Another information asymmetry that can cause problems in markets is moral hazard, which exists when the information gap involves the inability of one party in an economic transaction to observe another party’s behavior. That last word—“behavior”—is important. Moral hazard is about how one party acts once an economic relationship has been entered into; that is, after an insurance policy has been sold, a contract has been signed, an employee has been hired, and so on.

A stark example of a moral hazard problem is fraud. Suppose you pay a mechanic to fix a problem with your car, and he pockets the money without making any repairs, hoping that you will be unable to detect this. He’s relying on the fact that his action is unobservable. If you either literally cannot see whether the work has been done—for instance, whether he actually replaced the valves in the engine or not—or if you don’t know enough about cars to be able to tell the difference, it’s difficult to rely simply on the car’s performance to determine the quality of the work, or whether anything was done at all. A lack of improvement in the car’s performance might indicate nothing was done, but there are also genuine reasons why no performance improvement might be observed even if the mechanic did the work as specified.

Fraud aside, numerous moral hazard situations are less malicious but still potentially damaging to markets. Insurance markets again offer many such cases. We talked about adverse selection in insurance, which deals with the unobservable riskiness of people seeking insurance coverage. Moral hazard is different. Moral hazard has to do with the effect insurance coverage has on individuals’ behaviors once they have it: specifically, that they will make fewer efforts to avoid having to make claims once they are already covered. Moral hazard is therefore a concern to insurers after a policy has been purchased.

628

An example will make this clear. (It’s slightly ridiculous in order to exaggerate the point.) Suppose movie producers could buy “box office insurance” that guaranteed them a specified gross revenue for a particular film if box office receipts did not reach this level. Producers wanting more coverage—that is, a higher minimum revenue guarantee—could obtain it for a higher premium.

Such policies could be very valuable economically. Movie revenues are unpredictable. Producers might prefer to have a more certain income flow for planning purposes among many other reasons, and risky movies tend not to get made. Insurance companies could find selling such policies appealing in principle as well: If the insurers spread their risk over enough movies, they could earn enough premium revenue from the successful movies that didn’t need to cash in their policies to both pay claims on disappointing movies and make a profit.

The adverse selection problem is that producers of lower-quality movies are more likely to want to buy this type of policy. But let’s ignore that for the moment; even if adverse selection weren’t a problem, moral hazard would be. To see why, consider the decisions a producer faces once he buys the insurance. Suppose the movie hadn’t yet been shot when the policy was purchased. What incentive would the producer now have to make a film that audiences found appealing? The producer is guaranteed the insured gross revenue, no matter what. Therefore, the producer’s optimal response is to make as low-cost a film as possible: The Stuff I Found in My Pocket: Part 2. (Now in 3D!) The producer will have spent virtually nothing on the film but will have earned a guaranteed amount of revenue. Even if the insurance contract was signed after filming, the producer would still have no incentive to spend resources marketing it.

Again, this is a somewhat absurd example to make a point, but these forces operate in many markets. Figure 16.1 shows one way to think about how moral hazard works more generally. This figure plots a potential policyholder’s marginal benefit and the marginal cost of taking actions that increase the chance of a “good” outcome occurring.

What we mean by “good” outcomes depends on the particular situation. This could include a movie doing well at the box office, a driver not getting in an accident, a person not falling into ill health, and so on. Importantly, however, actions that raise the likelihood of such good outcomes are not costless to the potential policyholder. Those actions may involve financial cost, such as hiring a decent cast and crew and designing a solid marketing plan for a movie, and almost always involve effort, such as driving with care and resisting the temptation to text friends while behind the wheel, and eating well and exercising regularly. The cost in terms of money and effort of taking a bit more of these sorts of actions is shown in Figure 16.1 as marginal cost MC. We assume this marginal cost rises with the amount of actions already taken. This assumption is realistic; making initial efforts to put together a decent movie or to not drive recklessly is probably not too difficult and thus not too costly, but after having implemented the easier actions, taking further actions to improve outcomes becomes increasingly arduous and costly.

629

The benefit of such actions is the gain that the potential policyholder obtains when a good outcome occurs. Thus, the marginal benefit of the actions is the incremental increase in the good outcome (or the probability of the outcome) that taking further action creates. In Figure 16.1, these marginal benefits are shown as MB. We assume they decrease as the amount of action taken increases, because it’s likely that the actions taken to create a good outcome have diminishing returns.

In a market with no insurance, the potential policyholder would take actions up to point A*. Here, his marginal benefit of taking further action just equals his marginal cost of doing so; taking any more or less action than this level would only reduce his net benefit. Action levels less than A* would decrease his expected benefit more than it saves him in cost. Actions beyond A* would cost more than they are worth in benefit.

Now suppose he obtains insurance against bad outcomes. If the movie is a flop, he gets in a car accident, or needs medical care due to unhealthy behavior, the policy will kick in. If the policy offers full insurance—that is, the policy pays off enough so that the policyholder is just as well off as he would have been had the good outcome occurred instead of the bad one—then there is no marginal benefit to taking actions that make the good outcome more likely. The policyholder is going to be just as well off regardless of what outcome happens. In this case, the marginal benefit curve shifts to MB_FI (for full insurance) in Figure 16.1. This curve lies along the horizontal axis because with full insurance there is no marginal benefit to taking actions that make the good outcome more likely. In this case, even the simplest actions involve a marginal cost greater than their marginal benefit, so the policyholder takes no action. This outcome is terrible for the insurer, because it makes it much more probable that the insurer will have to pay a claim on the policy.

Even without full insurance, the existence of a policy that pays off in case of a bad outcome will reduce the policyholder’s marginal benefit of taking actions to make good outcomes more likely—whatever payouts the insurance company makes mean that any bad outcome won’t be as bad as it would have been without the insurance. In situations of partial insurance, the marginal benefit curve might shift to something like MB_PI in Figure 16.1. In this case, the policyholder still takes actions to raise the likelihood of a good outcome occurring, but takes fewer actions than he would have if he had no insurance. Specifically, he takes action level A*_PI, where his marginal cost equals his now lower marginal benefit of action. Therefore, bad outcomes are more likely than they would have been if he weren’t insured.

These cases exemplify the moral hazard problem in insurance: Being insured against a bad outcome actually leads the insured party to act (or not act) in ways that increase the probability of the bad outcome. If the loss associated with having a low-performing box office film is removed by insurance, for example, the policyholder has no motive to prevent a lousy box office gross from happening. Or, if you know your car repairs will be paid at least partially if you get in an accident, you may be less careful when driving. These actions leave the insurer stuck with a higher likelihood of having to pay a claim.

630

The role of information asymmetries in the moral hazard problem involves the insurer’s inability to observe and verify the actions of the policyholder. If the insurer could specify that the policyholder take certain actions and be able to observe that the policyholder follows through, moral hazard becomes less of a problem (more on this in the next section). An insurance company underwriting a movie’s box office receipts, for example, might specify certain production requirements: cast and crew members, minimum budget, running length, marketing expenditures, and so on. But as a practical matter, it is impossible to observe and verify every single action taken by a producer that affects a movie’s revenue. Thus, there will always be some moral hazard problem. Because such a large fraction of producers’ actions are unverifiable, the problem is so difficult that it would likely destroy the market for box office insurance altogether.

Wrecked markets aside, moral hazard is still an issue in existing insurance markets. Many have argued that the United States’ National Flood Insurance Program encourages homeowners to build—and sometimes rebuild—too close to water. This program is administered by the U.S. government. It covers damages to homes caused by flooding (private insurance rarely covers such losses). The program’s provisions do not do a good job of matching premiums to risk, however. Even policyholders with large prior claims can obtain coverage relatively cheaply. As you might expect, knowing one’s beach house will be fully insured in case of a storm surge doesn’t do much to discourage the construction of beach houses in vulnerable locations. There are several cases where benefits were paid out on properties that had been destroyed by storms multiple times, and then rebuilt each time in the same location. A Government Accountability Office report in 2010 calculated that 1% of the properties covered by the program had experienced repetitive losses, accounting for 25 to 30% of total claim costs.8

631

Auto insurers are always concerned about their policyholders’ unobserved driving habits. Insurance is usually priced by the period (per six-month term) rather than the intensity of driving. Sometimes rough premium adjustments for mileage are made, but they are based on the policyholders’ reported “typical” mileage numbers, not actual use. Additionally, there is no adjustment for aggressive driving, like jackrabbit starts and stops or tailgating, which are associated with an increased probability of being in an accident. Once covered, then, the driver doesn’t bear the full marginal cost he imposes on the insurance company when driving more miles or more aggressively. Drivers therefore have too little incentive to avoid actions that raise the likelihood of their being in an accident. (The standard structure of auto insurance policies may be changing, however; see the Application “Usage-Based Auto Insurance” later in this chapter.)

Unemployment insurance, while offering some financial relief to workers who have lost their jobs by partially replacing their lost wages, can also reduce unemployed individuals’ incentives to look for work. While unemployment benefits are tied to the recipient actively looking for work, the agencies that administer the program cannot fully observe recipients’ true efforts to find employment. The intensity of the job search and the individual’s effort in interviews are difficult to monitor.

Moral hazard problems aren’t restricted to insurance markets. They arise between lenders and borrowers in financial markets, too. Lenders often loan funds to borrowers to use for a particular purpose. Suppose a borrower asks for a loan to invest in new equipment for a business. Once he has the money, however, he may find expenditures on other items more appealing. Perhaps he would like some fancy antique furniture for his office. If the borrower uses the loan for the antiques and then skimps on equipment, and the lender can’t fully observe how the funds are used, there is moral hazard. By spending part of the loan on unproductive antiques and shortchanging expenditures on productive capital, the borrower reduces the probability that he will be able to repay the loan. This is bad news for the lender and it results from the lender’s inability to observe exactly how the borrower uses the loaned funds.

Discussions about policy responses to the financial crisis of 2008 have involved moral hazard concerns. Governments around the world have bailed out banks and other financial institutions in an effort to stave off a collapse of the financial system. There is concern, however, that such policies could backfire by encouraging overly risky behavior by banks in the future. This criticism is based on moral hazard arguments. Specifically, a bailout keeps institutions that played too fast and loose in the run-up to the crisis from bearing the full cost of their bad decisions. If the bailouts cause banks to expect similar rescue actions should another collapse happen in the future (perhaps because the banks perceive themselves as “too big to fail”), they may again take overly risky actions. After all, if their bets do not pay off, the banks will probably be covered anyway. This scenario, while taking place outside the insurance market, is in many ways similar to the moral hazard problem facing insurers. Here, taxpayers are the insurance company. If financial institutions know they will be bailed out by the government if things go badly, they are in essence insured against such outcomes. They therefore take fewer actions to avoid bad outcomes (such as engaging in less risky behavior) as a result.

632

Unobserved actions play an important role in labor markets, too. Employers cannot typically observe all the actions of their employees. Employees may wish to engage in work-time activities that are not in the firm’s best interest, like surfing the Internet or gossiping with co-workers. This presents employers with the difficulty of figuring out how to induce employees to do their jobs even when some of the workers’ activities cannot be monitored. These employer–employee relationships apply at all ends of the manager–worker hierarchy: from those between foreman and line workers to those between shareholders and the firm’s CEO. There is actually a special class of economic analyses—called principal–agent relationships—that deal with these sorts of issues. We discuss them in more detail later in the chapter.

Just as with adverse selection and the lemons problem, many market mechanisms have developed to reduce moral hazard. One possibility for insurance markets that we hinted at above is the insurer’s specifying certain actions that must be taken by the policyholder as a condition of coverage. This is often followed up by the insurer verifying that the actions have, in fact, been taken. For example, commercial property insurance companies may require smoke detectors and firefighting equipment to be installed and maintained in buildings they insure. These insurers may then send inspectors to verify compliance with such regulations. Many life insurance policies include an exemption from paying benefits if the policyholder commits suicide.

These approaches seek to head off moral hazard problems directly, by specifying and monitoring what the policyholder does. That is, these methods recognize that while the actions of the policyholder might not be perfectly observable, key behaviors that greatly impact the insurer’s payoff can be specified by contract (assuming that the insurer maintains the ability to monitor and verify the policyholder’s actions).

A related approach is for insurers to structure policy contracts to give policyholders incentives to take actions that reduce risk. Homeowners typically get a break on their policies if they install smoke detectors, dead-bolt locks, or modernize their electrical system. Life insurance policy rates drop for those who quit smoking. Auto insurance policies offer discounts to drivers who maintain a clean driving record. Insurers can also reduce moral hazard by structuring policies that align policyholders’ incentives with those of the insurer. This is done by giving insured individuals some “skin in the game.” That is, devices are used to directly tie together the policyholders’ and the insurer’s payoffs. There are several ways this can be done. These devices—deductibles, copayments, and co-insurance—are all common.

Deductibles are the portion of claims that the policyholder must pay out of his own pocket. A person with a $500 deductible on his auto insurance who causes an accident leading to damage of $5,000 will only obtain $4,500 from the insurer for repairs. By imposing some cost of claims directly on the policyholder, the insurer gives him an incentive to take actions that reduce the likelihood of claims. Copayments work similarly. These are payments (most commonly applied in health insurance markets) that the policyholder must dole out whenever making a claim. A $5 fee you have to pay for each prescription you obtain through a prescription drug plan, for example, is a common type of copayment. In co-insurance, the responsibility for paying claims is split between the insurer and the policyholder on a set schedule. Many traditional health insurance policies, for instance, pay 80% of the cost of services. The policyholder remains responsible for the other 20%. The purpose of each of these three devices is to give a policyholder some incentive to reduce the size or likelihood of his claims.

These and other practices reduce the impact of moral hazard on insurance markets, preserving much of the economic gains from their existence. It’s important to remember, however, that even when damped by these institutions, moral hazard can still affect the structure of the markets in which it is a factor.

633