The Impact of Health Status and Out-of-Pocket Medical Expenditures on Annuity Valuation

This paper describes how differences in health status at retirement can influence the decision to purchase a life annuity. We extend previous research on annuitization decisions by incorporating the effect of health differentials via differences in survival throughout the latter portion of life. Next, we consider how precautionary savings motivated by uncertain out-of-pocket medical expenses influence annuitization decisions. Our results show that annuities become less attractive to people facing uncertain medical expenses. While full annuitization would still be optimal if annuity markets were truly complete and both lifeand health-contingent, lacking this, annuity equivalent wealth values are much lower for those in poor health, as compared to persons in good health. Authors’ Acknowledgement This research was conduct with support from the Social Security Administration via the Michigan Retirement Research Center at the University of Michigan and the Pension Research Council at the Wharton School of the University of Pennsylvania. Additional support was provided to the first author by Grant AG10168 from the National Institute on Aging, Samuel H. Preston, Principal Investigator. Useful comments were provided by Irma Elo, Silvia Matos, Alex Muermann, Samuel Preston, Sara Rix, and members of the Wharton IRM Study Group. We are especially grateful to Jeffrey Brown for sharing his optimization code and for his many useful suggestions. Any errors are the authors’ responsibility. ©2004 by Turra and Mitchell. All rights reserved. 1 The Impact of Health Status and Out-of-Pocket Medical Expenditures on Annuity Valuation Cassio M. Turra and Olivia S. Mitchell The primary purpose of annuities is to protect people against the risk of outliving their own financial resources in old age. Prior research has shown that annuities should be of substantial value to risk-averse people who face an uncertain date of death (Yaari 1965), yet relatively few people seem to purchase annuities at the point of retirement (Johnson, Burman, and Kobes 2004). A growing body of research has focused on the factors that may explain this puzzle, factors that include a desire to leave bequests, the existence of adverse selection in the annuity market, the over-annuitization of retirement wealth, and the need for liquidity. Further, researchers have found that people use private information about their survival chances to make the decision of purchasing an annuity, and those who anticipate living longer are more likely to buy an annuity (Petrova 2003).Yet currently we have a poor understanding of how private information on health status is related to the demand for annuities. Investigations of this kind are challenging for a number of reasons, including the multi-dimensional aspect of health, and the absence of long-term nationally representative panel data on health at older ages. The objective of this paper is to address this gap in knowledge by describing how differences in health status at retirement can influence the decision to purchase a life annuity. To do this, we use dynamic discrete choice estimation which is based on standard models of economic behavior. We propose two models which incorporate the effect of health differentials on annuitization valuation. One approach incorporates the effect of health via differences in survival throughout the life-cycle. However this approach does not consider precautionary savings motivated by uncertain out-of-pocket medical expenses, so in an extension, retirees in