Projecting how long you think you will live is a crucial exercise in retirement planning. An estimate of life expectancy could determine, for example, your savings rate, a portfolio allocation, or whether or not you should buy an annuity. The standard theoretical model predicts that individuals make unbiased estimates of their life expectancy based on personal, relevant information, such as family history, illness, lifestyle choices, and so on. But a new paper, funded by our consumer finance working group, suggests that life expectancy estimates can be, at least in part, also affected by irrelevant context factors -- in this case, the way survey questions are worded.

For their newly published article, John W. Payne, Namika Sagara, Suzanne Shu, Kirstin C. Appelt and Eric J. Johnson conducted experiments in which they asked half of the respondents to provide probabilities of their living to a certain age, and the other half to provide probabilities of their dying by a certain age. Ostensibly, these questions are asking the same thing, but the results yielded a surprising result: Those given the "live-to" question reported significantly higher chances of being alive at ages 55 through 95 than those who answered the "die-by" question. In fact, in the first two surveys, which included nearly 2,000 respondents, the mean life expectancy was 8.68 years higher in the live-to frame than the die-by frame.

The authors are careful to note that the wording of the question is not the only factor that respondents considered in answering the question; relevant factors that affect longevity (age, gender, health status, etc.) also clearly did have some impact on life expectancy projections. But why should the frame -- the words "live to" or "die by" -- have any effect at all? To investigate, the authors conducted an experiment in which they asked respondents to list their thoughts as they considered the probability of living to (or dying by) 85 years old or older (or younger). Participants then later coded these thoughts as being more about life/living, death/dying, or neither, as well as more about being positive, negative, or neither. The authors' analysis suggests that the live-to/die-by frame affects what participants retrieve from their memory. When given the living to frame, respondents had significantly more thoughts about living to age 85 and fewer thoughts about dying. The live-to framing also significantly increased respondents' optimism; they had more positive thoughts about being alive at 85 years old and this, in turn, increased their estimate of the probability that they would in fact be alive at age 85.

The study has important implications for those trying to measure consumer preferences and beliefs. In addition, as more American baby boomers near the end of their professional careers, this study shows that consumer "biases" can affect retirement planning; the authors found evidence that preferences for a life annuity could change depending on the order in which life expectancy questions were given. Could using different life expectancy question "frames," as this study did, also affect consumers' decisions about when to claim Social Security benefits? Could framing and life expectancy tasks have a similar impact on other consumer decisions? More research will surely be needed.

Read the study (subscription required) here. You can find more information about consumer finance working group here.