We talked earlier in the chapter about ways that sellers of high-quality goods can convince buyers of this fact. One involved offering warranties. The notion was that, since low-quality products are especially expensive to warranty (because they break more often), the very fact that a seller offers a warranty suggests that he or she is likely selling a high-quality product.

It turns out this basic idea applies across many economic settings, enough, in fact, to have its own name: signaling. Signaling lets one party in a transaction communicate to another party something that is otherwise not immediately observable. It’s one way in which asymmetric information problems can be solved in markets.

Signaling was first formalized by economist Michael Spence in the early 1970s and he won the Nobel Prize for it.10 The basic idea of signaling is that there is some action, the “signal,” that conveys otherwise unknown information about the signal sender. For the signal to be meaningful—that is, for the receiver to actually be able to learn something about the signal sender—the net economic benefit of sending the signal must vary across different “types” of senders. That is, sending the signal must be less costly for some types of senders than others. In the warranty example, sending the signal of providing a warranty is less costly for producers of high-quality products than for producers of low-quality products. This difference makes it more likely that high-quality products will have a warranty, allowing consumers to infer something about a product’s quality, which would be otherwise unobservable at the time of purchase.

Signaling can have surprisingly powerful effects. Even if a signal is inherently worthless—that is, if the action itself has no economic value of its own other than as a signal—the information it conveys can make a big difference for equilibrium prices and quantities. To see how this can happen, consider the classic signaling example: the correlation between more schooling and higher wages. To be concrete, we focus specifically on the well-documented fact that college graduates earn more, on average, than workers with only high school diplomas.

640

Suppose that going to college does absolutely nothing to improve your productivity on a job. (This is not what labor economists who have studied the issue have found, but it will show how powerful signaling can be.) Let’s assume that productivity is a function of personal organization, perseverance, and the ability to concentrate and learn new things quickly. Employers would like to find workers with these qualities. They want to hire productive workers and pay them more. But, the problem of asymmetric information gets in the way. Employers cannot tell from an interview and résumé if a potential employee has these characteristics.

Highly productive workers would like to convince potential employers they would be good hires. But, applicants can’t just say, “Hey, I’m very productive.” That’s cheap talk. Less productive workers will say the same thing. Productive workers need some sort of “expensive talk” instead.

That’s what a college degree offers. Why? College is costly, and, importantly, it’s more costly for some than others. It’s costly in a monetary sense, but also in the sense that it takes a lot of effort. To be able to expend the necessary effort, a student has to have certain qualities, like being organized, a willingness to persevere, and the ability to concentrate. The very same attributes that make a worker productive also make it easier for him to finish college. College is hard, but people without those attributes find it extremely hard. If you are a productive worker, therefore, a good way to demonstrate this to employers is by completing a college degree. You have made a costly choice that indicates something about your qualities that would otherwise be difficult to observe. That is, you have sent a signal. Here, the signal is aimed at employers with the intent of demonstrating your productivity.

The signal in this example won’t be meaningful if unproductive workers also receive college degrees. Employers can’t distinguish between high- and low-productivity workers in that case. This is why it’s important that the signal is more costly for low-productivity workers. If the gap is large enough (we’ll describe in a minute how large that gap has to be), low-productivity workers won’t obtain college degrees.

When more productive workers finish college and less productive workers don’t, employers can tell the difference between them. These employers are willing to pay the college graduates more, because they are more productive. We end up with the empirical fact we talked about earlier: College graduates earn more pay, even if they didn’t learn a thing in school.

Signaling: A Mathematical Approach Let’s use some specific numbers to make this result clearer. Suppose there are two types of workers, high-productivity and low-productivity. Each year of higher education costs high-productivity workers the equivalent of $25,000. This includes the monetary cost equivalent of going to class, finishing assignments, studying for exams, and so on. For low-productivity workers, each year of higher education costs the equivalent of $50,000 because of the higher cost of effort they face when attending school. Therefore, we can write the total costs of going to college as

C_H = $25,000y

C_L = $50,000y

where y is years in college, and C_H and C_L are the costs of attending school for high- and low-productivity workers, respectively.

Let’s suppose that, over the course of a worker’s lifetime, a high-productivity worker produces $250,000 worth of value for employers, while a low-productivity worker produces $125,000 of value. Because of the production of this extra value, employers are willing to pay high-productivity workers up to $125,000 more in wages.11

641

To be willing to pay high-productivity workers more, however, employers need to be able to tell them apart from low-productivity workers. Suppose for a moment that employers use completion of a college degree as a way to do that. That is, employers pay workers for whom y ≥ 4 extra wages to the tune of $125,000 over the course of their lifetime. For employers to use this payment strategy, only high-productivity workers better be finishing college. Otherwise, employers would face the prospect of paying extra wages to workers who weren’t creating the extra value to justify the higher wage.

Do only high-productivity workers finish college? Compare the cost and benefit of completing four years of college for each type of worker. For high-productivity workers, finishing college—that is, receiving at least y = 4 years of higher education—costs $100,000:

C_H = $25,000y = $25,000 × 4 = $100,000

Low-productivity workers face a larger cost of finishing college, $200,000:

C_L = $50,000y = $50,000 × 4 = $200,000

The benefit to each worker of finishing college is actually the same. Employers pay a total premium of B = $125,000 in wages for completion of a college degree. Remember, because employers can’t tell directly how productive a worker is, they are trying to rely on college completion to make the distinction.

Therefore, the net benefit (NB) to completing college for high-productivity workers is

NB_H = B – C_H = $125,000 – $100,000 = $25,000

while for low-productivity workers, it is

NB_L = B – C_L = $125,000 – $200,000 = –$75,000

High-productivity workers are $25,000 better off if they finish college, so they do. Low-productivity workers are actually worse off because the cost of finishing is higher for them. They will not choose college.

Signaling: A Graphical Approach With this example, we’ve shown that the employers’ strategy makes sense. They pay a wage premium to college graduates on the expectation that they will be high-productivity workers, even though the employers can’t observe productivity directly. Their expectation turns out to be correct because low-productivity workers aren’t willing to pay the extra costs that completing college requires of them.

This outcome can be seen in Figure 16.3, which shows workers’ costs and benefits of college as a function of the number of years in college. Low-productivity workers’ costs are shown by curve C_L; they increase $50,000 for every year of schooling. High-productivity workers’ costs are shown by curve C_H. They are lower because these workers only pay a total cost of $25,000 per year. The wage premium (benefit) of college to both types of workers is shown by curve B. Workers with fewer than four years of college receive no wage premium. Those with four or more years, however, earn a premium of $125,000, which is why the wage premium jumps from 0 to $125,000 at y = 4.

For low-productivity workers, no amount of college creates a positive net benefit. Going for fewer than four years offers no wage premium but imposes costs. Attendance for exactly four years results in a wage premium of $125,000, but this isn’t worth the $200,000 cost. And, more than four years only raises costs further without any additional wage premium. For high-productivity workers, however, four years of college does make sense. The four years costs them $100,000, but earns them a wage premium of $125,000, leading to a $25,000 net benefit. In this example, in fact, high-productivity workers don’t just have an incentive to finish college, they have an incentive to finish it in only four years, as more schooling creates extra costs without the benefit of extra pay.

642

This example demonstrates the potential power of signaling. Here, something that has no real benefit to society (because we assumed education does not enhance worker productivity) actually determines how much income each worker earns. The productive workers pay to go to college, but society gets nothing economically useful in return. A college degree doesn’t make these workers more productive, because they were already more productive before attending college. Yet, using a degree as a signal allows productive workers to indicate their productivity to employers, so these workers are paid more. However, four years of college is a rather expensive signal! If these workers could somehow find a cheaper way to signal their greater productivity, society would be better off.

Economists sometimes use “money burning” as a shorthand phrase to refer to the wasteful expenditure of resources for the purposes of sending a signal. In fact, literally burning money can itself be a signal in some cases. Suppose wealth were imperfectly observable (perhaps it’s difficult to show people all your assets at once or to convince them of their value), and that a person wished to signal his wealth to another. He could do so by starting a cash bonfire. While burning money is obviously costly, it is less costly—relatively speaking—for wealthier people. Therefore, wealthy individuals could show that they are rich by demonstrating that they literally have money to burn.

Education and Productivity As we said, economists have found a lot of evidence that, in reality, college (and more years of education in general) does have positive effects on actual on-the-job productivity. Students aren’t wasting all their tuition and effort just to prove that they will make good employees. There is some evidence, however, that part of the wage differences between workers with different education levels comes from signaling effects. For example, evidence exists of what is called the “sheepskin effect”—students with four years of college and a degree to show for it earn higher wages than students who have also finished four years of college but who did not receive a degree because they were just short of some graduation requirement. If time in school is what confers skills that are useful in the labor force, and employers pay workers based on which of those skills they observe in workers, then whether a student actually holds a degree shouldn’t affect his pay. The fact that having the degree does affect a worker’s pay suggests that signaling is at work; employers are, in part, relying on the existence of the degree itself to determine how much to pay workers.

643

This sheepskin effect has shaped the recent debate about college costs and whether the actual productivity benefits of higher education are overstated. After all, if tuition levels double not because college has actually made graduates more productive but instead because students are trying harder to signal their abilities by going to highly regarded (and more expensive) colleges, this situation is like an arms race among students, that is, a situation where competitors just try to stay ahead of their competition without any real improvement in rank. (The term is often used in military discussions as countries manufacture or purchase more and more weapons to try to stay ahead of one another.) In this example, students are spending large amounts of resources without any real change in their job market prospects because they’re sending the same signal as before, but it’s now more expensive. In addition, the increased spending does not provide any additional benefits to society in the form of more productive workers.

The evidence also suggests, however, that signaling effects are largest right after a worker is hired and quickly disappear thereafter. This isn’t surprising; after employees work for a while, firms can directly observe their actual productivities. The employees’ wages respond in turn. More productive workers—whether they have finished college or not—end up being paid more. Less productive workers—again, college graduates or not—earn lower pay. Because the employees’ performances have directly revealed information about their productivities, the signal is no longer necessary.

To sum up, there are two reasons why schooling is important and increases wages: (1) because it raises productivity and (2) because it works as a signal that the person is more productive. While both matter in the labor market, the evidence suggests that Reason 1 plays a bigger role in explaining the relationship between income and schooling than does Reason 2.

Signaling is present in all sorts of economic situations beyond our example. As we mentioned earlier, for instance, warranties signal that a good is of high quality.

Buying an engagement ring for a fiancée can signal one’s commitment to getting married. If, as sometimes happens, the woman keeps the ring should the wedding be called off, only men expecting the wedding to occur will be willing to pay for the ring. The ring, as with attending college and offering warranties, allows men to substitute “expensive talk” regarding their (imperfectly observed) true feelings for what would otherwise be cheap talk.

The choices people make about the products they use or the ways in which they live their lives can be signals. They may want to inform family, friends, neighbors, or even strangers about some aspect of their personalities that would otherwise be difficult to convey. Some monks take vows of poverty. This, in part, signals their devotion. Dressing up to go to work can be a signal of one’s commitment to and seriousness about a job. Consumption patterns can be used to signal income or wealth to one’s social network.

These examples only scratch the surface of the wide span of signaling that occurs in our society. They also indicate both the frequency of asymmetric information in economic interactions as well as the power signaling can have to reduce it.

644

645