Income Inequality and Individual Health: Exploring the Association in a Developing Country

We use individual and multi-level data from Zambia on child nutritional health to test the absolute income hypothesis (AIH), the relative income hypothesis (RIH) and the income inequality hypothesis (IIH). The results confirm a non-linear positive relation between economic resources and health, confirming the AIH. For the RIH we find sensitivity to what reference group is used. Most interestingly, while the IIH predicts that income inequality, independent from individual income, will affect health negatively, we find higher income inequality to robustly associate with better child health. The results suggest that the relationship between inequality and health in developing contexts might be very different from the predominant view in the existing literature mainly based on developed countries, and that alternative mechanisms might mediate the relationship in poor countries. * The authors are thankful for comments and suggestions from Carl Hampus Lyttkens, Jesper Roine, Mireia Jofre-Bonet, Pernilla Johansson, Björn Ekman and seminal participants at Lund University for useful comments and suggestions. Financial support from SIDA/SAREC, and Jan Wallander and Tom Hedelius foundation is gratefully acknowledged. † Lund University and Research Institute for Industrial Economics (IFN). [email protected] ‡ Research Institute for Industrial Economics (IFN) and Lund University. [email protected] 2 INTRODUCTION There is an on-going debate as to whether health is negatively affected by income inequality within a society. This issue has received abundant research interest in several disciplines (cf. Rodgers 1979, Wilkinson 1992, Kawachi and Kennedy 1997, Mellor and Milyo 2001, Deaton 2003, Ram 2006, Babones 2008 and Karlsson et al. 2010). The idea that inequality causes poor health originates from an often noted negative correlation between various income inequality measures and the average health status of a population (e.g. Babones 2008, Ram 2006, Waldman 1992). From Rodgers (1979) and Gravelle (1998) we know that such aggregate associations might stem from a non-linear relationship between income and individual health, making it vital to use individual or household level data. Such data enable the separation between different theoretical hypotheses, all consistent with a negative aggregate association between inequality and population health (Wagstaff and van Doorslaer 2000). Empirical studies using individual level data have produced largely contradictory conclusions. Consistent evidence of a negative association between income inequality and individual health is found in the United States (e.g. Kennedy et al. 1998, Subramanian et al. 2001, Lopez 2004). In contrast, data from other developed countries often find no such correlation (e.g. Shibya et al. 2002 on Japan, Gerdtham and Johannesson 2004 on Sweden, Jones et al. 2004, Gravelle and Sutton 2009 on the UK), suggesting that a general association between inequality and health does not exist. However, we have limited knowledge of the relationship between inequality and individual health in low-income countries. This article examines the relationship between income inequality and individual health in a less developed context. Using data from Zambia we test three hypotheses: the absolute income hypothesis (AIH) – stating that individual health is determined by individual income and that the positive effect from higher income is subject to diminishing return; the relative income hypothesis (RIH) – assuming that health is influenced by the relation of individual income to the average income in a reference group; and the income inequality hypothesis (IIH) emphasizing that individual health status is impacted by inequality in the distribution of income. Our dependent variable is child nutritional status expressed by height-for-age. Research increasingly emphasizes the important role of child health as a major factor influencing future economic outcomes (cf. Currie 2009, Bengtsson and Lindström 2000, Maluccio et al. 2006). Moreover, anthropometrical indicators are objective, relatively precise, and consistent across subgroups (Heltberg 2009). We test the RIH and the IIH by using average household expenditures and expenditure Gini coefficients calculated at three geographical levels; provincial, district and constituency. Taking complex survey design into account and using different econometric techniques we confirm the AIH, and also find some for the RIH at the constituency level. In contrast to the IIH, we find a positive association between contextual inequality and child health, robust to measuring inequality at different geographical levels, to alternative inequality measures and alternative specifications. The next section reviews suggested theoretical relationships between income, income inequality and individual health and shortly reviews existing empirical evidence. The third section describes the data and discusses methodological choices. The fourth section presents the empirical results, while the final section interprets the findings and concludes.