For most of the past century, the financial-services industry has used actuarial tables to design life-insurance policies, pensions, and other products based on predictions of human life span. These “life tables” rely on historical death rates to predict the future longevity of broadly defined population groups. But human life expectancy has increased dramatically—from 47 years in 1900 to 77 today in the United States, with similar surges around the world. That’s led to skyrocketing pension and health-care costs. What’s more, sizable variations in longevity have emerged among different subgroups. Thus the financial-services industry no longer considers life tables adequate, as they leave too much room for companies to lose money.
A growing number of corporations and governments are turning to an emerging group of lifespan modelers. These experts are studying the living in an attempt to predict who will make it well into old age—and who won’t. “Life tables are crude and based on the past,” says S. Jay Olshansky, a professor of epidemiology at the University of Illinois at Chicago and cofounder of GD Analytics, a longevity consulting firm. Olshansky says we now need to “generate much more finely grained estimates of survival.”
A couple of years ago, Olshansky traveled around the world lecturing J.P. Morgan Chase executives on the rapidly changing face of human longevity. It was an eye-opening lesson for many at the firm. “He had striking visuals showing the growth of obesity in the Western world,” recalls Guy Coughlan, a managing director in J.P. Morgan’s London office. “He showed us how individual trends specific to a country, social class, or occupation all play into longevity.”
For example, Americans between the ages of 55 and 64 are not as healthy as the generation that preceded them into retirement, Olshansky says. People born in the United Kingdom in the 1930s—often called the “golden cohort”—have lower death rates than the generations that came before and after them. Insights like these are vital to J.P. Morgan, which is among a growing number of Wall Street firms that offer “longevity swaps”—insurance policies that pension funds use to hedge the risk that pensioners will outlive their funds’ reserves.
In addition to correlating health conditions with longevity, life-span modelers drill far deeper into individual traits than traditional actuaries do. So, says Olshansky, he’s not just a 56-year-old white male in the United States. “I’m also a Jewish male with a given level of education,” he says. “Education and religious background actually play an important role, as do the state you live in and the duration of your parents’ lives.”
Other companies are creating models based on what people buy online, what magazines they read, and even what hobbies they pursue. According to a recent Wall Street Journal report, Deloitte Consulting uses models that predict individuals’ risk of specific illnesses on the basis of their exercise habits, shopping patterns, and hankering for fast food. Future models may even include genetic test results and data from wireless health monitors. Scrutinizing all these factors yields insights about specific subgroups of the population, which financial professionals can then use to predict the longevity of individuals and the average longevity of groups.
Earlier this year, the Society of Actuaries published a report called “Health Expectancy,” which explored specific population cohorts by age, health condition, gender, and other factors. “Health expectancy” was defined as the number of years specific population groups are likely to stay healthy and independent, as opposed to time they can be expected to spend, say, in assisted-living facilities. Among the society’s pieces of advice to life-span modelers: A healthy 75-year-old man can expect to make it another 10.52 years without needing assisted living or full-time nursing care. But a diabetic 75-year-old man can only expect 7.92 more healthy years. A 75-year-old woman who has osteoporosis but has suffered no fractures and doesn’t have diabetes can enjoy an additional 8.16 good years.
The quest to perfect life-span modeling could ultimately benefit various industries in addition to insurance and health care. Organizations that maintain pension funds or offer lifelong health benefits have been struggling to keep pace with the expanding population of elderly retirees. General Motors, for example, disclosed in the spring that its pension fund is underfunded by $27 billion. Insurance companies commonly assume that death rates will decline 1 percent a year. But if their forecasts are off by as little as 0.25 percent, they can lose billions across many funds at a time of severe economic distress, says Olshansky.
There are now several artificial-intelligence programs that can model individuals’ longevity on the basis of their health status, pharmacy records, lifestyle, and habits—and then predict the potential impact of behavioral changes. Some health plans are using this data to target members who face an increased risk of developing chronic diseases because of lifestyle liabilities such as obesity and smoking. “They can tell specific members ‘If you change your lifestyle, we’ll lower your premium,’” says Anand Rao, a principal with PricewaterhouseCoopers’ Diamond Advisory Services unit.
Fine-tuned life-span models also benefit the growing market for “life settlements”—life-insurance policies that investors purchase from individuals. “That investor makes a bet on how much longer you have to live, pays the premium until you die, and then gets the full amount of the policy,” explains Robin Willi, owner of Rigi Capital Partners, a Swiss company that purchases and manages life settlements. They’ve become popular investments because they produce yields of 12 to 20 percent, making them more attractive than bonds, which yield as little as 4 percent, Willi says. But investing in them is a fine art. The ability to make a good return, Willi says, “depends on old people dying in a predictable manner.”
So life-settlement investors make assumptions about life span based on data they collect on past holders of life-insurance policies. For example, such investors assume that an upper-middle-class person who has a $5 million policy and resides in Florida is likely to live longer than a Midwesterner with a $500,000 policy. “If you can afford a $5 million policy, and you can afford to live in a gentle climate, then you can also afford good medical care,” Willi says. “It sounds cruel, but it’s just reality.”