Will the American research university make it to its 100th birthday?

I’m sure the title makes little sense to some readers. We are talking about revered, globally unique public and private universities that have developed over the last 75 years into the most productive and diverse research institutions in the world. Their faculty lead a critical portion of the nation’s research and scholarship, and their students become innovators and leaders of society and the world’s largest economy. How can research universities not flourish in a nation that leans heavily on advances in science and technology and focuses on improving the human condition?

My goal here is to present facts that, when taken together, paint a worrisome picture of the future of our research universities. But first, I will describe where our research universities came from.

The birth of the US research university

I set the birth of US research universities to coincide with the launch of the National Science Foundation (NSF) in 1950, as recommended in the landmark report by Vannevar Bush, Science — The Endless Frontier opens in new tab/window. The NSF was created to support faculty-led research in recognition of the contributions of science and engineering to winning World War II and as a prescient investment in the future. In effect, May 10, 2025, will be the 75^th anniversary of the modern research university.

With the NSF came peer review for judging the merit of research proposals, effectively spreading research and doctoral education among institutions. Even with this distributed funding model, a specific goal was to create a number of globally leading universities. This aim is clear in a 1960 report by President Eisenhower’s Science Advisory Committee opens in new tab/window that stated, “We must hope that where there were only a handful of generally first-rate academic centers a generation ago and may be as many as fifteen or twenty today there will be thirty or forty in another fifteen years.” Implicit in their report is a call for determining who is first-rate; however, no ranking was intended as the free market of quality ideas and peer reviewed grants would suffice.

From this beginning, other agencies joined the NSF funding model, and research thrived with the universities also buoyed by tuition revenue from more students going to college. Student demand flattened in the 1970s, and the need for STEM graduates moderated in a recessionary economy. Simultaneously, the government investment shifted from physical science toward human health, and the National Institutes of Health budget grew threefold in that decade.

American research universities developed from these early years. By the 2021 accounting (a new version is due in 2025), there were 146 Carnegie R1 universities (doctoral universities with very high research intensity) out of the over 2,000 research universities in the country. The academic research enterprise is enormous; the 2022 NSF HERD survey opens in new tab/window reported that universities collectively spent $97 billion on research, of which $54 billion was federal support, which totals less than 20 percent of the government’s investment in the nation’s GERD (gross domestic expenditures on research and development). Not unexpectedly, federal funding skews toward the 20 biggest universities, which receive a third of the support.

Universities face mounting pressures

So, what is the problem? Aren’t our research universities vibrant enterprises?

Hidden under the announcements of new discoveries and the pictures of young researchers in laboratories are mounting financial pressures that threaten the system. Why?

I recently published an analysis in Issues in Science and Technology that uses data on research and undergraduate education to compare, not rank, 70 universities taken from the top of the 2024 US News & World Report rankings. My goal was to use data science to reveal the differences between research universities.

The analysis uses Principal Component Analysis (PCA), a well-established method for reducing the dimensionality of a data set by combining variables (or features) to capture its variance, essentially measuring how research universities statistically differ. Using just two PCA components captures over 60% of the variance. The most important component (PCA1) is weighted toward features that measure perceived excellence, and the second (PCA2) toward institutional size and undergraduate access by having lower tuition and fees. This biplot is reproduced from my article.

Each principal component analysis (PCA) component has values for the 18 features; the blue lines stretch from the origin (0.0) to the end point representing the value of the feature in the two components, with the origin representing the median of each feature. The features are standardized, and all have been scaled, so that a larger positive value is considered optimal; features with an inverted scale are labelled in red (e.g., a high value for “acceptance rate” feature corresponds to a low acceptance rate). The projection of individual universities onto the two components is shown by the dots with state universities in purple and private ones in green. (Each PCA component is normalized; it would be more meaningful to scale PCA2 with the ratio of the variances (25/37 = 0.68), but this squashes this component toward the horizontal axis and makes the plot more difficult to read.) Figure reproduced from Issues in Science and Technology. 41(2), p. 31.

Several observations follow from the analysis. First, public (state-supported) universities are separated from privates by the scale of their educational programs and by comparatively lower values of undergraduate metrics such as acceptance rate, student retention and graduation. Although interpreted in rankings as measures of quality, many of these differences may simply reflect the unique missions of public and private universities. Not surprisingly, the analysis suggests that enhancing access for low-income students and being the largest research enterprise (having the most doctoral students and research dollars) are not indicators of quality.

The importance of money — measured by endowment support — is clear, especially for private universities. Institutions with large endowments relative to their student body clustered around high values of quality metrics, like peer evaluations, Field Weighted Citation Indices (FWCI), low acceptance rates and high graduation rates.

Because the features in the analysis are standardized, the analysis spreads the universities creating “long tails” stretching to negative values along the PCA1 and PCA2 axes. Adding more universities to the study potentially would give the lagging institutions company, but it is not necessary to expose the underlying causes of the separation.

First, sluggish growth of federal support of research since the Global Financial Crisis of 2008-09 has not kept up with demand, while the cost of research has grown. The resulting increasing competition for federal dollars has driven up the cost of university self-supported research (which has almost doubled in the last decade) to the point that in total, universities report spending 45 cents for every federal dollar. This is up from 34 cents a decade ago and amounts to an almost $8 billion increase. Couple this with the higher salaries and lower teaching loads needed to attract and retain the very best researchers, and we may reach a point where many research universities become financially unsustainable.

Without a large endowment or very ample and sustained state support, increasing tuition or teaching more students become the only paths to sustaining the enterprise. The second strategy has been used by many R1 universities over the last two decades, where the average growth in undergraduate and graduate students has been 40% and 80%, respectively. During the same period, there has been only a 3% increase in tenure-track faculty, increasingly leaving contract faculty and graduate students to do the teaching. Fewer students are seeing faculty in the teacher-scholar model that has been a hallmark of the US system.

If these trends continue, will today’s research universities survive for another quarter century? This is especially a concern for private universities, which rely on student tuition and which must balance research support against financial aid for student access and offer high quality education to justify their sticker prices.

In the coming decade, these institutions are facing unprecedented challenges as they deal with the decline of college-age students and their changing demographics. The student body growth of the last two decades is not sustainable. And these forces are coupled with what is likely to be massive pedagogical and operational upheavals caused by AI, most likely accompanied by the expectation that the cost of education will come down with chatbots and copilots replacing staff.

One view is that we are simply experiencing an over-supply of research universities and faculty, and the system will shrink to a new equilibrium. This would not be the first time this has happened.

During the stagflation of the 1970s, Allan Cartter, former Chancellor of NYU and the foremost economist studying academic labor markets, was concerned about the downturn in the commitment to research and graduate education when he wrote opens in new tab/window, “It is urgently necessary for the federal government to identify a category of ‘national universities,’ perhaps 75 to 100 in number, and guarantee certain minimum support levels for graduate education, research and student aid.” His call for defining “national universities” led to the first ranking of entire universities opens in new tab/window to define which is “best” — the precursor of today’s ranking industry.

The need to support merit

How do we define the best? Our society is fixated on simple rankings that become easily articulated bragging rights. Alumni, pundits, prospective students and parents love this simplicity even though research universities don’t lend themselves to being reduced to a single number. Universities do need critical and transparent assessments, but these need to be more descriptive than the rankings. Although still very crude, the PCA analysis is a starting point.

A special danger of rankings in today’s environment is the resurrection of Cartter’s call for national universities. A worrisome outcome would be to give up on the free-market research system and use a list to decide who gets support, following countries like China whose “double first-class university strategy opens in new tab/window” is meant to propel 42 selected universities toward “world class,” fueled by enormous new resources. An obvious concern with this approach is that the lists become politicized, discounting objective criteria for evaluating universities, rewarding certain universities while others whither without this privileged designation. Going down this path would destroy the American system.

If the system needs to re-equilibrate, it still will require continued government support based on merit, not simply on institutional wealth or political whim, so that the very best minds and creative ideas are supported. This is the only chance for remaining the leading knowledge-based economy in the world — a precondition for our quality of life and, perhaps, our democracy.

A final note

When this essay was drafted, I never conceived of the precipitous damage that was possible by thoughtless policy changes — like last week’s retraction of federal support by mandating a cap on NIH’s reimbursement of facilities and administrative costs. This move would ravage research universities in a single stoke and would essentially throw in the towel on US leadership in science and technology. It may not be a question of research universities surviving another 25 years; all that we have been building could be undone in less than 25 days.