Health Insurance, Habits and Health Outcomes: A Dynamic Stochastic Model of Investment in Health

I develop a dynamic stochastic model of individual choices about health insurance, exercise, smoking, alcohol consumption and medical treatment. The primary objective is to estimate the parameters of the model to conduct counter-factual health policy experiments. The model is estimated through maximum likelihood using data on 3671 males from the Health and Retirement Study. A kernel smoothed probability simulator is developed to solve an initial conditions problem. Preliminary estimates show that the model does well in matching the means, frequencies and transitions in this sample for all the choices and states except for the insurance choices in the first half of the life cycle (ages 22 to 54). It needs to be emphasized however that the fit of the insurance choices is improving with the improvement of the estimates. The estimates are then used for two radically different counter-factual health policy experiments. In the first experiment that simulates the provision of comprehensive health insurance coverage, every individual is mandated to be on a health insurance plan that charges a premium of $1000.0 per annum (that is comparable in cost to the options in the data) but covers all out of pocket costs. The simulations suggest that the average health outcomes do not change much over the life cycle in the new regime. However the proportion of individuals smoking and consuming alcohol falls slightly, especially for those older than 60 years, with the decrease being as much as 0.8% in both instances. The proportion of individuals seeking medical treatment increases by as much as 49%. Average consumption of a composite commodity also rises by upto 7% providing partial evidence that the individuals are better off under the new policy. In another experiment that simulates the withdrawal of provision of subsidized medical care, all individuals are denied health insurance. Simulations reveal that the average health outcomes do not change much from the status quo. However the proportion of individuals smoking and consuming alcohol increases by as much as 3.2% and 0.5% respectively. The proportion of individuals seeking medical care falls by as much as 95%. The two experiments taken together suggest that provision of subsidized medical treatment tends to increase demand for medical care but fosters healthier behaviors. On the contrary withdrawal of subsidized treatment chokes demand for medical services but leads to increases in unhealthy behaviors. Of particular interest is the fact that the model provides no evidence of the existence of a moral hazard problem associated with the provision of subsidised medical care on habits like smoking and alcohol consumption.