Study: Social Security overestimates death rates

Reuters reports on a new study that concludes that the Social Security Trustees and actuaries underestimate improvements in life expectancies, particularly due to declines in smoking. If health improvements due to lower rates of smoking are included, authors Haidong Wang of the University of Washington in Seattle and Samuel Preston of the University of Pennsylvania argue, future life spans may be longer and Social Security's long-term deficit larger than currently projected. (Here's a link to the full paper.)

For instance, Wang and Preston conclude that while current SSA mortality projections assume a 50-year-old man in 2034 has only a 39 percent chance of living to age 85, an estimate that includes smoking would put the likelihood of surviving to 85 at 57 percent.

A couple thoughts: first, Wang and Preston's paper builds smoking into the Lee-Carter model of mortality, which already projects longer life spans than does SSA. Perhaps a third of the difference between the Wang-Preston estimates of life expectancies and SSA's are simply due to starting with a model that already projects lower mortality. The remaining two-thirds or so is based on Wang and Preston's modeling of smoking.

Second, it's not easy to translate the results in the Wang-Preston paper into an estimate of how much larger Social Security's deficit would be were Wang and Preston's estimates to prove correct. A reasonable guess is that adding smoking to the baseline Lee-Carter estimates would place mortality reduction close to the "high cost" projections of Social Security's trustees (their main projections are known as "intermediate cost").

Were the high cost projections for longevity to hold, and all other economic and demographic projections remain at their intermediate cost values, Social Security's 75-year shortfall would increase from 1.7 percent of taxable payroll to around 2.3 percent, an increase of 35 percent in the size of the long-term deficit. At the least, this study shows a) that considerable uncertainty exists regarding improvements in future life spans, and b) that Social Security's finances are very sensitive to changes in longevity, more so than to other factors such as economic growth.