The Surprising Importance of Grades

Their models said nothing about other types of information, such as grades, letters of recommendation, and other measures of “reputation.” This led economists to predict that students of varying abilities would tend to end their education at different points in the journey from high school to a Ph.D., and that the more capable students would pursue more education to differentiate themselves from less capable students

In reality, there are many factors that influence how many degrees people tack on to their résumés. But despite the limitations of the models, attempts to understand how ability is communicated through educational attainment have led to a general mathematical understanding of the way that one party conveys or signals information to another. It is a framework that can be applied to everything from online rating systems to car warranties. Now Brett Green, an assistant professor of finance at the Kellogg School of Management, and Brendan Daley, an assistant professor at Duke University, have extended this model of signaling to include something economists typically ignore: the implications of grades.

When Green talks about grades, he does not mean only the kind that students receive on their report cards. He includes many kinds of information about a person or entity that come from an independent evaluation of quality, which includes seller ratings on eBay, bond ratings handed out by Moody’s, and product reviews on CNET or Amazon.com.

Sending Signals
The traditional notion of signaling in economic theory is that, in the context of asymmetric information—which means I have information that you do not, such as knowledge about my likely value as an employee—I have to engage in costly activities to prove my value to you. Those efforts could include the pursuit of more education, extracurricular activities, or any other endeavors that cost me time and energy. The idea is that, if I am smart and capable, these things are easier for me to attain than if I am dull and incompetent—thus, they are credible signals about my abilities.

From this premise, economists who have studied signaling have concluded that, when signaling their quality to potential employers, high-quality candidates will pile on as many degrees, volunteer hours, and positions in student government as needed to distinguish themselves from the less-capable. But that is not what happens in the real world.

In the real world, grades matter, or at least they do to the extent that they accurately measure of a student’s ability. Previous research has suggested that information conveyed by grades is irrelevant, because if capable and less-capable students were perfectly sorting themselves out by the level of costly education they were able to attain, any additional “noisy” information can be safely discarded.

When grades are incorporated into the old models, their predictions change. Green and Daley’s model, which adds the independent assessments they call grades to the costly signals that can be transmitted in instances of asymmetric information, yields many explanations of everyday phenomena in markets. Its primary implication is that when grades are added to the information an employer has about a potential employee, both high- and low-quality potential employees tend to pool in the middle of the achievement field. In other words, because employers have an additional assessment of a candidate’s value (grades), the high-quality students do not have to spend as much time and energy earning graduate degrees or participating in extracurricular activities to distinguish themselves from low-quality students, They can partially rely on their superior grades to demonstrate their abilities.

The Upside of Grades
Green and Daley’s finding that capable and not-so-capable students tend to pool in the middle of a distribution of potential levels of educational attainment—that is, many more of both high and low capability will stop at a bachelor’s or master’s degree than would have been expected absent the information contained in grades—stands in sharp contrast to results derived from older models of signaling, in which the only things that were considered were costly signals like time spent in school. One implication of this finding is that when grades are absent from the information a potential employer has—as with business schools for which either school administration or the student body has decided to conceal grades on the grounds that it creates a more collegial atmosphere—students must get more education or participate in more extracurricular activities in order to signal their value to employers.

“One policy implication for schools with grade non-disclosure policies is that you’ll see people expending more resources to signal their ability than would be the case if you had grades,” says Green. Such behavior reduces students’ welfare, leads to overall inefficiency, and in general suggests that grades are a good thing because they enhance transparency in a market, he adds.

Incorporating grades into the model changes one other major counterintuitive prediction from the traditional models: that the seller’s reputation does not matter. In previous models, just as employers disregarded information about students’ grades, so too did buyers completely disregard their prior assessments of a seller’s type—in other words, their reputation. In these models, even if receivers were 99.9% certain that the sender of information—the seller—was high quality, the sender would expend the same amount of costly resources signaling as he or she would if the receivers’ initial beliefs were much less favorable. In Green and Daley’s model, having a good reputation leads to exactly the opposite of what the old models predicted.

A good example of this prediction are automobile warranties. Green points out that Hyundai offers a much better warranty than Honda because Honda has a better reputation than the upstart Hyundai, even though the latest models of both companies cars get similar reliability ratings (or grades) from reviewers like Consumer Reports. In traditional signaling theory, the expectation would be that any car company with a good record of reliability would offer the maximum warranty it could because, as a costly signal, this would let consumers know just how reliable the manufacturer believed its cars were.

In reality, Honda offers a less comprehensive warranty than Hyundai, which accords with Green and Daley’s model. Their prediction is that the “grades” of Honda’s cars, which have accumulated over the years from ratings services like Consumer Reports and Edmunds, result in Honda’s not having to expend as much capital on warranties. Consumers have already ascertained, through independent means, the quality of their cars. In other words, sellers with higher reputations need not signal as vigorously as sellers whose reputations are not as strong.

Green believes that the explosion of channels for grades on the Internet has changed how both consumers and sellers solve the problem of asymmetric information. “There’s just so much information out there on the Internet that to think that the only information you have is what the person selling you the car wants you to have is ridiculous,” says Green. “Gathering information from the Internet is cheap and easy and changes the way markets function.”