Family income and child health in the UK

Abstract

Recent studies examining the relationship between family income and child health in the UK have produced mixed findings. We re-examine the income gradient in child general health and its evolution with child age in this country, using a very large sample of British children. We find that there is no correlation between income and child general health at ages 0–1, that the gradient emerges around age 2 and is constant from age 2 to age 17. In addition, we show that the gradient remains large and significant when we reduce the endogeneity of income. Furthermore, our results indicate that the gradient in general health reflects a greater prevalence of chronic conditions among low-income children and a greater severity of these conditions. Taken together, these findings suggest that income does matter for child health in the UK and may play a role in the intergenerational transmission of socioeconomic status.

Introduction

A large amount of literature shows a positive correlation between socioeconomic status and health in adulthood (Adler et al., 1994, Blaxter, 1990, Deaton and Paxson, 1998, Deaton and Paxson, 1999, Marmot and Bobak, 2000, Van Doorslaer et al., 1997, Wilkinson and Marmot, 2003, Winkleby et al., 1992). Recent research initiated by Case et al. (2002) investigates whether the gradient in general health observed in adulthood has antecedents in childhood. Understanding the determinants of child health is important because health in childhood affects human capital accumulation, and health and labor market status in adulthood (Currie, 2008). Findings firmly establish that family income is positively related to children's general health in Australia (Khanam et al., 2009), Canada (Currie and Stabile, 2003), Germany (Reinhold and Jürges, 2012), and the US (Case et al., 2002, Condliffe and Link, 2008). Moreover, the correlation between family income and children's general health strengthens as children grow older in Canada and the US, meaning that the disadvantages associated with parental income accumulate as children age (Case et al., 2002, Currie and Stabile, 2003). These authors argue that the steepening of the gradient with age can be due to two mechanisms: (1) either children from poorer families are more likely to be subject to health shocks than their wealthier counterparts (prevalence or incidence effect), or (2) poorer children are less able to respond to health shocks, and so health shocks are more severe for them (severity effect). The distinction between these two mechanisms is important because they have different implications from a policy perspective: the first mechanism implies that the gradient may be reduced by addressing the reasons why poorer children are more likely to get chronic conditions, whereas the second mechanism means that a policy should improve access to health care services for poorer children. In the US, the strengthening of the gradient is due to a combination of a prevalence and a severity effects (Case et al., 2002), whereas in Canada, it is only due to a prevalence effect (Currie and Stabile, 2003).

Findings on the gradient in general health for British children are not firmly established. Currie et al. (2007) and Case et al. (2008) analyze the evolution of the gradient as children grow older, using cross-sectional data from the Health Survey for England (HSE), the same variables, and the same methods. Specifically, they estimate the gradient for four age groups (children ages 0–3, 4–8, 9–12, 13–17) and compare the estimates between the age groups to depict the evolution of the gradient with age. In spite of these similarities, their conclusions are different. Currie et al. (2007) highlight that there is a gradient in general health, that it increases between 0-3 and 4-8 and stops increasing afterwards, using six waves of the HSE. In contrast, Case et al. (2008) conclude that the gradient in general health does increase with age from birth to age 12, using three additional years of data from the HSE. In addition, Propper et al. (2007) suggest that when maternal health and behaviors are included, there is almost no correlation between family income and child health, for a cohort of British children less than 7 years of age. This means that the gradient may not reflect any causal effect of family income on child health.

The previous literature on the UK uses relatively small datasets, which could explain why the results are somewhat contradictory. A larger sample of British children may shed more light on the gradient in general health. In addition, the previous literature on the UK investigates the evolution of the gradient in general health using four age groups, which makes it impossible to examine the turning points in the evolution of the gradient with age. We suggest to compare the gradient between ages, instead of age groups, to get a precise description of the evolution of the general health/income relationship with age. Finally, in a small sample like the HSE, it is not possible to study the role of rare chronic conditions in the general health gradient: the analysis of rare chronic conditions requires large sample sizes.

This paper re-examines the general health/income gradient in childhood in the UK, using a large sample of approximately 78,000 children drawn from the Family and Children Survey (FACS). First, we exploit the large sample size of the FACS to investigate the evolution of the gradient with child age in a more detailed manner. Specifically, we estimate the effect of income on health separately for children of each age, instead of each age group. Second, we examine whether the association between family income and child health could represent causality running from income to child health, as opposed to reverse causality or the omission of third factors. We adopt two strategies. On the one hand, we take advantage of the information we have on the influence of child health on family income in the FACS, to reduce reverse causation. As far as we are aware, we are the first to deal with this issue in a precise manner. On the other hand, we expand on the number of controls to address the omission of factors. Third, we examine the role of specific health problems, in particular some rare chronic conditions, Special Educational Needs, and the attention deficit hyperactivity disorder (ADHD), in the gradient in general health. Fourth, we investigate the channels through which family income could have an impact on child health, focusing on the use of health care services, housing conditions, nutrition, and clothing.

We find that there is a very small or negligible effect of family income on general health for children ages 0-1 and a large and significant effect for children above 2. In addition, the gradient remains constant as children grow older, from age 2 to age 17. This description of the gradient is very different from that given in the earlier literature on the UK, which highlights an increase in the gradient with age between birth and age 8 at least. We also show that our results are robust to various procedures that mitigate the bias due to the endogeneity of income. The paper also finds that the gradient in general health could be explained both by the prevalence and severity of specific health problems among low-income children, which implies that policies should address the reasons why low-income children are more likely to obtain specific health problems and why the severity of these specific problems depends on income. Finally, we show that the effect of family income on child health is not accounted for by differences in the use of health care services, housing conditions, nutrition, and clothing between low and high-income children. However, housing conditions, nutrition, and clothing do have a large independent effect on child general health.

The rest of the paper proceeds as follows. In Section 2, we begin by discussing the contributions of the previous literature and highlight the originality of our approach. Section 3 provides an overview of the data. Section 4 investigates in details the evolution of the income gradient in general health with age and discusses the endogeneity of income. Section 5 focuses on the role of specific health problems in the gradient in general health. Section 6 examines whether the use of health care services, housing conditions, nutrition, and clothing are important channels through which family income influences child health. The Section also contains additional results on the role of maternal education in child health. Lastly, Section 7 offers some concluding remarks.