Estimation

My goal is to estimate the impact of retirement on health-seeking behavior. The questions of interest for this paper include: Do individuals utilize more, less, or the same degree of medical care after they retire? Also, are individuals willing to pay more out of pocket after they retire?

The main dependent variables in this paper include total medical expenditures, medical expenditures that are covered by health insurance, and out-of-pocket medical expenditures. Y1i is defined as total medical expenditures, Y2i is defined as medical expenditures that are covered by health insurance, and Y3i is defined as out-of-pocket medical expenditures. Di is a binary variable indicating whether an individual is retired or not. Si measures the age of the individual.

A fundamental problem for the estimation of these equations(1.1)(1.3) is that the retirement de­cision itself Di is endogenous. Some underlying unobservables might drive either the retirement decision or the medical expenditure variables, or both. However, as the Chinese mandatory retire­ment rule is in place, the age threshold for retirement ( 50 for females and 60 for male) provides a credible source of exogenous variations in retirement status. Fig 1 and Fig 2 show the proportion of retirees over the life cycle. The overall retirement rate rises about 20% at age 50 for females, and rises about 28% at age 60 for males.

The mandatory age rules differ for males and females, and the medical expenditures are measured at the individual level. Therefore, this paper has chosen to apply the estimations to male and female samples separately.