweight rates collected from 1970 to 1995 for children under five years of age.1
Variables, Data, and Methods Employed
The study used the conceptual framework in Appendix 1, in which the determinants of child undernutrition are broken into three levels of causality: immediate (most proximate), underlying, and basic (the deepest level). It focused on the three underlying determinants: food security, care for mothers and children, and health environment. It also considered the roles of two basic determinants: per capita national incomes and democracy.
The dependent variable was national child underweight rates, mainly drawn from the WHO Global Database on Child Growth and Malnutrition. Four explanatory variables were employed to represent the underlying determinants. These were national food availability (for food security), women’s education and status relative to men (for care), and access to water (for health environment). The operational measures of these variables were per capita dietary energy supplies (DES), female secondary school enrollment, a ratio of female to male life expectancy at birth (a measure of women’s status relative to men), and population with access to safe water. For the basic determinants, the operational measure of national income was gross domestic product (GDP) per capita in purchasing power parity U.S. dollars, and that for democracy was a seven-point index of political rights and civil liberties.
The analysis employed “country fixed-effects” multivariate regression using data from at least two points in time over the study period for 63 countries (n = 179). This method allows estimation of the effects of any hypothesized determinant while controlling for both other factors that change over time and time-invariant country-specific factors that influence underweight. Examples of these latter variables may be agro-climatic features and entrenched social norms. Separate regression equations were specified for the underlying and basic determinants so that the magnitude of the variables’ independent effects could be estimated. A series of tests was conducted to ascertain as far as possible that the relationships identified are causal rather than merely associative and that the estimated magnitudes of the effects are as accurate as possible.
All of the determinants considered were found to have statistically significant, negative effects on underweight rates. Per capita dietary energy supplies and per capita national incomes were found to have a declining marginal effect such that as they increase the strength of their impact weakens. The effects of the basic determinants were found to occur mainly through the underlying determinants (national incomes via all four underlying determinants; democracy via health environment improvements and increased food availabilities).
TABLE A6.1: Contributions of various determinants to reducing the developing-country underweight rate, 1970-95
Drawing on the regression results, the study estimated the contribution each underlying determinant made to the decline in the developing-country underweight rate from 1970 to 1995. Two factors were taken into account. First, the strength of each determinant’s impact was measured by the size of the increase required to reduce the underweight rate by 1 percentage point. The smaller the required increase, the greater the strength of impact. These numbers are given in the upper panel of Table A6.1, column 1.
It would take only a 4.6 percentage-point increase in female secondary school enrollment to reduce the child underweight rate by 1 percentage point, implying that women’s education up to the secondary level has a very strong impact. In fact, from an absolute standpoint, all of the underlying determinants considered have a fairly strong impact. A scale-neutral measure that allows consideration of the relative strength of impact of the variables is given in column 2. It indicates that for the developing countries as a whole, women’s education has the strongest effect on underweight rates among the underlying determinants, followed closely by countries’ per capita dietary energy supplies. Women’s relative status comes in third, followed by access to water.
The second factor taken into account in estimating the determinants’ relative contributions is the amount that each actually changed over the 25 years, which is partially a matter of public policy choices. Column 3 of the table reports the increase of each determinant, with the numbers again standardized by range, so that a scale-neutral comparison can be made. Water access increased by far the most, followed by female secondary school enrollment and per capita dietary energy supplies. The determinant that changed the least was women’s relative status. Based on the numbers in columns 1 and 3 in the table, the estimated contribution of each determinant to the total reduction in the developing-country underweight rate is given in column 4. Per cent contributions are illustrated in Figure A6.1.
The study found that increases in one of the proxies for care, women’s education, have made the greatest contribution, being responsible for 43% of the total reduction caused by underlying determinants. This is due to both its strong effect and fairly large increases over the period. Improvements in food availability contributed to 26% of the reduction; health environment improvements, to 19%. Partly because there has been little improvement over the 25 years, the contribution of women’s relative status, important for both care provision and food production, while still substantial, was estimated to be the lowest (about 12%).
The lower panel of the table gives the same numbers for the two basic determinants considered. Column 1 indicates that per capita national income has a strong impact on child undernutrition, while from an absolute standpoint, democracy also has the potential to make a substantial difference.
A key message of the study is that any comprehensive strategy for attacking the problem of child malnutrition must include actions to address both its basic and underlying causes. If national income is not improved, the resources necessary for investing in health environments, women’s education and status, and food availabilities will not be available.
Actions at both levels will bolster the crucial efforts of more direct nutrition interventions, such as breastfeeding promotion, nutrition communications, food fortification, and micronutrient supplementation.
[Ukrainian] [English] [Russian]