A common manifestation of asymmetric information in markets is the lemons problem. This problem exists when the sellers know more about the quality of a good they are selling than the buyer does. The used car example in the introduction is one such case. It is, in fact, the origin of the problem’s name, because poor-quality cars are sometimes referred to as “lemons.” The issue was first formalized by the economist George Akerlof in 1970 and contributed to his winning the Nobel Prize.3

To see how damaging the lemons problem can be to markets, consider the following example. (It is overly simplistic but makes clear what the effect is, the way it works, and why it can cause such problems.) Suppose there are two types of used cars: good ones (we’ll call them “plums”) and bad ones (“lemons”). Half of used cars are plums, and the other half are lemons. Potential buyers value plums at $10,000, but lemons have no value to them. Sellers value plums at $8,000 and also have no value for lemons.

Let’s start by thinking about this market if the quality of cars were observable to both the sellers (the current owners of the cars) and the buyers. Because sellers value plums at less than buyers do ($8,000 vs. $10,000), the half of used cars that are of good quality will sell at prices between $8,000 and $10,000. (Any price in this range will work.) Both parties are better off because of the trade: Buyers value the cars more than their former owners did, and the cars are sold at a price that makes both the buyer and seller no worse off than had no trade occurred (in fact, at least one of them must be better off). Lemons have no value to either their current owners or potential buyers, so no lemons will be sold, and no one is better or worse off as a result of this outcome.

Thus, when quality is fully observable to everyone, buyers and sellers in the market are at least as well off—and many are better off—because of the transactions in the used car market. This is the way in which well-functioning markets work and raise the welfare of the markets’ participants.

Now consider what happens under asymmetric information, when sellers know if their cars are plums or lemons, but buyers do not. (It’s reasonable that car owners would know more about the quality of their own cars because they’ve experienced all of the cars’ problems.) All that buyers know is that half of used cars available for sale are plums and half are lemons. They realize that the probability that any particular car is a plum is 50%.

What is the most that a buyer will pay for a used car? Because she values a plum at $10,000 and a lemon at $0, and a car has a 50% chance of being either, the most a buyer would be willing to pay for a used car is ($10,000 × 0.50) + ($0 × 0.50) = $5,000. Anything more than that and she will be worse off in expectation: Any one car only has an expected value of $5,000.

617

But now, think about the owner of a plum deciding whether to sell it. This seller values his car at $8,000. Because buyers can’t know for sure whether or not his car is a plum, he could never get more than $5,000 by selling it. As a result, the seller won’t sell a plum car, only a lemon. But, here’s the problem: Buyers are not stupid. They recognize that the owners of good cars won’t sell when the price is $5,000, so buyers know that any used car available for sale must be a lemon. Because buyers don’t want lemons for $5,000, they are not willing to buy any used car that is actually offered for sale.

This is the tragedy of the lemons problem. We have a situation in which we know that if information were equally known by both buyers and sellers, there would be a large set of used car sales that would make people better off (because buyers value the plum cars more than their former owners did). With enough asymmetric information, however, no sales take place. There is no market. All those potential gains from exchange are destroyed because one side knows more than the other.

This example brings up two important points about asymmetric information and the lemons problem. The first is that the existence of quality differences is not, by itself, the cause of the difficulty. If there were complete information in the market and the quality of products were common knowledge, higher-quality products would have higher prices and lower-quality products would have lower prices. (Here, those prices would be between $8,000 and $10,000 for high-quality cars and $0 for lemons.) Consumers with a high valuation for quality would pay for it. But, there’s no unnecessary reduction in exchange in that case: Every buyer who is willing to pay more for a good than it is worth to the seller will be able to buy it. In other words, the market will efficiently allocate goods.

The lemons problem arises when information about quality is not known equally by the buyer and seller. This asymmetry leads to poor-quality goods being disproportionately put up for sale. That is, the average quality of items that are offered for sale is lower than the average quality of all such items, including those not for sale. This is exactly what happened in the simple example above: The average value of used cars to buyers was $5,000, but the average value of those actually offered for sale was $0. This prevents mutually beneficial trades from happening that otherwise would if everyone had full information.

When there are stronger incentives for “bad” types of a product to be involved in a transaction than “good” types of the product, we call that adverse selection (so named because the selection of types into the market tilts adversely toward those of bad quality). In our used car case, lemons are the “bad” types. They are disproportionately likely to be put up for sale because buyers can’t tell the difference between lemons and good cars. This makes the price buyers are willing to pay for a used car of uncertain quality too low to induce owners of good cars to sell. The sellers who are willing to offer their cars for sale at this price must therefore be lemon owners. (Adverse selection on the buy side of the market can also cause problems. We’ll see that below in our discussion of the insurance market.)

The second important point in this example is that information asymmetry harms not only those with less information, but also those who have more information. Both sides lose because information asymmetries keep sales from taking place that would have benefited both buyers and sellers. In our example, buyers and owners of good used cars suffer because they miss out on trades that would happen if each side knew the car’s quality. Paradoxically, rather than their information edge being an advantage for the sellers, it is a hindrance.

Could sellers get around this problem by telling potential buyers that their car is indeed a plum? The problem is that lemon sellers have an incentive to lie about the quality of their cars and try to pass them off as plums. If buyers recognize this incentive to lie, then a seller saying the car is a plum won’t be informative about its true quality. Without additional information, a buyer won’t be able to tell a plum seller (claiming a high-quality car) from a lemon seller (also claiming a high-quality car) any better than he can tell a plum from a lemon. However, there are ways that sellers of good cars can credibly convey this information to buyers, as we discuss below.

618

The specific example we ran through above was, as we said, purposely simplistic. Lemons problems usually don’t completely destroy markets. After all, we see plenty of cars purchased in the used car market in the real world. However, adverse selection can still reduce trade to a level below what is economically efficient. The vicious cycle of uncertain quality still reduces buyers’ willingness to pay, leading in turn to low-quality cars being disproportionately offered for sale, further reducing buyers’ willingness to pay and exacerbating adverse selection, and on and on. In our simple example, this feedback process was so strong that buyers were unwilling to pay for any car brought to market, since it was sure to be a lemon. But, real-world factors usually slow down the adverse-selection feedback before it completely spins the market into oblivion.

For example, in reality there is a much broader distribution of quality among used cars than the two-type distribution in our example. Furthermore, buyers and sellers don’t all have the same value for each type. These factors tend to diminish the disastrous effect of information asymmetries on markets. There will be disproportionately more “worse” types offered for sale, and the amount of trade will still be inefficiently low, but cars will be bought and sold.

Adverse selection is why, when considering buying a used car, you are likely to have the type of concerns about quality discussed in the introduction to this chapter. You recognize, for example, that the late-model Kia you are considering is likely to be of lower quality than the average late-model Kia, because owners satisfied with their current Kias are less likely to put them on the market. This holds down your willingness to pay for such cars, reinforcing in turn the reluctance of owners of good-quality Kias to sell them.

Potential lemons problems abound in the economy. We’ve discussed the most well-known example—used cars—but “lemons” can exist in any market where the quality of a product is better known by the seller than by the buyer. Used goods other than cars are obvious examples. Online sales might be especially vulnerable to lemons issues: Transactions are not face-to-face, and potential buyers typically cannot physically inspect the product beforehand.

Adverse selection is common in services and factor (input) markets, too. People seeking to remodel their homes confront the lemons problem, for example. It is difficult to judge the quality of workmanship to expect from a contractor. This concern might make homeowners reluctant to offer adequate compensation to contractors who bid on a job. But, this reluctance would, in turn, reduce the likelihood of a high-quality (but expensive) contractor taking the job, making it instead more likely that a contractor who would take a job at this lower pay level is incompetent.

In input markets, lemons problems can show up in used capital goods sales. Trucks, machine tools, and even buildings are often bought from sellers who know more about their quality than buyers. Labor markets can have their own adverse selection issues. Suppose workers differ in, and know, their “quality”—which could be their willingness to work hard, their honesty, or their knowledge of the tasks necessary for the job. If a firm could readily observe the quality type of any worker, it could pay him his marginal product (which will be higher for higher-quality workers), and everything would be fine. But, if the employer can’t easily tell good workers from bad, it will only be willing to offer a wage equal to an average of the marginal products of the different worker types. This result can make higher-quality workers less likely to apply for a job, and lower-quality workers will be adversely selected in the applicant pool.

619

Because lemons problems can destroy so much economic value by keeping beneficial exchanges from happening, many institutions in real-world markets have been set up to lessen information asymmetries, if not eliminate them altogether. We start again with the used car market, but the lessons we draw from this can be much more general. Many of the mechanisms that work to reduce adverse selection in the used car market act similarly in many other markets. We talk about a few specific cases later on.

Institutions and market mechanisms that address asymmetric information in the used car market basically work in one of three ways. One set acts to solve the asymmetric information problem directly, by giving buyers the ability to observe a car’s quality before any transaction. The second set of institutions works by punishing sellers who try to sell cars they know are lemons but still pass them off as higher quality. The third set reduces adverse selection by increasing the number of high-quality used cars brought to market. Note that the second and third sets of mechanisms don’t allow buyers to observe car qualities directly. Instead, the second set creates incentives for sellers to be truthful about quality, even if it’s still not observable before the transaction. The third set increases the average quality of used cars offered for sale by creating incentives for owners to put good-quality used cars on the market. We’ll discuss some examples of each type of mechanism in this section.

Reducing Asymmetric Information Directly There are several ways potential buyers can learn about the quality of a used car. Buyers can have a trusted mechanic who is not affiliated with the seller check out the car before agreeing to purchase it, for example. Owners can make their cars’ ownership and maintenance records available. (Remember our earlier discussion about why sellers may take the initiative to close the information gap, even though they know more about quality than buyers. Both parties lose—even the one with more information—when the lemons problem keeps mutually beneficial exchanges from happening.)

Some companies have made a successful business model out of reducing information asymmetries. Autocheck and Carfax specialize in providing information about used cars. Customers can type a car’s unique Vehicle Information Number into either company’s Web site and immediately see the ownership history of the car, and sometimes the maintenance records, too. This is valuable information for buyers, but sellers also benefit from the service. In fact, a large fraction of Autocheck and Carfax searches are paid for not by buyers, but instead by auto dealers who have bought subscriptions that allow their customers to do background searches on cars the buyers are considering.

Incentives for Truthful Quality Reporting The second set of institutions reduces the lemons problem by making it costly for sellers to be dishonest about their cars’ quality. These mechanisms can work even if they don’t actually give the buyer all the information about the car that the seller knows.

A classic example of this type of mechanism is reputation, a potential buyer’s perception of a seller. Businesses that operate honestly (by selling good-quality cars or, when selling lower-quality cars, by disclosing this and selling at an appropriately discounted price) will develop a reputation for doing so. This makes consumers more likely to do business with them. Dishonest dealers, on the other hand, will suffer when word gets out about their practices.

Reputation incentives work best when a business expects to operate for a long time, because a good reputation pays off in increased business in the future. A “fly-by-night” seller may be willing to cheat to make a quick buck today, knowing he will have moved on to something else before his loss of reputation would really matter.

Warranties can also reduce adverse selection. High-quality car sellers can communicate the value of their cars to buyers by offering to fix any problems that occur after the sale or even return the buyer’s money if he is unsatisfied. Just as with reputation, warranties don’t solve the lemons problem by getting rid of the asymmetric information. They instead give sellers an incentive to be truthful about the quality of their cars. It’s expensive—too expensive—for a seller to offer a warranty on a car he knows is a lemon. The seller has, as they say, “put his money where his mouth is.” Recognizing this, if a buyer sees that a car will be sold under warranty, he can reasonably infer that it is a higher-quality car. Cars without warranties are likely to be lower-quality. (And that’s fine, again as long as the price reflects their lower-quality levels.4)

620

Well-designed legislation and regulation can also counteract adverse selection. Most states have “lemon laws.” As the name suggests, these laws focus on conditions that must be met in auto sales. The laws essentially mandate short-term warranties for dealer-sold used cars under specified age and mileage limits. These warranty mandates help keep low-quality cars out of the market. Note that it is not necessarily efficient to keep low-quality used cars from being sold altogether. There may be people who would prefer inferior cars as long as these cars are priced appropriately. What’s important is not that low-quality cars are removed from the market, but rather that all potential buyers recognize them as such.

Increasing the Average Quality of Cars Placed on the Market The lemons problem can be lessened if something reduces adverse selection. Bringing more plums onto the market, even if the quality of individual cars is still unobservable, would achieve this. Raising the average quality of cars available increases buyers’ willingness to pay for used cars. This increase in demand further encourages owners of higher-quality used cars to sell. In the end, the amount of trade in the market rises, making both buyers and sellers better off.

Leasing programs are one way to make this happen. With a lease, a buyer or “lessee” takes ownership of a car for a set period of time, with the option to return the car to the seller (the “lessor”) at the end of the contract period. By encouraging the return of vehicles at a predetermined date regardless of their quality, leasing increases the number of higher-quality cars on the used car market and reduces adverse selection.

Beyond Used Cars These examples show that the lemons problem, while potentially very damaging to markets, also offers powerful incentives for people and firms to take actions that limit its reach. The examples discussed all involve the used car market, but it is easy to find similarly structured mechanisms working in other markets with asymmetric information.

For instance, we see similar institutions at work in our home repair services and labor market examples discussed earlier. In home repair, referral networks enable homeowners to learn more about contractors’ performance and specialties before they hire them. Angie’s List is a company that compiles grades on contractors from their previous clients. Homeowners can use these grades when considering which contractor to hire. (Such information is valuable. Over a million people have purchased memberships allowing them to view these grades.) Angie’s List and other business rating organizations like the Better Business Bureau also enhance the benefits to contractors of preserving a good reputation.

Referrals and reputations play a key role in labor markets, too. You normally list references on your résumé. Firms often check with these references and previous employers about a potential employee’s performance. Other institutions—schools, trades associations, and so on—act as third parties that certify competence at various tasks. There are even “warranties” of sorts in labor markets. Employees are often hired on a provisional basis at pay rates below their position’s typical pay, or they work for some period as a probationary employee, a status that makes it easier for the firm to break off employment. These types of arrangements let firms “try out” a worker at relatively low cost before committing to a longer-term agreement.

621

These examples are good news. Even though the lemons problem can potentially destroy markets, it has spurred the creation of institutions that moderate its impact. In some cases, these institutions are so effective that they remove the effects of information asymmetries on trading. Nevertheless, it’s important to remember that even then, the lemons problem still influences those markets. Buyers remain concerned about the quality of the used car they buy. It’s just that the buyers are better able to obtain a good sense of the car’s quality before purchase and pay the appropriate price for it.

622

In lemons-problem situations we’ve discussed so far, the seller had more information about the quality of the good or service than the buyer. But, information asymmetries where buyers know more than sellers can cause problems, too. Insurance markets are an important case in point.

In insurance markets, buyers may know more about their risk of needing to make claims on their insurance policies. Think about what insurance is: a good that compensates a policyholder when particular events occur. Depending on the type of insurance, these events could involve the policyholder getting sick, being in a car accident, having a tree fall on his house, or even death. Individuals expecting such events to be more likely will obviously assign a greater value to being insured. If you know you need expensive braces for your teeth, for example, you would greatly value dental insurance because it would help pay for your orthodontia.

But this means that buyers are adversely selected in insurance markets. From the standpoint of insurance companies, potential customers differ in “quality” because they have different likelihoods of making claims on their policies. (The expected sizes of their claims can be different, too.) Risky drivers are more likely to make claims on their auto insurance policies; unhealthy individuals are more likely to make health insurance claims.

Just as with the “seller-side” lemons problems discussed earlier, mere quality variation is not a problem in and of itself. If insurance companies could observe potential policyholders’ inherent riskiness, they could charge policy premiums commensurate with that risk. Customers likely to make more frequent or larger claims would pay higher premiums. There would be no economic efficiency loss in this world: Risky customers would pay more, but would at least be able to find coverage. The higher premiums would compensate their insurers for the higher expected payouts.

624

The adverse selection problem occurs when the customer knows more about their expected claim behavior than insurers do. Insurance companies realize that the riskiest customers will be those most likely to buy insurance, and will buy more insurance than low-risk individuals. They, after all, have the most to gain from being insured. If the insurance companies can’t tell the good risks (those customers unlikely to make many claims) from the bad, they have to charge premiums high enough to account for the large fraction of riskier customers.

This starts the vicious cycle again: The high premiums price some low-risk customers out of the insurance market. These low-risk buyers don’t want to pay high premiums for a policy they are relatively unlikely to need. That, in turn, makes the high-risk individuals an even greater share of those seeking insurance. This necessitates that the insurance company charge an even higher premium, and so on. In a worst-case scenario, that can result in the destruction of the market like the used car example discussed above.

Adverse selection on the buyer side of a market can therefore be as damaging as adverse selection on the seller side. And just as in the cases above, the party with more information can be hurt just as much as the party with less information. In particular, low-risk insurance customers could lose the ability to buy insurance if they cannot convince insurers that they are, in fact, less risky. At the same time, insurance companies suffer from adverse selection because it makes it more difficult for them to write policies for a consumer segment that they would very much prefer to insure, if they could see everyone’s true risk type. (There’s another asymmetric information problem in insurance markets when insurers cannot observe all of the actions of their policyholders: Once someone is insured against a particular event, he is less likely to take steps to prevent that event from occurring. Someone who has auto insurance may not drive as carefully, for example. This type of problem is called moral hazard. We discuss moral hazard problems more extensively later in the chapter.)

Just as we saw in the markets with seller-side lemons problems, many market institutions have arisen to mitigate adverse selection in insurance.

Group Policies One example is the writing of group insurance policies. These are general policies written for members of a specified group, often defined as employees of particular firms. In the United States, most people get their health insurance this way, for example. Why does group insurance help with adverse selection? By tying insurance to employment status, the insurance company removes much of the link between individuals’ riskiness and their insurance purchases. That is, it pools together a wider range of risk types. The good risks are added to the bad, reducing the correlation between individuals seeking to buy health insurance and their chances of falling ill. This is in many ways similar to the way in which leasing reduces the lemons problem in the used car market by breaking the tie between a car’s quality and its likelihood of being resold. (It also doesn’t hurt, from the insurance company’s perspective, that unhealthy people are less likely to be employed. This, too, reduces adverse selection.)

Screening A second adverse selection reduction tactic is screening. Insurance companies vet potential customers for the likelihood of their submitting claims by observing as many risk factors as possible. People seeking life insurance, for example, typically have to complete a medical questionnaire and submit to blood and urine tests, and sometimes a full medical exam. Having a risk factor present doesn’t necessarily mean that a potential policyholder will be denied coverage. Once his risk is known by both the buyer and seller, the customer might still be offered coverage, but he would have to pay higher premiums that match his riskiness. For example, life insurance premiums are higher for smokers, diabetics, or those with high blood pressure. Screening for risk factors isn’t constrained to direct influences on expected claims. If you have purchased auto insurance, in addition to asking questions about your driving habits and running a records check for moving violations and accidents, the insurer may have asked you to supply an academic transcript to take advantage of a good-student discount. Even though academic performance is not directly related to driving, good academic performance is historically linked in the data to lower accident rates. The insurer wants to account for as many observable risk factors as possible and adjust premiums to these factors accordingly. (Of course, all of these relationships between observable risk factors and actual policy claims hold on average, not individual-by-individual.)

626

Denying Coverage Insurers can also try to head off adverse selection directly by outright denying coverage to individuals with certain risk factors. Health insurance policies often will not cover conditions that already existed at the time the policy was purchased. If you could buy insurance after you became ill, many people would not buy it beforehand. Insurers would pay out massive claims while receiving little premium revenues. If this happened often enough, no one—healthy or unhealthy—might be able to obtain coverage.

This problem explains the economics behind the “mandate” rules in the Affordable Care Act (aka Obamacare) and the Massachusetts health care insurance law on which it was patterned. Both policies have provisions that forbid health insurers from denying anyone coverage, but also mandate that everyone must buy health insurance. While controversial, the mandates help solve what could otherwise be extreme adverse selection problems. The mandate is a lot like a government-driven form of the benefits from group insurance we discussed above. Requiring everyone to buy insurance solves the adverse selection problem by eliminating adverse selection entirely. Under a mandate, the average insurance customer must have the average insurance risk in the population. That’s why states require all drivers to get auto insurance.

Summing Up These mechanisms and others like them moderate adverse selection in insurance markets for the same reasons that the seller-side mechanisms discussed earlier arose: In the absence of such mechanisms, asymmetric information between parties in the market could destroy considerable gains from exchange.