By Richard Zeckhauser
Insurance plays a central role in the functioning of modern economies. Life insurance offers protection against the economic impact of an untimely death; health insurance covers the sometimes extraordinary costs of medical care; and bank deposits are insured by the federal government (see financial regulation). In each case, the insured pays a small premium in order to receive benefits should an unlikely but high-cost event occur.
Insurance issues, traditionally a stodgy domain, have become subjects for intense debate and concern in recent years. How to provide health insurance for the significant portion of Americans not now covered is a central political issue. Some states, attempting to hold back the tide of higher costs, have placed severe limits on auto insurance rates and have even sought refunds from insurers. And ways to cover losses from terrorism have become a major issue. Temporarily, in response to the massive losses of 9/11, the federal government adopted a heavily subsidized three-year program for reinsuring terror-related building losses. (The program was extended.) In theory, the government can recoup some losses after the fact by levying a surcharge on the premiums of surviving firms.
An understanding of insurance must begin with the concept of risk—that is, the variation in possible outcomes of a situation. A’s shipment of goods to Europe might arrive safely or be lost in transit. B may incur zero medical expenses in a good year, but if she is struck by a car they could be upward of $100,000. We cannot eliminate risk from life, even at extraordinary expense. Paying extra for double-hulled tankers still leaves oil spills possible. The only way to eliminate auto-related injuries is to eliminate automobiles.
Thus, the effective response to risk combines two elements: efforts or expenditures to lessen the risk, and the purchase of insurance against whatever risk remains. Consider A’s shipment of, say, $1 million in goods. If the chance of loss on each trip is 3 percent, the loss will be $30,000 (3 percent of $1 million), on average. Let us assume that A can ship by a more costly method and cut the risk by one percentage point, thus saving $10,000, on average. If the additional cost of this shipping method is less than $10,000, it is a worthwhile expenditure. But if cutting risk by a further percentage point will cost $15,000, it sacrifices resources.
To deal with the remaining 2 percent risk of losing $1 million, A should think about insurance. To cover administrative costs, the insurer might charge $25,000 for a risk that will incur average losses of no more than $20,000. From A’s standpoint, however, the insurance may be worthwhile because it is a comparatively inexpensive way to deal with the potential loss of $1 million. Note the important economic role of such insurance: without it, A might not be willing to risk shipping goods in the first place.
In exchange for a premium, the insurer will pay a claim should a specified contingency—such as death, medical bills, or, in this instance, shipment loss—arise. The insurer—whether a corporation with diversified ownership or a mutual company made up of the insureds themselves—is able to offer such protection against financial loss by pooling the risks from a large group of similarly situated individuals or firms. The laws of probability ensure that only a tiny fraction of these insured shipments will be lost, or only a small fraction of the insured population will face expensive hospitalization in a year. If, for example, each of 100,000 individuals independently faces a 1 percent risk in a year, on average, 1,000 will have losses. If each of the 100,000 people paid a premium of $1,000, the insurance company would have collected a total of $100 million. Leaving aside administrative costs, this is enough to pay $100,000 to anyone who had a loss. But what would happen if 1,100 people had losses? The answer, fortunately, is that such an outcome is exceptionally unlikely. Insurance works through the magic of the law of large numbers. This law assures that when a large number of people face a low-probability event, the proportion experiencing the event will be close to the expected proportion. For instance, with a pool of 100,000 people who each face a 1 percent risk, the law of large numbers says that 1,100 people or more will have losses only one time in one thousand.
In many cases, however, the risks to different individuals are not independent. In a hurricane, airplane crash, or epidemic, many may suffer at the same time. Insurance companies spread such risks not only across individuals, but also across good years and bad, building up reserves in the good years to deal with heavier claims in bad ones. For further protection, they also diversify across lines, selling both health and homeowners’ insurance, for example.
The risks normally insured are unintentional, either due to the actions of nature or the inadvertent consequences of human activity. Terrorism creates a new model for insurance for three reasons: (1) The losses are man-made and intentional. (2) Massive numbers of people and structures could be harmed. (Theft losses fall in the first category, but not in the second.) (3) Historical experience does not provide a yardstick for assessing likely risk levels. Nuclear war presented equivalent challenges in the twentieth century. Had there been a significant nuclear war, insurance companies simply would not have paid. The losses would have been too massive to pay out of assets, and many of the assets underlying the insurance would have been destroyed. In time, appropriate insurance arrangements for this new category of massive risk will be developed.
The Identity and Behavior of the Insured
An economist views insurance as being like most other commodities. It obeys the laws of supply and demand, for example. However, it is unlike many other commodities in one important respect: the cost of providing insurance depends on the identity of the purchaser. A year of health insurance for an eighty-year-old costs more to provide than one for a fifty-year-old. It costs more to provide auto insurance to teenagers than to middle-aged people. If a company mistakenly sells health policies to old folks at a price appropriate for young folks, it will assuredly lose money, just as a restaurant will lose if it sells twenty-dollar steak dinners for ten dollars. The restaurant would lure lots of steak eaters. So, too, would the insurance company attract large numbers of older clients. Because of the differential cost of providing coverage, and because customers search for their lowest price, insurance companies go to great pains to set different premiums for different groups, depending on the risks each will impose.
Recognizing that the identity of the purchaser affects the cost of insurance, insurers must be careful to whom they offer insurance at a particular price. Those high-risk individuals whose knowledge of their risk is better than that of the insurers will step forth to purchase, knowing that they are getting a good deal. This is a process called adverse selection, which means that the mix of purchasers will be adverse to the insurer.
What leads to this adverse selection is asymmetric information: potential purchasers have more information than the sellers. The potential purchasers have “hidden” information that relates to their particular risk, and those whose information is unfavorable are thus most likely to purchase. For example, if an insurer determined that 1 percent of fifty-year-olds would die in a year, it might establish a premium of $12 per $1,000 of coverage—$10 to cover claims and $2 to cover administrative costs. The insurer might naively expect to break even. However, insureds who ate poorly or who engaged in high-risk professions or whose parents had died young might have an annual risk of mortality of 3 percent. They would be most likely to purchase insurance. Health fanatics, by contrast, might forgo life insurance because for them it is a bad deal. Through adverse selection, the insurer could end up with a group whose expected costs were, say, $20 per $1,000 rather than the $10 per $1,000 for the population as a whole; at a $12 price, the insurer would lose money.
The traditional approach to the adverse selection problem is to inspect each potential insured. Individuals taking out substantial life insurance must submit to a medical exam. Fire insurance might be granted only after a check of the alarm and sprinkler systems. But no matter how careful the inspection, some information will remain hidden, and a disproportionately high number of those choosing to insure will be high risk. Therefore, insurers routinely set high rates to cope with adverse selection. Alas, such high rates discourage ordinary-risk buyers from buying insurance.
Though this problem of adverse selection is best known in insurance problems, it applies broadly across economics. Thus, a company that “insures” its salesmen by offering a relatively high salary compared with commission will end up with many salesmen who are not confident of their abilities. Colleges that insure their students by offering many pass-fail courses can expect weaker students to enroll.
Moral Hazard or Hidden Action
Once insured, an individual has less incentive to avoid the risk of a bad outcome. A person with automobile collision insurance, for example, is more likely to venture forth on an icy night. Federal pension insurance induces companies to underfund (see pensions) and weakens the incentives for their employees to complain. Federally subsidized flood insurance encourages citizens to build homes on floodplains. Insurers use the term “moral hazard” to describe this phenomenon. It means, simply, that insured people undertake actions they would otherwise avoid. Stated in less judgmental language, people respond to incentives. In the above salesman example, not only are low-quality salesmen enticed to join, but all salesmen, even those of high quality, are given an incentive to be less productive.
Ideally, the insurer would like to be able to monitor the insured’s behavior and take appropriate action. Flood insurance might not be sold to new residents of a floodplain. Collision insurance might not pay off if it can be proven that the policyholder had been drinking or had otherwise engaged in reckless behavior. But given the difficulty of monitoring many actions, insurers accept that once policies are issued, behavior will change adversely, and more claims will be made.
The moral hazard problem is often encountered in areas that, at first glance, do not seem associated with traditional insurance. Products covered under optional warranties tend to get abused, as do autos that are leased with service contracts.
The same insurance policy will have different costs for serving individuals whose behavior or underlying characteristics may differ. Because these cost differences influence pricing, some people see an equity dimension to insurance. Some think, for example, that urban drivers should not pay much more than rural drivers to protect themselves from auto liability, even though urban driving is riskier. But if prices are not allowed to vary in relation to risk, insurers will seek to avoid various classes of customers altogether, and availability will be restricted. When sellers of health insurance are not allowed to find out if potential clients are HIV-positive, for example, insurance companies often respond by refusing to insure, say, never-married men over age forty.
Equity issues in insurance are addressed in a variety of ways in the real world. Most employers cross-subsidize health insurance, providing the same coverage at the same price to older, higher-risk workers and younger, lower-risk ones. Sometimes the government provides the “insurance” itself, although the federal government’s Medicare and Social Security programs are really a combined tax and subsidy scheme—one that gives a bigger benefit to those who live longer. The government’s decision not to tax employer-provided health insurance as income acts like a subsidy. In pursuit of equity, governments may set insurance rates, as many states do with auto insurance. The traditional public-interest argument for government rate regulation is that it serves to control a monopoly. But this argument fails with auto insurance: in most regulated insurance markets, there are dozens of competing insurers. Insurance rates are regulated to help some groups—usually those imposing high risks—at the expense of others. The Massachusetts auto insurance market provides an example. High-cost drivers are subsidized at the expense of all other drivers. Thus, inexperienced, occasional drivers in Massachusetts paid, on average, $1,967 for insurance in 2004 compared with $1,114 for experienced drivers. In contrast, in neighboring Connecticut, where such cross-subsidies were not imposed, the respective rates are $3,518 and $845.
Such practices raise a new class of equity issues. Should the government force people who live quiet, low-risk lives to subsidize the high-risk fringe? Most people’s response to this question depends on whether they think people can control risks. Because most of us think we should not encourage people to engage in behavior that is costly to the system, we conclude, for example, that nonsmokers should not have to pay for smokers. The question becomes more complex when it comes to health care premiums for, say, gay men or recovering alcoholics, whose health care costs are likely to be greater than average. Moral judgments inevitably creep into such discussions. And sometimes the facts lead to disquieting considerations. Smokers, for example, tend to die early, reducing expected costs for Social Security. Should they, therefore, pay lower Social Security taxes? Black men have shorter lives than white men. Should black men pay lower Social Security taxes?
Government’s Role in Insurance
Government plays four major roles with insurance: (1) Government writes it directly, as with Social Security, terrorism reinsurance, and pension guarantees—via the Pension Benefit Guaranty Corporation (PBGC)—should a corporation fail. (2) Government subsidizes insurance: quite explicitly in some programs, such as federal flood insurance, but only de facto in other cases (e.g., the PBGC has a large projected deficit). (3) Government mandates a residual market for high risks (e.g., Florida’s program for hurricanes or many states’ programs for high-risk drivers). Governments hold down prices in such markets either by creating a state fund to cover losses or by requiring insurers who participate in the voluntary market to pick up a certain portion of this high-risk market. (4) Government regulates matters such as premiums, insurance company solvency (to make sure that insureds get paid), and permissible criteria for pricing insurance (e.g., for auto insurance, race and ethnicity are banned everywhere; Michigan bans geographic designations smaller than a city).
Property liability insurance is regulated at the state level, providing many opportunities to compare the efficacy of alternative approaches. The three main regulatory approaches to pricing have been: (1) prior approval (regulators must approve rates before they go in effect); (2) use and file (companies set rates, but regulators can disallow them subsequently if they are found excessive); and (3) open competition (a market-based system in which rates are deemed not excessive as long as there is competition). Empirical studies conflict as to whether regulation leads to lower prices.
Government participates far more in insurance markets than in typical markets. The two great dangers with government participation in insurance arise when, as is common, the goals for participation remain vague (e.g., promoting the insured activity, redistributing income, or spreading risk effectively), or when its expected cost is not recognized in budgets. With insurance, as with all government endeavors, the citizenry deserves to know both the rationale and the cost.
The traditional role of insurance remains the essential one recognized centuries ago: that of spreading risk among similarly situated individuals. Insurance works most effectively when losses are not under the control of individuals (thus avoiding moral hazard) and when the losses are readily determined (lest significant transactions costs associated with lawsuits become a burden).
Individuals and firms insure against their most major risks—high health costs, the inability to pay depositors—which often are politically salient issues as well. Not surprisingly, government participation—as a setter of rates, as a subsidizer, and as a direct provider of insurance services—has become a major feature in insurance markets. Its highly subsidized terrorism reinsurance provides a dramatic example. Political forces may sometimes triumph over sound insurance principles, but such victories are Pyrrhic. In a sound market, we must recognize that with insurance, as with bread and steel, the cost of providing it must be paid.