When the 2010 Affordable Care Act’s insurance mandate was before the Supreme Court, Justice Antonin Scalia asked, “If a paternalistic government could force people to buy health insurance, could it also force people to buy broccoli?” Astounded that Scalia seemingly did not understand the difference between a selection market good and an ordinary good, the authors declare, “That was the moment we realized we had to write this book.”
Adverse selection / Selection markets are precarious. Einav, Finkelstein, and Fisman illustrate this with a non-insurance product: AAirpass, which American Airlines rolled out in 1981. Buyers paid $250,000 for “unlimited first-class travel for life.” Management underestimated the enthusiasm some buyers would have for flying. Some buyers arranged multiple-city trips. Some flew to international destinations and back every other day. To stem the losses, American increased the price to $1 million. This made matters worse, as many lower-cost buyers dropped out of the market while higher-cost buyers remained. The moral of the story is that “the customers who are willing to pay the most are sometimes the ones you want the least.” American held out for a while but quit selling AAirpasses in 1994.
In contrast to the AAirpass, divorce insurance failed quickly. The authors tell the story of John Logan, who offered a policy called WedLock. Einav, Finkelstein, and Fisman identify two selection problems: couples who know their marriage is doomed—a classic “adverse selection” scenario—and schemers who plan to divorce. Logan attempted to avoid those problems by including a waiting period before a claim could be filed; as the authors note, “Waiting periods are a common technique that insurers employ to deal with selection.” WedLock charged a premium of $1,900 per year, with the earliest payoff of $12,500 after four years. The payoff would rise by $2,500 per year so long as the policyholder continued to pay the $1,900 premium.
Yet, WedLock failed “within two years of its introduction” because of a lack of demand. The authors explain, “The payout was sufficiently modest that a New York Times story on Wedlock suggested that couples would be better off putting the money they would have spent on divorce insurance premiums into a savings account instead.” Interestingly, the Times assessment is incomplete. The interest rate on the savings account would have to be at least 21 percent to beat WedLock’s $12,500 payoff after four years. Neither the couples who knew their marriages were doomed nor the schemers should have been deterred by the four-year waiting period. The situation was different if a couple stayed married for 10 years. If they had put $1,900 in the bank at the beginning of a decade and every year for another nine years, they would accumulate $25,093 after 10 years at an interest rate of 5 percent. That’s closer to WedLock’s payoff of $27,500 in the event of divorce after 10 years. The interest rate on the savings account would only have to be about 7 percent to beat the insurance payoff. The Times critique of WedLock makes more sense the longer a couple remains married. But it is similar to the owner of an automobile complaining about paying auto insurance premiums because he’s never been in an accident and made a claim.
One of the authors, Fisman, tells the story of his learning that he would need costly dental work. He did not have dental insurance at the time; he planned to enroll at the next opportunity and let the insurance cover the cost. He then learned that dental plans do not pay for major work because insurers intend to avoid customers who want to enroll when they know they will run up a big bill.
In addition to storytelling, the authors introduce recent scholarship to illustrate selection problems. Economist Marika Cabral documented that workers at Alcoa Corporation who paid for a dental plan that would cover more procedures subsequently had more procedures. She specifically found that “in the month following an upgrade, total dental spending of employees who switched into the higher-cap plan was, on average, about 60 percent higher and remained elevated for half a year.” If your dental insurance does not cover expensive procedures, that is because “insurers have for the most part decided that it just isn’t worth it to offer ‘real’ insurance.”
Long lives / Annuities—financial instruments that give beneficiaries a regular payment for as long as they live—are a form of insurance. As they do with other forms of insurance, the authors begin their discussion of annuities with a human-interest story and then share the latest academic research. They tell the tale of Frenchwoman Jeanne Calment, who agreed to give up the flat she lived in, upon her death, in return for the equivalent of $500 per month in 1965—an arrangement that today we would call a reverse mortgage. Although Calment was 90 years old at the time, she got the better end of the deal, living to be 122. The authors suspect that she knew she would live a long life. Recent academic work supports that possibility.
According to Einav, Finkelstein, and Fisman, economists recommend that savers annuitize “most” of a retirement portfolio. But few savers do. The authors ask, “Why don’t more people behave the way economic theory says they should and annuitize their savings?” Blame selection. The Society of Actuaries tallies the life expectancies of buyers of annuities, while the Social Security Administration tallies the life expectancies of the general population, which represents those who don’t buy annuities. It turns out that people who buy annuities live longer than those who do not. “According to calculations by the economists Jim Poterba and Adam Solomon,” for example, “a sixty-five-year-old male annuitant had a one-year mortality rate that is half that of the sixty-five-year-old male population at large.” Individuals who somehow know they will live long lives buy annuities. It follows that “the higher survival rates of annuitants at any age translate directly into higher costs to insurance companies.” Thus, insurance companies have raised the price of annuities to offset the higher costs. The high prices are why the public refrains from buying annuities.
Recall that when insurance providers raise prices to manage the selection problem, customers that cost less to serve will exit the insurance pool and render the business unprofitable. The authors do not explicitly explain why the market for annuities survives despite the high prices. They do explain that “insurance is valuable peace of mind,” which “can keep even some of the ‘good customers’ (those with lower costs) from dropping out and help the market survive.” That explanation appears to conflict with their statement that “Almost no one buys annuities if they can help it.”
Assessing risk / In a selection market, the price adjusts to overcome the selection problem. Take the market for car insurance. Safe drivers cost less to insure than unsafe drivers. Drivers know whether they are safe or unsafe. Insurance companies seek to differentiate the safe from the unsafe, set lower prices for the former and higher prices for the latter, and thereby reduce the selection problem. They excel at assessing a driver’s risk of getting into an accident. For instance, they know that drivers with higher credit scores are less accident-prone. Insurance companies are so adept at assessing risk that Einav, Finkelstein, and Fisman state that “the detailed data and advanced pricing algorithms get rid of most selection.” Still, some selection persists, so they declare that “the price isn’t right.”
Three factors explain why collecting information and adjusting price fail to eliminate adverse selection: “technological limitations, fear of consumer backlash, or legal restrictions.” Let’s stick with the market for car insurance. It is technologically possible to monitor a driver’s activity with an “onboard tracker.” However, the technology might be unreliable. The authors cite customer reviews that give the impression that “this type of insurance can lead to higher rates even if you’re not that bad a driver.” Furthermore, consumers balk at agreeing to an onboard tracker in return for a lower premium “because it feels creepy.” Fairness is the basis of legal restrictions. “If someone is living on the wrong side of the tracks,” the authors ask, “should they be penalized with higher auto insurance rates than those offered to others with identical driving histories who live in nicer parts of town?” So, state regulators weigh in on what is fair. Even though drivers with higher credit scores are safer, regulators in California prohibit insurance companies from using that information to set premiums. The insurance companies endeavor to make the market more efficient, and increase their profits, but regulators block them on grounds of fairness.
Regulations are paradoxical. Einav, Finkelstein, and Fisman state, “Laws that aim to promote equal access in the name of fairness may exacerbate adverse selection.” According to them, “most economists” reckon that “mandates are the best and most straightforward solution to selection problems.” Then they admit that “mandates aren’t a panacea.” Requiring a minimum level of insurance does nothing to prevent high-cost customers from purchasing as much insurance as they can, which is half the selection problem. Regulators might require too much insurance simply because at some point the marginal benefit of additional insurance becomes less than the marginal cost. The authors do not claim to know the right level of insurance to require. Also, regulators face the difficult decision of how to penalize those who violate a mandate.
The fundamental tradeoff is between efficiency and equity. In the market for health insurance, the tradeoff is between reducing adverse selection and the idea that “one price is fair, in the sense that no one is being penalized for being born sick or disabled.” Einav, Finkelstein, and Fisman illustrate with numbers. They assume that “healthy” consumers cost $60 per month to insure, and they will pay at most $70 per month for a policy. “Sick” consumers, who cost $100 per month to insure, will pay at most $150 per month. The most interesting outcome is based on the assumptions that “no one can afford monthly premiums above $80” and that the government “can only afford a subsidy of $10 to one group.” If the government subsidizes the healthy and insurers set a price of $80, both the healthy and the sick sign up. An insurer will gain $70 + $10 – $60 = $20 for every healthy consumer and lose $80 – $100 = –$20 for every sick consumer. An insurer will do this “because the $20-per-customer losses from paying the medical bills for the very sick ($100) is offset by the $20-per-customer profits after paying the medical bills of the healthy ($60).” The counterintuitive result is that “the best way to help the sick is to subsidize the healthy.” The authors immediately inform the reader that this optimal policy “might not play well in the court of public opinion.” Note that the analysis omits any consideration of a distortion caused by taxing people to finance the $10 subsidies.
Conclusion / Einav, Finkelstein, and Fisman succeed in illuminating the selection problem and making insurance markets interesting. Read the book to understand the puzzling features of several insurance markets as well as financial and labor markets. Also, read the book to appreciate the tradeoffs involved in regulating insurance markets. Sounding like Thomas Sowell, they reason, “There are no right or wrong answers, only trade-offs.”
The authors reveal plans for two more books: Riskier Business and Riskiest Business. Judging from the quality of this effort, if they find the time to write the sequels, the time will be well spent.
