Understanding wealth-based inequalities in child health in India: A decomposition approach

Abstract

India experienced tremendous economic growth since the mid-1980s but this growth was paralleled by sharp rises in economic inequality. Urban areas experienced greater economic growth as well as greater increases in economic inequality than rural areas. During the same period, child health improved on average but socioeconomic differentials in child health persisted. This paper attempts to explain wealth-based inequalities in child mortality and malnutrition using a regression-based decomposition approach. Data for the analysis come from the 1992/93, 1998/99, and 2005/06 Indian National Family Health Surveys. Inequalities in child health are measured using the concentration index. The concentration index for each outcome is then decomposed into the contributions of wealth-based inequality in the observed determinants of child health. Results indicate that mortality inequality declined in urban areas but remained unchanged or increased in rural areas. Malnutrition inequality increased dramatically both in urban and rural areas. The two largest individual/household-level sources of disparities in child health are (i) inequality in the distribution of wealth itself, and (ii) inequality in maternal education. The contributions of observed determinants (i) to neonatal mortality inequality remained unchanged, (ii) to child mortality inequality increased, and (ii) to malnutrition inequality increased. It is possible that the increases in child health inequality reflect urban biases in economic growth, and the mixed performance of public programs that could have otherwise offset the impacts of unequal growth.

Introduction

Despite improvements in average levels of child mortality and malnutrition in India in recent decades, inequalities in child health persist (International Institute for Population Sciences & Macro International, 2007). The importance of household-level determinants to child health is well-established (Mosley & Chen, 1984). Naturally, inequality in the distribution of these determinants may contribute to inequality in child health. Much has been written about economic inequality in India, particularly in the years since the ‘liberalization’ of the Indian economy in the early 1990s. One such study presents an extensive review of the literature on inequality trends in India between the mid-1980s and the mid-2000s (Pal & Ghosh, 2007). To summarize, there was an increase in inequality between states from 1987/88 to 1999/2000, even when looking separately at rural areas and urban areas. There was also an increase in intra-urban and intra-rural inequality at the all-India level (first a decrease from 1987/88 to 1993/94, and then an increase from 1993/94 to 1999/00) (Deaton & Dreze, 2002; Sen & Himanshu, 2005). Urban areas were more unequal than rural areas, with an increase in inequality between urban and rural areas. There is some evidence to indicate a continuation in earlier patterns in inequality after 2000 – there was an increase in inequality within states, as well as an increase in intra-rural and intra-urban inequality (Mishra & Ray, 2010; Topalova, 2008).

In this paper, we examine the possible role that economic inequality plays in producing child health inequality. We analyze the contributions of wealth-based inequality in the distribution of household-level determinants to wealth-based inequality in child health. Corresponding to the trends in economic inequality discussed above, we analyze births that occurred between 1987 and 2006. The data are drawn from the National Family and Health Surveys that were conducted in 1992/93, 1998/99, and 2005/06. At approximately the time that the first NFHS was conducted, the Gini coefficient in India at the national level was .303. Close to the time of the third NFHS, the Gini was .325 (Topalova, 2008). These increases in economic inequality presumably impacted the distribution of household-level determinants of child health.

A second factor that may have exerted an indirect influence on child health and its distribution is the imbalanced sectoral pattern of economic growth in India. For our analyses, we started with a sample of births that occurred in the five years before each survey (1987–1992, 1993–98, 2000–2005). Table 1 shows averages of growth rates in the primary, secondary, and tertiary sectors of the economy for the five years prior to years that the births are drawn from, as well as the entire lagged period. Growth in manufacturing and services far outpaced growth in agriculture. Further, there was a divergence between growth in services and growth in agriculture. This pattern of sectoral growth then suggests an urban bias in economic growth, which could have had differential moderating effects on the importance of household-level factors in child health in urban versus rural areas.

Apart from increases in economic inequality and imbalances in economic growth, there were distinct features of programs during these years that could have exerted an influence on child health inequalities. Public programs in theory subsidize households' investments in child health, thus making household wealth a less important determinant in the production of child health than it would have been otherwise. Effective programs – particularly ones targeted towards the poor – can thus successfully reduce inequalities in child health. The programs relevant to child mortality in India are the Child Survival and Safe Motherhood Program, which was launched in 1992 and then successfully incorporated into the larger, more ambitious Reproductive and Child Health Program in 1997 (World Health Organization, 2010). The programs relevant to malnutrition are the Integrated Child Development Scheme (ICDS) and the Public Distribution System (PDS). The former program explicitly targets child malnutrition, while the latter is aimed at general food security and is India's most far-reaching safety net in terms of coverage and also the most expensive in terms of public expenditure (Radhakrishna & Subbarao, 1997). In 1997, the government introduced the Targeted Public Distribution System (TPDS) with the aim of redistributing income by allocating a greater proportion of the distributed food to the poor, and at lower prices than to non-poor beneficiaries. The success or failure of these programs could have moderated the importance of household determinants to child health, and thus their contribution to child health inequality. However, the moderating effects may differ by outcome and by urban/rural area because programs for mortality and malnutrition had varying success, and performed differently in urban and rural areas.

The inverse relationship between economic status and ill health has been widely documented in varied settings. The World Bank produced a comprehensive report documenting socioeconomic differences in a wide variety of indicators in 56 developing countries (Gwatkin et al., 2007). The report clearly indicates wealth-based inequalities in more than a 100 demographic, health, and healthcare variables, related behaviors, and underlying determinants. As part of their voluminous findings, the authors present poor/rich ratios and concentration indices for infant mortality, under-five mortality, severe stunting, and severe underweight in India. Their analysis is limited to a shorter time frame than the present analysis – they used the first two NFH surveys – and it was conducted only at the national level. They find that inequality in infant mortality decreased between the first two surveys, while inequality in under-five mortality remained relatively unchanged. They also find an increase in inequality in both severe stunting and severe underweight. In their study using all three NFH surveys, Subramanyam, Kawachi, Berkman, and Subramanian (2010) analyze the relationship between household wealth and underweight and stunting. Their analysis, restricted to the national level, shows an increase in wealth-based inequalities in child malnutrition between surveys (Subramanyam et al., 2010). Pathak and Singh (2011) also use all three NFHS's to show inequalities in malnutrition, as measured by proportion underweight. Their results indicate a rise in economic inequalities in underweight during the period between 1992 and 2006 (Pathak & Singh, 2011). The next set of papers goes one step beyond documenting inequalities, to identify the sources of these inequalities.

To investigate how inequalities in the determinants of child health contribute to inequalities in child health outcomes, Wagstaff and others develop a method which they illustrate using data from Vietnam (Wagstaff, van Doorslaer, & Watanabe, 2003). They examine the causes of both levels and changes in inequalities in child malnutrition between 1993 and 98 and conclude that levels of inequalities in height-for-age in Vietnam in 1993 and 1998 are largely accounted for by inequalities in consumption and in unobserved commune-level influences. They then conclude that rising inequalities are almost entirely accounted for by inequalities in household consumption and commune-level features.

The application of this technique to DHS data from Iran reveals that nearly 50% of wealth-based inequality in infant mortality in Iran is determined by inequality in household economic status and mother's education. The authors conclude that socioeconomic inequality in infant mortality in Iran is determined not only by health system functions but also by factors beyond the scope of health authorities and care delivery system (Hosseinpoor et al., 2006).

This decomposition technique was also applied to NFHS-3 data (Mazumdar, 2010; Pradhan & Arokiasamy, 2010). Pradhan and Arokiasamy (2010) provide an assessment of the contributions of socio-economic determinants to inequality in under-2 child mortality. Decomposition results reveal that inequalities in household economic status, mother's illiteracy, and rural residence are the major observed sources of socio-economic inequalities in child survival at the national level. The contribution of economic status is relatively smaller in 5 states that are advanced in health transition. Mazumdar (2010) explores the linkage between poverty and malnutrition inequality. The decomposition reveals that poverty alone explains more than 50% of the inequality in malnutrition, and parental education another 25% of it.

This paper builds on the extant literature first by identifying changes over time in the sources of child health inequality. Secondly, by applying the same method to the same samples but examining different outcomes, we go beyond the broader label of child health. We investigate possible differences in the contributions of inequality in household determinants to inequality in child mortality and malnutrition, which are qualitatively different outcomes. By separating the analysis for urban and rural areas, we investigate other possible differences in child health inequality and its determinants.