More equal but heavier: A longitudinal analysis of income-related obesity inequalities in an adult Swedish cohort☆
Introduction
Since the 1980s obesity has increasingly become a public health concern. Between 1990 and 2000, obesity prevalence increased by 67 per cent in Sweden, reaching a prevalence of 9.2 per cent. Corresponding increases for France, England and the US were 56, 50 and 30 per cent, respectively (Höjgård, 2005 pp. 7–8). From an economic point of view, obesity is an important research area due to the increased social costs that are generated by poorer health among obese people, including direct costs due to increased utilisation of health care caused by higher disease risks, and indirect costs from, for example, potentially decreased productivity in the labour market.
While genetics partly determines obesity prevalence, it is unlikely that genetic evolution explains the rapid increase in obesity that has been observed in recent decades (Hedley Vickers et al., 2007, Höjgård, 2005 p. 15). Instead, both medical and psychosocial factors have been suggested to contribute (Hedley Vickers et al., 2007, Wamala et al., 1997). There is also a fair consensus in the literature that although obesity is partly determined by these factors, technological progress and economic and societal aspects also contribute to an extent that should not be ignored. Examples of such technological factors are easier and relatively cheaper access to food, more sedentary lifestyle and lower food prices as a result of cheaper production (Chou et al., 2004, Costa-Font and Gil, 2008, Goel Rajeev, 2006, Hedley Vickers et al., 2007, Lakdawalla et al., 2005, Propper, 2005, Smith, 2005). While modern economic times have undeniably changed our lives for the better in many ways, they have also generated new problems, with obesity being one of them.
The existing economic literature on socio-economic determinants of obesity generally suggests a negative relationship; higher income (or education or social status) is related to a lower risk of obesity (Costa-Font and Gil, 2008, Nayga, 1999, Wamala et al., 1997, Zhang and Wang, 2004, Zhang and Wang, 2007). Using longitudinal data on young American adults (being 16–23 and 39–46 years old in the first and last period, respectively), Baum and Ruhm (2009) find that the socio-economic gap in obesity widens with age.
Although obesity is not a new topic in the economic literature, inequality in obesity has not received much attention. The purpose of this paper is to analyse income-related obesity inequality and how the inequality changes as the population ages. We use Swedish longitudinal data containing information from three points in time (1980/1981, 1988/1989 and 1996/1997) and follow a random sample of the Swedish population in 1980/1981 over a 17-year period. We primarily focus on long-run inequality, using a long-run measure of income. Our main questions are: 1) Does inequality in obesity disfavour or favour the poor? 2) What explains the inequality in the cohort at different time periods? 3) How can the development of inequality over time be explained?
Obesity can be viewed as a health dimension that reflects avoidable health aspects more than general health itself. For example, decreasing health with age is partly reasonable whereas there is no such reasonable reason for obesity to increase with age. Moreover, weight is directly affectable by the individual herself, whereas health may partly be of a more unaffectable and complex nature. Health inequality has been discussed at great length recently, and is of interest for public health policy makers. Therefore, from a policy perspective, knowledge about obesity is of great interest. The phenomenon of the increasing prevalence of obesity is not unique for Sweden but is shared by most other countries in the world, and there is evidence of a negative socio-economic obesity gradient in many countries. This study is therefore also relevant for future research in other countries besides Sweden, and the conclusions should be useful also for non-Swedish public health policy makers.
Having access to panel data we have a good opportunity to provide useful information on the problem of obesity inequality. The study adds various aspects to the literature. First, as opposed to inequalities in general health outcomes, analysis of obesity inequality is sparse. Second, while there is a small amount of literature on inequality in obesity using cross-section data (Costa-Font and Gil, 2008, Zhang and Wang, 2007), by use of longitudinal data we are able to investigate long-run inequality. This may be very different from cross-section samples. Third, individual heterogeneity ought to be an important factor when dealing with obesity. The panel data allows us to take this aspect into account, leading to a more realistic probability function for obesity. Fourth, this study focuses on obesity inequality in an ageing cohort, giving insight into the interrelationship between age, income and obesity.
In brief, the study is carried out as follows. First, we calculate obesity concentration indices and estimate a reduced form probability model for obesity. This model is then used in a decomposition analysis of obesity inequality in order to enable investigation of the driving forces behind the inequality. Thereafter we investigate the sources behind changes in the obesity concentration index over time. Because of space limits and that we cannot a priori exclude gender differences, we focus on women. An initial analysis of the male sample confirmed the gender difference concern. However, we briefly discuss the results for men in the results section.
The paper proceeds as follows. First, we discuss the relationship between ageing, obesity and inequality. The methods section describes the concentration index, the decomposition techniques, the data and the model that we use for the decompositions. The following section contains the results, and the final section offers a discussion.
Section snippets
Ageing, obesity and inequality
Both income and risk of obesity tend to change as an individual ages. Regarding obesity, our data shows a right-skewed inverse U-curve; obesity rates for women tend to increase steadily with age and reach a peak among the 70–75 years old. At older ages, obesity tends to become less common, but it does not revert back to the youngest age group's level. Socio-economic inequality in obesity in cross-section samples may differ from inequality in the long run, in particular if changes in obesity
Concentration index
The concentration index (C) and decomposition thereof, as proposed by Kakwani et al. (1997) and Wagstaff, van Doorslaer, and Watanabe (2003), respectively, is a method that has been used frequently when analysing socio-economic health inequality. The C takes on values from −1 to +1, where a negative (positive) value emerges when the health variable is concentrated among the relatively poor (rich). The current study analyses ill health; we see obesity as an undesirable health status.
Obesity inequality based on current and mean income rankings
Although our main interest is in long-run inequality, it is illuminating to compare these results with obesity inequality when based on a short-run income measure (current income) instead. Table 2 reports the obesity Cs (normalized and non-normalized). The degree of inequality differs depending on which income measure, i.e., mean or current income, the individuals are ranked by. Moreover, depending on the income measure, the changes in inequality over time differs. While inequality increases
Discussion
Income-related obesity inequality among Swedish women is pro-rich; obesity tends to be less common among the relatively rich. Over time, when the cohort ages, the obesity inequality decreases.
It may be tempting to infer a pleasing development from the reducing obesity inequality, perhaps even a success of Swedish health policy. However, we strongly argue that this is a good example of how equality should not be an isolated goal, and how one should be careful about focusing too strongly on
References (34)
- et al.
Age, socioeconomic status and obesity growth
Journal of Health Economics
(2009) - et al.
An economic analysis of adult obesity: results from the behavioural risk factor surveillance system
Journal of Health Economics
(2004) - et al.
What lies behind socio-economic inequalities in Spain? A decomposition approach
Food Policy
(2008) - et al.
Socioeconomic inequalities in health: measurement, computation, and statistical inference
Journal of Econometrics
(1997) - et al.
Determining the impact of food price and income changes on body weight
Journal of Health Economics
(2008) Are time preference and body mass index associated? Evidence from the National Longitudinal Survey of Youth
Economics and Human Biology
(2005)- et al.
On decomposing the causes of health sector inequalities with an application to malnutrition inequalities in Vietnam
Journal of Econometrics
(2003) - et al.
Determinants of obesity in relation to socioeconomic status among middle-aged Swedish women
Preventive Medicine
(1997) - et al.
Socioeconomic inequality of obesity in the United States: do gender, age and ethnicity matter?
Social Science & Medicine
(2004) The impact of obesity on wages
The Journal of Human Resources
(2004)
Do life-saving regulations save lives?
The Journal of Risk and Uncertainty
Instrument relevance in multivariate linear models
The Review of Economics and Statistics
Obesity: an economic and financial perspective
Journal of Economics and Finance
Development programming of obesity and type 2 diabetes
Fetal and Maternal Medicine Review
Vad kan staten göra åt fetma?
Does income-related health inequality change as the population ages? Evidence from Swedish panel data
Health Economics
Cited by (0)
- ☆
We thank Barbara Eberth and participants in the Health Econometrics Working Group within the Health Economics Research Unit (University of Aberdeen), in the Health, Labour and Family Economics seminar group at the Department of Economics (Lund University) and in the Centre for Economic Demography Research School seminar (Lund University) for helpful comments and suggestions. We also thank Stephen Birch and our two referees for their important comments. Financial support from the Swedish Council for Working Life and Social Research (dnr 2006-1660 and dnr 2007-0318) is gratefully acknowledged.