This thesis concerns the modelling of individuals’ health over the life course, within the framework set by the now substantial international literature on the relationship between socioeconomic status and health. The focus is on people with long term illnesses and related disabilities, on inequalities in health by socioeconomic status (SES) and on the impact of health on employment.¶ The main tool of analysis is a dynamic microsimulation model of the Australian population which tracks the demographic, socioeconomic and financial characteristics of individuals and their families over the life course. Its original form, developed at the National Centre for Socioeconomic Modelling, University of Canberra, is based on a one per cent representative sample of the Australian population (around 150,000 individuals), with a series of life course events simulated for individuals and their families up to 2050 - such as births, deaths, migration, taxes, education, labour force participation, earned income, wealth accumulation and government transfers. The model is written in the C programming language and was initially used on a UNIX system. The dramatic increases in the speed and memory size of PCs over the past five years has led to a PC version now being available.¶ Despite their relatively short existence and long development phases, dynamic microsimulation models are now used in many developed countries – for example, the USA, UK, Canada, France, Sweden, Norway and Italy. In recognition of their ability to analyse distributional and financial issues in considerably greater depth than what is possible with traditional methods, their use by government for policy analysis is rapidly increasing.¶ In this thesis two new modules were added to the original Australian dynamic microsimulation model – namely: a Health_SES module and a Health State Transitions module. The former makes the study of health inequalities across socioeconomic groups possible. The latter provides a link between health status and the ability of individuals to carry out every day activities as the severity of their ill-health increases with age. A major advantage of adding these new modules to an existing main model is that it allows much more comprehensive studies over the life courses of individuals than the alternative would allow – that is the building of two stand-alone models developed exclusively for ‘health-SES’ and ‘health state transition’ types of applications.¶ The main data sources used to construct the two new modules were an extract from the Australian Institute of Health and Welfare’s Mortality database covering the 1995-97 period, and the Australian Bureau of Statistic’s 1998 survey of Disability, Ageing and Carers. The analysis of the mortality data was handled using EXCEL, and that of the much larger Disability survey unit record dataset - over 40,000 individuals and 100s of variables – using the SAS programming language.¶ While most of the methodologies used in constructing the new modules are in line with what became the norm for dynamic microsimulation model development, the thesis contains several innovations. The main ones are: a quantitative assessment of the suitability of different types of SES indicators for studies of health inequalities; the modelling of the progression of people’s health from illness-free status to mild and severe disability; the development of a methodology for estimating health state transition probabilities from cross-sectional data (in the absence of longitudinal data); and the linking of health status to individual’s ability to stay in the labour force.¶ As with most models, there are a number of limitations. These are discussed in the thesis, together with areas of possible future improvements.¶ The thesis also presents two novel and topical – though at this stage illustrative – applications of the enhanced dynamic microsimulation model. The first simulates the impact of a narrowing in health inequalities in Australia as health is lifted nationally to the level currently enjoyed by the most affluent 20% of the population. The findings are that, if such a policy change were implemented, close to half a million fewer Australians would be disabled, around 180,000 life years would be saved, health care costs would be around A$1 billion lower per year and the government could save close to A$700 million on the Disability Support Pension.¶ The second application quantifies the likely impacts of longer working lives in future, which may arise from changes such as: more favourable labour market conditions; government incentives to remain in the labour force longer (eg the lifting of the pension age); and general improvements in health. This application estimates the probability that Australians aged 65-70 would work more than 15 hours per week, had such changes eventuated. The decision to retire is modelled as a function of each individual’s own health, socioeconomic status, age, sex and family composition. The impacts are simulated in a world in which current patterns of health by age, sex and SES remain unchanged over time – the Base case; and a world replicating the narrower health inequalities scenario of the first application. Under the Base case an additional 450,000 persons aged 65-70 years were estimated to remain in the workforce - with the related earnings totalling up to $20 billion in 1998 ($35 billion in 2018) and savings by government on the age pension of around $2 billion ($4 billion in 2018). Under the narrower health inequalities scenario the numbers working, their earnings and the related savings on the age pension were estimated to be around 7% higher. Much of the original research carried out for this thesis has appeared, or is yet to appear, in refereed publications.¶

The aim of this thesis is to explore the quantitative potential to reduce energy requirements and CO2 emissions through changed patterns of consumption, given unchanged levels of consumption expenditure. The thesis question is analysed using a systems analysis approach which in this case means that life cycle assessment data on energy requirements and CO2 emissions related to household consumption are combined with a financial and behavioural analysis to make sure that the budget constraint is kept and that both the first and second order effects of adopting a green consumption pattern are analysed. The budget constraints are kept using a general linear model. By using marginal propensities to spend to direct the reallocation of saved or deficit money calculated utility is maintained as far as possible. Further, investigations explore the impact of individual household demographic characteristics and geographic context on household consumption patterns, energy requirements and CO2 emissions. The key result of this thesis is that changed household behaviour, choosing “green“ products and energy efficient technology will not make a big difference. What can be achieved in the short time perspective by adopting an almost completely green consumption pattern and energy efficient technology is a reduction of energy requirements by around 8% and CO2 emissions by around 13%. With a longer time perspective and further technological change that provides additional possibilities to move consumption patterns in a greener direction, the effect on energy requirements and CO2 emissions is still fairly small. By 2020, the potential to reduce energy requirements is around 13% and CO2 emissions around 25%. In the most extreme scenario (2050), the scope for reducing energy requirements is 17% and for CO2 emissions 30%. All these reductions will be outpaced by growth in income almost as soon as they are implemented. Of policy relevance the results reveal that very limited impact can be expected by a policy relying on greener consumption patterns, whether adopted voluntarily or as a result of incentives such as tax changes. Such a policy cannot achieve more than a small and temporary reduction to growth in energy requirements and CO2 emissions. It is also shown that, prescribing specific consumption patterns as a means of reducing energy requirements and CO2 emissions has to be done with care. This is illustrated by one of the experiments in which adopting a partly green consumption pattern, a green diet, in fact increased total energy requirements and CO2 emissions. This, and the results of all the other experiments show the importance of applying a systems approach. It demonstrates that life cycle data alone are irrelevant for assessing the total effects of adopting green consumption patterns. Further research on the potential to reduce energy requirements and CO2 emissions thus primarily needs to better capture system wide effects rather than to improve on, and fine tune the measurement of the energy requirements and CO2 emissions related to individual products.