Clay Shirky on Mega-Universities and Scale
The following is a guest post by Clay Shirky that grew out of a conversation he and Phil had about IPEDS enrollment data. Most of the graphs are provided by Phil.
Even now, it’s not clear what effect the internet is going to have on college in the United States. At the beginning of this decade, higher education looked as if it might become winner-take-all, like Facebook in social networking or Amazon in ecommerce. Following the launch of Sebastian Thrun and Peter Norvig’s online AI class in 2011, people began to predict the collapse of traditional schools and the rise of a few enormous universities:
Though predictions of imminent, extensive restructuring were kicked off by these massive online courses, recently the conversation has turned to “mega-universities”, universities operating at disproportionate scale:
Were half a dozen institutions to dominate the online learning landscape with no end to their expansion, or shift what Americans seek in a college degree, that would indeed be one of the greatest transformations in the history of American higher education. The available data, however, casts doubt on that idea.
Though much of the conversation around mega-universities is speculative, we already know what a mega-university actually looks like, one much larger than any university today. It looks like the University of Phoenix, or rather it looked like Phoenix at the beginning of this decade, when it had 470,000 students, the majority of whom took some or all of their classes online. Phoenix back then was six times the size of the next-largest school, Kaplan, with 78,000 students, and nearly five times the size of any university operating today.
From that high-water mark, Phoenix has lost an average of 40,000 students every year of this decade. You can see how dramatic the Phoenix collapse has been in Phil Hill’s chart of the 30 schools with the largest online enrollments from 2012-2017:
For mega-universities, the big story of this decade isn’t the rise of Western Governors and Southern New Hampshire. The big story is the collapse of Phoenix, the largest university the country has ever seen, by a considerable margin. In just the five years shown in Hill’s chart, while WGU and SNHU added 130,000 students between them, Phoenix alone lost 145,000. Much non-profit enrollment has been at the expense of for-profit contraction or collapse, and no matter what happens to Phoenix in the next 5 years, it does not have another 145,000 students to lose.
The mega-est university in U.S. history, one genuinely operating at Walmart scale relative to its peers, peaked nearly a decade ago and has since imploded. This complicates talk of future mega-universities, both because the original mega-university failed so rapidly, and because the reversal of Phoenix’s fortunes turns out to be a general pattern. Expanding online enrollments at state schools like UT Arlington, Central Florida, and UMUC/Maryland have been largely offset by losses by for-profits, like Ashford, DeVry, and Corinthian. Among the top 200 colleges with online enrollment in the last five years, these two forces have come close to cancelling each other out. Many individual schools grew or shrank sharply, but overall, online education among the top 200 grew approximately 2.3% per year. Not nothing, but not the kind of environment that will produce an educational Amazon either, especially since the top 30 actually shrank by 0.2% a year.
The assumption that there is “no end to [mega-university] expansion in sight” is also complicated by the tradeoff between size and growth. Here is Hill’s scatterplot of the top 200 schools by enrollment:
The X-axis is school size, from the smallest of the 200 to the largest (Phoenix, just over 100,000) for number of students taking at least one online course. The Y-axis is growth from 2016 to 2017, from significant shrinkage of online headcount (e.g. Drexel) to fastest growth (e.g. Kennesaw State) The trendline is in gray.
That trendline shows an inverse correlation between size and growth: the trend among small schools was positive, but the trend among large schools was negative, driven by the collapse of the for-profits. Of the 8 schools with 50,000 or more students, just three — Grand Canyon, SNHU and WGU — grew more than 10% last year. (The secret seems to be “Advertise on television.”) Meanwhile, of the 8 schools with over 50% growth, only one had over 10,000 students. If these patterns persist, online enrollment will see reversion to the mean, where growth slows with scale.
The largest schools are also not sorting themselves into some distinct group coming to “dominate the online education landscape.” When you graph colleges by online enrollment, the surprise is how smooth the curve is:
The blue line, school sizes in rank order, is very close to a power law distribution, the pattern underlying 80/20 dynamics. The best-fit power law distribution is shown here in red. Unlike the days when Phoenix was 600% larger than its nearest competitor, there is no significant discontinuity across the current distribution of institutional size; the largest gap in enrollment between any two adjacent schools is just 30%.
Since the founding of William and Mary, the country’s second college, higher education in the U.S. hasn’t been a winner-take-all market, and it isn’t one today. We are not entering a world where the largest university operates at outsized scale, we’re leaving that world; Phoenix was far more dominant in 2011 than any school will be in 2021. There will always be a biggest school, and it will offer mainly or solely online education, but as online instruction and marketing become more widespread, constraints on mega-universities’ growth are likely to prevent any future school from achieving Phoenix’s disproportionate scale.
At the beginning of the decade, most online students were enrolled in fully online degrees at for-profits. The catalyst for this for-profit dominance was regulatory change, first in 1998, allowing for-profits to get up to 90% of their revenue from Title IV funding, followed, in 2006, by a law that removed limits on Title IV funds for fully online degrees. For-profit enrollments doubled in the five years after 2006.
That growth has now been reversed, and as for-profits shrink or vanish, students mixing online and in-person classes have grown to outnumber students pursuing fully online degrees. This ‘mixed online and off’ trend clearly favors schools with urban and branch campuses, characteristics most common to public institutions.
You can see this advantage for public institutions in this comparison by Hill, charting growth by program type in two categories.
The top chart, fully online programs, shows public and private 4-years at nearly identical growth between 2012 and 2017, with community colleges showing more modest growth (and for-profit collapsing, as noted.) The bottom chart, though, of mixed on- and offline programs, shows public 4-years dramatically outgrowing all other school types. As student preference for this mixed mode increases, successful regional recruiting by public schools threatens to limit the future growth of mega-universities. State schools have always been the majority educator of the American public, and the next decade could see them return to that role online.
Something similar could happen at community colleges. Although early studies have generally concluded those students perform worse in online than in-person classes, issues of cost and scale create an enormous incentive to close that achievement gap. If the successful investment in online student success at the California community colleges turns out to be repeatable, community colleges, like state colleges, could end up enrolling a significant number of students interested in online classes, further capping the growth of mega-universities.
A 2016 report from the Parthenon Group examining colleges at risk of closure or merger found two risk factors which travel together: schools that enroll fewer than a thousand students, and school with no online classes. The essential competition of the 2020s will not be mega-universities versus large- and medium-sized schools. It will be schools with any online enrollment versus those with none, and the schools with none tend to be small.
If schools with online offerings start offering credible, low-cost alternatives to classes at institutions like the Tennessee College of Applied Technology or the Santa Barbara Business College, it may usher in the götterdämmerung of the thousands of small schools that have no national reputation or defensible niche. Whether the closure or merger of many small schools would count as a transformation of higher education depends largely on whether you are counting institutions or students.
If the smallest thousand Title IV schools were to vanish tomorrow, those closures would affect only 20,000 students. (You read that correctly. The smallest thousand Title IV schools, mostly local trade schools, enroll an average of 20 students each.) If online competition destroyed all of the smallest schools, it would displace considerably less than 1% of the roughly 20 million enrolled. Even with a dramatic increase in college closures or mergers, the number of institutions of higher education will easily remain in the thousands, and most students, almost by definition, will not be enrolled in schools that close.
New models appear with some pioneering institution — women’s education at Mt. Holyoke, doctoral education at Johns Hopkins, community colleges at Joliet, online coursework at UIUC. Eventually the new ideas lose their power to shock, and end up being widely copied. Institutional transformation starts as heresy and ends as a section in the faculty handbook. That pattern is now repeating itself. Though online education was mostly scaled up by startups like Western Governors or the Capella system, it is now difficult to find even medium-sized schools that don’t offer education online.
What the mega-university story gets right is that online education is transforming higher education. What it gets wrong is the belief that transformation must end with consolidation around a few large-scale institutions. The decade now ending has seen a dramatic reduction in the size of the largest online school, both in total enrollment and relative to all other schools. Instead of an “Amazon vs. the rest” dynamic, online education is turning into something much more widely adopted, where the biggest schools are simply the upper end of a continuum, not so different from their competitors, and not worth treating as members of a separate category.