Volume CXXXIII, Number 13 January 25, 2002 f

Grade inflation: an economic analysis
GREGORY DeCOSTER AND JIM HORNSTEN
FACULTY CONTRIBUTORS

The 2001 Nobel Prize in Economics was awarded to three economists for their contributions to our understanding of the functioning of markets when either buyers or sellers have an informational advantage. One of the winners, Michael Spence, was specifically recognized for his insights regarding signaling, the opportunity for market participants to undertake actions that could lead others to infer a desirable attribute that cannot be objectively verified. A vivid example of signaling is student grades.

If high ability students possess a distinct advantage in the competition for grades, then attaining high grades can be a straightforward means to signal ability level to interested outside observers (prospective employers, graduate schools, etc.). However, for grades to provide a significant amount of information to outsiders, it is necessary that lower ability students not be able to feign high ability by achieving a high grade point average. Unfortunately, this is exactly what is currently happening at Bowdoin due to rampant grade inflation.
Grade inflation results in fewer distinctions between students since nearly everyone receives high grades regardless of ability (see accompanying statistics). Consequently, the information content of grades is severely degraded. About the only information communicated by grades in such an environment is that the few students whose transcripts notably lack A's are quite unlikely to be high ability. Thus, high grades are reduced to a necessary condition for serious consideration for employment, or admission to attractive graduate programs.

The information loss associated with grade inflation has a number of undesirable effects. Not only are high ability students harmed by the destruction of the signaling function of grades but employers, and others, lacking the objective information that could be provided by non-inflated grades, must turn to other strategies for identifying high ability individuals. The importance of connections is enhanced. Statistical discrimination is a more likely occurrence. (Statistical discrimination occurs when employers make hiring decisions based not on an individual's qualities, but on the statistical characteristics of the individual's gender, race, ethnic origin, athletic affiliations, etc.) In addition, graduate school admissions will be based more heavily on letters of recommendation, which emphasizes whom, as opposed to what, a student knows.

It is also probable that high ability students will recognize that there is little payoff to excelling. Their attention to academic pursuits will likely be diminished resulting in a decline in the intellectual environment at the College. These consequences of grade inflation harm us all.

Any information provided by grades in a college-wide inflationary environment is further degraded by differential grade inflation. When departments across campus employ dramatically different grading standards, course selection as much as ability determines grade point averages, and it becomes impossible to render comparative ability judgments based on grades.

That differential grade inflation is a problem at Bowdoin is clear (see accompanying statistics). Of course, it might be argued that the higher grades in some departments are due to a greater concentration of high ability students in those departments. The logic of a signaling model, however, suggests that the opposite is almost certainly the case.

Relatively weaker students most need opportunities to mask their true ability level through attainment of high grades. Hoping to enhance their GPAs, they will gravitate to departments offering the combination of high mean/low variance grading. Conversely, high ability students benefit from the opportunity to distinguish themselves, and so, on the margin, will be attracted to departments offering higher variance grading.

The predicted result is, therefore, an inverse relationship between grades and student ability across departments. Unfortunately, outsiders do not possess the information necessary to distinguish departments based on grading practices. So important information that could be conveyed, for example, by major, is not available to those who need it. Again, when the objective information grades could provide is eliminated, employers, and others, will resort to more subjective, and potentially problematic, alternative screening mechanisms.

It is worth noting that some have argued that current high mean/low variance grading practices are due to a secular increase in student quality, and thus not a problem. Even if student quality has improved, however, the need for outsiders to acquire information regarding relative ability remains. That is, employers and others may be quite pleased that the overall pool from which they select is of enhanced quality, but be assured that they still have a strong interest in identifying an individual's relative ranking in the pool. So the fact that student quality may have improved does not eliminate the important signaling function of grades.

Neither Bowdoin, nor American higher education more generally, possesses any sort of general exit evaluation of graduating seniors. Among other things, this assures that empirical verification of the preceding analysis is essentially impossible. The theoretical conclusions are, however, unambiguous. To the extent that signaling analysis is applicable, grade inflation must have highly undesirable implications.

The obvious solution to grade inflation is institution of a rigid grading curve. However, we suspect that such a proposal would be unpopular. The most feasible alternative is to increase the information content in grades by attaching benchmark information to student transcripts. Providing interested outsiders with information about grade distributions in a student's major department, as well as the College, would enhance signal quality and enable departments to offer students the benefits of high variance grading.

Interestingly, signaling theory implies that this strategy could produce grade deflation as departments attempted to attract the high ability students who most benefit from a tough grading policy. Perhaps counter-intuitively, the winners would be Bowdoin's most able and diligent students since our grades would provide the outside world with information not available regarding students at other schools.