When I began teaching at my current university, I received precise instructions governing grades for multisection calculus classes and no guidance for grading upper level classes. The multisection classes have many sections with different instructors but share a common syllabus and a common final exam. Department policy aims to ensure that grading is uniform across all sections; students with comparable performance should receive comparable grades, even if they are in different sections. The method for approximating this uniformity of grades is simple. Final exams are graded in common, and the exam grades are submitted to an administrator. The administrator looks at historical grade distributions, consults with senior instructors, and sets a curve for the final exam. Individual instructors are then required to set a grade distribution for their sections which approximates their sections' grade distribution on the final. Thus, if 10% of my students make an A on the final, I should award A's to approximately 10% of my students (not necessarily the same 10% who scored an A on the final). Grade inflation is minimal in courses subject to this grading regimen.
Upper level courses are an entirely different game. When I began teaching here, I approached prior instructors of my assigned upper level courses and asked them what grade distributions they had assigned when they taught the course. Upper level courses typically have a computational component, a knowledge component, and a proof component. Grading proofs has a significant subjective element. It is easy to decide that an argument is wrong; incompleteness or clumsiness, however, are much more subjective. There is no universal meter stick to use to decide what grade to assign. It is, however, usually easy to partially order the students. This clump is better than that clump. What I have to decide is whether the clumps are B's and C's or A's and A-'s. So, when I teach calculus classes, I feel virtuous for keeping grade inflation so far below what we see in the humanities. When I teach upper level courses, I sometimes have grades concentrated in the nosebleed territory between B- and A. I don't know how I would distribute grades in a humanities class where so much of the grading is subjective.
What is the harm in grade inflation? The most interesting argument I have heard is an economic one: if grades are inflated, students are not pushed to migrate to majors where they hold a relative competitive advantage. Thus a student who is not gifted in, say English, may major in English and receive grades on a par with more gifted writers, rather than being pushed into a business or econ class, where they may earn a lower grade than in their English class, but outperform more of their peers. I also imagine grade inflation makes college transcripts less useful for employers in hiring recent graduates. When I write letters of recommendation for unspectacular students, I always tell the employer not only what grades the student earned in my classes but also where he ranked in the class, in order to obviate the grade inflation problems. I suppose we would need more data about the correlation between academic success and job success to understand whether grade inflation inconveniences employers.
If grade inflation is accompanied by a watering down of the course material, the harm to students is obvious. In my department, I have seen little evidence of decreasing course rigor. In fact, I believe that mathematics courses have become more rather than less rigorous during my time here.
What are the causes of grade inflation? Economists offer one of the most amusing explanations of the fact that grades in the humanities are so much higher than in mathematics, chemistry, and economics departments. They observe that the average mathematician's midcareer salary is approximately 50 % higher than that of a psychology major. They conclude that a good grade in a psych class is worth less than a good grade in a mathematics class. The good grade in psych should therefore cost less. The relevant currency is student effort. This leads to inflation of grades if psych is to retain enrollments. I like the argument because it makes math look good (or at least profitable), but I have met few students who know average salaries by major and fewer still who don't think they will be the exceptions. Hence, I don't believe the economic argument explains the reality. On the other hand, it is very believable that instructors inflate grades in order to keep enrollments up and course evaluation numbers high. When faculty have low enrollments or poor evaluations from their students, the department administration notices and seeks explanation. Consistently poor enrollment and evaluation numbers can lower chances of promotion of junior faculty and can depress salaries for tenured faculty. Strict grading policies can lead to low enrollment and lower evaluations. The resulting pressure to inflate grades is real.
Because many of the incentives to inflate grades stem from worries about administration response to the possible consequences of tougher grading, the administration can easily tighten grading standards if it chooses to do so. If the departmental administration requests that its faculty assign certain grade distributions, then enough faculty will cooperate to slash grade inflation. Departmental administrations, however, also worry about overall departmental enrollments. Hence they are unlikely to act alone unless they are over enrolled. If the university requests that all departments enact tighter grading standards, then much departmental resistance would disappear. Unfortunately, there is a significant faction of the humanities faculty that opposes tighter grading for political/philosophical reasons which I have never understood and will not, therefore, try to explain here.
Though, grade inflation is real, I am unsure of its negative consequences. It is, however, easily addressed if the university administration is willing to buck the politics of a minority of faculty opposed to rigorous grades.