Most of us, if we’re lucky, will grow old someday, and may need someone to help us get through the day. One in three of us will enter a nursing home, at a price tag of $92,000 a year on average — or far more for one with high-quality care, conveniently located. One in six will need daily care for at least three years. And Medicare does not pick up that tab.
Yet surprisingly, only 10% of elderly Americans have any sort of insurance for long-term care. Altogether, existing policies cover just 4% of spending on such care nationwide. The result is a uniquely American tragedy: Many ailing seniors spend down all their assets — intentionally becoming indigent dependents — to qualify for a bed in a Medicaid ward.
Chris Tonetti, an economics professor at Stanford Graduate School of Business, knows something about that: As a kid, his family had no other recourse when his grandmother’s dementia reached a point that she needed in-patient care. The cost was simply beyond their means. “I’ve seen the risks up front in my own life,” he says.
Given the odds, why don’t more people buy insurance? “It’s a puzzle,” Tonetti says. “If you take the numbers at face value, you conclude that people just don’t care that much what their quality of life will be in old age. I found that pretty hard to believe.”
For one thing, he says, working-age people do save a portion of their earnings in IRAs and other retirement accounts. Why would they do that if they didn’t care about their future comfort? Some economists countered that people save in order to leave a bequest, citing the fact that seniors who enter retirement with a nest egg of, say, $400,000 in assets often die with $400,000 in assets.
Tonetti still wasn’t convinced. “Sometimes,” he says, “the data the world presents to you just can’t answer the question you’re asking.” So he, along with four coauthors, took an unusual approach to solve that puzzle in a pair of recent papers: They engineered their own data.
“Economists usually look at market outcomes — how much was actually sold — and try to infer consumer preferences,” he says. “We went at it from the other end. We went to the consumers and asked them a lot of probing questions to estimate their preferences directly. Then we made up a simple insurance product, with actuarially fair terms, and predicted the market for it.”
What they found is that 60% of consumers in their model would buy that policy. The difference? The insurance available in the real world is nothing like Tonetti’s idealized version. In fact, the current options don’t eliminate the financial risk of late-in-life care.
“The demand is there for good long-term care insurance,” Tonetti says. “The problem is, the industry isn’t offering it.” In short, the market is failing, and that failure has a tremendous hidden cost.
Finding the Missing Customers
To construct their model, Tonetti and his colleagues were able to use a panel of nearly 1,000 Vanguard investment clients who had volunteered for studies under the Vanguard Research Initiative. Each participant was run through a rather grueling survey in which they had to choose between pairs of high-stakes outcomes, like something out of Let’s Make a Deal.
“For example, you can have $100 or you can trade it for a lottery ticket with a 50% chance of winning $120 and a 50% chance of getting only $90,” Tonetti explains. “Which would you take?” If a test subject took the lottery, the questions kept upping the risk until the person folded.
Another one: It’s the last year of your life, you need care, and you have $100,000 in the bank — just enough for a private room in a facility. How much do you spend on yourself and how much do you leave for a bequest? What if you have $200,000?
In this way, the researchers were able to disentangle people’s motives for saving. It became clear that for most respondents, leaving money to heirs was seen as a luxury, not a necessity. On the other hand, most were very worried about the cost of late-in-life care, as Tonetti had expected — it was their primary reason for saving.
“That can explain why someone might die with their wealth intact instead of spending it,” he says. They weren’t hoarding it for a bequest, as economists had thought. Rather, it was a fund held, even to the end, against the risk of ruinous expenses. “Essentially, they’re self-insuring.”
But even better, those precisely quantified hypotheticals allowed the researchers to measure exactly how people would trade off competing goods — like current versus future consumption — and their tolerance for risk. Those are, of course, key factors in the market for insurance.
Using these estimates of preferences, the researchers could now generate demand for long-term care insurance. Not just a scatter of data points showing actual purchases, as you’d get from market data, but the full underlying demand curves. What’s more, they could see the demand for each individual separately, revealing how interest varied with a person’s circumstances.
Solving the Puzzle
The final step was to create a simple, hypothetical insurance product: Customers would pay fixed premiums, and if they ever needed help with daily activities, they’d get a certain amount of money each month to pay for it. “It’s what we call a state-contingent asset,” Tonetti says. “If this state of the world comes up, you get the benefit, no questions asked, to use as you see fit.”
With premiums set at an actuarially fair level, based on each individual’s age, gender, and health, the model showed that around 60% of consumers would buy that policy — far more than the 10% who hold long-term care insurance today.
What’s more, because they had individual demand curves, the researchers knew not just what people paid but how much they actually valued it. The difference, known as “consumer surplus,” was large, anywhere from $100,000 to $500,000 for many individuals. They didn’t just want insurance, they wanted it badly.
But that potential surplus remains unrealized in the market today, due to the industry’s failure to offer an adequate product. Multiplying the individual loss by the number of lost customers — a large percentage of the American population, whatever the exact figures might be — yields a measure of the enormous loss in public welfare.
What’s Wrong Today
The flaws in existing policies are many. One common gripe is that premiums are too high relative to benefits. But Tonetti’s model shows that demand for long-term care insurance isn’t very sensitive to price — increasing premiums by 30% over the actuarially fair price had little effect on purchases.
The bigger deterrent, surely, is that the policies one can buy today don’t actually eliminate risk. “Those earlier studies basically assumed we all have access to a state-contingent asset and choose not to buy it,” Tonetti says. “But these aren’t state-contingent assets at all. They work on a reimbursement model. You pay for the care yourself and then hope to get your money back.”
Stories abound of insurance companies denying claims or dragging out the process. “It can get adversarial,” Tonetti says, “and you might be in no shape to fight back or might be dying and have a short horizon.”
Short stays in a facility, the most common case, are not covered because of deductibles. Long stays, often needed for patients with cognitive decline — the most expensive case — are not covered because benefits end after one to five years. Within those bounds, there are limits on the services paid for and where they can be delivered. And, oh, your premiums might be raised at any time; fail to pay and you lose your coverage.
Tonetti says those flaws don’t entirely explain the under-insurance puzzle. When the better policy was explained to test subjects, not all those predicted to want it said they’d actually buy it. But that gap arose mainly among the wealthiest individuals, who can rely on their own resources.
For the majority of elderly Americans, the introduction of an improved form of long-term care insurance would offer a tremendous increase in quality of life, not to mention peace of mind. And by lightening the load on Medicaid, it would be a relief for state and federal finances as well.
That’s not to say it would be easy. These papers don’t analyze why the market appears to be failing, but fears of “adverse selection” are likely a factor; that’s when coverage is purchased mainly by people who expect to cash in on the benefits, making it unprofitable. But Tonetti and his colleagues have convincingly demonstrated that there’s an unmet demand for long-term care insurance — a big opportunity for any insurer who can figure it out.
Christopher Tonetti is an assistant professor of economics at Stanford Graduate School of Business. His coauthors on the papers “Long-Term Care Utility and Late-in-Life Saving” and “Late-in-Life Risks and the Under-Insurance Puzzle” are John Ameriks, Vanguard; Joseph Briggs, New York University; Andrew Caplin, New York University & NBER; and Matthew D. Shapiro, University of Michigan & NBER.