Not too long ago, economist Michael Keane sat across a desk from the chief executive of a leading insurance company. In a playful mood, Keane tossed out one of those rascally jokes that's so close to the truth it's downright cringe-worthy. Let's just cut to the bottom line, he said to the executive. You want to sell insurance to people who don't need it.

The executive sat back in his chair, looked Keane straight in the eye and said, Yes. That's exactly right.

In fact, any insurance seller could have made that statement. That's the name of the game, and it's why the term adverse selection can cause insurance folks much angst.

Adverse selection is what happens when the people who need protection most -- those, for example, with the greatest health problems or worst driving records -- buy lots of coverage.

For decades, insurance companies have been pricing policies based on thebelief that adverse selection comes into play among their customers. But Keane, a professor of economics at the W. P. Carey School of Business, says it does not. There are no empirical data proving adverse selection, he says. Other factors are operating and, in fact, insurance companies often benefit from advantageous selection, in which the best risks also are their best customers.

Proof goes poof!

Several studies have looked for the existence of adverse selection in insurance buyers, and many have found it missing.

Researchers in France determined that young drivers who picked comprehensive policies had no more accidents than drivers who picked the legal minimum coverage. In U.S. life insurance markets, researchers discovered that mortality was lower for men who bought policies than it was for those who decided to meet the Grim Reaper without greenbacks to follow. It's the same thing with annuities. You ought to want to buy annuities if you're likely to live a long time, but you don't see that pattern, Keane says.

And, he adds, it's a mystery. In fact, it's a mystery that has led to a new direction of research, one where scholars investigate the possibility ofadvantageous selection that may result from risk aversion.

According to Keane, popular opinion has it that those who are most averse to risk take better care of themselves, and they also demand more insurance. If the more risk-averse people take better care of themselves and are healthier -- and they buy more insurance -- you end up with healthy people feeding insurance-company coffers.

This was the belief that Keane held when he teamed up with Duke University's Hanming Fang and the University of Michigan's Dan Silverman. The three economists, none of whom is focused on health issues, nevertheless ventured into this area to see if risk aversion explained why adverse selection is an empirical boogeyman, not a real and serious threat to health insurers' profit margins. The team examined the question by looking at seniors in the U.S. who were eligible to purchase the Medigap insurance that supplements Medicare.

As it turns out, this research earned the team an Arrow Award, a prize conferred annually on writers of the best paper in health economics by the International Health Economics Association.

All three of us were expecting risk aversion to be the answer, Keane admits, but it wasn't. Overall, the team found that high-risk or less-healthy seniors didn't buy the supplemental insurance for one simple reason: They probably didn't understand it.

A model of good health

In exploring the mysterious but apparent lack of adverse-selection, Keane and his colleagues decided to narrow their research to seniors who are eligible for Medigap insurance, the insurance that supplements Medicare. The team considered Medigap ideal for a number of reasons. First, insurers can't deny coverage or charge more for pre-existing conditions for six months if the policy applicant has reach age 65 and enrolled in Medicare Part B. What's more, the policies are relatively standardized. All of this leads to a relatively simple decision for new Medicare enrollees: to supplement or not.

Along with the simplicity of the decision, the researchers chose this particular set of insurance buyers because there is a lot known about them. Specifically, the team tapped Medicare administrative data. Medicare has very good information on health and expenditures for individuals. Keane explains. There is a list of hundreds of possible health conditions someone might have. There also are data on a person's healthcare costs. This allowed the team to develop a good model for predicting the likely medical costs for a person with any given set of conditions.

What's missing are data on less tangible traits people have, such as how cautious folks might be. To discover that, the team examined the Health and Retirement Study (HRS), a long-term study of a large group of senior citizens. The HRS contains extensive measures of health, just like the Medicare data. But in addition, the HRS also contains a number of questions that ask people to say if they'd be willing to take certain gambles. This measures risk aversion.

The survey also gives respondents little math and reasoning tests, so you can measure cognitive ability, Keane notes. Plus, it asks people how far in the future they'll plan ahead. It's a whole battery of psychological questions.

Keane et al. took these data, married them with the Medicare facts and figures, and then built a model that pinpointed expected healthcare costs for each person examined in the research.

Once we had this measure of expected costs for every person, you would think that the people with higher expected costs would be more likely to buy the extra insurance, Keane says. But, in fact, they're less likely. As it turns out, the people who bought Medigap policies had expected annual healthcare costs equaling some $4,000 less than the expected costs for those who stuck with Medicare's minimal coverage alone.

Once the researchers factored in the psychological variables -- especially cognitive ability -- it turns out that those who purchased Medigap coverage did have higher healthcare expenditures -- about $2,000 more per year than those without the supplemental policies. For the population as a whole, adverse selection isn't operating, Keane says. But, once you consider cognitive ability, it is the less healthy who buy more insurance.

The huh? factor

So, what makes the general population of seniors disprove the theory of adverse selection? As noted earlier, the older folks probably don't understand the policies or their value. At least, that's what Keane's research indicates.

Actually, Keane said he probably should have figured this out earlier, based on research he conducted with health economist Katherine Harris. Done in 1998, the Keane/Harris study also looked at adverse selection and, here again, found it wasn't the why behind a buy. We expected to find that the people who were less healthy would want the supplemental plans with the most coverage, Keane says. We found there were other things that were more important. The single most important thing for seniors is whether they can stick with their long-time doctors.

When Keane conducted this research, his primary interest was econometrics, or the statistical research methodologies that allow scholars to delve into such questions. In pursuit of new research tactics, he'd created a model that examined decision-influencing attributes of healthcare plans -- such as physician choice or required co-pays -- as well as perceived attributes, as opposed to real ones.

What I found in the paper with Harris is that people who wanted low co-pays chose to have Medicare alone because they thought Medicare had the lowest co-pays, he says. But, in fact, Medicare had the highest co-pay. Seniors didn't understand the plans.

Similarly, Keane, Fang and Silverman found, after looking at HRS data that measure cognitive functioning, folks with higher reasoning ability were both healthier and more likely to buy Medigap insurance. Meanwhile, seniors with low cognitive ability tend not to buy the added coverage. They don't get the value of supplemental insurance, Keane says.

You can't have adverse selection if people don't understand which plans provide more coverage, he adds.

Policies on policies

The fact that senior citizens have difficulty understanding insurance plans has implications for public policy, he continues. What it suggests is that it's important to set up mechanisms to help people make choices more easily.

For instance, Keane envisions software that's similar to popular tax-preparation packages. You could plug in information about yourself, and it would tell you what kind of insurance to buy, he says. Perhaps there also will come a day when people consult insurance advisors, just as they might get tax help at a local tax-preparation shop like H&R Block.

Adding to Keane's concern over insurance bafflement among oldsters is a study conducted by his wife, health policy expert Laurel Hixon. She examined Medigap prices in the state of Florida, and found that sometimes the same coverage costs twice as much from one insurer to the next.

According to standard economic theory, this shouldn't happen, Keane notes. The law of one price states that, in efficient markets, identical products should have identical price tags. When consumers are confused, the law of one price can break down, he observes.

Given this, plus his own findings that those who don't understand insurance don't tend to buy it, Keane says elders could be losing out. You can get huge welfare losses with people not buying what they might use or buying more expensive coverage than they need, he notes. Losses to seniors could be very substantial.

On a broader scope, Keane also sees how his research findings could have implications relevant to today's health care debates in Washington. If a goal of health care reform is cost containment, Keane reminds us that health maintenance organizations didn't appeal to seniors, because the seniors cared so much about provider choice. They wanted to stay with their own doctors. And, he notes, Co-pays can't work to control costs when people don't understand them. The research seems to invalidate the two main approaches to cost control in the U.S.

According to Keane, You have to understand how people make choices before you can make policies aimed at influencing behavior. He maintains that the U.S. government has, to date, ignored this reality. And, it is continuing to do so in the current reform effort, he concludes.

Bottom Line:

  • Adverse selection is when those who need insurance most also buy the most of it.
  • Insurance companies operate on the assumption that adverse selection occurs.
  • Research offers little empirical evidence to support that assumption.
  • Other factors drive insurance purchases. In the case of seniors and Medigap policies that supplement Medicare, researchers found buyers were less likely to chalk up healthcare costs. In fact, advantageous selection was at play.
  • Seniors, however, may be missing out. The research also indicates that the reason those who are less healthy don't buy Medigap policies is because they probably don't understand them or their value.