A common criticism of the faculty reward system is that it tends to value research over teaching. A just-released working paper from the National Bureau of Economic Research offers new evidence in support of that contention, suggesting that the number of students a professor teaches has relatively little to do with their compensation.

Disciplines with bigger class sizes do tend to offer better pay. But the highest-paid faculty members within departments tend to teach fewer undergraduates and fewer undergraduate courses than their lower-paid colleagues. The paper also suggests that changes in faculty pay over time have more to do with discipline than number of students taught, and that universities adjust to various cost pressures by increasing class size and other means.

Yet the paper asserts that universities behave “rationally” in making such decisions, and suggests that prizing research output over teaching doesn’t necessarily affect educational quality. Over all, the paper seems to dispute assertions that higher education spending — at least on instruction — is wasteful or inefficient.

The paper, by Paul N. Courant, Harold T. Shapiro Collegiate Professor of Public Policy and professor of economics and information at the University of Michigan, and Sarah Turner, university professor of economics, education and public policy at the University of Virginia, is part of a coordinated series of studies (most of which have not yet been formally released) that the economic research bureau is sponsoring focused on the topic of productivity in higher education.

The papers, collectively, aim to bring the emerging techniques of empirical economics to gauging how “productive” colleges and universities are, which the project defines as the benefits the institutions deliver to individuals and society as measured against their total costs. The studies focus, not surprisingly, on economic benefits, which tend to be easier to quantify, even though the researchers acknowledge that many of the full range of benefits higher education produces are harder to measure.

Courant and Turner’s primary purpose was to examine the extent to which research universities “deploy” faculty members efficiently for research and instruction.

The authors acknowledge that achieving efficiency in higher education is more complicated than it may be in other kinds of organizations because a chemistry professor can’t simply be moved to, say, a philosophy department if it makes economic sense. And although the production of some degrees surely costs more than others, they say, tuition is not usually dependent on one’s major.

But universities can and do move other resources between departments and schools, and between teaching and research within departments. The authors predict that because “scholarly reputation and output” at research-intensive institutions are shaped by largely by research, highly paid faculty members within a department “do relatively little teaching, on average.” And whatever teaching they do “has relatively high consumption value, either directly or as an input into research.”

That’s indeed what they find in an empirical study of compensation for tenure-track and tenured arts and sciences (not professional) faculty members in 11 disciplines at Michigan and Virginia, which are arguably representative of many other research universities. The Michigan data span from 2002-15, while the Virginia faculty salary data are from the last three years.

The authors’ first set of questions focused on department-level variation in salary and teaching loads over time between academic units. The second part of their analysis examined variations in compensation and teaching within departments.

At both Virginia and Michigan, English had the lowest student course enrollment-to-faculty ratios, at 35.2 and 30.5, respectively. Chemistry and economics, meanwhile, had some of the highest, with student course enrollments four to five times higher than in English. The central finding of the interdepartmental analysis is that overall salary levels are negatively correlated with the cost of providing a course seat across disciplines, meaning that departments that tend to pay professors more have lower per-student costs.

Regarding educational quality, the paper says that compensating for higher salaries “via larger class sizes will vary as a function of the way in which disciplines produce and share knowledge.”

So in the humanities, it’s “often the case that being able to express knowledge is inextricably bound up with the knowledge itself, in which case good pedagogy requires substantial writing (or filming, or podcast creating) with careful evaluating and editing on the part of the instructor.” In contrast, the paper says, more quantitative fields can be taught and assessed “without close interaction among the material, the student and the instructor.”

It notes that it’s possible that increasing class size may come at a cost to quality, but that “the terms of the trade-off may differ greatly by field.”

For their intradepartmental comparisons, the authors performed a regression analysis on only full professors across disciplines, in an attempt to control for various factors, including productivity. They found that an increase of salary in $10,000 per year leads to a reduction in the number of undergraduate courses taught of about 5 percent of a course per year, and a reduction in the number of undergraduate students by about 3.5 per year. In psychology, economics and chemistry, it’s about six fewer students per year. In English and sociology, it’s about 1.5.

“The results suggest that superstars whose salary is $100,000 or more than the mean teach half an undergrad course less and about 35 fewer undergraduate students” per year, reads the paper. Additionally, “coefficients for graduate students and graduate courses are positive and significant, consistent with the idea that graduate teaching has amenity value for faculty, or is part of the production of research, or, most likely, both in some combination.”

Source: NBER

Put another way, the number of undergraduates taught falls as full professor salaries increase, but the number of graduate students taught rises.

“Departments in research universities (the more so the more elite) must pay high salaries in order to employ research-productive faculty,” the authors conclude. “These faculty, in turn, contribute most to the universities’ goals (which include teaching as well as research) by following their comparative advantage and teaching less, and also teaching in ways that are complementary with research — notably graduate courses.” So the university “pays these faculty well because they are especially good at research. It makes perfect sense that they would also have relatively low teaching loads,” along with relatively high research expectations, which the study doesn’t consider directly.

John Barnshaw, director of research and public policy for the American Association of University Professors, said he agreed with the study’s general implication that academic units know how to allocate their resources. Yet he said he questioned some of the paper’s underlying assumptions — including that teaching undergraduates doesn’t contribute to a professor’s research agenda. Undergraduates tend to look at information with “fresh eyes,” he said, and often ask questions that graduate students already steeped in a discipline might not. Moreover, he said, while top research universities can maintain their reputations by offering graduate students access to esteemed professors, institutions or departments that are trying to build prestige often do so by offering undergraduates more access to such professors.

“One of the things they seem to assume is that if a faculty member is doing research in a department, that’s a good thing, and if they’re conducting instruction, it’s — they don’t necessarily use the term ‘bad,’ but there’s a kind of balance here that I want to push back on,” Barnshaw said.

Turner said in an interview that the paper considered averages, and that it’s entirely possible superstar researchers who enjoy teaching undergraduates can and will do so. More than make value judgments, she said, the paper attempted to explain how universities allocate scarce resources to fulfill their dual missions of teaching and research.

“There’s not very much opportunity to redeploy faculty across fields, and with those constraints, what we see is universities adjusting class sizes in response to difference in salaries between fields,” she said. “And those disciplines that tend to have higher salaries tend to have higher class sizes — which means that net differences in cost per student at the course level are much smaller, and in some cases reversed, relative to differences in average salaries across fields.”