Fall is the season of open enrollment, when Americans with employer-sponsored health insurance have the opportunity to reassess their coverage and change providers. For tens of millions of Americans who don’t have health insurance, employer-sponsored or otherwise, this fall has also brought the ability to choose health insurance for the first time with the roll out of the Affordable Care Act (ACA).
“For people who already have insurance, this can be a big decision,” says Professor Eric Johnson, who has a well-established career path studying how consumers assess and make financial (and other) decisions, yet confesses he needs a quiet room and closed door to make sense of his own benefits statements. “So I think that people who have never bought healthcare insurance before would find it difficult. And my coauthors and I wanted to learn if people could make this decision well, and if not, whether we could help them.” Johnson worked with Allison T. Bajger of Columbia University, Tom Baker of the University of Pennsylvania, Ran Hassin of Hebrew University of Jerusalem, and Galen Treuer of the University of Miami.
The average person in the target market for the healthcare exchanges tends to be on the lower end of the income spectrum and is likely to spend 5 to 10 percent of her income on health insurance. If she doesn’t make the best, least costly choice for her needs, that cost represents a real liability. And, since many will buy coverage using government subsidies, making expensive mistakes on the healthcare exchanges will also cost other taxpayers.
The researchers designed a simple choice set from which study participants would choose from four or eight insurance policies, each varying in price and with different premiums, deductibles, and co-pays. Participants were asked to pick the least expensive coverage for their situation, which researchers provided them. For example, a participant might be told to assume she would make 3 doctor visits and spent $1,200 on out-of-pocket costs for things like x-rays, prescription drugs, or lab tests.
The results were shocking. “People chose very poorly,” Johnson says. “They were picking very close to chance — almost as if they were throwing darts at a dartboard.” These poor choices cost hundreds of dollars relative to the least expensive choices.
In the next study — one of several subsequent attempts to uncover which kinds of prompts would yield the least expensive choices — they paid people to pick the right policy. “We thought, maybe people are being lazy, and incentives would help,” Johnson explains. “But there was almost no difference from the first study. People still chose barely better than chance would dictate.” (Curiously, when participants were offered financial incentives, they not only spent a lot more time making their choice, they were more confident that they had chosen well — even though they had not.)
In a third variation, the researchers included a calculation of what each policy would cost, which cut down on errors significantly. But a surprising proportion of people still made poor choices, prompting the team to wonder if making what at first seemed a relatively straightforward choice was essentially impossible, especially for the typical participant, who did not have formal math or finance skills. So the team tested how well MBA students did. This population chose the least expensive policy about 75 percent of the time, making far fewer mistakes than other participants. When the researchers asked the students how they made their decisions, most students reported, perhaps unsurprisingly, that they’d plugged the numbers into an Excel spreadsheet to calculate the least expensive option.
“We ended up with this quandary,” Johnson says. “Ordinary people — those without finance or business backgrounds — don’t do this math.” To see if they could prompt laypeople to chose the right insurance policy, they began combining choice strategies in their subsequent experiments. They started by teaching the participants how to calculate the relative cost of each policy, like multiplying the monthly premiums by 12, and adding the smaller of their deductible or out of pocket costs, and adding in the other costs based on the number of doctor visits they’d been told to estimate, and so on. This just-in-time education improved choices a little, but when the researchers combined education with completed calculations, the average mistake decreased from $500 to $200.
Up to this point, the researchers hadn’t used one of the newer tools for designing choices — setting up the right default option. On a theoretical live healthcare exchange this might work as: a tool on the exchange site asks you how many times you visited the doctor last year and what your total healthcare expenses were, or how many times you expected to visit the doctor this year, and then recommends the least expensive policy for those circumstances. In one study, researchers mimicked this kind of set up by pre-checking the cheapest policy for each participant, who remained free to choose any policy they wanted. Results were good, yet 20 percent of participants still picked a different, more expensive policy.
The very best results came when the researchers included the pre-checked default with the calculator — under this design, participants chose as well or better than the MBA students. “The combination of telling people the cheapest option and why they should choose it actually turned out to be a big win,” Johnson says.
Why do consumers choose so poorly on their own? “People seem to have a problem combining the different costs,” Johnson says. “There’s a monthly premium; there’s an out-of-pocket for each visit to the doctor. They are not doing all the math and instead appear to be overweighting deductible and out-of-pocket costs. As a result, policies with high deductibles tend to be selected much less often than they should be,” Johnson says. “People who have only very basic math skills seemed to benefit more from this choice architecture, so we think this could help people who need it the most.” (Johnson notes much additional evidence that people over-weight deductibles when choosing similar financial products, such as auto insurance.)
While the researchers embarked on this project wondering how they could help people make less costly choices, during the study they realized their results would let them estimate the dollar value of improving those choices. Without any kind of decision aid, people made mistakes of about $570 on average. “That’s a lot for an individual, and if that individual’s choice is subsidized by the government, it’s a lot out of the federal treasury,” Johnson says. “When we present the least expensive choices to individuals, they make mistakes of less than $100 on average.” The researchers’ made a rough calculation that if 20 million people buy insurance on the exchanges, the right choice architecture could help consumers and the treasury could save between 9 and 10 billion dollars.
And how do insurers fare? “Insurers could, in fact, try and take advantage of these mistakes, and so I could imagine an insurer offering a low deductible policy that had a higher premium, which would raise their profitability,” Johnson says. “But if I’m a good provider with a good cost structure, I want to make sure the market is efficient and people are picking the policy that’s best for them, and so some insurers will, I think, be very thrilled if the choices are made more efficiently by consumers.”
Johnson expects that while small businesses will do the math to make smart choices, they might also benefit from similar choice architecture. “If I were running an insurance brokerage, I would try to make it easier for small businesses to make this decision. That would be a huge competitive advantage.”
The researchers have been busy sharing their results with the various state-run exchanges and are negotiating with some exchanges to study the decision-making processes of consumers in the real world. “We hope that allows us to learn even more about what is and isn’t effective,” Johnson says. “It’s going to be an amazing experiment — a sort of laboratory of democracy.”
Listen to Eric Johnson discuss this research in this Columbia Ideas at Work podcast:
Eric Johnson is the Norman Eig Professor of Business in the Marketing Division and director of the Center for Decision Sciences at Columbia Business School.
Eric Johnson is a faculty member at the Columbia Business School at Columbia University where he is the inaugural holder of the Norman Eig Chair of Business, and Director of the Center for Decision Sciences. His research examines the interface between Behavioral Decision Research, Economics and the decisions made by consumers, managers, and their implications for public policy, markets and marketing. Among other topics...
Read the Research
Eric Johnson, Ran Hassin, Tom Baker, Allison Bajger, Galen Treuer
"Can Consumers Make Affordable Care Affordable? The Value of Choice Architecture"