Sunday, August 30, 2015

Why the Cost of College Has Tripled

It’s back-to-school time, so again people are talking about the rising cost of college. I wrote about this issue two years ago, and produced a plot showing how college tuition has increased faster than virtually any other component of the U.S. Consumer Price Index. Here’s an updated version of that plot, showing the relative cost of various types of goods and services compared to the overall CPI, since 1978 (the first year for which college tuition has its own CPI category):

As I said before, it’s not hard to understand the basic economics shown in this plot. Manufactured goods have become cheaper over time, as manufacturing has been automated and outsourced. The cost of professional services has therefore risen in comparison. College is often the ticket into high-paying service professions, so the demand for college and the willingness to pay for it have risen even more.

But even if we understand why people are willing to pay ever-higher tuition, this fact doesn’t tell us where all that money is going. Has the actual cost of educating a student more than tripled since 1978 and if so, how is that possible?

The answer to this question depends on whether we’re talking about public or private colleges (and universities). We can separate the two sectors, and also look 15 years farther back in time, by going to the Education Department’s Digest of Education Statistics. Here’s the Digest’s tuition data in constant (2013-14) dollars:

Obviously the private colleges charge much higher tuition than the public ones. Notice also that tuition gradually decreased, in real dollars, from the mid-1970s through the early 1980s, probably because colleges lagged in keeping up with the double-digit inflation of that era.

If you look closely at this second graph, you’ll see that since the 1970s tuition has increased slightly faster, in percentage terms, at the public schools than at the private ones. And even at the public schools the increase has been only about 200%, slightly less than what’s shown on the CPI graph. I don’t know the reason for this slight discrepancy, but the fact remains that tuition has roughly tripled over the last 35 years. Again, where is all this money going?

Let me first answer the question for the public colleges, which currently enroll 72% of all students and 69% of full-time students. Based on the data I’ve found (described below), it appears that the cost of an education at these schools has increased since the late 1970s, but only by about 20% (after accounting for inflation). However, these schools receive a great deal of their revenue from state appropriations, and that revenue, on a per-student basis, has declined by about 25%. Amazingly, the combination of these two 20-25% effects has resulted in a tuition increase of roughly 200%.

To show how this is possible, let me present a grossly simplified “toy” model that uses rounded numbers and ignores a variety of complications as well as all the little bumps and dips in the actual data:

In today’s dollars, the actual annual cost of educating a full-time student was about $10,000 back around 1980 and has increased about 20%, to about $12,000 today. Meanwhile, state funding of higher education has declined, on a per-student basis, by about 25%, from $8000 to $6000. This means that the average tuition has had to triple, from about $2000 to $6000. Simple arithmetic has combined 20% and 25% to yield 200%.

To construct this toy model I relied on the tuition data shown above, along with data from The College Board’s annual Trends in College Pricing reports. Figure 18A of the latest Trends report shows that state and local appropriations currently cover about half the cost of education at public colleges (more at two-year schools but less at four-year schools), and that this share has been decreasing in recent years. Figure 16B shows the history of state appropriations in more detail back to 1983-84, and the corresponding figure in the 2010 Trends report goes back to 1979-80. Here I’ve plotted state funding relative to its value in 1979-80, comparing the total amount to the amount per student:

The decrease in per-student funding from 1979-80 to 2013-14 was almost exactly 25%, so that’s the number I used in my toy model. But the bumps in the data (caused mostly by economic ups and downs) have been large, so you can get very different overall changes by choosing slightly different starting and ending years.

It’s important to note, meanwhile, that total state funding of higher education has increased over time, even after allowing for inflation. As you can see, the increase since 1980 has been about 25%. The decrease in per-student funding has been caused by a combination of two further effects. First, the U.S. population has grown by about 40% since 1980, and the working-age population has grown by about the same amount, so state funding for higher education has not kept up with the growth in the population or the tax base. Second, college enrollments have grown faster than the overall population (and also faster than the college-age population). Here is a graph of full-time-equivalent enrollments as a percentage of the total population, since 1950:

Whereas attending college was once the privilege of a small elite fraction of Americans, it is now commonplace among the middle class. And while most of us celebrate this transformation, we need to realize that it doesn’t come for free. The increasing number of college students has caused the total cost of educating these students to grow to become a substantial chunk of the U.S. economy. Somehow society has to pay that cost.

In any case, the toy model shown above is based on actual (rounded) data for the current levels of tuition and state funding, the decline in state funding per student, and the observed growth in tuition. From those numbers it’s a simple matter to calculate that state funding provided about 80% of the total cost in 1980, and that the total per-student cost of education has increased by about 20% since then. (It would be nice, of course, to corroborate these results with independent data, but I don’t know where to find such data.)

And why has the per-student cost of education increased, even if only by 20%? Probably for many reasons, which I hope to explore more carefully in a later article. In brief, it appears that expenditures for faculty salaries have been almost unchanged (on a per-student basis, after allowing for inflation), although there has been a significant rise in the number of part-time faculty. Meanwhile, there has also been a steep rise in the number of professional staff, as well as a steep rise in the cost of medical insurance for all full-time employees. Other possible factors are non-staff expenses such as academic and nonacademic buildings, library books, journals, computers, software, and student financial aid. The important thing to remember is that even small increases in any of these expenses have had amplified effects on tuition (or on mandatory student fees, which are included in the tuition statistics), because state funding has not increased to absorb any of the increases.

Finally, what about the private colleges and universities? Given that they never had any state funding to begin with, you might expect their tuition to have increased by only about 20%, to absorb the same increased expenses as at the public schools. Yet they’ve actually raised tuition nearly as much as the public schools: about 150% (above inflation) since the late 1970s. Where is all that money going?

There’s good data to show that faculty salaries have been increasing faster than inflation at the private colleges, so that’s one difference. It also seems likely that the private schools have been spending increasingly more than the public ones on almost everything else: professional staff, buildings, computers, and so on. It would be interesting (but difficult) to explore whether these disparate expenditures have affected the relative quality of private vs. public education over the years.

A critical difference, meanwhile, is that the more expensive private colleges tend to provide large amounts of need-based financial aid to many of their students. In other words, the advertised “sticker price” applies only to those who can afford to pay it, and these wealthy families subsidize students who are more needy. Perhaps one could construct a toy model of the interplay between this practice and rising costs and tuition over time.

But let’s not lose sight of the big picture here. Private colleges enroll only 30% of all college students, and they couldn’t get away with raising tuition by 150% if the public colleges weren’t raising it by 200%. That increase is being driven by a variety of modest cost increases, amplified and greatly exacerbated by the decline in state funding per student.


  1. This is, as expected, a cogent and insightful analysis.

    When Dan lists the factors of "buildings, library books, journals, computers", I just want to say that in 1978 (one year after I graduated from college), it was not unusual for each college to own one computer. (I remember that Mount Holyoke College had four time-sharing terminals hooked to a CDC at the University of Massachusetts -- Amherst, plus the Physics Department owned a PDP5 as big as a breadbox and as capable as a 1990's pocket calculator.) Now, most colleges don't even keep track of how many computers they own.

    Now, businesses also own more computers than they did in 1978. But businesses use computers to replace tasks that were done by hand in 1978, so they actually save money by using computers. Colleges use computers do to things that they didn't do in 1978.

  2. Thanks, Dan.

    I started college in 1980, and immediately got a work-study job writing software for the registrar's office to handle student transcripts. At that time they were already handing the current term's registration and grades on their PDP 11/60 (I think) computer; then at the end of each term they'd print the grades on sticky labels and manually peel them off and stick them on students' transcript cards. I wrote programs to pull the data instead into a new database file that stored each student's entire academic record, and another program to print out the transcripts from the data in that file. My salary to write this software cost them almost nothing, of course. Whether it saved the staff a significant amount of time I'm not sure, but I don't recall that they cut staffing after the system was in place. It did make the transcripts a whole lot prettier!

  3. Here are a couple of columns by professional pundits on this subject. The first, written by law professor Paul Campos and published in the New York Times last April, is an especially good example of accurate facts deceptively presented. The second is a great response by Jordan Weissmann of Slate, and includes a very nice graph of per-student state funding and tuition from 1990 to 2012. The graph is attributed to Andrew Gillen, but I can't find it at the link that Weissmann provides.

  4. Thank you Dan for an exceptional analysis. First one I've read that explores the details of rising student tuition and compares public to private colleges.

  5. I think your last sentence gets the causal arrow backwards. Increased demand is the driving factor here: it enables schools to charge more tuition, and with new money burning holes in their pockets they (necessarily) find ways to spend it. That's why private schools have increased tuition nearly as much as public schools, despite less apparent need - that need is not what's driving things.

    And private schools have increased faculty salaries more than public schools have because the latter have an extra pot to dump windfall revenues into: they can rebate taxpayers by tolerating lower government subsidies, which translates either into tax cuts or spending increases elsewhere in government budgets. Or maybe it's the result of political pressure to restrain faculty salaries at public schools which you don't see at private schools.

    1. Thanks for your comment; you've certainly got a point. To sort out cause and effect would be difficult, but one way to start would be to try to classify the increased spending at colleges into "necessary" and "unnecessary" categories. For example, I would put the increased cost of employee health insurance and the cost of ADA compliance into the "necessary" category. I'd also want to turn the question around and ask if there are understandable reasons why people were (apparently) willing to pay so much less for college in the past.

  6. Danielle Allen recently published a column in the Washington Post focusing on the difference between "cost" and "tuition" at elite private colleges. During a subsequent Twitter conversation, she kindly read my article above, and pointed out that one of my remarks near the end may be misleading: I said that at expensive private colleges, "wealthy families subsidize students who are more needy," whereas Allen claims that even the wealthy students are getting a subsidy, presumably from endowment revenue and other private donations, at these schools. It looks to me like she's mostly correct, although I think her Post column exaggerates the case by citing "cost" figures at Amherst and Princeton that are somewhat inflated by the inclusion of some questionable categories.


Not registered? Just choose "Name/URL" and enter any name you like; you can ignore the URL field.