Abstract
The Affordable Care Act includes an individual mandate whereby persons are required to carry health insurance. This mandate will bring currently uninsured persons into the insurance pool. The uninsured are a heterogeneous group that includes persons with diverse risk preferences. It is important, therefore, to understand the role risk preference plays in (1) the likelihood of being uninsured and (2) the health care expenditures. Using the recently available data from the Medical Expenditure Panel Survey (MEPS), we analyse eight years of US data using multivariate regression and quantify the role of risk preference in insurance and expenditure equations. The results provide evidence that a person with high risk preference is less likely to hold health insurance, and spends less on healthcare even when controlling for insurance.
Notes
1 The estimation of the insurance decision and health expenditures raises a number of econometric issues. First, the relationship between insurance coverage and health expenditures is potentially endogenous. The decision to acquire health insurance coverage can be influenced by the expectation of the amount of health care needed (Frank, Citation2001; Gemmill et al., Citation2006). Second, health expenditures contain a significant number of observed zeroes. Sample selection may be an issue: observed zeroes can reflect that either (1) a person does not require medical care or (2) a person requires medical care but chooses not to purchase it (for example, an unfilled prescription). Therefore, a sample selection model was estimated (results are available upon request). Consistent with the Gemmill et al. (Citation2006) model of inpatient care, the selection model estimation indicates that sample selection is not present in the model (rho is insignificant). Given this result, this article utilizes ordinary least squares on the sample of persons with positive expenditures and includes the predicted insurance from the logit model.