X hits on this document





6 / 22

cholesterol lowering drugs (statins) that are intended to be taken long term (Australian Institute of Health and Welfare 2007). Strategies of primary prevention such as improving diet and exercise combined with secondary prevention and treatment strategies to reduce the morbidity of those with disease offer opportunities to improve health and productivity of the workforce.

The aim of the current study is to determine the influence of the two major preventable chronic diseases in high income countries - diabetes and cardiovascular disease - on labour supply. The study uses an endogenous multivariate probit model with a recursive simultaneous structure and unit record data from a stratified random national survey and physical examination of 11,247 individuals in Australia. In Section II we describe some of the recent literature on the relationship between health status and labour market outcomes. In Section III we describe a theory and the empirical estimation of a joint model of labour supply and chronic disease. Section IV describes the AusDiab dataset: unit record data from a stratified random national survey with a physical examination of 11,247 individuals in Australia. In Section V we report the results of 3 separate univariate probit models for labour force participation, diabetes and cardiovascular disease and a 3 equation recursive simultaneous multivariate probit model. The latter allows for both unobserved heterogeneity and the simultaneous determination of diabetes and cardiovascular disease.


Recent literature

A large number of studies have estimated the relationship between health status and labour market outcomes. The vast majority of these studies have examined the relationship between a self reported general health status measure and labour outcomes. A small number of studies has looked at the effect of particular conditions such as obesity (Morris 2007), diabetes (Brown and Pagán Elena Bastida 2005), problem drinking (Feng, Zhou et al. 2001; MacDonald and Shields 2004) using either instrumental variable methods or recursive bivariate probit techniques.

The theoretical notion in all of these studies is that health like education can be considered an endowment of human capital that deteriorates over time but is capable of enhancement as a result of household production (Becker 1964; Becker, Zamagni et al. 1965; Lancaster 1966; Grossman 1972; Currie and Madrian 1999; Grossman 1999). This implies higher returns from work for healthy workers and that healthy people are more likely to work. Poor health may also impact directly on an individual’s preference for paid employment through the relative utility of work or leisure as well as reduce the total amount of time available to earn money. In addition, sickness in most Western countries gives an entitlement to income from welfare benefits conditional on not working (Grossman 1999; Cai and Kalb 2006; Disney, Emmerson et al. 2006). There are then a number of theoretical links between health status and work that suggest not only that better health improves labour outcomes but that that poorer health is likely to be associated with lower labour supply. This is not to deny that in some circumstances environmental factors at work may contribute to physical or mental ill health but in the case of chronic diseases studies here that kind of reverse causality is less likely.

Most studies in this literature rely on global self reported health as a proxy for latent individual health. Like many self reported measures, health status suffers from the potential for measurement error and endogeneity that could result in the overestimation of a health effect on labour market outcomes (Anderson and Burkhauser 1985; Stern 1989; Bound 1991; Bound and et al. 1999; Dwyer and Mitchell 1999; Campolieti 2002). The potential for endogeneity arises for at least 3 reasons. Poor health may be associated with unobserved household characteristics such

Chronic disease and labour force participation in Australia: an endogenous multivariate probit analysis of clinical prevalence data


Document info
Document views68
Page views68
Page last viewedMon Jan 16 17:57:52 UTC 2017