Day 9: The Demographic Cliff
A series following my new class: The Economics of Higher Education
Today we’re talking about The Demographic Cliff, or why the enrollment game we talked about on Day 7 is about to get harder.
Today’s readings
Grawe, N. D. (2021). The agile college: How institutions successfully navigate demographic changes. JHU Press. Chapters 1 and 2.
Grawe: The Agile College, Chapters 1 and 2
If you haven’t heard of the demographic cliff, well then boy have I got some unpleasant news for you. Basically, people aren’t having a lot of babies anymore. This is true in the US, true in Europe, true in South Korea, true in China, true in most places around the world.
The chart below shows the crude birth rate in the US from 1960 to 2023, or the number of live babies born per 1000 people. You can see that there was a big drop from 1960 to 1973ish, there was a slight increase from 1973ish to 1990, then a decline from 1990-1996 and then a leveling off from 1996 to 2007, and then a long decline from 2007 to 2020. There’s some evidence that birth rates have leveled off since 2020, but the recently published number for 2023 suggests that the downward trend may continue.
Falling birth rates are a problem for society for lots of reasons. The viability of social support programs like old age pensions and medical care (Social Security, Medicare, and Medicaid to pay for nursing homes) are at the top of my mind, but a shrinking population is also bad for innovation and economic growth. It’s also a problem for a higher education industry that has arranged itself to rely on ever increasing numbers of students.
“But wait,” you say, “there have been drops in birth rates before and it wasn’t a problem for higher ed. Why is this drop different?”
Remember that Education Pays 2023 report we looked at for Day 4? They have a nice chart that shows the fraction of recent high school graduates who enrolled in college from 1980 to 2020.
Higher education survived the drop in births from 1960-1973 and the resulting drop in the number of 18-year-olds from 1978-1991 because traditional-age enrollment rates started that period around 50% and then increased to around 60%. The total number of kids going to college increased, even though the relevant population size decreased. That was also a time when the returns to a college education were increasing (remember the charts we looked at on Day 4?), so there were more “non-traditional” (i.e. older than 22 years old) students enrolling also. That’s also how higher ed survived the dip in births from 1990-1996 and the resulting fewer 18-year-olds from 2008-2014, in part due to increasing enrollment rates among 18-year-olds and also increased enrollment of older adults who were out of work because of the Great Recession.
I suppose we could survive this next drop in 18-year-olds that starts this fall by increasing traditional-age enrollment rates further. There’s certainly room to improve enrollment rates among Black and Hispanic/Latino populations. But enrollment rates in those groups have been falling in recent years, the returns to college have leveled off so there’s less demand for education (and that’s before AI has been incorporated into the economy), and the zeitgeist has soured against higher ed. Depending on whatall the Trump administration does with tariffs, there could be a recession on the horizon that could increase non-traditional age enrollment, but that’s a terrible thing to hope for. I’m not holding my breath that we’re going to see increasing college enrollment over all.
Nathan Grawe has been writing about how the demographic cliff will affect demand for higher ed for several years now. His calculations are way more sophisticated than what I showed you about the overall birthrate in the US. Essentially, his calculations take account of migration (both into the US and between states) and the college-going patterns of different groups. He uses data from the Census/American Community Survey to get at populations by census tract by year, and data from the High School Longitudinal Survey of 2009 (high school first years in 2009-10, high school seniors in 2013-14) to look at college-going patterns by group. What do I mean by a group? Grawe breaks up the population by location, sex (limited by Census categories into a binary), race (again limited by Census categories: non-Hispanic White, non-Hispanic Black, Hispanic, Asian American, and Other, which includes 2 or more races), and parental education (no BA parent, one BA parent, two BA parents) to look at the likelihood that each sub-group attends a two-year institution, a regional four-year institution, a national four-year institution, or an elite four-year institution (universities and liberal arts colleges in the top 100 of their respective rankings). He uses those population estimates and college-going likelihood estimates to predict the number of students who will attend each type of higher education institute by state.
It’s important to note that all of his analysis is predicated on the assumption that migration and college-going patterns will be the same in future years as they were in past years. This is obviously a huge assumption and limitation of his model. Migration patterns are clearly changing due to the political environment. College-going patterns are changing both from the demand side (i.e. what students want to do) as well as from the supply side (i.e. college admission policies). Grawe’s estimates essentially pretend that migration and college-going patterns are stuck in time.
Anywho, what does his model predict?
Essentially Grawe predicts that for a college like mine (an “elite” school), the number of students who want to enroll will increase over the next ten years.
“Wait a second,” you say, “I thought you said that the total number of 18 year-olds is falling.” Yes, that’s true, but Grawe’s model predicts that the number of 18 year-olds born into families that are likely to send their kid to an “elite” school has increased. Basically, privileged families have been having more babies, while disprivileged families have been having fewer. Therefore demand for “elite” colleges is going up, while demand for college overall is going down.
There’s also quite a bit of regional variation in Grawe’s estimates. While the demand from privileged families is up overall, the number of 18 year-olds in privileged families is down in the Northeast and Midwest but increasing on the West Coast and Mountain West.
I was curious to make Colorado College specific predictions using Grawe’s analysis, so I downloaded his data from his website. I combined Grawe’s state-by-state predictions about numbers of “elite” college going students with IPEDS data about where recent CC first-years have come from. Colorado College tends to overperform in the Western States, drawing more students from Colorado, Wyoming, Montana, and Arizona relative to the population of 16-24 year-olds in those states. We also do well in New England and Minnesota. Grawe predicts that the elite-college-going population is going to increase in the Rocky Mountain West and Minnesota while its expected to decline in New England.
Crunching all this together, this means that demand for a Colorado College education should be fine, so long as we remain “elite.” If our reputation falls such that families think of us more like a National 4 year school instead of an “elite” college, then our demand should just hold steady. If we predict our demand based only on the size of the young people population in the states we tend to draw from, ignoring the demographic features that predict which students go to which types of colleges, then our demand will fall.
This of course assumes that our “elite” competitor schools (like those in New England) don’t change their admission patterns. If Amherst’s or Colby’s or Middlebury’s acceptance rates increase, demand for CC could fall.
This also assumes that students will continue to demand an “elite” education at the same rates and with the same demographic patterns that they did in 2014. But 2014 was a long time ago. How sure are we that college-going patterns haven’t changed? Well, there was a High School and Beyond Survey of 2022 that had gotten underway with 9th graders who were then expected to graduate from high school and go on to college in 2026 so we could actually answer this question. However, this is one of the many studies cancelled or at least delayed by the Trump Administration. What a waste.
Day 10 was an independent work day for the students, so I’ll be skipping a post for Day 10. And that wraps up Week 2 of our 3.5 week block!
To recap where we are and where we’re going: on Day 6, we discussed peer effects in higher education and why colleges care who goes to their school. On Day 7, we talked about Admissions and Enrollment, thinking about how many of which types of students a college needs to admit in order to meet both their peer group goals AND their revenue needs. Today we talked about how finding enough students may be just fine for schools like CC, or it may get more difficult in future years. On Day 8, we examined Colorado College’s financial statements, to see for ourselves how much of the money comes from tuition and room and board, and how much money comes from other sources.
Next up, we’ll get into where some of that other money comes from. Stay tuned for Day 11: Fundraising.