SMOKING DECISIONS, HEALTH SHOCKS, AND EX POST MORAL HAZARD
In this paper, we suggest that smokers experiencing a health shock may not only update their beliefs in the health domain but also in the financial domain, through out-of-pocket medical spending. The onset of a severe condition can have substantial financial consequences that are difficult to predict. Market features related to medical care (e.g., limited price transparency) and peculiarities of these goods (e.g., diversity in types of treatments for a single illness, heterogeneity in price and treatment intensity, and the existence of health care provider discretion over treatment decisions) make it exceedingly costly to obtain information on potential out-of-pocket expenditures prior to illness, particularly for low probability but highly complex diseases. Thus, a given smoker may have poorly formed priors about her true financial risk exposure related to medical care. Receiving a health shock provides new information, which can then lead to different smoking decisions going forward.
To explore this potential mechanism, we focus on the degree of financial risk exposure to medical spending at the time of disease onset. We suggest that while health beliefs are similarly updated across insurance coverage groups after a shock, the signal in the financial domain is much stronger for individuals with less generous insurance coverage. Using data from the US Health and Retirement Study, we start by providing stylized facts on the smoking response to health shocks in various insurance groups (e.g., those experiencing an uninsured spell, those with Medicare basic coverage, and those with Medicare supplemental coverage) but then go on to model the differential impact of health shocks by financial risk exposure in a triple differences framework. The main empirical challenge is the underlying selection mechanism into insurance. To partly deal with this issue, we exploit the panel nature of the data to estimate individual-level fixed effects models and use the timing of severe, cardiovascular health shocks as a source of useful variation. Specifically, we use the age 65 Medicare-eligibility threshold to examine any differential effect of the health shock by pre- and post-Medicare status among otherwise similar individuals. We also separately analyze acute and chronic shocks, as learning is likely more immediate for the former.
Preliminary results provide consistent evidence that experiencing a cardiovascular health shock increase the probability of smoking cessation. However, the effect is smaller among individuals who face lower financial risk exposure to subsequent medical care needs. Market insurance appears to be a sufficient substitute for self-protection after a health shock. As more policy efforts are made to increase the generosity of insurance and shift away from risk-rated coverage, a potential unintended side effect (and externality) could be worse health behaviors among those who have already experienced the onset of chronic disease.