Child health and the income gradient: Evidence from Australia
Introduction
A growing literature documents a strong positive correlation between household income and child health (see for example, Caase et al., 2002, Caase et al., 2007, Caase et al., 2008, Currie and Stabile, 2003, Propper et al., 2007, Chen et al., 2006, Currie et al., 2007, Dowd, 2007). Two pioneering and influential papers, by Case et al. (2002) and Currie and Stabile (2003), using US data and Canadian data, respectively, established that the gradient is greater for older than for younger children. This finding of an increasing income–child health gradient is supported by another two recent studies by Condliffe and Link (2008) and Murasko (2008); although these two papers found smaller effects of income on child health than those reported by Case et al. (2002) and Currie and Stabile (2003). The studies that have examined the income–child health gradient after those pioneering papers (Caase et al., 2002, Currie and Stabile, 2003) have not, however, always produced corroborative evidence of an age-increasing income–(child) health gradient. For example, although Chen et al. (2006) documented a very significant effect of income on child health using same data set as Case et al. (2002) they did not find that the income–health gradient steepened with child age. Recent studies by Currie et al. (2007) and Propper et al. (2007) using the 1997–2002 Health Surveys of England (HSE) and the Avon Longitudinal Study of Parents and Children (ALSPAC) respectively also found no evidence that the income–health gradient increased with age in their sample of British children. Several other English studies have also documented a relationship between socio-economic status (SES) and health that presents in childhood, but which either flattens or disappears in adolescence, only to reappear in adulthood (see for example, West, 1997 and West and Sweeting, 2004).
Notably, Case et al. (2008) recently re-examined the HSE data and compared their findings with those of Currie et al. (2007). They established that the apparent differences in the income–health gradients for American and English children are less striking than those presented by Currie et al. (2007) when data from the same time period are compared. Case et al. (2008) used an expanded English sample by adding 3 more years of data from the HSE (1997–2005), and compared the results with those from American NHIS data for the period 1998–2005. Their results showed that the income–health gradient for children does indeed increase with age in both the US and the UK. The income–health gradient for children was, however, smaller for the English sample than for that of the United States but slightly greater than that which was uncovered by Currie et al. (2007) for the UK. Thus, the literature presents mixed results on the hypothesis that the income–child health relationship is increasing in child age. Furthermore, the existing literature suggests that a gradient exists even in countries (e.g., the UK, Canada) with universal health care financing and delivery schemes.
There are many possible mechanisms via which income may affect child health even if health care is essentially “free” at the point of care. Greater income, ceteris paribus, creates greater opportunities for households to consume health and non-health inputs. The latter have been shown to be important sources of cross-sectional variation in health status in developed countries, where the marginal product of medical care may approach zero, but the marginal product of investments in education, etc. are still positive. In the only study to consider the effect of marginal health care services with an experimental design, viz. the Rand Health Insurance Experiment (Newhouse and the Insurance Experiment Group, 1993), there was little difference in pediatric health status by health plan. In relation to children in particular, rich households may be more efficient at producing child health. This may be due to a correlation between income and education, the latter of which enables greater allocative efficiency in health input selection; and/or the opportunity to buy more (or better) market inputs for the production of health. Alternatively, or additionally, higher incomes may be correlated with healthier environments (e.g., the physical environment, including housing), more nutritious diets, or more active lifestyles. On the other hand, higher incomes may also be correlated with health production “bads”. For instance, assuming that the opportunity cost of time is higher for parents from higher-income households, market inputs may be substituted for parental-time inputs to a greater extent and the marginal health product may be lower as a result. Examples may include the substitution of bought meals for home-cooked meals, or market childcare services for parental care. The empirical direction of the influence of such effects, which may be correlated with income are, in large measure, indeterminate a priori. Thus, there are good reasons to conduct an empirical investigation of the relationship between household characteristics – many of which may be correlated with income – and child health.
The empirical evidence on the mechanism(s) via which higher incomes produce better child health is also far from settled, although a small number of studies have explored this issue. Case et al. (2002) found that insurance, health at birth, and simple genetics could not explain the association between health and income in their sample, and concluded that the mechanisms underlying the income–child health association required further exploration. Currie et al.’s (2007) work answers this call by using the Health Surveys of England (HSE) to examine the effect of child nutrition (as measured, e.g., by fruit and vegetable consumption by children) and family lifestyle (as measured, e.g., by parental exercise) choices on child health. Interestingly, the inclusion of nutrition and family lifestyle in their analyses did not reduce the magnitude of the income–health gradient, suggesting that the roles of nutrition and lifestyle are important, possibly independent, determinants of child health status. Propper et al. (2007) found evidence that parental behaviour, and especially maternal health, also influences child health and, importantly, that the relationship between household income and child health disappeared when controls for parental health were used. Notably, the mother's health, particularly her mental health plays an important role in their models and effectively reduces the estimated effect of income per se to zero. In contrast, Dowd (2007) finds no significant mediator of the relationship between household income and child health. Therefore, the mechanisms by which income transmits to better health remain unresolved. This question is important to resolve for several reasons, not least of which is the potentially important role of health in the intergenerational transmission of economic status (Currie, 2008).
Thus, in this paper we examine the income–health gradient in young Australian children using two recent waves of data from the Longitudinal Study of Australian Children (LSAC). Of particular interest to us is this question of whether or not the income gradient increases with child age in our sample (i.e., from early- to mid-childhood). We address this question using parent-reported measures of overall health status and parental reports of chronic conditions that are likely to have been physician-diagnosed. We then direct our focus to an examination of the question of whether other child characteristics (e.g., child's diet) and parental attributes (e.g., health states) or behaviours (e.g., diet and exercise) attenuate the income–health relationship for children in our sample.
We contribute to the existing empirical literature in several ways. First, we produce the first econometric estimates of the income–health gradient for Australian children. Second, by using panel data we were able to account for the past investment made into child health (or cumulative effect of health) in the child health production function which have not been used extensively (with the exception of Murasko, 2008) in this literature, to examine the income–child health gradient. In fact, the previous literature on this topic such as Case et al. (2002) and Currie et al. (2007) have utilised cross-section data and hence were unable to account for the cumulative effect of health/health care used in the past on child health production. This represents an important addition to the literature, that is consistent with the conventional theoretical model of human capital accumulation Grossman (1972). Third, using the appropriate econometric techniques, we explore the relationship of some further variables that, in theory, could affect child health and examine whether or not these measures moderate the apparent income–health relationship for Australian children. Specifically, we present evidence on the roles of child's nutrition and parental health on health states of children. Thus, we are able to control for some variations in household characteristics that were not observable (i.e., may have constituted unobserved heterogeneity) in other influential studies. Finally, we compare our specifications of the model with those used in work of Case et al. (2002), Currie and Stabile (2003) and Currie et al. (2007) by estimating analogs of their models for our Australian birth cohorts. Doing so provides an insight into how the Australian results compare with those generated by other influential studies in this field.
In summary, our results represent novel empirical evidence on (i) the income–child health gradient for parental- and physician-reported child health, (ii) the mechanisms via which household income may affect child health status, and (iii) the relative gains that may be produced by applying sampling weight and clusters for robust estimates. This paper makes significant contribution to our understanding about the relationship between child health and family income. For example, the results of this paper show that the income–child health gradient is much smaller in Australia than that of the USA, and even Canada and the UK; the latter two of which, like Australia, have long-standing universal and compulsory health care financing schemes. This result underscores a fundamental point of health production that was originally made by Grossman (1972): health production is a multivariate production process. By extension, one should not assume that access to health care services that are heavily subsidised or zero-priced nullifies the influence of income on health.
Our results show that the child health–income gradient is sensitive to the omission of confounders and controls, and the choice of age break. Furthermore, when we include a richer set of controls, including parental health, we find no evidence of an income–child health gradient at all. Our results indicate that parental health and, in particular, the mother's health play a significant role in this regard, reducing the income coefficient to zero when we account for it. Thus one important contribution of this paper is to show that parental health, particularly the mother's physical and mental health are factors that explain the univariate (and restricted multivariate) result of a positive relationship between child health and income in Australia.
Section snippets
Household production of child health
Our theoretical model for the analysis of child health production derives from household production theory, which originated in the work of Becker (1965) and Becker and Lewis (1973), and was adapted by Grossman (1972) to analyse the accumulation and depreciation of health capital. The health production model, in which health capital is conceived as the output of a multivariate production process (Grossman, 1972, Behrman and Deolalikar, 1988, Liebowitz and Friedman, 1979, Strauss and Thomas, 1994
Data sources
This study utilises the data from the first two waves of the nationally representative Longitudinal Study of Australian Children (LSAC) (Australian Institute of Family Studies, 2007). The LSAC has so far involved two waves of data collection for more than ten thousand children. The LSAC collects data on these children every 2 years and will follow them until 2010 or beyond. The LSAC was conducted using both face-to-face interviews and survey instruments that were sent and retrieved via mail.
Econometric model
An empirical formulation of the dynamic health demand function, Eq. (6) can be written aswhere Hit is the stock of health of child i in period t (in this case, the LSAC Wave 2), Hi(t−1) is the stock of health of child i in period t − 1 (in this case, the LSAC data Wave 1), I represents average income of the family (i.e., log of average CPI-adjusted family income in Wave 1 and Wave 2) and is our proxy for permanent income, Zit is a set of exogenous
Results and discussion
In this section we first estimate specifications that are close analogs of the models invoked by Case et al. (2002), Currie and Stabile (2003) and Currie et al. (2007) to examine income–child health gradient using similar variables as Case et al. and Currie et al., on cross-sectional analyses. We refer to these specifications as “Specification 1” and “Specification 2”. In addition to these two specifications, we estimate another specification (“Specification 3”) to account for some additional
Conclusions
This paper contributes to a growing literature on the income–child health gradient. This literature is advancing, in part due to the availability of high-quality data and advances in econometric methods. The current paper presents the first Australian econometric evidence on the income–child health gradient and the mechanisms via which child's nutritional and parental health may affect child health, independently of the household's income. It also presents comparisons of the empirical estimates
References (46)
- et al.
The income gradient in children's health: a comment on Currie, Shields and Wheatley Price
Journal of Health Economics
(2008) - et al.
Socioeconomic status and health: do gradients differ within childhood and adolescence?
Social Science and Medicine
(2006) - et al.
The child health/family income gradient: evidence from England
Journal of Health Economics
(2007) Early childhood origins of the income/health gradient: the role of maternal health behaviors
Social Science and Medicine
(2007)The family as producer of health—an extended Grossman model
Journal of Health Economics
(2000)Health inequalities in the early years: is there equalisation in youth?
Social Science and Medicine
(1997)- et al.
Evidence on equalisation in health in youth from the west of Scotland
Social Science and Medicine
(2004) - et al.
Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations
Review of Economic Studies
(1991) - Australian Bureau of Statistics, 2008. Consumer Price Index, Australia, June 2008. Australian Bureau of Statistics....
- Australian Institute of Family Studies, 2007. Longitudinal study of Australian children, wave 2 data release....
On the interaction between the quantity and quality of children
The Journal of Political Economy
A theory of the allocation of time
Economic Journal
Crowding: risk factor or protective factor for lower respiratory disease in young children?
BMC Public Health
Economic status and health in childhood: the origins of the gradient
The American Economic Review
Parental Behaviour and Child Health
Health Affairs
Analysis of Survey Data
Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences
The relationship between economic status and child health: evidence from the United States
American Economic Review
Balancing the number and size of sites: an economic approach to the optimal design of cluster samples
Controlled Clinical Trials
Socioeconomic status and child health: why is the relationship stronger for older children
The American Economic Review
Cited by (103)
The rise and fall of SES gradients in heights around the world
2023, Journal of Health EconomicsThe “Long-arm” of chronic conditions in childhood: Evidence from Canada using linked survey-administrative data
2023, Economics and Human BiologyNatural resources, child mortality and governance quality in African countries
2023, Resources PolicyHealth effects of cooking fuel transition: A dynamic perspective
2022, EnergyCitation Excerpt :Many researchers have discussed the influence of income on health from two aspects: income level and income inequality. First, income level has been recognized as an important incentive factor for health [36]. Economic inequality may adversely influence health as well.
The socioeconomic gradient in child health and noncognitive skills: Evidence from the Czech Republic
2021, Economics and Human BiologyWealthy, healthy and green: Are we there yet?
2021, World Development