Duke Economics Working Paper #00-15
Using four waves of the Health and Retirement Survey (HRS) this paper tests whether longevity expectations predict actual mortality at the individual level. The results suggest longevity expectations do predict mortality reasonably well. Serious health shocks and new activity limitations do reduce longevity expectations. Given one is prepared to accept that other unobserved causal factors have the same means for those who die and those who survive in each wave it is possible to test whether longevity expectations can serve as a sufficient statistic. The test findings imply that they do not appear to reflect all that individuals know about their personal odds of living to seventy-five.
JEL: I12, J14, D80
Published in American Economic Review, Vol. 91, No. 4, September 2001, pp. 1126-1134.