Health Information, Subjective Probability of Survival and Health Behavior

Hamermesh(1985)provided the first study about whether people’s longevity expectations conform to life tables. Using two different samples, he found that subjective longevity perceptions do not accurately correspond to actuarialdistributions. The subjective distributions were flatter and exhibited greater variance.

Moreover, respondents tend to base their beliefs disproportionatelyon relatives’ longevity. In recent years, several studies examine the evolutionof subjective longevity expectations over time based on the panel data from the Health and Retirement Study(HRS)and the Asset and Health Dynamics among theOldest-Old Study (AHEAD). Hurd and his colleagues provide a consistent evidence on the predictability of the subjective survival probabilities on actual mortality.

Hurd and McGarry(1995)found that respondents with higher socioeconomic status(education, income and wealth)give higher probabilities of survival, whereas respondents who smoke give lower probabilities. Hurd and McGarry(1997)use two waves of e HRS and suggested that subjective beliefs decline with the death of a parent and respond to only one type of new health shock–cancer. Hurd, McFadden and Merrill(1999)focus on an older population(i.e., aged 70 or older)by using the panel structure of the AHEAD.

These author also confirmed that subjective beliefs do respond to the onset of health conditions (cancer, high blood pressure, diabetes and depressions).On the other hand, Smith et al.(2001)use a risk-updating framework to investigate how exogenous health shocks impact people’s longevity expectations. Their result supports the conclusion that smokers have a different risk perception process. Smokerswere only sensitive to their own smoking related illness while former smokersand those who never smoked react to a wider range of health signals.

Further,Smith, Taylor and Sloan(2000)use the most complete version of the HRS available (four waves)and found that serious health shocks and new activity limitations do reduce longevity expectations, which is consistent with the previous findings. We will address three issues in this study. First, we investigate the determinants of subjective probability of living to 75 or 85.

Second, we examine how longevity expectations respond to new health shocks. Third, using a simultaneous framework, we explore the roles of health information(including health shocks and health knowledge)in determining health behaviors. To obtain detailed information on health behaviors and expectations, we will conduct a survey of adults who take the physical checkup by free-of-charge at Mackary Memorial Hospital.

Our analysis is based on the information of pre-physical checkup and post-physical checkup interview as well as the reports on health evaluations. The measures of health shocks will be constructed as the incidence of health conditions according to the evaluation reports. Prior work have found that unrealistic stated subjective survival probabilities are associated with low cognitive ability.

To control for the focal-point responses, we examine the determinants of subjective survival probabilities by using the probit model with sample selection. Our second test recognizes that all the determinants of subjective probabilities must be considered in estimating the risk equivalent associated with new information.Following the approach developed by Viscusi (1990), we can evaluate how new information, acquired through exogenous health shocks, affects people’s longevity expectations.

The impact of health shocks on health behaviors can be examined after treating health knowledge as an endogenous explanatory variable in the health behavior equations.