Socio-economic status, health and lifestyle
Introduction
Analysts of the causes of inequalities in health have long recognised that variation in medical care utilisation cannot fully explain observed health differences (e.g. Auster et al., 1969, Evans et al., 1994). A common feature of this literature has been the growing use of the concept of lifestyle by epidemiologists, sociologists, and economists, in order to categorise behavioural patterns and explain observed health inequalities. In an influential work, Fuchs (1986) argues that beyond a fairly low level in the provision of food, hygiene and basic health care, it is personal lifestyle that causes the greatest variation in health. McGinnis and Foege (1993) estimated that the three leading external (nondegenerative or directly genetically determined) causes of mortality in the US in 1990 were tobacco, diet and activity, and alcohol consumption. They estimated that these lifestyle variables explained around 38% of premature mortality, and also noted that a dramatically reduced quality of life is associated with many of the diseases related to these behaviours. Other authors have concluded that, with the exception of tobacco consumption, lifestyle factors do not affect the widely observed relationship between socio-economic status and health substantially (Borg and Kristensen, 2000, Lantz et al., 1998, Power et al., 1998; Marmot et al., 1997; Lynch et al., 1996).
Although the focus of research into the determinants of health has shifted, it is not easy to define lifestyle both comprehensively and empirically. The World Health Organisation (1986) considered a number of meanings, and adopted a broad definition; ‘… the term ‘lifestyle’ is taken to mean a general way of living based on the interplay between living conditions in the wide sense and individual patterns of behaviour as determined by sociocultural factors and personal characteristics’. In this paper, we adopt a narrower (and operationalisable) definition of lifestyle which focuses on health related behaviour and accords with the epidemiological literature on the determinants of health (e.g. Lynch et al., 1996, Lynch et al., 1997, Marmot et al., 1997). We define a lifestyle as a set of behaviours which are considered to influence health and are generally considered to involve a considerable amount of free choice. In using this definition, there is no implication that other characteristics of an individual’s environment, both natural and social, are inconsequential. We adopt an economic approach which recognises that individuals are making decisions that reflect the constraints of their circumstances, as well as their preferences. We develop a static model to identify interactions between health related behaviour and self-assessed health (SAH) status, given other factors that are observable, such as socio-economic status, and unobservable heterogeneity.
Descriptive analysis of our data suggests a number of interesting features. Firstly, using the baseline assumption that lifestyle choices are independent and that all individuals have the same probabilities of each lifestyle choice, we can compare the expected number of individuals behaving completely ‘healthily’ relative to the observed number. The expected value is around 2/3 the number of observed values. This suggests that health-related behaviours are not randomly distributed, but rather that healthy behaviours cluster together in certain individuals. This clustering may be due to observed or unobserved factors. Secondly, average self-assessed health increases as the number of healthy behaviours increases. Thirdly, some of the variation in lifestyle choices appears to be related to observed characteristics of individuals. For example, the proportions of individuals in higher social groups gradually increase as we move from completely unhealthy to completely healthy lifestyles. Conversely, the proportions in lower social class groups gradually decrease. However, while these observations are indicative, they are correlations only; the assessment of causality requires the use of appropriate econometric techniques.
We estimate the structural parameters of a health production function, together with the reduced form parameters for the lifestyle equations using panel data from the Health and Lifestyle Survey (HALS) conducted in the United Kingdom in 1984 and 1991. This is achieved using Maximum Simulated Likelihood (MSL) for a multivariate probit (MVP) model with discrete indicators of lifestyle choices and self-assessed health (SAH). In addition to the substantive empirical results, this paper demonstrates the applicability and computational feasibility of models with flexible heterogeneity structures in the presence of multiple discrete outcomes in health economics.
The structure of the paper is as follows. Section 2 surveys previous economic literature in this area. Section 3 explains the theoretical model which forms the basis of our empirical analyses. Section 4 presents the UK Health and Lifestyle Survey (HALS) dataset. Section 5 describes our estimation strategy, while the empirical results are discussed in Section 6. Section 7 contains a short conclusion.
Section snippets
Previous economic literature
An example of the generic approach to the estimation of health production and input equations is Rosenzweig and Schultz (1983). Rosenzweig and Schultz used instrumental variable techniques to examine the effect of health inputs on birth weight in the presence of unobservable heterogeneity. Their concern was to obtain consistent estimates of the parameters of the child health (birth weight) production function, while recognising the difficulties created by input choices being influenced by
A simple model of lifestyle and health production
Becker’s (1965) seminal work on the allocation of time provides our starting point. He focuses on the distinction between technology and preference orderings in the production and consumption of fundamental commodities. In our model, the fundamental commodity adult health is produced by health related behaviours and other inputs, and also provides consumption benefits.
Data
The Health and Lifestyle Survey (HALS) is a national representative sample of adults living in private households in Great Britain. Carried out by Social and Community Planning research, the first wave of data (HALS1) were collected between Autumn 1984 and Summer 1985 during two home visits; firstly an hour long interview, followed by a nurse visit to collect physiological measurements and data on cognitive function. The available sample has information on 9003 individuals, although some gave
Estimation strategy
A consistent estimator of the health production function must account for the endogeneity in Eq. (2), introduced by the existence of lifestyle components as regressors and correlations between the errors of the models determining lifestyle choices and that which determines self-assessed health. In the absence of the complication that our endogenous variables are binary, many easy to implement estimators would be available for the linear model such as 2SLS, 3SLS and full information maximum
Lifestyle equations
Table 1 shows selected partial effects for the reduced form lifestyle models estimated using the MVP specification of the full recursive system.19 A slight gradient in the probability of
Conclusion
We developed a simple model to identify interactions between health related behaviour and self-assessed health status, given other observable and unobservable factors. Unobservable heterogeneity may reflect underlying causal factors such as correlations in the direct marginal utilities of health, income, and lifestyle choices which may in turn be related to differences in genetic characteristics, childhood circumstances, attitudes to risk and the rate of time preference. It may also reflect
Acknowledgements
The authors wish to thank participants in the York Seminars in Health Econometrics and two anonymous referees for their comments. Data from the Health and Lifestyle Survey (HALS) were supplied by the ESRC Data Archive. Neither the original collectors of the data nor the archive bear any responsibility for the analysis or interpretation presented here. We are grateful for research funding from the ESRC (award no R000238169).
References (52)
The psychological benefits of moderate alcohol consumption: A review of the literature
Drug and Alcohol Dependence
(1985)- et al.
Relationship of physical health status and health practices
Preventive Medicine
(1972) - et al.
Social class and self-rated health: can the gradient be explained by differences in life style or work environment?
Social Science and Medicine
(2000) - et al.
Smooth unbiased niultivariate probability simulators for maximum likelihood estimation of limited dependent variable models
Journal of Econometrics
(1993) The meaning of death: some simulations of a model of healthy and unhealthy consumption
Journal of Health Economics
(2001)- et al.
Simulation of multivariate normal rectangle probabilities and their derivatives: theoretical and computational results
Journal of Econometrics
(1996) Working conditions and the relationship between schooling and health
Journal of Health Economics
(1987)- et al.
Why do poor people behave poorly? variation in adult health behaviours and psychosocial characteristics by stages of the socioeconomic lifecourse
Social Science and Medicine
(1997) - et al.
Social inequalities in health: next questions and converging evidence
Social Science and Medicine
(1997) - et al.
Inequalities in self-rated health: explanations from different stages of life
The Lancet
(1998)
Monte Carlo evaluation of multivariate normal probabilities
Journal of Econometics
The production of health: an exploratory study
Journal of Human Resounce
Preference parameters and behavioural heterogeneity: an experimental approach in the health and retirement study
Quarterly Journal of Economics
A theory of the allocation of time
Economic Journal
Schooling, self-selection, and health
Journal of Human Resources
Self-evaluated health status and smoking behaviour
Applied Economics
Self-rated health: is it as good a predictor of subsequent mortality among adults in lower as well as in higher social classes?
Epidemiology and Community Health
Analysis of multivariate probit models
Biometrika
Uncertainty and the demand for health care
Journal of Health Economics
A model of the demand for longevity and the value of life extension
Journal of Political Economy
Cited by (236)
Household energy price resilience in the face of gas and electricity market crises
2024, Energy EconomicsCo-benefits from health and health systems to education
2024, Health PolicyGovernment drivers of gastric cancer prevention: The identification of risk areas and macro factors in Gansu, China
2023, Preventive Medicine ReportsThe relationship between physical activity, cognitive function and health care use: A mediation analysis
2023, Social Science and MedicineRelative deprivation, inequality and the Covid-19 pandemic
2023, Social Science and Medicine
- 1
Tel.: +44-1904-433-766; fax: +44-1904-433-759.